Rahman

Digital Surface Model (DSM) Construction and Flood Hazard
Simulation for Development Plans in Naga City, Philippines
Muhammad Zulkarnain Bin Abdul Rahman

March, 2006
Digital Surface Model (DSM) Construction and Flood Hazard
Simulation for Development Plans in Naga City, Philippines
by
Muhammad Zulkarnain Bin Abdul Rahman
Thesis submitted to the International Institute for Geo-information Science and Earth Observation in
partial fulfilment of the requirements for the degree of Master of Science in Geo-information
Science and Earth Observation, Specialisation: Natural Hazard Studies
Supervisors:
Drs. Dinand Alkema (First Supervisor)
Drs. Nanette Kingma (Second
Supervisor)
Thesis Assessment Board

Prof.Dr. F.D. van der Meer (Chair)
Prof.Dr. V.G. Jetten (External Examiner)
Drs. Dinand Alkema (First Supervisor)
Drs. Nanette Kingma (Second
Supervisor)
Observer:
Dr. P.M. van Dijk (Programme Director)
INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION

ENSCHEDE, THE NETHERLANDS
Disclaimer
This document describes work undertaken as part of a programme of study at the

International Institute for Geo-information Science and Earth Observation. All views and
opinions expressed therein remain the sole responsibility of the author, and do not necessarily
represent those of the institute.
Abstract
Cities in developing countries are seeing fast and unpredictable urban development. With the
increasing of population and value of property, the potential loss from prolonged and severe flooding
can easily extend into huge number of losses. This makes proper flood hazard assessment very
important for instance by using a 2D-hydraulic flood propagation models. These models require
accurate elevation data and detailed analyses on the impact distribution in the surrounding areas. One
of the main problems is the frequent changes of land use in major cities, where frequent updating of
the digital terrain model (DTM) for flood modelling might be needed. On the other hand the
assessment should be based on realistic flood hazard indicator that would help to reflect the real
impact of urban development on the surrounding areas. This paper presents an example of assessing
the impact of flood for future developments in Naga City, the Philippines. The elevation data is
constructed through integrating various elevation data derived from many sources. The development
impact assessment begins with the detailed observation on changes in flood characteristics. This is
supported by the analyses on the community-based flood risk perception and investigation on
changes of flood hazard (based on the flood velocity and depth).
In the DTM construction the natural terrain is separated from the man-made terrain. The
geostatistical approach is used to investigate the effect of integrating multi-sources of elevation data
by evaluating the nugget values. The data sources are prioritized based on the nominal horizontal and
vertical accuracy, and form of data. In this paper, there are 4 interpolation methods used, namely
Australian National University's Digital Elevation Model algorithm (ANUDEM), Kriging,
Polynomial and Triangulated Irregular Network (TIN). The assessments are based on percentile
vertical accuracy assessment, error point’s distribution and visual assessment. As a result, the kriging
interpolation method has produced the best DTM and it full-filled the requirements for hydrological
flood modelling purpose. Finally the Digital Surface Model (DSM) of the study area was constructed
by integrating both man-made and natural terrains. The DSM was also generated to simulate the new
developments in Naga City. The 1D2D SOBEK flood model was used to simulate flood events for 2,
5, 10 and 17.5 years return period flood. In addition, the flood depths and flood extent during
supertyphoon Nanmadol were used in flood model calibration.
In the development impact assessment, it was found that the impact of the developments is larger for
a larger flood magnitude. Furthermore the pattern of the changes in flood behaviour depends on the
location from the main developments. The Almeda Highway acted as a barrier, that obstructs the
flood water from go farther. In addition the small scale construction, for instance the Drainage
System in Barangay Triangulo had played a major role in changing the flood behaviour, especially in
a small magnitude flood. Moreover, the flood risk perception in 10 years return period flood has quite
good relationship with the flood hazard map. Through this study, it was proved that by simply
elevating ground terrain only can solve the flood problem in a particular area. However, the flood
problem is transferred to another area.
Keywords: 2D-hydraulic, geostatistical, DTM, DSM, ANUDEM, TIN, Polynomial, Kriging, 1D2D
SOBEK
i
Acknowledgements
Alhamdulillah, praises to ALLAH for giving a strength and opportunity to finish my MSc study at
ITC. I would like to acknowledge here, my family in Malaysia, especially my father, Mr. Abdul
Rahman Bin Yusof, and my mother Aisah Binti Abdul Rahman for a huge support throughout my
study. Besides, thanks to Universiti Teknologi Malaysia for giving an opportunity to me to further
my study abroad.
My special thanks go to Dr. C.J Van Westen who had allowed me to join this project in Naga City,
The Philippines. Special grateful goes to my supervisors, Drs. Dinand Alkema and Drs. Nannete
Kingma and to Dr. D.G.Rossiter for spending their valuable time, efforts and sharing their knowledge
throughout my MSc study here. Without them, I can’t imagine what would happen with my thesis. I
have my special regards to Mr. Paul Hofstee, Dr. Erik de Man, Graciela and Saut Sagala for their
helps and encouragements.
Special gratitude to Mr. Jesse M. Robredo, Naga City Mayor, Mr. Ernesto Elcamel and other staffs
from the Naga City Mayor Office, friendly staffs at EDP and City Engineer Office, staffs at Public
Work Department, Mr. Edgar De La Cruz and his family and my dear friend, France De La Cruz. I
will never forget their supports, helps and joyful life in Naga City. Salamat Po’.
Thanks to all lecturers in EREG department, and to all my lovely international friends; Jimee,
Manuel, Tommy, Pieter, Hendro, Kung, Sheila, Samuel, Fekerte and Edgar. Not forgotten to
Malaysian students in Holland, Edna, Badd, Halik, Rozie and Ina thanks for your support.
Finally special gratitude to all the Indonesian friends, especially Mr. Amin Budiarjo, Yuda, Tommy,
Hendro, Saut, Nining, Nunung, Mas Trias and family, Ibu Tias, Dewi and other PPI’s friends.
WWy h h blcss Ws∫

WWhWmmW3 Wl§WrnWin in hb3 QWhmWn
4nschc3c∫ thc ðcthcrlWn3s
i
Table of contents
1 INTRODUCTION......................................................................................................................1
RESEARCH BACKGROUND...................................................................................................................................................................................... 1
RESEARCH PROBLEM............................................................................................................................................................................................... 2
RESEARCH QUESTIONS............................................................................................................................................................................................ 3
RESEARCH OBJECTIVES........................................................................................................................................................................................... 4
RESEARCH METHODOLOGY................................................................................................................................................................................... 4
LIMITATION OF THE STUDY................................................................................................................................................................................... 5
THESIS OUTLINE........................................................................................................................................................................................................ 8
2 STUDY AREA.........................................................................................................................10
Climate and weather 11

Topography 12
Soil condition 12
DATA NEED ANALYSIS.......................................................................................................................................................................................... 13
DATA COLLECTION................................................................................................................................................................................................ 14
Elevation data 14
Landuse or Landcover and development plans............................................................................................16
Flood emergency plan 16
Rainfall and floodwater depth........................................................................................................................16
Flood risk perception 17
Digital GIS data 17
3 LITERATURE REVIEW..........................................................................................................18
DIGITAL ELEVATION MODEL (DEM)...................................................................................................................................18

Digital Terrain Model (DTM) for flood modelling............................................................................................19
DEM interpolation method.............................................................................................................................21
VERTICAL AND HORIZONTAL ACCURACY OF TOPOGRAPHICAL MAP.................................................................................................... 30
Vertical accuracy 30
Horizontal accuracy and resolution...............................................................................................................31
Traditional land surveying accuracy..............................................................................................................32
DEM QUALITY ASSESSMENT.............................................................................................................................................................................. 32
RMSE and other error assessment methods................................................................................................34
The percentile method of Accuracy Assessment..........................................................................................36
Spatial error distribution 39
HYDRODYNAMIC MODELING.............................................................................................................................................................................. 39
1 dimensional (1D) hydrodynamic modelling................................................................................................39
2 Dimensional (2D) hydrodynamic modelling................................................................................................40
1D-2D Flood model 45
FLOOD HAZARD MAPPING.................................................................................................................................................................................... 45
COMMUNITY RISK PERCEPTION IN NAGA CITY............................................................................................................................................ 48
4 DTM AND DSM GENERATIONS...........................................................................................50
OVERALL METHODOLOGY................................................................................................................................................................................... 50
ELEVATION DATA................................................................................................................................................................................................... 52
Data sources 52
Nominal data accuracy 53
Fieldwork data 54
ii
Table of contents
DATA PREPARATION AND ANALYSIS............................................................................................................................................................... 55
Semi-variogram analysis for each elevation dataset.....................................................................................55
Elevation dataset integration.........................................................................................................................57
GROUND TERRAIN MODELLING......................................................................................................................................................................... 63
Krigging interpolation 64
Higher order interpolation, TIN – based terrain modelling and ANUDEM interpolation.........66
DTM quality assessment...............................................................................................................................66
MAN-MADE TERRAIN MODELING...................................................................................................................................................................... 73
Current situation of Naga City........................................................................................................................73
Future developments 74
Terrain model integration...............................................................................................................................76
5 FLOOD MODELLING.............................................................................................................77
INTRODUCTION............................................................................................................................................................................................ 77
BOUNDARY CONDITION....................................................................................................................................................................................... 78
SURFACE ROUGHNESS.......................................................................................................................................................................................... 80
BUILDING STRUCTURE......................................................................................................................................................................................... 81
MODEL SCHEMATIZATION.................................................................................................................................................................................. 82
MODEL CALIBRATION.......................................................................................................................................................................................... 84
6 DEVELOPMENT IMPACT ASSESSMENT............................................................................89
DETAILED CHANGES IN FLOOD CHARACTERISTICS..................................................................................................................................... 91

COMMUNITY-BASED FLOOD RISK PERCEPTION............................................................................................................................................ 97
10 years return period flood...........................................................................................................................99
FLOOD HAZARD 103
17.5 years return period flood......................................................................................................................104
10 years return period flood.........................................................................................................................106
ANALYSIS ON THE RELATIONSHIP BETWEEN FLOOD RISK PERCEPTION AND FLOOD HAZARD.................................................. 109
7 CONCLUSION AND RECOMMENDATION........................................................................111
DTM AND DSM GENERATION BASED ON MULTI SOURCES ELEVATION DATA............................................................................... 111
FLOOD MODEL CALIBRATION.......................................................................................................................................................................... 112
DEVELOPMENT IMPACT ASSESSMENT.......................................................................................................................................................... 112
RECOMMENDATION FOR NAGA CITY MUNICIPALITY............................................................................................................................ 115
RECOMMENDATIONS FOR FUTURE STUDY.................................................................................................................................................. 116
8 REFERENCES.....................................................................................................................118
9 APPENDICES.......................................................................................................................121
APPENDIX A (BOX PLOT FOR EACH ERROR DATASET OF THE INTERPOLATION METHODS)..............................................121
APPENDIX B (ERROR DISTRIBUTION FOR EACH DTM INTERPOLATION METHOD)...........................................................123
APPENDIX C (SEMI-VARIOGRAMS FOR 15 M AND 25 M LAGS).........................................................................................124
APPENDIX D (SEMI-VARIOGRAMS OF THE POLYNOMIAL INTERPOLATION METHODS).....................................................125
APPENDIX E (MINIMUM-MAXIMUM PLOT FOR ELEVATION VALUE AGAINST LANDUSE/LANDCOVER).126 9.6.
APPENDIX F (FIELDWORK PLANNING FOR DATA COLLECTION)..............................................................127
APPENDIX G (5 YEARS RETURN PERIOD FLOOD (PRESENT SITUATION))...........................................................................129
APPENDIX H (10 YEARS RETURN PERIOD FLOOD (PRESENT SITUATION)).........................................................................129
APPENDIX I (17.5 YEARS RETURN PERIOD FLOOD (PRESENT SITUATION)).........................................................................130

i
List of figures
Figure 1-1: Overall research methodology..............................................................................................7
Figure 2-1: Naga City, the Philippines..................................................................................................10
Figure 2-2: Average monthly rainfall in Naga City (www.naga.gov.ph).............................................11
Figure 2-3: Geodetic levelling...............................................................................................................14
Figure 2-4: Zones for elevation measurement.......................................................................................15
Figure 2-5: Landuse map of Naga City (2005).....................................................................................16
Figure 3-1: Inverse Distance Weighted (IDW) interpolation method; Weighting function with
normalized distance in X axis and weight value in Y axis (a), predicted point marked as “A” (b).....22
Figure 3-2: The location of a query point that needs an interpolated height relative to the position of
the TIN elements (a) ; The interpolation found the closest surrounding nodes in all direction to the
query point and establishes a relationship to them for the use in height estimation (b) (Maune, 2001)
................................................................................................................................................................23
Figure 3-3: Cross section of Spline interpolation surface (Maune, 2001)............................................23
Figure 3-4: Experimental variogram.....................................................................................................24
Figure 3-5: Cross section of a surface from kriging (a), Cross section of the same surface with one of
the data points changed, depicting the interpolated surface does not exactly confirm to the data points
(Maune, 2001)........................................................................................................................................26
Figure 3-6: TIN structure.......................................................................................................................28
Figure 3-7: ANUDEM-based DTM generation incorporates stream line data in interpolation...........29
Figure 3-8: Graphs that combine flood velocity, depth to define the initial level of flood hazard (a)
and the final flood hazard level which was derived through the adjustment of the initial flood hazard
level with the relative evacuation time (b) (Ramsbottom et al., 2003).................................................47
Figure 3-9: Graph of flood hazard categories (The Geological Society of Australia, 2005)...............48
Figure 4-1: The conversion of contour lines to points with 50 m interval (a) and 100 m interval.......50
Figure 4-2: Overall methodology for DTM and DSM generations......................................................51
Figure 4-3: Available elevation data and newly....................................................................................54
Figure 4-4: Contour line from Naga Drainage Plan (a) and Contour line for the whole Naga City (b);
The datasets contain high nugget value.................................................................................................55
Figure 4-5: Semi-variograms of the elevation datasets, constructed with 150 m lags.........................56
Figure 4-6: The second dataset selection step.......................................................................................58
Figure 4-7: Close semi-variogram constructed with 10 meters lags for 50 meters contour lines to
point conversion interval (Dataset B); with high nugget datasets (a) and without high nugget datasets
(b); the nugget value in figure b is still more than 1.0,.........................................................................60
Figure 4-8: Semi-variogram with 150 m lag for 50 m line to point conversion interval (a) Semi-
variogram with 150 m lag for 100 m line to point conversion interval (b)..........................................61
Figure 4-9: Averaging elevation data within 5 m distance interval......................................................61
Figure 4-10: Semi-variogram with 150 m lag for 5 m average block dataset......................................62
Figure 4-11: Box plot between elevation and land use or land cover classes; No clear elevation
separation between landuse or landcover classes..................................................................................64
Figure 4-12: Residual variogram with 10 m (a), 50 m (b) and 150 m (c) lags; the residual of the has
no strong spatial dependence trend........................................................................................................65
v
Figure 4-13: Variogram models for 100 m line to point conversion distance interval (a) and with 5 m
averaging (b); lag 150 m (these datasets will be used in Ordinary Kriging interpolation)..................66
Figure 4-14: Histogram of the error dataset produced by the TIN-based terrain modelling................69
Figure 4-15: Errors distribution for TIN (a), Ordinary Kriging with 5 m block average (b), Ordinary
Kriging (100 m line to point conversion) (c) and ANUDEM (d); TIN has the least error within the
study area, followed by block ordinary kriging with 5 m, Ordinary Krigging for 50 m line to point
conversion interval and ANUDEM.......................................................................................................70
Figure 4-16: Natural terrain derived from 6th Polynomial degree (a), TIN-based terrain modelling (b),
ANUDEM (c), Ordinary Kriging (100 m conversion interval) (d), and Ordinary Kriging (5 m average
block) (e), visualized in 3D...................................................................................................................71
Figure 4-17: Road elevation compared to the residential area in the roadside.....................................73
Figure 4-18: Examples on Buildings in CBD II....................................................................................73
Figure 4-19: Almeda highway connecting Naga City and Almeda......................................................74
Figure 4-20: New mall which is proposed to be built next to the Almeda highway............................75
Figure 4-21: Micro drainage project at Barangay Triangulo................................................................75
Figure 4-22: DSM of present situation of Naga City............................................................................76
Figure 4-23: DSM of future situation of Naga City and the elevated areas (right window)................76
Figure 5-1: Overall methodology for flood modelling..........................................................................77
Figure 5-2: Boundary condition for the Bicol River Upstream............................................................79
Figure 5-3: Boundary condition for the Naga River Upstream.............................................................79
Figure 5-4: Boundary condition for the Bicol River Downstream.......................................................79
Figure 5-5: Discharge and Water Level relationship (QH)...................................................................80
Figure 5-6: Landuse of Naga City in future (a), and in 2005 (b)..........................................................81
Figure 5-7: Building structure as solid block (a), and rough surface (b)..............................................81
Figure 5-8: Model Schematization in SOBEK 1D2D...........................................................................82
Figure 5-9: River cross-section input window......................................................................................83
Figure 5-10: The original cross section data (a), Constant interpolation (b) and Linear interpolation (c)
................................................................................................................................................................83
Figure 5-11: The distribution of the flood depth samples (Saut Sagala and Peters Guarin Graciela)..85
Figure 5-12: The histogram of maximum water depth prediction error for flood modelling with rough
surface (a), and solid block (b) as a building structure.........................................................................86
Figure 5-13: Difference in maximum flood water depth and flood extent for 10 years flood modelling
treated building structure as solid block and rough surface..................................................................86
Figure 5-14: The comparison of water velocity for different representation of building structure; solid
block and rough surface at University of Neuva Caceres (open area) (a) and Residential area in
Barangay Tinago(b)...............................................................................................................................87
Figure 5-15: The comparison of water velocity for different representation of building structure;
rough surface (a) and solid block (b) after 56th hour simulation (the small right windows show the
inset of the study area)...........................................................................................................................87
Figure 5-16: The histogram of maximum water depth prediction error for flood modelling with
Manning coefficient set 1 (a), and Manning coefficient set 2 (b).........................................................88
Figure 6-1: General work flow of development impact assessments...................................................89
Figure 6-2: Barangays located inside the study areas...........................................................................90
Figure 6-3: Flow chart for development impact assessment based on detailed changes in flood
characteristics.........................................................................................................................................91
vi
Figure 6-4Changes in maximum water depth (a), changes in maximum water velocity (b) changes in
warning time (c) and changes in inundation period for 17.5 years return period flood......................92
Figure 6-5: Changes in maximum water depth (a), changes in maximum water velocity (b) changes
in warning time (c) and changes in inundation period for 10 years return period flood.....................93
Figure 6-6: Changes in maximum water depth (a), Changes in maximum water velocity (b) Changes
in warning time (c) and changes in inundation period for 5 years return period flood........................94
Figure 6-7: Histogram of changes in flood characteristics...................................................................96
Figure 6-8: Flow chart for development impact assessment based on community-based flood hazard
perception...............................................................................................................................................97
Figure 6-9: Flood risk perception maps of 10 years return period flood for present situation (a) and
after development situations (b) in Naga City.......................................................................................99
Figure 6-10: Areas classified into flood risk perception for present (a) and after development
situation (b) in Naga City (10 years return period flood)....................................................................100
Figure 6-11: Flood impact map for 10 years return period flood.......................................................100
Figure 6-12: Flood risk perception maps of 5 years return period flood for present (a) and after
development situations (b) in Naga City.............................................................................................101
Figure 6-13: Areas classified into flood risk perception for present (a) and after development
situation (b) in Naga City (5 years return period flood)......................................................................102
Figure 6-14: Flood impact distribution (negative and positive impact) represents changes in
community flood hazard perception (5 years return period flood).....................................................102
characteristics (a) and hazard level (after Ramsbottom et al. (2003))................................................103
Figure 6-16: Flood hazard map for present (a) and the situation after development (b); the worst case
hazard for 17.5 years return period flood............................................................................................104
Figure 6-17: Areas classified into flood hazard classes for present (a) and after development situation
(b) in Naga City (17.5 years return period flood)...............................................................................105
Figure 6-18: Flood impact map for 17.5 years return period flood....................................................105
Figure 6-19:Flood hazard map for present (a) and the situation after development (b); the worst case
hazard for 10 years return period flood...............................................................................................106
(b) in Naga City (10 years return period flood)..................................................................................106
Figure 6-21: Flood impact map for 10 years return period flood.......................................................107
hazard for 5 years return period flood.................................................................................................107
(b) in Naga City (5 years return period flood).....................................................................................108
Figure 6-24: Flood impact map for 5 years return period flood.........................................................108
Figure 6-25: Overall methodology of the relationship analysis between flood risk perception and
flood hazard.........................................................................................................................................109
Figure 6-26: The comparison results between flood risk perception and flood hazard for present
situation (a) and after development situation (b); 10 years return period flood.................................110
Figure 7-1: The impact of the developments on hazard area (a) and flood risk perception (b).........113
Figure 7-2: Areas classified into negative and positive impact (based on the hazard map) for 10 years
return period flood (a) and (b) and 17.5 years return period flood (c) and (d)...................................114
vi
Figure 7-3: Risk management cycle; the results of this study can be used in HAZARD mapping and
MITIGATION process. Source: Smith (2001)...................................................................................115
Figure 7-4: Landfill approaches; 1) External fill (b) and 2) Internal fill (c). Source: Lecture notes by
Kingma (2006).....................................................................................................................................116
v
List of tables
Table 3-1: Optimum pixel size for various model applications (Tennakoon, 2004)............................20
Table 3-2: Equations and number of point required for each polynomial interpolator (ILWIS help). 28
Table 3-3: Vertical accuracy of topographical map based on contour interval (Maune, 2001)...........31
Table 3-4: Horizontal accuracy of topographical map based on map scale (Maune, 2001)................31
Table 3-5 : Relative Vertical Accuracy Standard for Conventional Vertical Survey (FEMA)............32
Table 3-6: Commonly used error reporting techniques........................................................................35
Table 3-7: Percentile error assessment method (Maune, 2001)............................................................37
Table 3-8: Manning’s coefficient definition methods (Sellin et al., 2003)...........................................42
Table 3-9: Manning’s coefficient used by Tennakoon (2004)..............................................................43
Table 3-10: Manning’s coefficient used by Werner et al.(2005) and Tennakoon (2003) for flood
model calibration...................................................................................................................................44
Table 3-11: Flood hazard categories (The Geological Society of Australia, 2005)............................47
Table 3-12: Community-based flood risk perception in Barangay Triangulo (Ongoing PhD research
by Peters Guarin Graciela Graciela)......................................................................................................48
Table 3-13: Community-based flood risk perception in Barangay Triangulo (Ongoing PhD research
by Peters Guarin Graciela).....................................................................................................................49
Table 4-1: Available data sources for DTM generation from previous study by Tennakoon (2003). .52
Table 4-2: Nominal horizontal and vertical accuracy of the data sources based on the NSSDA
standard..................................................................................................................................................53
Table 4-3: Elevation datasets with the corresponding priority level based on the nominal horizontal
and vertical accuracy.............................................................................................................................57
Table 4-4: The final priority level for each elevation dataset based on the first and the second
selection steps........................................................................................................................................58
Table 4-5: The value of Semi-variogram model parameters for each dataset; the 2nd and the 3rd
datasets will be used for the DTM interpolation...................................................................................62
Table 4-6: Normality test for each error dataset....................................................................................67
Table 4-7: Percentile accuracy assessment method for each interpolation method (after Maune (2001))
................................................................................................................................................................68
Table 4-8: The summary of the DTMs quality assessment...................................................................72
Table 5-1: Manning’s coefficient (After Tennakoon (2004)); This coefficient will be used in model
calibration..............................................................................................................................................80
Table 5-2: Skew and kurtosis value for error datasets of flood model with rough surface and Solid
Block building structures.......................................................................................................................85
Table 5-3: Z- test between Manning coefficient datasets prediction error; the result are not significant
(less than 1.96).......................................................................................................................................88
Table 5-4: Overall results of the flood model calibration.....................................................................88
Table 6-1: Flood risk perception for the residential area in Barangay Triangulo (ongoing PhD
research by Peters Guarin Graciela)......................................................................................................98
Table 6-2: Flood risk perception for the residential area in Barangay Mabulo (on going PhD research
by Peters Guarin Graciela).....................................................................................................................98
Table 6-3: Probable flood risk perception for other residential areas in Naga City.............................98
Table 6-4: The comparison between flood hazard classes and flood risk perception........................109
i
List of abbreviations
1D One-dimensional
2D Two-dimensional
ANUDEM University’s Digital Elevation Model Algorithm
DEM Digital Elevation Model
DHI Danish Hydraulic Institute
DSM Digital Surface Model
DTM Digital Terrain Model
EDP Electronic Data Processing
FEMA Federal Emergency Management Agency
GIS Geographical Information System
GPS Global Positioning System
IDW Inverse Distance Weighted
LIDAR Light Detection and Ranging
NMAS National Map Accuracy Standard
NSSDA National Standard Data Accuracy
RMSE Root Mean Square Error
SRTM Shuttle Radar Topography Mission
TIN Triangulated Irregular Network
VMAS Vertical Map Accuracy Standard
x
DIGITAL SURFACE MODEL (DSM) CONSTRUCTION AND FLOOD HAZARD SIMULATION FOR DEVELOPMENT PLANS IN NAGA CITY,
PHILIPPINES
1 Introduction
Research background
Rapid and uncontrolled urbanization in developing countries has become one of the major
issues in hazard and risk management. This is certainly one of the major environmental problems in
the developing world, today and in the years to come. The rapid growth of urban areas is the result of
two factors: natural increase in population (excess of births over deaths), and migration to urban
areas. Migration is defined as the long-term relocation of an individual, household or group to a new
location outside the community of origin. Migration is often explained in terms of either “push
factors” (conditions in the place of origin which are perceived by migrants as detrimental to their
well-being or economic security) or “pull factors” (the circumstances in new places that attract
individuals to move there).
A huge concentration of people, business activities and properties has made hazard
management in urban area more difficult and complex. With an increased value of property, for
instance, buildings and other structures, potential damage from prolonged and severe flooding can
easily extend into the million of dollars. Besides, flooding in crowded area due to rapid urbanization
would dramatically increase the loss. Urbanization has a great influence on rainfall runoff and flood
behaviour. The flow of the floodwater becomes complex as a result of complicated buildings
distribution and structures in an urban area. Heavy rainfall is easily converted to run-off over paved
surface, and due to improper urbanization planning, water will accumulate and increase the potential
of flooding. As a consequence, it is a great challenge to forecast urban flooding and calculate the
potential damage. Improper development planning, in developing countries, might ignore its impact
on flood hazard and risk to the surrounding community. Proper development planning in urban area
requires numerous analyses and studies on the impacts to community and economy. Therefore, local
municipality tends to take the easiest approach in preventing flooding, for instance by elevating the
corresponding areas. This approach won’t solve the problem and the flood problem might be
transferred elsewhere.
Advancement in computer processing power, accurate terrain data acquisition and the
integration between 1 dimensional (1D) and 2 dimensional (2D) flood modelling make possible to
model dynamic flooding in a complex urban environment. The 2D flood modeling requires
information on terrain, which quality depends on the acquisition techniques and the terrain of the
study area. Complex and densely populated urban areas require more detailed terrain elevation data
compared to rural area. In urban areas, the characteristics of floodwater flow are controlled by the
distribution of buildings, roads, elevated area and etc. This requires the basis on defining the spatial
resolution and accuracy of Digital Terrain Model (DTM). Changes in urban area as a result of
urbanization can be simulated trough modification of the existing DTM and/or land use or land cover
1
PHILIPPINES
information. Future changes can be simulated through several scenarios as implemented by Alkema
and Middelkoop (2005) in investigating the impact of new compartmentalization within the riverine
polders in the Dutch Rhine-Meuse delta. In this study, the flood events were simulated through 28
flood scenarios and the resulting damage and risk were calculated for each scenario.
The concept of flood hazard management includes flood control management and flood plain
management. Traditional flood control measures are generally referred to various engineering type
projects aimed at controlling floodwaters, such as building of dikes and traditional flood plain
management. Efforts should be directed towards comprehensive flood hazard mitigation planning
covering operation of an overall program of corrective and preventive measures for reducing flood
damage. In flood preventive measures, flood emergency planning is crucial in reducing flood impact,
especially in urban area. Through several flood scenarios obtained from flood simulation, potential
damage and risk can be calculated and used as a basis for environmental impact assessment for
development project.
Research problem
This study focuses on flood problem in Naga City, the Philippines. Naga City lies in the low
lying area, in which area both Naga River and Bicol River are connected. Thus, the substantial
discharge and heavy rainfall during monsoon commonly causes severe flood in this city. According
to the National Statistics Office Republic of the Philippines (2002), the total population of Camarines
Sur, as of May 1, 2000, was 1,551,549 persons. This result implied an increase of 118,951 persons
over the 1995 census figure or an average annual population growth rate of 1.72 percent slightly
lower than the annual population growth rate recorded during the 1990 to 1995 period (1.75 percent).
It was stated that among 2 cities and 35 municipalities in Camarines Sur, Naga City has the largest
population size accounting for about 8.9 percent of the total population of the province. Naga city is
expanding very fast with city economic growth exceeds 6.5 percent. For the next ten years the annual
estimated growth rate of the population within the city is more than 1.6 percent (Naga City
Government Philippines Business for Social Progress, 2001). In general the developments in the
Naga City are shared by large and fast growing of commercial land, religious centre and educational
institutions.
Commercial, industrial/agro- industrial and tourism area covered about 24 percent of the
whole area of Naga City. This followed by residential area (19.77 percent) and agricultural area
(42.06 percent). Naga City has 6-growth zones; Central Business District I (CBD I), Central Business
District II (CBD II), South Riverfront Growth Area, Concepcion Growth Corridor, East Highland
Tourism Zone and Naga City Agro-Industrial Zone. Some of these zones are located in flood prone
area, for instance CBD II. The CBD II area is located on formerly paddy field area, which acted as a
temporary water retention area during floods (Naga City Government Philippines Business for Social
Progress, 2001). This low lying area was elevated and paved for the purposes of development.
However, as a consequence floodwater is transferred to the surrounding area, especially to the
vicinity the residential areas.
2
PHILIPPINES
Recently, new developments have taken place in Naga City to provide better facilities for the
community. The Almeda area is now connected to Naga City by a 4.5-kilometer by-pass Almeda
highway. This is paralleled with the vision and plan of Naga City Municipality to develop Almeda as
a continuation of the developments in Naga City. The CBD II was inaugurated in 1996, was designed
for medium-end commercial development. It was dubbed as Naga’s Transport exchange. This area
consists of bus terminal Naga City Coliseum, Jeepney stations and satellite market. Beside that, the
Micro Drainage System in Barangay Lerma and Triangulo project is being implemented as an
initiative to contain the flood problem in both areas. As one of the future developments in Naga
City, a new mall will be constructed next to the Almeda highway within next few years. All these
developments might change the terrain of the corresponding areas by elevating the ground and
increase the impervious areas. Improper development planning in Naga City, will in consequence
transfer flood problem from one to another areas.
In general, changes in flood behaviour would change and increase community vulnerability
on flood. Without any clue on behaviour of flood after some major developments, the normal flood
emergency plan will become ineffective and increase the probability of damage and loss on
properties and lives. Thorough study on flood behaviour could be used as a guideline in
reformulating or restructuring the proposed development plans in order to prevent future severe loss
during flood. This study focuses on the development impact assessment through detailed
investigation of flood behaviour. Further analyses will concern on the community flood risk
perception in residential area, which based on the ongoing PhD study by Peters Guarin Graciela
Graciela in Naga City. Finally, the development impact assessment ends with the investigation on
changes of flood hazard (based on flood velocity and depth). The final result can be used as a
guideline in defining the impact of particular development projects on the surrounding area in Naga
City.
Research questions
1. To update and improve the digital elevation model (DTM) for flood modelling in Naga City
a. How many points should be used to improve the DTM? Where the
additional elevation observation should be collected?
b. What is the proper DTM resolution for urban flood modelling?
c. Which interpolation method should be used for DTM generation?
d. What are the recent and future developments that should be taken into account
in generating DTM for present and future situation of Naga City?
e. What are the DTM quality assessment methods that should be used in
evaluating the DTM quality for flood modelling?
2. To simulate 2-year, 5-year , 10-year and 17.5-year return period flood events in Naga City
a. What are the appropriate parameters for model calibration?
b. What are the significant changes in flood behaviour if buildings are represented as
solid block and rough surface?
3
PHILIPPINES
3. To collect information on the recent flood event caused by super typhoon Nanmadol and to
calibrate the flood model
a. What is the return period for the recent flood event which caused by super typhoon
Nanmadol?
b. Are there flood maps prepared by the Naga municipality for this flood event?
c. What is the available information on the flood event that can be used for flood
modelling and calibration?
4. To assess the impact of developments to the surrounding areas based on detailed flood
characteristics, then followed by the community flood risk perception and flood hazard
based on flood depth and velocity.
a. What is the community perception on flood risk?
b. What is the impact of recent and future development to the surrounding areas?
Research objectives
The main objective of this research is to generate DTM and DSM of the study area, followed
by simulating the flood events and development impact assessment. The specific objectives of this
study are:
1. To update and improve the digital elevation model (DTM) for flood modelling in
Naga City
2. To simulate 2-year, 5-year, 10-year and 17.5-year return period flood events in Naga City
3. To collect information on recent flood event caused by super typhoon Nanmadol and to
calibrate the flood model
4. To assess the impact of developments to the surrounding areas based on detailed flood
characteristics, then followed by the community flood risk perception and flood hazard
based on flood depth and velocity.
Research methodology
This study is divided into 5 main phases, namely, 1) data requirement assessment 2)
fieldwork for data gathering and data preparation, 3) DTM and DSM modelling, 4) flood model
calibration and modelling and 5) development impact assessment. Data need assessment required
detailed investigation on the available data and the additional data that needs to be collected during
fieldwork. Therefore data requirement list was prepared before fieldwork (Appendix E). The 5 weeks
fieldwork duration was used to collect data and information regarding on terrain elevation, recent and
future developments in Naga City, landuse or landcover, information on the recent super typhoon
Nanmadol and other supporting data. The second phase of the research methodology concerns on the
construction of DTM and DSM based on different sources of elevation data, which were derived
from
4
PHILIPPINES
both primary and secondary sources. The primary data collection was aimed to fill the gaps in the
available data and to update terrain changes in the study area as a result of recent developments. In
general, the elevation data was derived in various forms, for instance points, line and polygons, and
these data are then aggregated into ground terrain and man-made features. Further aggregation is
made to separate the elevation data into two terrain situations; present and after development
situations. Elevation data for DTM construction was prioritized and integrated based on the data
collection approach, map scale and contour interval, and data form. The DTMs are produced using
different terrain interpolation methods. The product of each approach is evaluated and the final DTM
is chosen based on the accuracy of the DTM, spatial error distribution and visual assessment. The
best product is then selected and the next step to integrate the DTM with man –made features to
produce the DSMs of the study area.
The 1D2D SOBEK flood model was used to simulate floods with 5 recurrence intervals. The
flood calibration was made based on flood depth information derived from recent field observations
(Saut Sagala and Peters Guarin Graciela Graciela) after the flood event caused by Super typhoon
Nanmadol (with an equivalent 10 years return period flood). The flood depth information was
collected through interviews with the local community in Barangay Triangulo and Barangay Sabang
few months after the event. The flood modelling will provide information on inundation depth, water
level, inundation duration and flood velocity. The flood calibration is based on two aspects; surface
roughness and building structure. The calibrated surface roughness and the suitable building
representation will be used for further flood modelling. The surface roughness value of the study is
based on landuse map provided by The Naga City Electronic Data Processing (EDP) department.
However, updating and verification process are done to ensure the validity of the landuse or
landcover as presented by the map. This updated information is used together with development
plans to create recent and future landuse or landcover in Naga City.
The final phase of this research emphasizes on the development impact assessments based on
detailed investigation on flood characteristics before and after the development. The next assessment
focuses on the community flood risk perception which is based on the ongoing PhD research by
Peters Guarin Graciela Graciela in Naga City. The assessment ends with the investigation on the
changes of flood hazard area before and after the development. This flood hazard is defined by
combining both flood velocity and depth. Figure 1-11 illustrates the overall research methodology.
Limitation of the study
This study contains several limitations that in general related to data availability, computer
hardware and etc. This is one of the major challenges working under sparse data environment. It
requires detailed investigation on the available data, and the proper techniques to produce the
required information.
5
PHILIPPINES
The efficient technique in deriving detailed and updated DTM of the study are is through
remotely sensed data, for instance, photogrammtery and LiDAR. However in this study the terrain
should be generated and updated through combining various sources of elevation data. The
conventional geodetic levelling is crucial to supply detailed terrain information and update terrain
changes. Thus, dense elevation measurement is needed to improve and update the existing terrain
model. However, due to limited time during fieldwork, the measurement was just made at the
selected area in Naga City.
The second constraint is regarding on the computer that used for the flood modelling. High
resolution DSM and long flood simulation period need good performance computer for the
simulation. Low computer performance had increased the computation time, for instance, flood
simulation for 10 years return period with 6 days simulation required about a week to complete. The
computation time increases for higher flood event magnitude. On the other hand, the inconsistent
computer network had stopped the simulation and in consequence, the modelling should be run
again. These problems had caused long processing time in flood modelling phase. Flooding in Naga
City is mainly caused by the river overflows and heavy rainfall. However, in this study only flood
triggered by river overflow was taken into consideration. Thus, the flood modelling results might
underestimate the flood characteristics, for instance flood depth.
The next constraint is on the non availability of flood extent record for the recent flood caused
by the Super Typhoon Nanmadol. Therefore, the flood calibration only relies on flood depth data
collected by Saut Sagala and Peters Guarin Graciela Graciela in Barangay Sabang, Metro Ville,
Barangay Triangulo and Barangay Mabulo. However, this information is not enough to cover the
whole area under consideration in this flood modelling. In general, flood depth and flood extent are
useful for flood model calibration and validation.
Detailed information on buildings, for instance building footprint and number of floor are
needed for accurately simulate the distribution of building in the DSM. The most efficient way to get
this information is through aerial photograph or high resolution satellite imagery. However, both data
were not available for Naga City. Thus, the updating and inventory were done through field survey
and air photo (from Phillipines Airline plane) taken by Peters Guarin Graciela Graciela.
6
PHILIPPINES
Figure 1-1: Overall research methodology
7
PHILIPPINES
Thesis outline
This thesis contains 7 chapters.
The first chapter touches the overview and the introduction of the research, which
accommodates explanations on research background, research problems, and research objectives and
questions and overall methodology used in this research.
Chapter two contains detailed explanation on the study area and data collection during
fieldwork. The explanation on the study area concerns on climate condition, topography and soil
condition of the study area. This chapter ends with a brief description on primary and secondary data
collecting during fieldwork.
Chapter three covers literature reviews focussing on topics and fields that crucial for this study.
In general, this chapter is divided into five main subtopics; Digital Elevation Model (DEM),
Topographical map accuracy, Hydrodynamic Modelling, Flood Hazard Mapping and Community
Risk Perception on flood risk. The first subtopic contains detailed explanation on the DTM and
focuses more on DTM requirement for flood modelling purpose. Apart from that, detailed discussion
is also put on DEM interpolation approaches, which is more specifically focused on methods used in
this research. The DEM quality assessment part reviews various commonly used DEM quality
assessments, which in general covered both quantitative and qualitative approaches. The second
subtopic accommodates explanations on the nominal accuracy of topographical maps. Further
reviews continued with the discussion on hydrodynamic modelling. This subtopic deals with reviews
and discussions on state of the art of 1 and 2 dimensional flood modelling. In general, this sub topic
will discuss the procedures and issues faced by previous researches in flood modelling. Flood hazard
mapping subtopic is devoted with some reviews on methods used for hazard mapping, in which
commonly based on the combination of flood characteristics. Finally, the community flood risk
perception in Naga City obtained from the ongoing PhD researcher (Peters Guarin Graciela Graciela)
is reviewed back in this sub topic.
Chapter four is devoted for DTM generation and evaluation. The DTM generation and
evaluation are made based on flood modelling perspective. This topic deals with detailed explanation
on data preparation and evaluation, data integration and the construction of DTM and DSM of the
study area. Several DTM interpolation approaches used and evaluated with series of quality
assessment methods. The final DTM, which in general full-filled the requirement of flood modelling,
is then used for the construction of the DSM. This chapter will explain detailed procedures in terrain
model construction, starting from data preparation and data description, natural terrain modelling,
man-made terrain modelling and terrain model integration. The final product is DSM representing
terrain of present and future situation in Naga City.
Chapter five deals with flood modelling aim at simulating flood behaviour before and after
development. Detailed explanation focuses on parameters used in flood modelling, for instance
8
PHILIPPINES
boundary condition, surface roughness and etc. Moreover, attention is also put on flood model
schematization in 1D2D flood modelling in SOBEK. Further discussion will concern on the model
calibration, flood result extrapolation and ends with discussion on the findings during the modelling
and the results.
Chapter six contains analyses on development impact assessment based on three assessment
approaches. The assessment begins with detailed investigation on flood characteristics. This is useful
to reveal minor changes in flood characteristics caused by the new development. Further assessment
supported by the flood risk perception of resident in the residential areas in Naga City. The changes
in each flood risk perception will be discussed and the all the changes are summarized into negative
and positive impacts. Finally, flood hazard map with consideration of wading capability of adults and
children is created for flood event in present and after development situations. Changes in flood
hazard zone (statistically and spatially) are used to support above mentioned impact assessments.
Recommendations and conclusions are discussed in the final chapter. This chapter contains
recommendations on development plan is Naga City and future study.
9
PHILIPPINES
2 Study area
Naga City is located in Bicol region, at the south-eastern tip of the Philippine island of Luzon.
Naga City, located about 377 km to the south of Manila, is well known as a fast-growing area. Naga
City has the largest population among 35 municipalities in Camarines Sur, which population covers
about 8.9 percent of the total population of the province. Naga city expands very fast with city
economic growth exceeds 6.5 percent. For the next ten years, the annual estimated growth rate of the
population within the city is predicted to be more than 1.6 percent (Naga City Government
Philippines Business for Social Progress, 2001). Naga is considered the heart of the Bicol region and
consists of 27 Barangays on the land of 7,748 hectares. Commercial areas, industrial/agro- industrial
and tourism activities cover about 24 percent of the whole area of Naga City. This is followed by
residential area with 19.77 percent and agricultural area with 42.06 percent.
1: 500,000 (approximate scale) 1: 2,000,000 (approximate scale)
Naga City
1: 30,000 (approximate scale)

Figure 2-1: Naga City, the Philippines
Naga City is listed as one of the areas, which falls within typhoon belt or typhoon prone area.
The Bicol region, including Naga City experiences in average 5 typhoons annually (Naga City
Government Philippines Business for Social Progress, 2001). During typhoon, Naga City collects
floodwater via the Bicol River from Mt. Mayon in Albay, Iriga of Riconada area, and other areas,
which then rushes to San Miguel Bay. Metro Naga (the neighboring municipality) and Naga City are
10
PHILIPPINES
located in the floodplain area that lies between 1 m and 5 m above mean sea level (msl). In
consequence, small volume of water, which even resulted from a small typhoon, could give
considerable threat to local communities in both areas. The main portion of The Naga city is located
in low and flat topography and usually inundated by flood when water from the Naga and Bicol
River overflow. Nonetheless, the Bicol River at Naga city is affected by inter-tidal action that could
elevate the normal river water level up to 3 m during long-term recurrence events (normally up to 1.2
m). The city is also affected by storm surges during typhoon events and inundates the main national
highway and major lifeline for commerce and delivery is shut down (Naga City Government
Philippines Business for Social Progress, 2001).
Climate and weather
Naga City is classified as a Type IV climate; a tropical climate with relative abundant rainfall
and sunny day. The average annual rainfall is 2,214.3 mm recorded from 1977 to 2002 and the
average temperature is 27C. The climate in Naga city is influenced by the northeast monsoon winds
and trade winds (Laud, 2001; OIDC, 1999). The Northeast monsoon wind starts affecting Bicol
region in late of October, peaks on January to February and reduces in April. The monsoon wind
creates low- pressure area in the Pacific that produces a tropical cyclone and affects the whole Bicol
region during the months of November and December (Laud, 2001). The monthly average rainfall or
rainfall during typhoon is high enough to trigger flooding in Naga City. The highest annual average
monthly rainfall is 297.1 mm (November) (figure 2-2) and maximum rainfall intensity during a
typhoon event can reach up to 400 mm within 24 hours. The highest monthly temperature experience
in this area is 31oC (May and June), and the lowest monthly temperature is 25oC (January).
Average Monthly Rainfall in Naga City
350
300
250
200
150
100
50
Rainfall
Month
J
a
M
M
A
F
n
ar
pr
Figure 2-2: Average monthly rainfall in Naga City (www.naga.gov.ph)
“Flooding in Naga City is always associated with the occurrence of typhoons” (Naga City
Government Philippines Business for Social Progress, 2001). Thus, at the same time the flood-prone
areas are also subjected to wind hazard due to typhoons. The impact of flooding in Naga City seemed
quite significant since the main economic and human resources are located in the flood-prone areas.
Many new commercial developments had been located in the flood prone area including the
11
PHILIPPINES
construction of the new diversion road (the Almeda Highway). According to the flood hazard
analysis in the Naga City Government Business for Social Progress (2001) most of the flooded areas
in Naga City are in the vicinity areas of the Naga River and the Bicol River. The inundation depth
varies from less than 1.0 m to over 2 m with duration les than 3 hours to a few days. The city centre
had been subjected to flooding and many commercial establishments have resorted to landfilling
their properties. However, this step has created pockets of low ground that are regularly flooded even
during moderate thunderstorms. Furthermore, it tends to induce new pattern of flooding in nearby
area. Commonly, Barangays Mabulo, Tabuco, Triangulo, Sabang, Igualdad and Albella experience
severe and prolong flooding due to their proximity to the Bicol River, tidal influence and the low
riverbank (Naga City Government Philippines Business for Social Progress, 2001).
Topography
Camarines Sur is generally characterized by hilly and mountainous area with flat central area
called Bicol plain. The Bicol plain is generally flat and located in the middle of the intersection
between mountain ranges with other small coastal plains on the northeastern coast. The reclamation
of low-lying areas, for instance the CBD II area, had obviously changed the topography of certain
parts of Naga City. This area was formerly covered by swampy and rice field and acted as a
temporary water retention area during heavy rainfall. The elevation of this area was between 1.0 m
and 1.5 m above mean sea level. Recent developments in this area elevated the original terrain up
to between
2.05 m and 3.24 m.
The elevation in CBD I area is between 2.0 m and 5.0 m and mainly consists of commercial
buildings, institutional buildings and residential areas. Some areas in Naga City were elevated for
new residential areas to reallocate residences from low lying areas in the southern part of Naga City.
(i.e. Metro Ville residential area). These areas were elevated up to between 2.0 m and 3.0 m above
mean sea level. In general, the elevation in Naga City is increasing one move towards Mount Isarog.
The middle part of Naga City is a floodplain area and consists of commercial and residential areas.
This flood plain area is commonly inundated by floodwater during flooding.
Soil condition
Naga City mainly consists of alluvium of alluvial deposits, pyroclastics, limestone and
volcanic flows. The alluvium is commonly found along Bicol River and surrounding areas of low
lying areas. This alluvium are transported during heavy rainfall and deposited along active river
channels. The flat terrain at the foot of the Mt. Isarog, are covered by deposits of pyroclastics flows
and volcanic debris. In addition cliff forming limestone is a very prominent geographic feature of the
study area (Laud, 2001).
12
PHILIPPINES
Active deforestation on the slope of Mt. Isarog has caused severe erosion. As a consequence,
huge amount of sediments were transported into the Naga River and it tributaries. Exposed soil
surface due to deforestation, are easily eroded by heavy rainfall. Siltation or sedimentation raised the
riverbed and decreased the flow capacity of Naga River. This problem was recognized at
municipality level. “The authorities reallocated several families to avoid deforestation but the
damage is not only produced by illegal occupation but also loyal forces opening routes in the
ongoing battle against insurgents hidden nearby”(Tennakoon, 2004).
Data need analysis
Data need analysis begins with gathering the available data from previous studies in Naga
City. The list of the new data requirement was prepared based on this and also the research objectives
and research questions.
1. Available data
a. 10 set elevation datasets were obtained from Tennakoon’s research. However,
further elevation measurements are needed to fill gaps in the study area.
b. Landuse or Landcover map for year 2000 was obtained from Tennakoon’s
research. This map need to be updated and verified since, there were a lot of
changes in Naga City due to recent developments.
c. Muhibuddin Bin Usamah’s study provided the boundary conditions for 2, 5,
10, 25 and 50 years return period flood. The flood calibration is based on 10
years return period flood (the Super Typhoon Nanmadol is equivalent to 10
years retun period flood (Peters Guarin Graciela Graciela)).
2. Required data
a. This research will simulate and investigate flood behaviour for present
situation and after development situation of Naga City. Therefore, this study
needs detailed development plans for both situations.
b. Rainfall data for recent flood caused by the Super Typhoon Nanmadol
c. Water depth for recent flood caused by the Super Typhoon Nanmadol that
useful for flood model calibration
d. Updated building footprint
e. Flood emergency plan
f. Flood risk perception
13
PHILIPPINES
Data collection
Data collection during fieldwork is divided into primary and secondary sources. Both data
were derived by the author himself and with helps from colleagues (Saut Sagala, Peters Guarin
Graciela and Pieter). The fieldwork focused on collecting primary and secondary data from various
sources. The data requirement list was prepared based on the available data, collected by the previous
students and new data required for this study. The data requirement covered some expectations on
data accuracy and precision. However it was changed and improved due to some limitations and
availability of the required information. Since all the required information is available in different
sources or offices, detailed information regarding on the sources for each data were prepared with
helps from the local officers.
The collected data can be divided into 6 major groups as follow:
1. Elevation or topographical data

2. Landuse or landcover
3. Recent and future developments
4. Flood emergency plans
5. Rainfall and floodwater depth during the Super Typhoon Nanmadol.
6. Flood risk perception
Elevation data
Elevation data were derived from various sources with different scale and accuracy. There are
4 sources of elevation data.
 Almeda Highway Project (Naga Diversion Road 2 Project) – This dataset contains detailed
information on physical structure on the construction.
More over, the construction plans also enclose
detailed elevation measurement for both original
ground elevation and final road elevation. The survey
process was made by private company (DMCI
company group) in the year 2000 and the plan was
derived from the Naga City Engineer Office.
 Micro drainage system in Barangay Triangulo and
Lerma – This dataset is a detailed construction plan of
the drainage. The construction plan included
information on the original terrain elevation and also
the dimension of the drainage. This project is founded
by the Community Organization of The Philippines
Enterprise Foundation Inc. (COPE), a non-
government organization.
Figure 2-3: Geodetic levelling
14
PHILIPPINES
 CBD II development plan – This dataset contains detailed information on building

footprints, roads and elevation at 0.25 m contour interval. The elevation information
recorded in the map was derived from the leveling process made by Naga City Engineer
Office Surveyors.
 Geodetic leveling for selected areas – The leveling was made continuously starting from
the 2nd to the 5th week of the fieldwork (figure 2-3).
More than 200 points were collected at the selected zones; 1) CBD II (Zone A), 2) Barangay
Tinago and Panganiban Drive (Zone B), 3) Penafrancia
Avenue (Zone C), 4) Barangay Bagumbayan Sur (Zone D),
5) Barangay Sta. Cruz and CBD I (Zone D), 6) LCC Mall
D
(Zone F) and 7) finally Barangay Sabang and Metro Ville
C
(Zone G) (figure 2-4). The measurement was constrained by
the non-accessible areas, weather and the Global Positioning E
B
System (GPS). The accuracy of the hand held GPS depends
GF
on many constraints for instance, site, sky view and number A
of GPS used for position calculation. Beside that, the
measuring process was stopped and disturbed by congested
traffic during peak hour and heavy rain. In general, the
selection of the zones is based on the following
considerations.
Figure 2-4: Zones for elevation
measurement
i. Area with new developments (i.e. Metro Ville and CBD I)

ii. Less density of elevation data
iii. Areas with high topography variations
Most of the new developments are located in Barangay Triangulo. Therefore, dense elevation
measurements were done including the elevation of the elevated terrain and also the original
terrain. In zone B, additional elevation data is needed to represent the depression area in the
middle part of the residential area in Barangay TINAGO. Less dense elevation measruments
in zone C, D, E and F. There is a new residential area in zone G.
15
PHILIPPINES
Landuse or Landcover and development plans
The landuse or landcover information was

derived from the Naga City Electronic Data Processing
Department (EDP). The information was verified and
updated through ground checking of some portions in
Naga City (figure 2-5). The information regarding on
the developments in Naga City were derived from
documents, detailed construction plan and through some
information told by the officers. The derived
information is listed as follow.
Figure 2-5: Landuse map of Naga City (2005)
 Detailed CBD II development plan

 Almeda Highway construction Plan (Naga Diversion Road 2)
 Micro drainage system in Barangay Triangulo and Lerma (construction plan)
 New Mall (development plan)
 Almeda developments (development plan)
Flood emergency plan
The flood emergency plan in Naga City is quite effective, since it combines the efforts from
the municipality, Barangay coordination and non-government organizations. The information on
flood emergency plan in Naga City was derived from, the Naga City Mayor Office and interviews
with corresponding officers in the Municipality and also Barangay level. Furthermore, detailed
information on location and type of evacuation centers was derived from the EDP.
Rainfall and floodwater depth
The rainfall for year 2005, including rainfall measurement during Super Typhoon Nanmadol
was obtained from the Philippine Atmospheric, Geophysical and Astronomical Services
Administration (PAGASA). The rainfall data was obtained from 7 measuring stations; 1) Camaligan,
2) Ombao, 3) Bato, 4) Buhi, 5) Ligao, 6) Ocampo and 7) Sipocot. The floodwater depth during super
typhoon was obtained from ongoing PhD research by Peters Guarin Graciela Graciela and Saut
Sagala from interviews conducted in Barangay Mabulo, Metro Ville, Barangay Sabang and Barangay
Triangulo.
16
PHILIPPINES
Flood risk perception
The flood risk perception was obtained from the ongoing PhD research study by Peters
Guarin Graciela Graciela. This information was derived through interviews with the residents in
Barangay Triangulo and Barangay Mabulo. The flood risk perception was defined based on flood
depth and flood duration.
Digital GIS data
There is other data derived from EDP and it contains additional information on Naga City, for
instance roads networks, building footprints, cadastral maps, topographical maps, drainage maps,
evacuation centers, landuse or landcover map and other supporting data. These data were attached
together with detailed description on map projection and map layout.
17
PHILIPPINES
3 Literature review
Digital Elevation Model (DEM)
Digital Elevation Model (DEM) is a generic term for both topographic and bathymetric data,
and can be found in various forms. A model means that the digital form of the elevation data can be
used, analysed and modelled directly by a computer in 3 dimension and in such way reduces the need
for labour-intensive human interpretation (Maune, 2001). The DEM implies elevation of the bare
earth surface without vegetation and man-made features. DEM term is usually used together with
DTM, but the later may also incorporate elevation of significant topographic feature on the land
together with mass points and breaklines that are regularly spaced for better representative of true
bare earth shape. The DTM term was originally come from Massachusetts Institute of Technology in
about 1955 to 1960, using digital data obtained from stereo models (Lodwick, 1983). According to
U.S Geological Survey (USGS), “a DEM is the digital cartographic representation of the elevation of
the terrain at regularly spaced intervals in x and y directions, using z-values referenced to a common
datum”(Maune, 2001). Meanwhile Digital Surface Model (DSM) is similar to DEM or DTM except
that the DSM includes the elevations of vegetation and man-made structures.
The resolution of the DEM will define terrain detail that can be represented by the elevation
model. Coarse terrain model resolution might miss terrain details. However, detailed terrain
properties represented by high-resolution terrain model require massive data storage and needs
longer processing time. For instance, in hydrological application, stream detail is crucial in
hydrological modelling, and according to above-mentioned constraint, the uniformly spaced DEM
will miss the streamline and fail to show the streamline in DEM. Therefore, a new form of hydro-
enforced DEM is now being produced by USGS, Federal Emergency Management Agency (FEMA)
and others who drainage need pattern to be appeared in DEM.
New emerging technologies have improved the quality of DEM for instance, laser altimetry
in particular LiDAR (Light Distance and Ranging). LiDAR can produce accurate and high resolution
DEM by direct measurement of ground surface. With 2000-5000 height measurements per second,
15 cm vertical precision and with 1 m horizontal resolution, LiDAR is considered as an expensive
method in acquiring DSM and DEM compared to any conventional method such as photogrammetry
and geodetic surveying. Nevertheless LiDAR data are decreasing in cost as techniques of acquisition
and processing (notably, filtering out vegetation and man-made structures) improve in efficiency and
economies of scale make the data more competitive in the marketplace. Another technique, in which
terrain accuracy and resolution are comparable to optical remote sensing method, is Radar
Interferometry or InSAR. InSAR requires simultaneous or repeated signal acquisitions by synthetic-
aperture mapping radar and delivered a near-global DEM at a uniform horizontal resolution of 90 m
18
PHILIPPINES
The Shuttle Radar Topography Mission (SRTM) obtained elevation data on near-global scale
generate the most complete high-resolution digital topographic database of earth (about 80 percent of
earth land surface). SRTM is an international project spearheaded by the National Geospatial-
Intelligence Agency (NGA) and the National Aeronautics and Space Administration (NASA). There
are two versions of SRTM elevation data, 90 m and 30 m horizontal resolution and they are available
for the whole globe and United States areas respectively. However at the early stage of SRTM data
acquisition, 1/5 of earth’s land area was excluded. Moreover, the average relative accuracy of the
new dataset is 10 m, and for 90 m dataset with serious error, the accuracy would potentially at 30 m
(Pike, 2002). Some reasons were laid on the limitation of radar in penetrating dense vegetation cover.
Furthermore, sometimes some elevation data filled with blank and water bodies may not appear flat.
Throughout this report, the term DTM will refer to bare earth surface.
Digital Terrain Model (DTM) for flood modelling
DTMs are used for many purposes, such as in terrain modelling, map visualization and
hydrologic modelling (Wechsler, 2003). The DTM resolution is highly dependent on application, for
instance high resolution DTM (between 2.25 to 3.25 m) is important for modelling processes at the
micro scale level. In general the resolution is also defined by the surface landscape where low relief
landscape being less sensitive to resolution impacts and larger range of relief needs higher resolution
(Anderson et al.). Accurate DTM is essential for urban flood modelling, since, many obstacles
involve in controlling the behaviour of floodwater. Structural elements in floodplain such as river
embankments and elevated road have a significant effect on flood extent and thus it is crucial to use a
good DTM in representing such obstacles (Bishop and Catalano, 2001; Werner, 2001). The irregular
nature of urban environment makes fixed grid the easier approach compared to unstructured grid in
representing topography (Bishop and Catalano, 2001). Urban flood modelling by Schmitt et al.,
(2004) has made sure that the DTM includes “distinct levels of street cross section, side-walks and
street curbs as well as the border line between public (street, side-walk) and private space”. However
in order to include all these elements, it would increase the computational time. On the other hand,
such detailed information is only possible to get through detail conventional ground elevation
observation or through LiDAR.
LIDAR can produce dense elevation information with vertical accuracy of 15 to 20 cm and
sub-meter planimetric accuracy (Cobby et al., 2001). Mark et al.,(2004) has suggested that DTM with
spatial size of 1 m by 1 m to 5 m by 5 m is sufficient in representing urban elements such as road
width, side-walk width, houses and building in urban flood analysis. Moreover, DTM with 1 m by 1
m horizontal resolution is not necessary in order to get accurate result in flood level, and 5 m by 5 m
DTM size could be used for quick assessment of the model results. Blomgren (1999) used DEM with
horizontal resolution of 5 to 10 m with less than 1.0 m vertical accuracy, to accurately represent
narrow features such as road embankments and dunes in his flood modelling. Major road should be
one of important elements in the DTM, since it would behave as an artificial channel during flood
(Mark et al., 2004). Buildings are one of the important elements in controlling floodwater behaviour.
Building treated as solid building, in one hand might overestimate flood extent and flood depth. On
19
PHILIPPINES
the other hand, building as bare earth with rough surface could underestimate flood depth and flood
extent. El-Ashmawy (2003), showed that buildings have a significant effect on flood extent when
built-up area is 10 percent or more, in fact treating building as solid block raises flood height twice as
much as if building is treated as partially solid block.
The flow pattern remains unchanged after solid block building structures were removed and
high surface roughness is used instead (Tennakoon, 2004). Building structures that are treated as
rough surface would help in reducing the overestimation of flood depth. Moreover, building
structures, which are treated as solid block in area with 10 percent building density, will increase
floodwater depth by 25 cm (El-Ashmawy, 2003). This concludes that, area with less than 10 percent
building density doesn’t need special representation of building structures (building structure with
partially solid structure, which allows intrusion of water during flood) (El-Ashmawy, 2003).
Roads and other obvious man-made features will influence the floodwater flow, thus features
less than final DTM resolution need to be increased to at least ™ (pixel resolution) (Tennakoon,
2004). However this assumption is only valid up to certain DTM resolution compared to features
size. Feature size exaggeration might overestimate the contributions of such elements in flood
modelling. Table 1-2 shows the optimum pixel size for DTM for various scales of flood modelling
applications.
Table 3-1: Optimum pixel size for various model applications (Tennakoon, 2004)
Applications 5m DTM 7.5m DTM 10m DTM
Flood hazard mapping in urban areas Adequate
Detailed studies related to velocity measurements,

Adequate
sedimentation and erosion.
Studies (EIA) relating to individual structures or
Adequate
reclamation of small lands
Small scale EIA studies in regional level for
investigating effects on new roads/ small More than 10m
scale reclamations projects etc.
The requirement on the DTM accuracy for flood modelling depends on the characteristics of
the study area. In his research, Maune (2001) has pointed out that for hydraulic and hydrologic
modelling, the hydraulic modelling for lower part area, for instance floodplain, high vertical accuracy
and high resolution of elevation data is required compared to hills and mountain areas. The direction
water flows is crucial and complicated at relatively flat area compared to hilly and mountainous area
where water direction is defined by the steep slopes.
20
PHILIPPINES
Fixed-size grid is a common approach in any Geographical Information System (GIS)

packages for representing continuous terrain surface. However, fixed-sized gridded DEMs suffers a
problem in sampling size (spatial resolution), where rough terrain is usually undersampled with
coarse sampling size, while on the other hand, over-sampling the flat terrain area (Raaflaub and
Collins, 2005). Inevitably, the generation of DEM is prone to invariability introduces by error or
artificial elevation samples. According to Raaflaub and Collins (2005) artificial pits or sink features
in DTM are hydrologically serious problem. In fact, “Pits generally appear in flatter area where even
a one meter error can be enough to produce a close depression”(Raaflaub and Collins, 2005).
Artificial pit can introduce serious problem during flood modelling, where water seems trapped and
stays longer than it should be. “However, most pits can be considered errors since fluvial processes
will not normally produce such features at the scale resolved by DEMs” cited as Band (1986) in
(Raaflaub and Collins, 2005). Therefore, pits should be analyzed and remove if necessary after the
DTM construction.
DEM interpolation method
Various interpolation methods have been used with the aim to fill the gaps of unsampled
measurements (Demirhan et al., 2003). The interpolation is based on the available measurement of
the surrounding areas. In DEM interpolation, a gridded surface of unknown elevation data is created
from known elevation measurements. Generally, the interpolation techniques can be classified into
several techniques such as classification models, trend surface analysis, fourier series, proximal or
nearest neighbours, moving averages and inverse distance weighting, splines and finally kriging.
Kidner (2003) pointed out that the DTM accuracy depends on the interpolation method. In his
research, he found that the interpolation techniques with consideration on the local terrain
neighbourhoods are more consistent and accurate. He had proved that the bilinear interpolation
method had reduced the root mean square error (RMSE) by up to 20 percent. However, different
interpolation approaches should be used for different surface characteristics, thus there is no single
interpolation method that is the best for all situations (Maune, 2001). On the other hand, a single
interpolation technique might have many parameters. Therefore different areas might require
different parameters values. Terrain interpolation for areas with different geomorphological classes,
for instance a coastal plain area might require different parameter values compared to floodplain and
mountainous areas. In his study, Tomaž Podobnikar (2005) pointed out that the quality of the DEM
might be different based on their purpose of use, quality of the source data, interpolation algorithm
and operator experience. Instead of that, one of the major challenges faced by most of the
interpolation algorithm is the presence of noise in elevation dataset (Demirhan et al., 2003). They
had also pointed out that, this noise might be originated from measurement error, overlapping or near
elevation measurement with obvious difference in elevation value and etc.
The basic principle of terrain interpolation is that elevation points that are closer together tend
to be more alike than points that farther apart (Demirhan et al., 2003; Maune, 2001). The changes of
elevations, from one point to another, are closely related to their distance or the elevation difference
is spatially dependence. Therefore, in most of the interpolation algorithms, the closest point to
the
21
PHILIPPINES
predicted location would have big influence in the interpolation. This principle is normally defined
by the weighting method. In this method the weight value decreases as the surrounding available
measurements are farther away from the point to be predicted. The interpolation method can be
classified into 2 main approaches namely, deterministic and probabilistic methods. In principle, the
deterministic interpolation methods use the surrounding elevation samples in interpolation. The
probabilistic method, on the other hand, shares the similar principal. In addition, the selection of the
surrounding elevation measurements will be based on their correlation with the predicted area. For
instance, the geostatistical based interpolation methods use the calculated autocorrelation values from
point pairs, to develop a spatial dependence variogram model. The weights will be assigned to the
surrounding measurements and it will be based on the distance and correlation between the predicted
points and the samples points.
Three interpolation techniques will be

discussed here: 1) inverse distance weighted
interpolation method (IDW), 2) Natural
Neighbour interpolation method, and 3)
Spline interpolation method.
F
a
Normalized distance b
Figure 3-1: Inverse Distance Weighted (IDW)
interpolation method; Weighting function with
normalized distance in X axis and weight value
in Y axis (a), predicted point marked as “A” (b)
The IDW interpolation method is among the simplest interpolation methods (figure 3-1). The
IDW uses linearly weighted combination of values from the nearby points to determine the value of
new points. The weight is defined as a decreasing function of distance. However this assumption is
not valid for terrain interpolation (Maune, 2001).
The second interpolation method is the Natural Neighbour interpolation method known as
“area stealing” which is based on area in calculating weight. The result will be a smooth and
conservative terrain surface. The key strength for the interpolation is laid on the method in finding
appropriate input samples (neighbours) for height interpolation at a given point. The nearest points
are chosen, based on Thiessen polygon, and this is crucial to ensure the interpolated value will
depend only on a local subset of data. There is no required searching radius distance in the
interpolation which becomes a problem when the input samples are not evenly distributed. The
weighting scheme is based on the amount of area that is picked from the Thiessen polygons
representing input samples. The samples, which are not selected as neighbourhood, would have zero
weight (figure 3-2). The linear-based interpolation scheme will produce a smooth interpolation
everywhere except at the sample points. While the non-linear based interpolation scheme will
produce smooth interpolated surface everywhere including the sample points.
22
PHILIPPINES
a b
Figure 3-2: The location of a query point that needs an interpolated height relative to the position of the
TIN elements (a) ; The interpolation found the closest surrounding nodes in all direction to the query
point and establishes a relationship to them for the use in height estimation (b) (Maune, 2001)
The Spline interpolation method is based on a mathematical formula and produces a surface
that minimizes the overall surface curvatures (figure 3-3). The product should be a smooth
interpolated surface including the sample points. Basically, the Spline method is like bending a sheet
of rubber through a set of points (Maune, 2001). This method is very useful to produce a gently
varying terrain with smooth slope transitions. This method is very sensitive to abrupt elevation
changes in a short distance. Thus, the terrain interpolation should avoid man-made terrain features to
be incorporated in natural terrain interpolation. Samples with sharp terrain elevation changes might
exaggerate the value of the neighbouring cells or areas. In general, the neighbouring cells should
have less variance than the farther cells.
Figure 3-3: Cross section of Spline interpolation surface (Maune, 2001)
Kriging interpolation in terrain modelling
A Geostatistical interpolation or kriging interpolation method is similar to a probabilistic

interpolation type, in which the weights are derived from the surrounding sample points. The weights
are not only based on the distance, but also on the strength of the overall correlation among the
measured points (Maune, 2001). The basic interpolation assumption is that, values at a short distance
are more likely to be similar than at a larger distance. The elevation value for each pair is compared
and expressed as variance or covariance. Then, the spatial structure could be analysed through a
semivariance method. This method defines the degree of spatial dependence among pairs at a
specific distance (equation 3-1).
23
PHILIPPINES
(h) = ?2  [(Z (xi + h) – Z (x)) Equation 3-1

2
]
Where, i=1
(h) = semi-variance at distance h
m = number of point pairs within h
distance Z (x) = elevation value at position
Z (xi + h) = value at h distance from i
The dataset is analysed through an experimental semi-variogram (figure 3-4) and it is used
to assign the weights for each elevation sample. An experimental variogram plot is a graph with
semivariance values of the elevation data (Y-axis) against the distance between elevation data (X-
axis). After building the experimental variogram, an appropriate theoretical function (e.g. Gaussian,
Exponential, spherical) is needed to model the spatial variation. The next step is to determine the
values of the required parameters; “Nugget”, “Sill” and “Range”. Nugget is a semivariance as the
separation between elevation data approaches zero. This value represents variability at a point that
can’t be explained by spatial structure. Sill is a maximum semivariance value that represents
variability in the absence of spatial dependence. Finally, range, represents the separation between
point-pairs at which the sill is reached (distance at which there is no evidence of spatial dependence).
In addition lag refers to the average distance of semi-variogram in x-axis.
Krigging interpolation is based on “a

Semivariance
statistical model of nature rather than a model
interpolation function” (Chiles and Delfiner, 1999).
Interpolation with the kriging method can’t get rid Sill
of the user experience on a particular application or
phenomenon (Maune, 2001).
Range
Nugget
Distance
Figure 3-4: Experimental
variogram
A unique feature of Kriging is that it provides an estimation of the error at each interpolated
point, providing a measure of confidence in the modeled surface. The trade-off for estimating the
optimal solution for each point by Kriging is computation time. Ordinary Kriging is commonly used
24
PHILIPPINES
for elevation data interpolation (Maune, 2001). Rossiter (in lecture note “An introduction to applied
25
PHILIPPINES
geostatistics”, 2005) had pointed out that the Ordinary Kriging is an optimal local predictor, since this
predictor has the following features:
1. The prediction is based on “linear combination of known data values”
2. The prediction is unbiased and exact at known points. This statement will be
discussed later.
3. Known measurements closer to the point to be predicted will have larger weights
4. Clustered measurements resulted by over-sampling in a small area, in general can

introduce bias. However, in the Ordinary Kriging, the clustered measurement will
be reduced to a single equivalent point to avoid the bias.
5. Closer measurement will “mask” the further measurements in the same direction
6. “The prediction error should be as small as possible”
The prediction variance that is derived together with the prediction result is based on the
configuration of the existing measurements around each point to be predicted. The kriging weight
must sum to one. The weight values are assigned based on the fitted variogram model (Peter et al.,
2005) with an optimization criterion to minimize the prediction variance. The estimated elevation
value (point to be predicted) is calculated using equation 3-2.
Zo = Σ Equation 3-2
λiZi i=1
Where,
Zo = The estimated value
λi = Weights for measurements participated in the estimation
Zi = Samples value
Kriging is used as an exact interpolator, where the interpolated surface passes exactly
through the input samples (figure 3-5 (a)). However, due to errors in measurements that are quite
away from the local trend of other samples, the Kriging interpolator will put attention to the local
trend rather than to such measurement values (figure 3-5 (b)). The Kriging interpolator is sensitive to
homogeneity of input data. The homogenous input dataset will produce nice and better model
(Dowman, 2002). Therefore, it’s quite a challenging job to fit a semivariogram model of a terrain
model that contains inhomogeneous terrain features (e.g. elevation data with natural terrain and man-
made terrain). Instead of that, the inhomogeneous elevation data can also be derived for area with
complex geomorphological features.
26
PHILIPPINES
a b
Figure 3-5: Cross section of a surface from kriging (a), Cross section of the same surface with one of the
data points changed, depicting the interpolated surface does not exactly confirm to the data points
(Maune, 2001)
Blomgren (1999), in his study used a rectilinear grid over the contour lines to transfer
elevation information from contour lines to point forms. The point was digitized as close as possible
to the overlaid grid. Therefore, more evenly distributed elevation samples were derived, and it
improved the interpolation performance (Blomgren, 1999). On the other hand, the arrangement of
elevation data will influence the shape of the variogram model. In this case, the clustered elevation
data, which were found around dunes, road embankments and other local abrupt changes in
topography, were removed from the dataset (Blomgren, 1999). The clustered elevation data will
produce high elevation variation at a short distance. By so doing, the local abrupt changes in
topography, road embankments and other man-made terrain cannot be expected to be preserved
during the interpolation. These features can be added later in the final DTM. Wilson and Atkinson
(2003) in their research, “Prediction the uncertainty of DEM on flood inundation modelling”, used
the ordinary kriging to interpolate the elevation data in the floodplain area. The elevation data was
the combination between the contour lines and the elevation points that were derived from GPS
measurement. The original experimental semi-variogram of the contour lines had quite general
shape. However this general shape or trend was reduced (increased variance at shorter lags than
globally) when the elevation points derived from the GPS was added to the dataset. As mentioned
previously, noise is one of the major challenges in spatial interpolation. Demirhan et al., (2003) in his
study which was concerned on the performance of several interpolation methods with presence of
noise and sampling pattern had pointed out that the “Ordinary Kriging is the most robust
interpolations method against noise”.
Nugget effect is a term in Kriging that refers to high nugget value in a semi-variogram. A
high nugget value tends to even the weight of the surrounding samples and thus the effect of distance
is reduced (Meijerink et al., 1994). In this case, the interpolated surface becomes smoother because in
term of statistical distance none of the samples is any closer to the point to be interpolated than any
other. In the Ordinary Kriging, the interpolation result of a short range variogram model will be
similar to that for a pure nugget effect (Meijerink et al., 1994).
Universal kriging is based on the trend of the dataset as a function of coordinates (Hengl et
al., 2003). The semivariances of the Universal Kriging are based on the residuals, and not the original
data. The model parameters for the residuals will be different from the original variogram model. The
variogram of the residuals often appear with lower sill and shorter range compared to the original
27
PHILIPPINES
variogram. Rossiter (in lecture note “An introduction to applied geostatistics”, 2005) has pointed out
the approach to obtain the semi-variogram of the residual from the original elevation dataset:
1. The best fit surface is calculated on the original dataset with the same model to be
used in the Universal Kriging
2. The residual is obtained by subtracting the original dataset with the prediction value
of the fitted model
3. Compute the semi-variogram of the residual dataset
The Universal krigging is classified together with kriging with external drift in hybrid kriging
interpolation group. The kriging with external drift uses auxiliary information, in which has a good
relationship with target variable (elevation) (Hengl et al., 2003). The auxiliary information should be
available at each elevation sample points. Detailed information on kriging can be found in (Isaaks
and Srivastava, 1989).
Higher order terrain interpolation
The deterministic interpolation approach tries to fit a mathematical function to a set of

elevation samples of known coordinate (x and y). The interpolation could be done either through an
exact interpolation or smooth interpolation (Meijerink et al., 1994). The smooth interpolation method
is suitable when one assumes that the point measurement data are regarded as trues value with less
error. On the other hand, with coarse data accuracy, the smooth interpolation scheme might be the
best way to level out the error to some degree. The fitting process uses various types of mathematical
functions usually known as polynomial functions at a certain degree of complexity to fit the surface
through the sample points. Once the polynomial or mathematical surface is calculated by a least
square fit, the prediction of the unknown elevation points will be based on this calculated
mathematical surface. The fitting process usually starts with a simple mathematical surface, for
instance simple plane surface and extends to more complex polynomial surface. The general idea of
the polynomial interpolation could be expressed by equation 3-3, where ci are the polynomial
coefficients and v is the polynomial degree.
v
F(x) =  ci x i Equation 3-3
i=0
The plane surface is the simplest and crudest assumption, which assumes that the point to be
interpolated is the arithmetic mean of its four nearest pixels. This approach might produce a
discontinuous surface compared to mathematical models with higher polynomial degree (Kidner,
2003). The higher degree interpolation is also based on the spatial dependency of the elevation
samples. Thus, initial investigation should be carried out on spatial correlation of dataset. Several
polynomial interpolation methods with different degree of polynomial are applied in the ILWIS
28
PHILIPPINES
software (table 3-2). There are indications on the minimum requirement on number of samples and
samples should be more than the minimum requirement for better result.
Table 3-2: Equations and number of point required for each polynomial interpolator (ILWIS help)
Minimum number of
Polynomial interpolator Equation
points required
Plane Z = a + bx + cy 3
2nd degree linear Z = a + bx + cy + dxy 4
2nd degree parabolic Z = a + bx + cy + ex2 + fy2 5
2nd degree Z = a + bx + cy + dxy + ex2 + fy2 6
3rd degree Z = a +…..+ gx + hx2y + ixy2 + jy3 10
4th degree Z = a + ….+ kx4 + lx3y + mx2y2 + nxy3 + oy4 15
5th degree Z = a + …+ px5 + qx4y + rx3y2 + …+ uy5 21
6th degree Z = a + … + vx6 + … 28
The TIN based terrain modelling
The nature of TIN modelling is splitting the surface into

triangular element planes (Meijerink et al., 1994). More detailed, “TIN
is a digital terrain model based on irregular array of points which
forms a sheet of non-overlapping contiguous triangle facets” (Maune,
2001). It is a vector model that supports lines, points and area-based
features in representing the surface morphology (figure 3-6). Since
there is no transformation to intermediate data structures is needed, the
original input data will be preserved in generating the terrain model.
The TIN structure is dynamic which allows changes in its resolution
according to surface complexity. For instance, a simple flat surface
can be represented with large or coarse TIN resolution, while the high
resolution TIN structure is needed to represent a rugged terrain.
Figure 3-6: TIN structure
The TIN model is a good model in representing terrain information that includes buildings,
roads and other man-made features. The TIN model is well adapted to the complexity of the terrain
in term of surface variations. It enables the incorporation of break lines along the triangle sides and it
is also avoid “saddle point” problem in contouring. Furthermore it can be used to calculate terrain
attributes (Meijerink et al., 1994). However, this model requires a large amount of high accuracy
29
PHILIPPINES
data, which are normally expensive (Maune, 2001) and not readily available. In addition, the TIN
model can be created by using many various types of data. It can also be generated at different stages
and for each stage, additional data can be used to refine the existing TIN. Breaklines area linear
features, or the boundaries of areal features, which represents important natural or artificial
discontinuities in slope contained in the landscape. There are 3 common breaklines used in TIN
generation, namely hard breaklines, soft breaklines and faults.
Soft breaklines will ensure that the known elevation values along a linear feature are
maintained in the TIN. Furthermore, it also can be used to maintain linear features and polygon edges
by enforcing the breakline as TIN edges. There is no elevation interruption defined by the soft
breaklines. Hard breaklines on the other hand, define location of abrupt surface change for instance,
streams, ridges, shorelines and building footprints. The fault breaklines are used to represent an
interruption in surface continuity that caused by for instance, a geological fault.
ANUDEM
The ANUDEM is a software package known as the Australian National University Digital
Elevation Model developed by Hutchinson (Geodata and Geoscience Australia, 2002). The
ANUDEM interpolator was designed and optimised to create a hydrologically correct terrain models.
“It’s unique in both input and output for building a good terrain model”(Maune, 2001). The input
data is not only confined to point data, but also lines which represent streams and ridges for drainage,
and polygons as a lake boundary to produce a DEM that is virtually free of spurious sinks and pits
(figure 3-7).
Figure 3-7: ANUDEM-based DTM generation incorporates stream line data in interpolation
Hydrographic (drainage) enforcement in ANUDEM will create a network of valleys and

ridges. The interpolated surface is created using Spline interpolation method while maintaining
consistent down-slope below. The advantageous of ANUDEM-based DTM interpolation are as
follows (Sinha, 2000).
1. Efficient interpolation processing time. For instance, DEMs with over a million
points can be easily interpolated using a computer workstation.
30
PHILIPPINES
2. The roughness penalty can be modified to preserve sharp changes in terrain

associated with ridges, streams and other features
3. A drainage enforcement algorithm that attempts to remove all sinks in the fitted DEM.
4. Drainage enforcement is further enhanced to incorporate streamline data supplied by
user in interpolation process.
5. ANUDEM can recognize and preserve sinks in the landscape.
6. User can specify the output resolution of the DTM
The additional information on drainage structure is useful to remove spurious sinks and pits.
Therefore, it has more or less overcome the problem faced by another interpolation methods, thus
increases the final DEM output quality for hydrological application purposes. Proper representations
of ridges and streams are deduced automatically from the input contour lines. The result of the
interpolation will be a smooth elevation surface. Furthermore, cliff information can also be fitted in
the final terrain product when the cliff line data is used as a reference in interpolation process. In
general, ancillary data is needed to guide this interpolation method to correctly construct the desired
terrain model. The ancillary data (e.g. global drainage condition) is useful to increase the
interpolation accuracy, when dealing with sparse elevation information. The size of the ancillary data
can be at least less than the number of points required in adequately describing elevation using
digitised contours. Moreover, this can minimize the expenses in terrain data acquisition, storage and
also time to remove spurious drainage features (Geodata and Geoscience Australia, 2002). Kenny
and Matthews (2005) have tested the performance of several DEMs on hydrological and watershed
applications. They have pointed out that DEM’s derived using ANUDEM were found as good as
DEM’s derived from Inverse Distance Weighting (IDW) and Kriging.
Vertical and Horizontal accuracy of Topographical map
Vertical accuracy
Most of the topographical maps users usually use contour interval as a definition of vertical
accuracy. However, the National Map Accuracy Standards (NMAS) in the US specifies vertical
accuracy in terms of contour interval at the 90 percent confident interval. The method is defined as:
“Vertical accuracy, as applied to contour maps on all publication scales, shall be that not more than
10 percent of the elevations tested shall be in error more than one-half the contour interval ” as
quoted in Maune (2001). This one-half contour interval under the NMAS is called the Vertical Map
Accuracy Standard (VMAS). However this technique has become obsolete due to the emerging of
digital mapping, which allows user to change a scale and contour interval of a map.
The National Standard Data Accuracy (NSSDA), in 1998, had introduced the vertical RMSE
(RMSEz) and vertical accuracy (Accuracyz) at the 95 percent confidence level which was defined as
1.9600 X RMSEz. This calculation assumes that the errors are normally distributed and all the
systematic errors have been eliminated as best as possible. However, the TIN-based DTM generation
31
PHILIPPINES
often doesn’t follow the normal distribution of errors, thus another more robust DTM accuracy
assessment technique should be used.
Table 3-3: Vertical accuracy of topographical map based on contour interval (Maune, 2001)
NMAS Equivalent NMAS VMAS 90% NSSDA RMSE NSSDA Accuracy at 95%
Contour interval (m) confidence interval (m) confidence level (m)
(m)
0.305 0.152 0.091 0.183
0.610 0.305 0.183 0.366
1.219 0.610 0.366 0.732
1.524 0.762 0.457 0.914
3.048 1.524 0.914 1.829
Horizontal accuracy and resolution
Users often refer to the published scale of the map to define the horizontal accuracy. The US
National Map Accuracy Standards has stated a guideline in defining the horizontal accuracy of the
topographical map with scale larger than 1: 20,000. In this guideline, the maps with scale larger than
1:20,000, it should not more than 10 percent of the points tested shall be in error by more than 1/30 th
inch, measured on the publication scale. On the other hand, for maps on publication scales of
1:20,000 and smaller, the standard is 1/50 th of an inch. Both standards are called the Circular Map
Accuracy Standard (CMAS). As stated above, due to the emerging of digital mapping technique,
NMAS was replaced by the NSSDA with statistical and testing approach to estimate the positional
accuracy of points on map and digital data with respect to clearly defined geoferenced ground
positions of higher accuracy. The CMAS was changed to Radial RMSE (RMSE r), and at 95 percent
confidence level the accuracy is defined as 1.7308 x radial RMSE r. This method assumes that the
errors are normally distributed, in which the systematic errors were eliminated as best as possible.
This statement assumes that the elevation points will fall within circle of tolerance within 95 percent
at all times. Horizontal resolution is lower compared to DEM resolution, which is called post spacing
(Maune, 2001). It refers to the density of elevation point that will be used in DEM interpolation, for
instance, horizontal resolution of 5 m supposes to have post spacing less than 5 m (more than a point
within 5 m).
Table 3-4: Horizontal accuracy of topographical map based on map scale (Maune, 2001)
NMAS map scale NMAS CMAS NSSDA RMSE NSSDA Accuracy at 95%
90% confidence (m) confidence level (m)
level (m)
1:1,200 1.006 0.671 1.158

1:2,400 2.042 1.341 2.316
1:4,800 4.054 2.682 4.633
1:6,000 5.090 3.353 5.791
1:12,000 10.150 6.706 11.582
1:24,000 12.192 8.016 13.899
32
PHILIPPINES
Historically, the contour lines were created manually by interpolating the spot heights
(Maune, 2001). Thus implies that the contour lines were less accurate compared to the spot heights.
However, more recently, the contour lines were compiled through the ptohogrammteric technique.
Traditional land surveying accuracy
The vertical accuracy of the traditional land surveying is expressed in terms of orders and
classes of survey. The Federal Emergency Management Agency (FEMA)(2006) had defined the
relative accuracy of the surveyed points to the geodetic reference points (table 3-5). In addition,
Tomislav et al (2003) had pointed out that the accuracy of the ground survey through levelling
system is approximately to 1 mm.
Table 3-5 : Relative Vertical Accuracy Standard for Conventional Vertical Survey (FEMA)
Vertical Survey Vertical Relative Accuracy between directly

Order Survey Class connected points
1st I Standard error = 0.5 mm * Sqrt (k)
1 st II Standard error = 0.7 mm * Sqrt (k)
2 nd I Standard error = 1.0 mm * Sqrt (k)
2 nd II Standard error = 1.3 mm * Sqrt (k)
3 rd N/A Standard error = 2.0 mm * Sqrt (k)

* k refers to distance in km between observed points and the reference point
DEM quality assessment
“In Science, the word error does not carry the usual connotations of the term mistake or blunder.
Error in a scientific measurement means the inevitable uncertainty that attends all measurements. As
such, errors are not mistakes; you cannot eliminate them by being very careful. The best you can
hope to do is ensure that errors are small as reasonably possible and to have a reliable estimate of
how large they are” (Taylor, 1997) as quoted in (Wechsler, 2005).
“It is important to note that for many applications, accurate representation of terrain shape is more
important than absolute elevation accuracy. Measures of accuracy of overall shape and drainage
structure, as principally determined by slope and aspect, are more difficult to quantify than standard
elevation errors”(Geodata and Geoscience Australia, 2002).
The elevation error at a single point in DEM depends on the resolution of the DEM, and the
roughness (here refer to uneven terrain surface) of the modelled surface, which is commonly
represented by the slope (Hutchinson, 1996) as quoted in Geodata and Geoscience Australia, (2002).
The DEM errors are generally grouped into 3 wide categories, namely 1) blunders, 2) systematic
errors and 3) random errors. Blunders are vertical errors that originate from the data collection
33
PHILIPPINES
process. Being classified as the error, blunders exceed the maximum absolute error permitted. Such
characteristics make blunders easy to detect and remove from dataset. On the other hand, systematic
errors can be also produced by errors in the measurement procedures and in the DEM generation
process. This error , commonly follow a specific pattern. Random errors are produced by an operator
error in recording elevation data during levelling process and etc. However, most of the errors in the
DEM, which are derived from the contour maps cannot be removed by any mean of processing (Pike,
2002). Furthermore, the map accuracy standards vary widely and do not provide a rigorous
evaluation.
Errors in terrain models are also produced by the interpolation procedure in digital terrain
modeling. The errors are attributed by several factors such as features of the terrain surface,
distribution and density of the elevation data, method of DEM surface modeling, specialty of DEM
surface and state of the vector data (Shi et al., 2003). The complexity of the terrain features are the
important factors among other factors affecting the DEM’s accuracy. The complexity of the terrain
features can be expressed by the slope of the terrain surface. The terrain data distribution can be
defined as the position, structure and direction. The density of the elevation points can be ascertained
by the distance between the elevation points and the quantity of elevation points in a unit area. In
general, denser elevation data is needed for accurately representing a complex terrain surface and less
data is needed for quite homogenous terrain. Terrain information represented by vector for instance,
contour lines contain large errors because of the selectively expression by contour (contour interval).
Instead of that, the accuracy of the original data can be altered by non-linear factors for instance,
scanning error and map transformation (Shi et al., 2003).
Podobnikar (2002) used three steps in data pre-processing to combine multi sources of
elevation data in; 1) elimination of the gross errors of reference points using reference DEMs, 2)
statistical evaluation of non-point sources with reference points and 3) systematic error elimination
from non-point data sources. The elimination of the gross error was done through a variance filter.
The filter was used to eliminate gross errors from reference points. The method compared the
absolute difference between reference points and reference DEMs (in grid) with threshold. The
reference points exceeded the threshold value were discarded. This dataset was further used to
evaluate the non- point source elevation dataset. Finally, this dataset was used to correct the
systematic errors in the non-point sources. However, in general, this technique requires a lot of
precise elevation measurement using conventional leveling process to correct the whole non-point
datasets. Furthermore, the reference points with high elevation variance might be introduced by the
detailed elevation measurements from the conventional leveling process (Wilson and Atkinson,
2003). They can’t be discarded from the dataset, since the good quality elevation dataset which are
derived from the conventional leveling process are useful to represent the real terrain characteristics
of the study area.
Through online survey analysis, Wechsler (2003) has revealed that “Over half of the DEM
users who responded to the survey “rarely” (24.9 percent) or “never” (27.3 percent) account for
uncertainty in the DEM”. Besides that, Root Mean Square Error (RMSE) reporting was the common
method, which accounts about 21.4%, followed by visualization (19.3 percent), error maps (12.8
percent), and finally simulation techniques with 8.4 percent (Wechsler, 2003). Users of DEMs should
sufficiently know what the definitions of good DEM for their applications are. Different application
34
PHILIPPINES
might have different definition of a good DEM, in terms of their evaluation aspects. DEM users
should at least know how to answer the following questions (Maune, 2001).
1. What are the performance parameters that are suitable to characterize various DEM?
2. What are the consequences of the DEM accuracy for their applications?
3. What are the suitable methods for measuring the DEM’s accuracy according to their
applications?
Maune (2001) has classified the categories of DEM performance parameters into two groups,
namely, data requirements and usability. The data requirements including vertical accuracy,
horizontal accuracy and post spacing tend to define the actual data requirement of the application.
Meanwhile the usability of the DEM could be defined by the vertical datum, the horizontal datum,
the data projection, the file format, the file size, the Metadata and the DEM editing process.
RMSE and other error assessment methods
RMSE is a common method in defining the vertical accuracy of the DEM (equation 3-4).
The RMSE is based on the assumption that the errors are normally distributed with zero mean
(Wechsler, 2003). This technique is valid when the errors are free from the systematic error and
random in nature. The horizontal accuracy concerns the accuracy of the two-dimensional positional
accuracy during the extraction of the elevation data. On the other hand, the geometric correction of
the remotely sensed imagery is the best way to define the horizontal accuracy of elevation data which
is derived from remote sensing approaches. On the other hand, the horizontal accuracy of the Global
Positioning System (GPS) is crucial when the elevation data is derived through a differential GPS
method.
Equation 3-4
Where Zi is the interpolated elevation value, Zj is a true value and n is number of samples.
The RMSE calculation is one of the most widely used error description and known as a
dispersion measures. However as discussed above, it is only valid for the normally distributed error
datasets and is being used by the USGS and the Ordinance Survey. Small RMSE value indicates a
small gap between the assumed true elevation data (check points) and the interpolated elevation
dataset. It simple calculation with one value of result makes this approach popular and understood by
most users. However, since the deviations from the mean value are attributed from the different set of
processes, this single value might suppress some important information. This single RMSE value
doesn’t include necessary information for instance, skewness information and the normality testing
value. Beside that, there is no information about the spatial distribution of the errors. Furthermore,
The RMSE calculation is made regardless of difference in relief, thus Wood (1996) has suggested
one
35
PHILIPPINES
method which considers the relative relief effects from measurement deviation. An error in surveying
benchmark is one of the examples for spatially dependence error and classified as a systematic error.
In the levelling process, the error propagates from one measurement to another measurement and the
non-spatial statistical method such as RMSE couldn’t represent such error (Wechsler, 2003). The
systematic errors are not easy to detect and can introduce a significant bias. However, this spatially
dependence errors can be analyzed through the residual semi-variogram.
Table 3-6: Commonly used error reporting techniques
Technique Equation Description

Accuracy Ratio a = sqrt[Z(Zi – Zj)2 / n] The Accuracy ratio method could
(Wood, 1996) Zi = interpolated elevation value eliminate relative relief effects
Zj = true elevation value from measurement deviation. The
RMSE is divided by a measure of
Zi = mean of interpolated elevation values.
relative relief. It can be used to
n = Number of samples
report the DEM uncertainty at
different spatial scales and
different terrain surfaces.
Bias Bias = Z(Zj – Zi) / n Defines the average deviation of
(Wechsler, 2005) Zi = interpolated elevation value the statistic
Zj = true elevation value

Relative Bias R-Bias = Z((Zj – Zi)/Zi) / n Defines the average percent
(Wechsler, 2005) Zi = interpolated elevation value deviation of the error dataset
Zj = true elevation value

Average relative ARAD = Z(|Zj – Zi|/Zi) / n Defines the average percent
absolute different Zi = interpolated elevation absolute deviation of the
(ARAD) value Zj = true elevation value interpolated elevation values
(Wechsler, 2005)
Relative root R-RMSE = SQRT[Z((Zj – Zi)/Zi) / n] Standardised the RMSE value
mean square Zi = interpolated elevation value
error Zj = true elevation value
(Wechsler, 2005)
Log RMSE L-RMSE = SQRT[Z(ln(Zj/ Zi))2 / n] The error dataset is transformed
(Wechsler, 2005) Zi = interpolated elevation into natural logarithm which
further normalize the data
value Zj = true elevation value
distribution
36
PHILIPPINES
Mean and dZij = (ZZi - ZZj2 )/ n

standard S =SQRT[ Z[(Zi – Zj) – dZij]2 / n]
deviation of error
Zi = interpolated elevation
dataset
value Zj = true elevation value
(Wood, 1996)
Skewness and dZij = (ZZi - ZZj )/ n
kurtosis Skew = Z[(Zi – Zj) – dZij]3]/ ns3
(Wood, 1996) Kurtosis = Z[(Zi – Zj) – dZij]4]/
ns4
According to Maune (2001), the DEM posting refers to the distribution of the elevation
samples and it can be used to define the resolution of the final DEM. However, the high resolution of
DEM doesn’t mean high DEM accuracy. Accuracy refers to the degree of conformity with a standard
(http://www.flatsurv.com/accuprec.htm). Thus, the accuracy of the DEM is a measurement that
defines the deviation between the measured elevations to the real elevations. For instance, the 5 m
resolution DEM is generally more precise compared to the 10 m DEM, however, the 10 m resolution
DEM might be more accurate, if it is derived using a more accurate procedure or technology.
Maune (2001) suggested that quick assessment on the DEM quality can be done through
analysing the minimum and maximum elevation value. These values are supposed to be within a
realistic range. Instead of that, the histogram of the constructed DEM might be useful to get the
information on the distribution of the elevation data. Furthermore, the analyses on the slope and the
hill shade maps are useful for visual DEM assessment. However, additional knowledge on the
geomorphology and terrain properties of the study area is required for such assessment. According to
Meijerink et al. (1994), contouring the final interpolated DTM is necessary since, it can be used to
evaluate the consistency of data. This technique is useful to check the artificial isolated depression or
peak, which is derived from the interpolation process. However, the isolated peaks and depression
should not be considered as error, unless it was proved. Meijerink et al. (1994) also suggested to
apply linear interpolation in assessing the possible oscillations of higher order polynomial
interpolation techniques. Maune (2001) has introduced the percentile method of accuracy assessment.
This method can be used in reporting the DEM error for non-normally distributed residual dataset
(error dataset) and is simpler than RMSE.
The percentile method of Accuracy Assessment
The percentile method requires error dataset to be ranked by their absolute value. It involves
5 steps in accuracy assessment and reporting. The error dataset is derived through subtracting the
interpolated elevations to the more accurate elevation measurements (the check points to define the
accuracy of the interpolated elevations).
37
PHILIPPINES
1st step
The error dataset is ranked from the worst to the best. The ranked values are then classified
into 100 percentile based on the total number of dataset. For instance, the largest error value will be
classified into 100 percent class down to 90 percent and so on (table 3-3).
Table 3-7: Percentile error assessment method (Maune, 2001)
Error in cm
Percentile Meaning (value from error dataset)
(example)
100th + 381.6 100% of the checkpoint errors are equal to or less than  381.6 cm
98 th + 49.6 98% of the checkpoint errors are equal to or less than  49.6 cm
93 rd + 24.6 93% of the checkpoint errors are equal to or less than  24.6 cm
92nd + 24.4 92% of the checkpoint errors are equal to or less than  24.4 cm
91st - 23.6 91% of the checkpoint errors are equal to or less than  23.6 cm
Table 3-3 shows the threshold of error for each percentage of the error dataset. The
description of the accuracy is clearly regarded to the corresponding percentile class.
2nd step: Accuracy suing RMSE computation

The simple normality test (for instance Shapiro-Wilk, Anderson Darling and Kolmogorov-
Smirnov) is carried out to determine whether the whole error dataset is normally distributed. In such
case, the accuracy can be simply calculated as equal to 1.9600 x RMSE. However, if the error dataset
is not normally distributed, which includes statistically outlier data, then the RMSE calculation is not
valid.
 The entire error dataset is used to calculate the mean and the standard deviation values
 The standard deviation is multiplied by 3 to establish the “3 sigma value”
 If one or more error values are outside of this threshold, then statistically, the error dataset
contains “outliers” and RMSE is not valid for accuracy description. Proceed to step 3.
 If none of the errors beyond the threshold (3 sigma), then compute the Accuracy = 1.96 X
RMSE. The error reporting is: “Tested cm vertical accuracy at 95 percent confidence
level using RMSE procedures with no outlier”. The error dataset is normally distributed.
38
PHILIPPINES
3rd Step: Accuracy using 95th Percentile

This step will determine whether the accuracy can be calculated from 95 th percentile error dataset
at 95 percent confidence interval. This method is valid up to 5 percent outliers.
 The best 95 percent of the error dataset is used to calculate the mean and the standard
deviation values
 The new standard deviation is multiplied by 3 to establish 3-sigma threshold
 If one or more error value beyond of this threshold, then statistically, the error dataset
contains some “outliers” and the RMSE is not valid for accuracy description. Proceed to step
4
 If none of the error beyond the threshold (3 sigma), then the 95 th percentile value will be used
in error reporting. The error reporting is: “Tested cm vertical accuracy at 95 percent
confidence level determined by the 95th percentile method”. The error dataset is not
normally distributed with 5 percent outliers
4th Step: Accuracy using 90th Percentile

This step will determine whether the accuracy can be calculated from 90th percentile error dataset.
This method is valid up to 10 percent outliers.
 The best 90 percent of the error dataset is used to calculate the mean and the standard
deviation values
 The new standard deviation is multiplied by 3 to establish the 3-sigma threshold
 If one or more errors are beyond this threshold, then statistically, the error dataset contains
some “outliers” and RMSE is not valid for accuracy description. Proceed to step 5
 If none of the error is beyond the threshold (3-sigma), then the 90 th percentile value will be
used in the error reporting. The error reporting is: “Tested cm vertical accuracy at 90
percent confidence level determined by the 90 th percentile method”. The error dataset is not
normally distributed with 10 percent outliers
5th Step
The 5th step is the last step, since the outliers are more than 10 percent. Maune (2001)
suggested that the data might have a systematic error due to limitations on technology used for data
acquisition. In this study, the accuracy assessment is further continued up to 20 percent outliers in
order to ensure that all interpolation techniques can be compared at the same level of error
assessment.
39
PHILIPPINES
Spatial error distribution
Beside of these measurers, it is quite crucial to observe the spatial pattern of error. The
spatial autocorrelation can be used as a tool to reveal the spatially correlated error. Experimental
variogram can be used to reveal the spatially correlated errors (Wood, 1996). The spatially correlated
might related to the systematic type of errors in elevation measurement. This information can be used
to refine the DEM interpolation process for instance in universal kriging interpolation method.
Hydrodynamic modeling
Today’s flood management practice has been enhanced through the emerging of a 1
dimensional and a 2 dimensional flood modelling. Both methods have their own advantages and
disadvantages, which in turn, encourages some efforts in integrating both of them. SOBEK was built
by the WL | delft hydraulics, founded in 1927. It is an independent consulting and research institute
located in the Netherlands. The SOBEK is a fully dynamic 2D hydraulic model, specifically for
floodplain flood modelling. The floodplain area is represented by the 2D grid, which is derived from
either the DTM or the DSM in representing the terrain of the floodplain. The computation or solution
used for the 2D floodplain modelling is based on the finite different method. The SOBEK also allows
the integration between the 1D channel flows and the 2D overland flow. Structures, for instance
bridge and etc. can be incorporated in the flood model through the 1D module.
1 dimensional (1D) hydrodynamic modelling
In the 1D flood model, terrain should be characterized as “a series of cross-section

perpendicular to the direction of flow” (Alkema et al., 2004). The information on the water depth and
the flow velocity can be calculated at each cross section and the measurements in between, have to
be interpolated from the known cross section. In fact, the water flow in the 1D model is assumed
parallel to the main direction of the river channel and no perpendicular flow is considered. This is the
major limitation of the 1D flood modelling. This model is good for modelling water flow in the
channel when water is still confined in the channel. Inevitably, in decision support system, the
computation time of the flood modelling is crucial. In general, the 2D flood modelling needs longer
computation time compared to the 1D flood model and at this point, the 1D flood models become
more attractive (Werner, 2001). Among the popular 1D flood models are HEC-RAS, which is
produced by the US Army Corps of Engineers and MIKE 11 from the Danish Hydraulic Institute
(DHI). However, this models have a severe problem in simulating water depth and water flow
velocity when over-bank discharge occurs during flood event (Tennakoon, 2004).
The flood map is crucial to show the flood extent and also for further flood spatial analyses.
In this case, the interpolation methods can be used to create the flood maps based on the 1D flow
results (Werner, 2001). Nevertheless, the 1D flow model can be integrated with the Geographical
40
PHILIPPINES
Information System (GIS), for further flood analyses especially for decision making. The accuracy of
the flood mapping using the 1D approach relies on the accuracy and the applicability of the
calculation for each reach under consideration. The 1D flood modelling may not suitable for highly
sinuous channel, especially when the sinuosity of the channel is greater than of the floodplain
(Werner, 2001).
2 Dimensional (2D) hydrodynamic modelling
The 2D flood model requires a continuous representation of terrain topography. In general,

the 2D flood model can provide the information on the flood extent, flood depth and flood velocity.
Therefore it becomes more popular in the flood modelling, especially in a complex topography of the
floodplain area (Bishop and Catalano, 2001). Furthermore, “a 2D nature flood map has promoted the
use of the 2D models in order to promote the synergy between distributed observation and
predictions, whereas point measure of stage or discharge are more compatible with the 1D model”
(Horritt and Bates, 2002). In general, the main inputs for the 2D flood models are topographical data,
hydraulic parameters and boundary conditions (Bishop and Catalano, 2001). All the inputs and the
outputs of the 2D models are assumed uniform within a single pixel. Therefore, the selection of the
appropriate DTM resolution is crucial to represent the real terrain characteristics of the study area.
Bishop and Catalano (2001) has suggested that the minimum grid size of the DTM should be defined
approximately to the size of the smallest features to be presented in the flood modelling. On the other
hand, the selection of the DTM resolution should consider the computation time, which in general
increases with the increasing of the DTM resolution. Nowadays, the DTM resolution between 2 m
and 10 m is commonly used for urban flood modelling and it is also depends on the scale of the area
of interest (Bishop and Catalano, 2001). Among the popular 2D flood models are FLS, LISFLOOD,
Telemac 2D and MIKE_21. The SOBEK model, produced by the WL|Delft Hydraulics is a
combination of the 1D and the 2D flood models. The 2D flood model in SOBEK was designated to
simulate the overland flow on the initially dry land and through complex topography (Alkema et al.,
2004).
Horritt and Bates (2000) in their study, has suggested that the performance of the flood
model can be evaluated through the consistency of the calibration values, for instance, the Manning
coefficient values with different flood event magnitudes. Therefore, this assessment can be used as a
basis of the justification on the flood model suitability for the practical use in the flood risk and the
flood hazard assessments. “Whatever approach seems the most appropriate for a given problem, the
application of such kind of deterministic modelling poses high demands on the data, the modelling
skills, computation power and interpretation capabilities of the hazard assessment team” (Alkema et
al., 2004). Furthermore, Werner (2002; 2001), has pointed out that the 2D flood model has a major
drawback in supporting rapid flood assessment as it requires long computation time in flood
modelling. In addition, the 2D model requires consistent data that sometimes are not readily possible
at the feasibility study stage.
41
PHILIPPINES
The hydrological modelling consists of a wide range of process. This modelling is related to
“land phase of hydrological cycle”. The flow produced by one catchments outlet may pass various
phases of the hydrological process, for instance surface run-off, sub-surface flow and exchange from
groundwater. The overland flow can be modelled through the laws of conservation of mass and
momentum. However, for a small and medium size of urbanized catchments, “the contribution of
sub- surface flow to an event-based flood hydrograph may be small”(Rehman et al., 2003). However,
there are options to incorporate the information on evaporation, wind and wave action in the flood
models. However these factors “are not commonly significant in most urban situation due to the short
duration of events and relatively small fetch distances”(Bishop and Catalano, 2001). Furthermore,
the turbulence factor can also be ignored since it is not critical in urban flooding.
Boundary condition
The boundary condition nodes represent the geographical location where the model is
separated from the real world. These nodes are needed to connect the model with the real world by
assigning hydraulic boundary condition in each node. Flow boundary condition can be represented
by constant or a series of discharge that flows in and out of the model. Positive flows mean water
flows into the model, while negative flows indicate flow out. Commonly, flow boundary is located in
the upstream edge of a river, where model receives water from the upstream rivers. Constant or
alternating water level boundary is usually located at river mouth (downstream boundary).
Inconsistent water level at river mouth can be represented by alternating water level. Finally, a
discharge-water level (Q-H) boundary is a common boundary for downstream edge of a river model.
Boundary nodes are needed to solve the water flow equation which comprises of continuity equation
and momentum equation.
Surface roughness
Surface roughness is used to represent the bottom friction either for the floodplain area or
channel. The roughness parameter for instance, Chezy, Strickler and etc are commonly used to
represent the surface roughness of the floodplain and channel (Tennakoon, 2004). Nowadays, the
remote sensing imagery can be used to obtain the roughness information through the land-use or
Land-cover classification. This information can be translated to the surface roughness value of the
floodplain area. Furthermore, LiDAR data can provide “information on vegetation biochemical
properties” that can be used in representing water friction during flood (Cobby et al., 2001). Verwey
(2001) and Priestnall et al (2000) used the LiDAR data in surface roughness estimation based on the
vegetation characteristics, for instance vegetation height and its variation, vegetation density and the
diameter of stem which together with the density determines the blocking area per unit area. The
surface roughness is an important parameter to estimate the flow velocity. Equation 3-5 shows the
velocity calculation with the consideration of Manning coefficient.
42
PHILIPPINES
V = 1/n R2/3 S 1/2

Equation 3-5
e
Where,
V = Mean velocity of flow (meters per second)
R = Hydraulic radius in meters
Se = Slope energy grade line (meters per
meter) n = Manning’s roughness coefficient
Sellin et al.(2003) has summarized several previous works concerning on the evaluation of
the surface roughness in a floodplain area. The approaches rely on the characteristics of the
vegetations. Manning’s coefficient in a floodplain area is a complex function of flow velocity and
depth, vegetation type and density, dimension and flexibility, which function can be expressed as a
function of age and season. In general, photographs and simple tables are simpler, quicker and less
cost approach (Sellin et al., 2003).
Table 3-8: Manning’s coefficient definition methods (Sellin et al., 2003)
Method Data required Advantages Disadvantages

Experience and Site visits Quick Requires experience
intuition Photographs Open to errors of
judgment
Photo matching Site visits Quick and widely used – No variation with
Photographs gives an approximate water level
figure
for any condition
Use of n value tables More detailed Systematic, familiar Apparent accuracy is
information on deceptive
vegetation type
density season etc
Rigid roughness Estimation of Gives variation with depth Only applicable to
formula variation of plant limited conditions: low
density with depth, velocity i.e. no bending
appropriate value
of Cd (discharge
coefficient for
roughness units,
hedge)
Hedge roughness CD (drag Accounts for flow Calibration required
coefficient for through hedges
vegetation) value
for hedge type
Flexible roughness CD (drag The most accurate Only applicable to
formula coefficient for method for flexible densely packed
vegetation) and vegetation in a high vegetation. Each new
flexural stiffness velocity flow variety requires costly
for all plants calibration
types present
estimation
of density
43
PHILIPPINES
In an urban environment, surface roughness is defined by vegetations, buildings,

embankments, roads, and other major obstruction, that would influence water velocity, speed and the
flow direction. Alkema and Middelkoop (2005), Bishop and Catalano (2001) and Tennakoon (2004)
used landcover information as a basis in defining Manning’s coefficient for their study area. In most
cases, for the land surface which is covered buildings in a commercial and residential area, can be
represented by high roughness value (Bishop and Catalano, 2001).
Table 3-9: Manning’s coefficient used by Tennakoon (2004)
Landcover type Manning’s coefficient

Roads 0.025
River 0.03
Commercial zone 0.032
Residential area 0.035
Agricultural 0.050
zones
Building 1.000
footprints
Model calibration and verification
Model calibration or optimization is a process in which certain parameters are tuned to

reduce the deviation between model results and the corresponding observed data (Mark et al., 2004).
On the other hand, model verification is a process in which the calibrated model quality is tested by
comparing the results of the modelling with the observed data. For example, the calibration of urban
drainage can be carried out through calibrating the surface run-off and pipe flow model towards the
measured flow and the water level at the specific locations. In a surface run-off calibration
hydrological parameters could be tuned for instance by adjusting time of concentration until
hydrograph agrees closely to the observed runoff. Calibration can be done through changing pipe
flow parameters for instance Manning number until calculated discharge and water level output
agrees closely to the observed data. Generally the main objectives of urban flooding models are to
match the flood extent and flood depth to the observed information (Mark et al., 2004). Tennakoon
(2004) and Hsu et al.,(2000) have used Manning coefficient to calibrate the model to reduce the
simulated flood extent with observed data during flood event. “Inundation extent data have also
provided a second source of observed data independent of hydrometry, allowing models to be
independently calibrated and validated” (Horritt and Bates, 2002).
In the 1D and 2D flood modelling, the friction coefficient should be defined for both
floodplain and channel. Werner et al.(2005) has found that, changing floodplain friction coefficient
doesn’t increase the model performance, thus, the calibration process should put more attention to
main channel roughness. The water flow in floodplain is not only disturbed by the roughness due to
vegetation cover. The roughness is determined by a significant effect of turbulent momentum
interaction between main channel and floodplain, inconsistent terrain surface, crossing flows and etc.
The contribution vegetation roughness to the overall roughness is not clear or not linearly correlated.
44
PHILIPPINES
Surface roughness is a localize factor, in which has relatively low impact on overall flood inundation
and extent (Pappenberger et al., 2005).
Table 3-10: Manning’s coefficient used by Werner et al.(2005) and Tennakoon (2003) for flood model
calibration
Mannings – n Landuse Initial run Trial 1 Trial 2

Land use type Residential 0.035 0.032 0.038
Min. Max.
Horritt Commercial 0.032 0.028 0.035 is
Mainand Bates (2002;
channel 2002) have
0.025 0.06 pointed out that the optimal parameter sets, which
Agricultural 0.040 0.035 0.042
obtained during model calibration might be different for different flood events, thus reducing
Weeds without structure 0.035 0.08 Open land /
predictive power of the model. On the other hand, it is River 0.030
also arguable that whether0.025 0.032
parameter sets
Weeds with structure 0.035 0.08
obtained from model calibration of certain magnitude isRoads
valid to be used0.025
for more0.022
extreme0.028
event
Cultivated fields 0.025 0.05 Buildings 1.000 1.000 1.000
(Horritt and Bates, 2002). A good flood model will have more or less stable calibration parameter
Pasture
values for different flood events0.025
and it is 0.05
reliable for hazard and risk assessment. However, the model
Marginal grassland 0.025 0.05
validation process is difficult to carry out due to sparse flood data and, furthermore it requires
comprehensive Bares
andsoil 0.02
detailed field 0.04
dataset. Somehow, the validation process should be at least use
commonHardwood woodland
flood depth 0.08 (Mark
and extent data 0.2et al., 2004).
Cultivated woodland 0.08 0.2
Uncertainties in the flood modelling can be classified into several groups, namely 1) structure
of flood model, 2) numerical setting in flood modelling, 3) topography, 4) hydrological input, and
parameters (Pappenberger et al., 2005). Structure of the flood models deals with the structural
elements to simulate or approximate the real world. In such circumstance, simplification is needed,
for instance the channel or river can be represented by a cross section. There are some other
assumptions and simplifications used by the flood models to make water behaviour in the channel
and the floodplain possible to be simulated. The second uncertainty concerns on the numerical
scheme, for instance calculation step used for simulation. A short simulation step theoretically will
produce detail and accurate result, but in fact, it is time consuming. Time step has a great impact on
model robustness, and it still unclear in choosing the optimal time step in the flood modelling. On the
other hand, topographical data has an important role in the 2D flood modelling. It was proved that
small errors in topography would significantly affect the flood modelling results. However, this issue
gets more complicated when the representation of infrastructure, for instance building structures are
included in the model. Further explanation on the effect of the building structure on flood modelling
45
PHILIPPINES
results can be found in El-Ashmawy (2003) and Tennakoon (2004). Another factor contributes in the
flood modelling uncertainty is parameter uncertainty, for instance surface roughness. The uncertainty
in the surface friction might give a significant effect at local scale, but only a small impact to the
overall inundation predictions
1D-2D Flood model
SOBEK has been developed by WL | Delft Hydraulics in partnership with the National Dutch
Institute of Inland Water Management and Wastewater Treatment (RIZA), and the major Dutch
consulting companies. SOBEK is a fully dynamic 2D hydraulic model for floodplain modeling. The
model capable to combined the 1D channel and hydraulic structures within the 2D domain. In
addition rainfall excess can be directly applied as input in SOBEK. The computation in SOBEK is
divided into 2 domains; 1) river network (1D) and 2) 2 Dimensional part which uses rectangular
computational cells to represent the floodplain terrain. The computational is based on the equation of
momentum balance and the conservation of mass. “For the momentum balance the 1D and the 2D
system remain strictly separated. In this case the velocities or discharges belong to either to the 1D
part or 2D. On the other hand, the conservation of mass, the appropriate 1D and 2D volumes are
combined and mean that they share the same water level. The integration of 1D and 2D in general
give wide opportunities for instance, modeling the complex urban flood scenario which commonly
takes into account the urban drainage, street flow and flow over (overland flow in parking places)
(Bishop and Catalano, 2001).
There are several advantages of combining the 1D and 2D modelling (Bin Usamah, 2005):
1. Combining the 1D and 2D model allow us to closely model the real situation of flood event.
Each model for instance the 1D model is simulating the flow in channel or in river with
extensively detailed calculation. On the other hand, the 2D model simulates the overland
flow which incorporates detailed information on terrain. The detailed off course depends on
the quality of the terrain model, for instance the high resolution terrain model capable in
representing small physical obstruction in the floodplain.
2. The combination of 1D and 2D flood model often allows the grid cells significantly larger
than the purely model with 2D model. In this case, the reason is that the river, streams,
channel are modelled separately from the 2D model. Narrow channels or streams, with
purely 2D model require very fine 2D resolution for better representation. This would make
2D system numerically less feasible and almost impossible to model
3. The modeller works in the same user interface for both pre-processing of data and post
processing the results
Flood hazard mapping
Hazard is “a potentially damaging physical event, natural phenomenon or human activity that
may cause the loss of life or injury, property damage, social and economic disruption or
46
PHILIPPINES
environmental degradation” (United Nations of International Strategy for Disaster Reduction).

However, hazard can be characterized in many ways based on the specific purposes. On the other
hand “Risk is the actual exposure of something of human value to hazard and it is often regarded as
the product of probability and lost” (Smith, 2001). Smith (2001) suggested that the impact of the
natural hazard can be recognized in several aspects as listed below:
1. Hazard to people – Death, injury, disease, stress

2. Hazard to goods – Property damage, economic loss
3. Hazard to environment – loss of flora and fauna, pollution, loss of amenity
In reality flood hazard consists of human and natural component. Flood maybe exacerbated
by fluctuation of climate, and also by human activities, for instance deforestation and modification of
river course. This chapter will discuss several ways of defining flood hazard through the combination
of flood characteristics. Factors that affect hazard and the disruption caused by a flood could be
grouped into four broad categories (Ramsbottom et al., 2003).
1. Flood behaviour (severity of flood, response time, rate of rise, depth, flow velocity,
duration, water quality and etc.)
2. Evacuation issues (evacuation routes and time for evacuation)
3. Population at risk (number and vulnerability of people, flood awareness)
4. Emergency management (flood forecasting, flood warning, flood response, evacuation
and recovery)
Flood emergency management can be divided into flood prevention, preparedness, response
and recovery (Emergency Management Australia, 1999). Flood preparedness, response and recovery
aim in reducing existing and future flood risk. One of the major concerns in flood preparedness is
flood evacuation plan, which require information of flood behaviour during flooding. Flood hazard
analysis is crucial in emergency planning in serving valuable information of level of flood hazardous
based on specific criteria. The assessment requires detailed analyses of all factors affecting the
hazard, such as flood behaviour, flood awareness and possible evacuation problems (Ramsbottom et
al., 2003). “Even a relatively crude analysis will identify the main hazard to consider in a flood
emergency and will provide valuable guidance on the deployment of emergency vehicles, etc.”
(Ramsbottom et al., 2003). Ramsbottom et al (2003) suggested a flood hazard map which was based
on the combination of flood velocity, depth and evacuation time (figure 3.5). The monitoring of
hazard level was focused along the evacuation routes, and the hazard map has taken into account the
wading capability of adults and children for a specific flood depth and flood velocity. The final
hazard level was obtained from the adjusted hazard estimate based on the initial hazard estimate and
the relative evacuation time (ratio of the time available for evacuation to the minimum time required
for orderly evacuation).
47
PHILIPPINES
b
Figure 3-8: Graphs that combine flood velocity, depth to define the initial level of flood hazard (a) and the
final flood hazard level which was derived through the adjustment of the initial flood hazard level with
the relative evacuation time (b) (Ramsbottom et al., 2003)
The relationship between flood characteristics and it adverse effects should be determined
carefully, in order to construct the meaningful flood hazard maps for emergency plan purposes. The
Geological Society of Australia (2005) suggested flood hazard categories based on Johnstone River
flooding event. The suggested hazard categories are shown in table 3-10 and figure 3-9.
Table 3-11: Flood hazard categories (The Geological Society of Australia, 2005)
Hazard Flood Characteristics

category return
period
Low 100 yr  Area that are inundated in 100 year flood, but the floodwater are
relatively shallow (typically less than 1 meter deep) and not
flowing with high velocity
 Adult can wade
High 100 year  The depth and velocity are sufficiently high that wading is
(wading not possible (risk of drowning)
unsafe)
High – 100 year  Areas where the floodwater are deep (>1m), but not flowing with
Depth high velocity
 Damage only to building contents, large truck able to evacuate
High 100 year  Typically areas where there is deep water flowing with high velocity
(Floodway)  Truck evacuation not possible, structural damages to light
frame houses, high risk to life
Extreme 100 year  Typically areas where the velocity is > 2m/s
 All building likely to be destroyed and high probability of death
48
PHILIPPINES
Most of the hazard maps have taken into account the vulnerability of the respective element
at risk. Thus it creates vague boundary between hazard and vulnerability, especially when the hazard
map is created for the specific purpose.
Figure 3-9: Graph of flood hazard categories (The Geological Society of Australia, 2005)
Community risk perception in Naga City
“In reality the environment is neither benign nor hostile. It’s neutral and it’s only human
location, needs and perception which identify resources and hazards in the spectrum of natural
events” (Burton et al.,1993) as cited in (Smith, 2001). On the other hand, the sensitivity of human to
natural hazards is a combination of the physical exposure (the range of potentially damaging events
and their statistical variability at a particular location) and human vulnerability (the extent of social
and economic tolerance to hazardous events) (Smith, 2001). In this study, the sensitivity is defined
by the flood risk perception of the residents in several residential areas. The community-based flood
risk perception in Naga City was derived through interview sessions with the residents by the
ongoing PhD researcher, Peters Guarin Graciela Graciela. This flood risk perception reflects the
coping mechanisms and also the level of acceptance on the flood problem in the residential areas
(based on the past flood problems). The flood risk perception was defined by the researcher (Peters
Guarin Graciela Graciela) based on the flood depth and duration. Table 3-11 and 3-12 show the flood
risk perception derived from the interviews.
Table 3-12: Community-based flood risk perception in Barangay Triangulo (Ongoing PhD research by
Peters Guarin Graciela Graciela)
Risk Perception (Barangay Triangulo)

Depth/ Duration 1 day (24 hrs) 1 to 2 days (24hrs – More than 2 days (more
48 hrs) than 48 hrs)
Ankle level (20 cm) Normal Normal Get used to the situation
Knee level (40 cm) Manageable Disturbing Unmanageable
49
PHILIPPINES
Waist level (80- 100 cm) Manageable Highly disturbing Disastrous

Breast level (140 cm) Unmanageable Disastrous Disastrous
Breast level and high Disastrous Disastrous Disastrous
wind
The Barangays Mabolo and Triangulo are the worst affected areas in Naga City, and always
experience high water depths during flood events (Palmiano-Reganit, 2005). The residential areas in
Barangay Triangulo are adjacent to the Naga River and furthermore some parts of the area are
located in the backswamp where also water accumulates from adjacent barangay drains. During
flooding, some parts of the areas were flooded with more than 2.0 m, especially in the areas with
marginalized community (Palmiano-Reganit, 2005). Barangay Triangulo is known as the lowest part
of Naga City. In Barangay Mabolo, on the other hand, the deepest flood depth ever experienced was
2.0 to 2.5 m.
Table 3-13: Community-based flood risk perception in Barangay Triangulo (Ongoing PhD research by
Peters Guarin Graciela)
Risk Perception (Barangay Mabulo)

Depth/ Duration 1 day 1 to 2 days (24hrs – More than 2 days (more
Ankle level (20 cm) Normal Normal Disturbing (skin
diseases)
Knee level (40 cm) Manageable Highly disturbing Disastrous
Waist level (80- 100 cm) Unmanageable Disastrous Disastrous
Breast level (140 cm) Unmanageable Disastrous Disastrous
Breast level and strong Disastrous (< 5 hours) Disastrous Disastrous
wind
In general, “Rapid urbanization has attracted migrants from countryside”, and buying house
in Naga City is far from their means (Ernesto T. Elcamel, Naga City Disaster Mitigation Project).
Thus, they were forced to live on marginal area; the banks of the Naga and Bicol River. These are
where most of the urban poor community in Naga City located. They are exposed to disaster, in
which most of the houses were not well constructed and subjected to destruction during typhoon and
flood.
Generally, community in these areas have high awareness on flooding in their areas. Areas
with high magnitude of flooding have higher response, especially during flash flood and flood during
monsoon season. In conclusion, most of the respondents in these areas consider that flooding is a
nuisance except in areas with low financial resources and had experienced of severe event of
flooding (Palmiano-Reganit, 2005). However, some had just accepted that flood is a natural process
in their area. They have chosen to live in the area, for economic and education reasons.
50
PHILIPPINES
4 DTM and DSM generations
Overall methodology
In this study, the DTM was generated in 4 major steps; 1) elevation data preparation and
analysis, 2) elevation data interpolation, 3) accuracy assessment and reporting and finally 4)
integrating the natural terrain with the man made terrain. Figure 4-2 shows the detailed flow chart
throughout the DTM and DSM generation process. The process begins with a detailed investigation
on each elevation data source. In this stage, each elevation dataset is assigned to the corresponding
nominal horizontal and vertical accuracy based on the NSSDA standard (see table 3-3 and 3-4). Next,
each elevation dataset is ranked and prioritized based on 2 selection steps, which will be further
discussed in section 4.3.2. In this study, the entire dataset need to be in point format for data
integration. Thus, all the contour lines are converted into points. The conversion is made based on 2
line-to-point conversion interval; 1) 50 m and 2) 100 m intervals (figure 4-1). After that, the
elevation datasets are separated into ground and man-made elevation data. Detailed discussion on
data integration will be in section 4.3.
50 m Contour lines 100 m
a b
Figure 4-1: The conversion of contour lines to points with 50 m interval (a) and 100 m interval
The elevation dataset of the ground terrain is used to construct the DTM of the study area.
The DTMs are generated through 4 interpolation methods; 1) Ordinary Kriging 2), ANUDEM 3),
TIN
3) and 4) Polynomial interpolation method. The results of the interpolations are evaluated through 3
DTM quality assessment approaches; 1) DTM accuracy assessment, 2) visual assessment and 3) the
distribution of interpolation errors. The final DTM is then selected based on these assessments. On
the other hand, the construction of the man-made terrain is done separately by using a second point
dataset (man-made elevation data). The man-made terrain includes information on recent and
proposed development in Naga City is constructed through the TIN-based interpolation method.
Finally, the man-made terrain model is added onto the ground DTM to produce the DSM for present
and future situations of Naga City.
51
PHILIPPINES
3
4
Figure 4-2: Overall methodology for DTM and DSM generations
52
PHILIPPINES
Elevation data
Data sources
The elevation dataset for DTM generation is derived through the integration of various
elevation data sources. One should keep in mind is that they vary in both horizontal and vertical
accuracies. As mention before, the user can do nothing to improve the accuracy; however,
something needs to be done to ensure that the interpolation methods used for terrain modelling could
keep further error as low as possible. Most of the data as listed in table 4-1 were derived from the
Naga City Municipality Office. Some of the data were derived from the developers for specific
projects for instance, spot heights derived from the Almeda highway plan and the drainage project in
Barangay Triangulo. Those plans provide good quality of ground elevation measurements, which
were derived through the conventional geodetic levelling approach. Its accuracy and reliability are
comparable to those derived from the field observation spot heights.
Table 4-1: Available data sources for DTM generation from previous study by Tennakoon (2003)
Contour interval and
Map scale and GPS Production Production
Source of data the precision of
positioning accuracy year method
geodetic levelling
Aerial
0.25 m contour photography and
1 Contour line south of Naga City 1:4000 1981
interval ground control
survey
Contour line for the whole Naga 1.00 m contour
2 1: 1000 1981 N/A
city interval
Contour line from Naga Drainage 0.25 m contour
3 1: 1000 1981 N/A
Plan (1981) interval
Ground control
Contour line from CBD II 0.25 m contour survey and
4 1: 1000 2000
development Plan interval contour line
interpolation
Ground control
Spot height from topographical survey for map
5 1:4000 N/A 1981
map south of Naga City with scale of
1:4000
Spot height along roads from the Ground control
6 1:1000 N/A 1981
Naga City Drainage Plan (1981) survey
Spot height from the Drainage
Ground control
7 Plan in Triangulo (ground and 1:1000 Less than 1 meter 2000
survey
drainage crown elevation)
Spot height from the Almeda
Ground control
8 highway plan (ground and 1:1000 Less than 1 meter 2000
survey
final road elevation)
Spot height for the whole Bicol Ground control
9 1: 10,000 N/A
region survey
Less than 5m (GPS Ground control
10 Field observation spot height Less than 1 meter 2003
accuracy) survey
53
PHILIPPINES
Most topographical maps were published in 1981, which are considerably old, thus
subjected to changes due to natural process (sedimentation, erosion and so forth) and man-made
changes. However, efforts had been focused to update the changes as detailed as possible, especially
those caused by the development projects in of Naga City. Each map is rectified to a common map
projection (Philippines Transverse Mecator), and the error in map geometry correction process was
reduced as low as possible (less than a pixel). According to the officer in the City Engineer Office,
all the topographical maps were referred to the same vertical datum. However, there is only one
topographical map (Contour line south of Naga City) that has a definition on the acquisition
techniques, while the rest were not defined. On the other hand, the spot height which was derived
from the drainage project in 1981 was observed through the geodetic levelling technique.
Nominal data accuracy
The nominal horizontal and vertical accuracies for each elevation dataset were assigned
based on the NSSDA standard, (see section 3.2). In this standard, the nominal horizontal and vertical
accuracies are defined based on the map scale and contour interval. According to the nominal
horizontal and vertical accuracies of all elevation data sources (table 4-2), the field observation spot
height has the lowest nominal horizontal less than 5 m. This is caused by the accuracy of the hand
held GPS used in the levelling process. On the other hand, the lowest nominal vertical accuracy is
from the contour lines of the topographical map for the whole Naga City, with accuracy between
0.366 m and 0.549 m.
Figure 4-3 shows the spatial coverage of the different datasets. The spot heights that were
derived from 1:10,000 topographical map only cover part of the study area. It is used to extend the
study area for the purpose of the DTM interpolation. In this case, this spot height (1:10,000
topographical map) will have less influence in the DTM generation. The geodetic levelling is one of
the most accurate elevation data acquisition techniques. The accuracy of the levelling system as
defined in the literature is 1 mm. However, in this study, the accuracy spot height derived from the
geodetic levelling is defined as less than 10 cm (after taking into account errors during the levelling
process). This information will be used in the data integration process (section 4.3.2).
Table 4-2: Nominal horizontal and vertical accuracy of the data sources based on the NSSDA standard
Nominal horizontal Nominal vertical
Source of data
accuracy accuracy
1 Contour line south of Naga City 2.316 to 4.633 m < 0.183 m
Contour line for the whole Naga
2 < 1.158 m 0.366 to 0.549 m
city
Contour line from Naga Drainage
3 < 1.158 m < 0.183 m
Plan (1981)
Contour line from CBD II
4 < 1.158 m < 0.183 m
development Plan
Spot height from the
5 topographical map south of Naga 2.316 to 4.633 m < 0.183 m
City
54
PHILIPPINES
< 0.1000 (geodetic

Spot height along roads from
6 < 1.158 m survey is assumed
Naga City Drainage Plan
less than 10 cm)
(1981)
Spot height from Drainage Plan < 0.1000 (geodetic
7 in Triangulo (ground and < 1.158 m survey is assumed
drainage less than 10 cm)
crown elevation)
< 0.1000 (geodetic
Almeda highway plan
8 < 1.158 m survey is assumed
(ground and final road
less than 10 cm)
elevation)
Spot height for the whole Bicol
9 3.353 to 6.706 m N/A
region
< 0.1000 (geodetic
< 5.000 m (hand held
10 Field observation spot height survey is assumed
GPS accuracy)
less than 10 cm)
Fieldwork data
Figure 4-3 shows an uneven distribution of elevation data over the study area. A field survey
was conducted to fill in the major gaps. More than 200 points were observed through the
conventional geodetic levelling. The measurements were made at 7 specific zones, namely 1) the
CBD II area, 2) Barangay Tinago, 3) Penafrancia Avenue, 4) Barangay Bagumbayan Sur, 5)
Barangay Sta. Cruz-CBD I, 6) Barangay Sabang LCC and 7) Metro Ville (figure 4-3). These areas
were selected based on the new development projects, less density of elevation measurement (based
on available data collected by Tennakoon (2004)), and high topography variation area.
5 2
7
6 1
Figure 4-3: Available elevation data and Spot Heights derived during
newly Collected spot heights during fieldwork fieldwork Field survey zones
55
PHILIPPINES
Available elevation
points Available
contour lines
56
PHILIPPINES
Data preparation and analysis
Semi-variogram analysis for each elevation dataset
The semi-variogram analysis for each dataset reveals the spatial dependency of the datasets.
The integration of all elevation datasets (will be discussed later in section 4.3.2) will rely on the
characteristics of each elevation dataset. In this case, the datasets with high nugget value might
influence the final nugget value of the integrated dataset. As discussed in section 3.1.2.1, the high
nugget value in the integrated elevation dataset could be caused by several factors including the
complexity of the terrain and the effect of integrating different sources of elevation data (vary in
scale and vertical accuracy). Thus, with the detailed semi-variogram analysis of each dataset, the
influence of these factors in the integrated elevation dataset can be roughly defined. This analysis
will be discussed later in section 4.3.2.
In this study, a semi-variogram was constructed for each elevation dataset (figure 4-5). All
semi-variograms, except the semi-variogram derived from “Contour line from Naga Drainage Plan”
dataset, show quite good spatial dependency. According to figure 4-5, most nugget values are less
than 1.0. However, high nugget values are observed in the “Contour line for the whole Naga City”
and “Contour line from Naga Drainage Plan” datasets. In this case, the high nugget value might be
caused by the high variation of elevation values in a short distance. The contour lines derived from
the Naga Drainage Plan dataset contains quite complicated terrain features, for instance natural levee,
swales and meandering of the river course (figure 4-4 (a)). This leads to abrupt change in elevation,
especially in the meandering river course. This data suffers “total nugget effect”, which contains high
nugget value with less spatial dependency elevation data. On the other hand, the contour lines
derived from the topographical map of the whole Naga City covers a large area. Therefore, as
illustrated in figure 4-4 (b), it contains different type of terrain features, for instance meandering river
course, rugged terrain area in the South-eastern part of Naga City.
b
Figure 4-4: Contour line from Naga Drainage Plan (a) and Contour line for the whole Naga City (b); The
datasets contain high nugget value
57
PHILIPPINES
Contour lines south of Naga City Contour lines for the whole Naga City Contour lines from Naga drainage
plan
Contour lines from the CBD II

Spot heights from the topographical Spot heights along roads from
development plan
map of the southern part of Naga City Naga City Drainage Plan
Field observation spot heights (2003) Field observation spot heights (2005) Spot heights for the whole Bicol
region
Figure 4-5: Semi-variograms of the elevation datasets, constructed with 150 m lags
58
PHILIPPINES
Elevation dataset integration
The problem of integrating elevation data from different sources with different scales and
accuracies lies on the fact that the elevation values in the combined dataset may lie close to each
other. The challenge is to identify the appropriate approach to prioritize the datasets, to identify
which of those datasets represent the true terrain elevation and to combine the entire datasets. In this
study, the datasets were prioritize based on 2 steps; 1) nominal horizontal and vertical accuracy and
2) three general selection steps as illustrated in figure 4-6. In the first step, all datasets need to be
ranked according to their nominal horizontal and vertical accuracies as discussed in chapter 4.2.2.
Table 4-3 shows the datasets with the corresponding level of priority.
Table 4-3: Elevation datasets with the corresponding priority level based on the nominal horizontal and
vertical accuracy
Source of data Priority level

1 Contour line south of Naga City 4
Contour line for the whole Naga 3
2
city
Contour line from Naga Drainage 2
3
Plan (1981)
Contour line from CBD II 2
4
development Plan
Spot height from the 3
5 topographical map south of Naga
City
Spot height along roads from 1
6
Naga City Drainage Plan (1981)
Spot height from Drainage Plan in 1
7 Triangulo (ground and drainage
crown elevation)
Almeda highway plan (ground 1
8
and final road elevation)
Spot height for the whole Bicol 6
9
region
10 Field observation spot height 5
* remarks 1 = Highest and 6 = Lowest
The spot heights derived from the construction plans have the highest priority level since they
are mapped in the large scale maps and produced by the conventional geodetic levelling techniques.
The spot heights derived from the fieldwork suffer low positioning accuracy of the hand held GPS,
thus ranked in the 5th place. Small scale topographical maps with large contour interval are ranked
with low priority compared to the large scale and small contour interval topographical maps. The
spot heights derived from the smallest scale map (“Spot height for the whole Bicol region”) is ranked
to the lowest priority dataset. In addition there is no information on the nominal vertical accuracy of
the spot heights.
59
PHILIPPINES
However, compared to other datasets, the spot heights derived from the fieldwork contain
clear information on the acquisition technique and also the reference points used. On the other hand,
according to the section 3.2.2, historically, the contour lines were interpolated manually based on the
spot heights. This implies that the contour lines is less accurate compared to the spot heights in the
topographic maps. Based on this information, the second step of data selection is carried out to give
the highest priority to the field survey spot heights, followed by the spot height from the
topographical map and construction plans and finally to the contour lines (figure 4-6).
Field survey points
Spot height
Contour lines
Figure 4-6: The second dataset selection step
The final priority level for each dataset was assigned based on the first and the second
selection steps. According to table 4-4, the highest priority is given to the spot heights derived from
the field survey. This is followed by the spot height from the construction plans. The spot heights
from topographical maps are assigned a higher priority level compared to the contour lines. Finally,
the spot heights derived from the small scale topographical map is assigned with the lowest priority
level.
Table 4-4: The final priority level for each elevation dataset based on the first and the second selection
steps
Priority
Priority level
level based Final priority
Source of data based on the 2nd
on the level
selection step
nominal
accuracy
Contour line south of Naga
1 4 3 6
City
Contour line for the whole
2 3 3 5
Naga city
Contour line from Naga
3 2 3 4
Drainage Plan (1981)
Contour line from CBD II
4 2 3 4
development Plan
Spot height from the
5 topographical map south of 3 2 3
Naga City
Spot height along roads from
6 Naga City Drainage Plan 1 2 2
(1981)
Spot height from Drainage
7 Plan in Triangulo (ground and 1 2 2
drainage crown elevation)
8 Almeda highway plan (ground 1 2 2
60
PHILIPPINES
and final road elevation)

Spot height for the whole
9 6 2 7
Bicol region
10 Field observation spot height 5 1 1
The next step is the elevation data integration. In this process the entire contour lines are
converted to points. The integration of the elevation datasets is based on the point’s integration.
Elevation data in points are easier to combine and edit. However, direct conversion from contour
lines to points might produce huge amount of points. Thus, the conversions are made based on 2 line
to point conversion intervals; 100 m and 50 m distance intervals (figure 4-1). These conversion
schemes are applied to all contour lines and it produces two sets of elevation point datasets. At this
point, all datasets are in the point elevation format. The next step is to combine all the datasets. The
integration is based on the data priority derived from the previous step (table 4-4). The dataset is
combined one by one, and at the same time the overlapping data between one dataset with another
dataset should be removed. The elevation point is defined as overlapping if they are within 3 m
radius distance. In this process the lower priority datasets will be removed if it overlaps with the
higher priority datasets. This integration will produce 2 datasets; the combined datasets with 100 m
and 50 m contour to line conversion intervals. After this, the 100 m and 50 m line to point conversion
interval will be called as dataset A and B respectively.
This step-by-step data integration technique aims at reducing overlapped data and to put the
most reliable elevation data in representing the true terrain elevation in the study area. However,
since the original elevation datasets were derived from various sources, with different scale and
accuracy, there must be some disagreements between the elevation points. According to the previous
literature, the disagreement between accuracy of the original data had increased the variation of the
elevation data set in a short range. On the other hand, complex terrain features usually exhibits abrupt
changes of the elevation in a short distance. These problems can be observed by mean of semi-
variogram analysis. The high variation elevation value in a short distance can be shown as high
nugget value in the semi-variogram plot. This effect is known as a “nugget effect”. Figure 4-5 in
section 4.3.1, shows that most of the datasets, except the two datasets with high nugget value, have a
nugget value less than 1.0. In this case, we have to ascertain the effect of complex terrain features
and the effect of data integration to the “nugget effect”. For this purpose, the original elevation
datasets with the “nugget effect” need to be removed from the combined dataset. This step will reveal
the influence of integrating various elevation datasets to the “nugget effect”
61
PHILIPPINES
Figure 4-7: Close semi-variogram constructed with 10 meters lags for 50 meters contour lines to point
conversion interval (Dataset B); with high nugget datasets (a) and without high nugget datasets (b); the
nugget value in figure b is still more than 1.0,
Figure 4-7 shows the semi-variogram for dataset B with and without the original datasets
that contain the high nugget values. The lines in the semi-variogram are plotted to show, the nugget
values fall. The semi-variogram in figure 4-7 (b) has low nugget value compared to figure 4-7 (a).
This implies that the original datasets with the highest nugget have a negative effect on the nugget of
the overall dataset. Figure 4-7 (b) shows that integrating multiple datasets with nugget of less than
1.0 still result in a dataset with a significantly higher nugget value, namely 2.2. However, the nugget
value in the semi-variogram illustrated in figure 4-7 (b) is more than 1.0. However, the original
elevation datasets cannot be removed because they are needed in the DTM generation. Both datasets
cover quite a large portion of the study area.
Figure 4-8 (a) and 4-8 (b) illustrate the semi-variogram plot for dataset B and dataset A
respectively. Both semi-variogram are fitted with Gaussian semi-variogram model. Both semi-
variograms are fitted at 4500 m distance (range) since the spatial dependency pattern only exists until
62
PHILIPPINES
this distance. The Gaussian semi-variogram model should be chosen based on the underlying process
under investigation. The Gaussian model with an exponential rise trend has strong close range
dependency and very smooth at each point. This trend fits with the landscape of the Naga City, which
starts with low lying flood plain areas, and the elevation is slowly increased towards Mount Isarog.
The lags for the semi-variogram is set at 150 m, since through the experience in the study area, the
Naga City is quite flat with less elevation variation in a short distance.
According to the parameters of the model in the datasets, the nugget and the sill value in
dataset A is lower than dataset B. This result shows that, the 100 m contour lines to point conversion
interval has reduced the high variation of elevation values in a short distance, thus reduces the total
variation of the dataset (decrease in sill value).
63
PHILIPPINES
Model: Gaussian Nugget: 2.9 Model: Gaussian Nugget: 2.2

Sill: 52 Sill: 42
Range: 4500 m Range: 4500 m
a b
Figure 4-8: Semi-variogram with 150 m lag for 50 m line to point conversion interval (a) Semi-variogram
with 150 m lag for 100 m line to point conversion interval (b)
Dataset A, with lower nugget effect compared to dataset B is used for further analysis. The
next step still aims at reducing the nugget effect of the integrated dataset by taking block average.
The next approach concerns on averaging the elevation dataset within 5 m distance (figure 4-9). At
this point, it is needed to be mentioned that because of flood modelling requirements, the final DTM
will have a pixel size of 5 X 5 meters. This corresponds also with the lowest horizontal accuracy of
the source datasets (the field survey, see table 4-2). Therefore the block for averaging should be
equal or less than 5 meters. According to the literature in section 3.2.2, the resolution of the DTM
should consider the post spacing of the elevation data. The post spacing represents the density of the
elevation data in a certain area. Thus, in this process, the averaging technique will not have a big
impact on the quality of the final DTM. The process is carried out using “MapRasterizePointCount”
and “MapRasterizePointSum” functions in ILWIS.
5m 5m
5m 5m
Figure 4-9: Averaging elevation data within 5 m distance interval
Figure 4-10 shows the semi-variogram of the dataset produced by the averaging technique.
The semi-variogram is fitted with Gaussian model, with 4500 m distance. The sill value has decrease
to 37, however the nugget value remains unchanged at 2.2. The lags of the semi-variogram is set to
150 m. Difference lag value might produce different shape of semi-variogram. Narrow lag produces
higher resolution semi-variogram and the nugget effect can be observed in a short distance. However,
according to table 4-5 the nugget value remains unchanged for lag value of 25 m and 15 m.
64
PHILIPPINES
After completing all the steps above, the final dataset with the lowest nugget value might
contain:
1. Less overlapping elevation data
2. Less complex terrain features (less clustered elevation dataset)
3. Good elevation data to represent the real terrain of the study area
4. Decrease in disagreement between datasets
In this approach we might loss detail terrain elevation, while trying to reduce the
disagreement between various datasets.
Model: Gaussian Nugget: 2.2

Sill: 37
Range: 4500 meters
Figure 4-10: Semi-variogram with 150 m lag for 5 m average block dataset
Table 4-5: The value of Semi-variogram model parameters for each dataset; the 2nd and the 3rd datasets
will be used for the DTM interpolation
Number
Dataset Nugget Sill Range Lag Model
of points
50 m contour lines to point 150 m Gaussian 11,131
2.9 52 4500
conversion interval (Dataset B)
100 m contour lines to point 150 m Gaussian 6889
2.2 42 4500
conversion interval (Dataset A)
5 m block elevation average 2.2 37 4500 150 m Gaussian 5566
5 m block elevation average* 2.2 16 2500 25 m Gaussian 5566
5 m block elevation average* 2.2 7.5 1500 15 m Gaussian 5566
* The semi-variogram is shown in appendix C
Table 4-5 shows the semi-variogram parameters value of the different final dataset. The
datasets with the lowest nugget value 100 m contour lines to point conversion interval and 5 m block
elevation data average) will be used in the ground terrain modelling.
65
PHILIPPINES
Ground terrain modelling
The next step focuses on the ground terrain modelling using different interpolation methods.
The interpolation method should preserve the detailed terrain information and reduce the effect
datasets disagreement. The interpolation of the ground terrain is done with 4 interpolation
techniques;
1) Kriging, 2) TIN, 3) Polynomial trend surface and 4) ANUDEM.
The kriging interpolation is not only based on the distance between the elevation points to be
predicted with the existing elevation points, but also on the strength of the correlation. This
interpolation is also known as an exact interpolator, where the interpolated surface will pass exactly
on the input samples surface. However, the kriging interpolation method is more robust to the
presence of noise in the dataset. Due to noise or errors in the datasets that are quite far from the local
trend, kriging will put attention to the local trend rather than to that measurement. However, kriging
is quite sensitive to the homogeneity of the surface, in which it tends to produce a smooth
interpolated surface. In this sense, the complex terrain features usually can’t be preserved in the final
interpolated surface.
TIN is also known as an exact interpolation technique. The advantage of the TIN structure is
the sampling density should vary with the complexity of the terrain. Less complex terrain will require
less elevation data, and dense samples are needed for complex terrain. However, this model requires
a large amount of high accuracy data. In general, the DTM surface will be constructed exactly based
on the input dataset. In this case, the TIN based terrain modelling is sensitive to the errors in the input
dataset.
ANUDEM interpolation technique relies on the ancillary data to produce good quality DTM
for hydrological applications. This interpolator uses ancillary information of the terrain in the
interpolation process. This interpolator was optimized to work efficiently in local interpolation
without losing the surface continuity of the global surface. Its performance is comparable to Kriging
and Spline interpolation methods. However, in this study, no ancillary information available on sink
point and stream network. Thus, the interpolator is forced to work alone in terrain interpolation.
The polynomial interpolation is based on fitting a mathematical function to the elevation

dataset of known coordinate. The corresponding polynomial function will define number of points
required in the interpolation. Higher polynomial function requires large number of elevation points
and it tends to produce quite complex interpolated terrain. On the other hand, low degree polynomial
function tends to produce smooth surface and required less number of elevation samples. The smooth
interpolation scheme may be suitable to level out the errors in the coarse elevation data accuracy.
The final DTM resolution is 5 m, since it is adequate to investigate the changes in flood
behaviour for a small reclamation area. As discusses in section 3.1.1 table 3.1 as suggested by
Tennakoon (2004), the 5 m resolution DTM is enough to represent the detailed features in urban
area.
66
PHILIPPINES
Krigging interpolation
Additional analysis on the variogram is made to assess the possibilities of using universal
kriging and kriging with external drift. Kriging with external drift requires good correlation between
the auxiliary data and the target variable (elevation). In this study, the landuse or landcover
information is used as an auxiliary data, with the assumption that this data will have a good
correlation with the elevation values. Thus the relation between them is first investigated by
analysing the box plots and for this purpose, all elevation dataset (ground terrain and man-made
terrain) are combined together.
1
1
Elevation
5
0
c1 c10 c11 c12 c2 c3 c4 c5 c6 c7 c8 c9

C1 - Agriculture (paddy
C4 – Dumpsite C7 – Open area C10 – Riverbank
field)
C5 – Industrial C8 – Parks C11 – Swampy area
C2 - Cemetery
C6 - Institutional C9 - Residential C12 - Transportation utilities (railway)
C3 – Commercial
Figure 4-11: Box plot between elevation and land use or land cover classes; No clear elevation separation
between landuse or landcover classes
However, according to figure 4-11, there is no clear separation or trend between elevation
values of each landuse or landcover class. Most of the elevation values of each landuse or landcover
class are overlapped. Thus, no further analysis made on this dataset. Further investigation is made on
the trend of the residual for the whole study area. The Universal kriging interpolation is based on the
trend of the residual as an auxiliary data in the interpolation. The trend of the residual is modelled as
a function of elevation points coordinate. The residual dataset can be obtained by subtracting the
original elevation dataset to the predicted dataset which derived from trend surface of the original
elevation dataset. In this study, the residual dataset was obtained automatically using R software.
Figure 4-12 shows the residual variogram for 3 lag distances (10, 50 and 150) for the whole study
area. However, it is shown that the residuals of the dataset have no strong spatial dependence trend or
not spatially dependence.
67
PHILIPPINES
a b
Figure 4-12: Residual variogram with 10 m (a), 50 m (b) and 150 m (c) lags; the residual of the has no
strong spatial dependence trend
Previous analyses have shown that land use or land cover and residual data can’t be used to
support the kriging interpolation. Since the elevation points are quite densely distributed over the
study area, the Ordinary Kriging is used for the interpolation. Both semi-variograms, which are the
100 m line to point conversion distance interval and with 5 m block average with 4500 m range are
used for the Ordinary Kriging interpolation (figure 4-13). Even though, attempts were made to
decrease the lag value to 15 m and 25 m to investigate the shape of the semi-variogram at a short
distance. However according to table 4-5, though the lag of the semi-variogram was reduced to 15 m
and 25 m, the nugget value remains unchanged. Furthermore, according to section 3.1.2.1, the shorter
range semi-variogram model will have the similar effect with pure nugget semi-variogram. In this
case, the Ordinary Kriging tends to smooth the interpolated surface.
In the elevation interpolation, only 90 percent of the whole data are used in the interpolation,
in which the remaining 10 percent of the elevation points will be used for the accuracy assessment.
The interpolation is done by using ILWIS 3.3. For the result, see figure 4-16 (d) and (e).
68
PHILIPPINES
Model: Gaussian Nugget: 2.2 Model: Gaussian Nugget: 2.2

Sill: 42 Sill: 37
Range: 4500 meters Range: 4500 meters
a b
Figure 4-13: Variogram models for 100 m line to point conversion distance interval (a) and with 5 m
averaging (b); lag 150 m (these datasets will be used in Ordinary Kriging interpolation)
Higher order interpolation, TIN – based terrain modelling and ANUDEM

interpolation
The next interpolations are made on the 100 m contour lines to point conversion interval. The
higher order interpolation technique is based on the 2 nd to the 6th polynomial degrees in ILWIS. On
the other hand, TIN and ANUDEM interpolation methods are done using Arc GIS 9.0. In TIN-based
DTM modelling, the interpolation starts with the construction of TIN structure. This structure is then
converted to 5 m raster resolution DTM. The ANDEM is integrated in ArcGIS under the option
“Topo To Raster”. It is based on the algorithm used in the ANUDEM software. The “Topo To
Raster” operation allows multiple input data processing (line, point and polygon) and each data can
represent information for instance, elevation data, sink, streamline and lake boundary. In this study,
only elevation data is used as an input.
DTM quality assessment
The DTM quality assessments are divided into accuracy assessment and visual assessment
and the spatial distribution of the interpolation errors. The error dataset is derived through
subtracting the interpolated elevations to the control elevation measurements. The control points are
the independent elevation points that were omitted from the interpolation. The control points were
randomly picked from the original data. However the selection didn’t include the elevation points
from the 1:10,000 scale map.
The suitable method for DTM accuracy assessment relies on the shape of the error dataset.
The conventional error assessment method requires normally distributed error dataset. However, in
most cases the error dataset may contain statistically outlier values and skew. In addition, the quality
of the DTM can be assessed through non-quantitative approach, for instance trough visualization of
69
PHILIPPINES
the generated DTM. In this case, the DTM should be able to represent the general landscape of the
area under investigation. Nevertheless, the definition of best DTM is depending on the application, in
which different applications may put constraint in different aspects. Thus, the optimum method is
through the combination of different DTM quality assessment approach. In this study, the final DTM
will be used in flood modelling. Thus the final DTM should full fill the following requirements.
1. Less error especially in the area of interest (less than 1.0 m)
2. Less artificial pits
3. Able to represent the real landscape of the study
area Thus, 3 DTM quality assessment approaches are used in this
study
1. DTM accuracy assessment
2. Spatial error distribution
3. Visual assessment
1) DTM accuracy assessment
The first step begins with the normality testing for each error dataset of the interpolators.
The normality test for each error dataset was carried out using Kolmogorov-Smirnov that is suitable
for large datasets. With 2 tailed test at 95% confidence level, the error datasets of all
interpolation method failed the test, which means that the datasets are not normally distributed (table
4-6). According to the box plots for each error dataset, it was found that the datasets contain
statistically outlier value (appendix B). In this case, it’s clear that the conventional DTM accuracy
assessment methods, for instance RMSE are not valid.
Table 4-6: Normality test for each error dataset
Interpolation approach Kolmogorov-Smirnov Z Significance

Ordinary Kriging with 5 meter block average 0.147 0
Ordinary Kriging with 100 m line to point conversion
0.127 0
distance interval
DEM derived from Triangulated Irregular Network (TIN) 0.353 0
2nd degree linear 0.071 0.001
2nd degree parabolic 0.081 0
2nd degree 0.071 0
3rd degree 0.076 0
4th degree 0.082 0
5th degree 0.077 0
6th degree 0.080 0
ANUDEM 0.145 0
According to this analysis, it is shown that the error datasets of the interpolation techniques
need to be analysed with more robust accuracy assessment methods. The next assessment method is
the percentile accuracy assessment method with no underlying normally distributed data in error
assessment. Detailed discussion on this method can be found in section 3.2.2. This method is based
on a simple statistical method whereby the accuracy statement depends on the maximum value of
error in each percentile. The DTM accuracy can only be made if there is no statistically outlier values
70
PHILIPPINES
found in the dataset at 3 standard deviation interval. However, the original method introduced by
Maune (2001) will only consider at maximum 10 percent outliers. In this study, the percentile error
assessment will be further continued until all error datasets can be compared at the same percentile
level (see section 3.2.2).
Table 4-7: Percentile accuracy assessment method for each interpolation method (after Maune (2001))
Error Boundary Num of

Interpolation techniques Percentile Mean Std
(m) (3 sigma) outlier
100 5.25 -0.03 1.02 3.06 >1
95 2.57 -0.04 0.72 2.16 >1
Ordinary Kriging with 5- meter block 90 1.69 -0.04 0.72 2.16 None
average 80 -1.00 0.00 0.39 1.17 None
70 -0.66 0.02 0.28 0.84 None
60 0.50 0.01 0.20 0.60 None
100 4.98 -0.04 1.15 3.45 >1
95 -2.75 -0.02 0.87 2.61 >1
Ordinary Kriging with 100 m line to 90 2.06 -0.04 0.68 2.04 >1
point conversion distance interval 80 -1.17 0.00 0.45 1.35 None
70 0.73 0.02 0.35 1.05 None
60 0.56 0.03 0.27 0.81 None
100 28.70 0.81 2.14 6.42 >1
95 4.61 0.41 0.80 2.40 >1
Triangulated Irregular Network 90 2.26 0.26 0.47 1.41 >1
(TIN) 80 0.88 0.12 0.19 0.57 >1
70 0.32 0.05 0.08 0.24 >1
60 0.15 0.02 0.04 0.12 >1
100 2.74 0.03 1.64 4.92 None
95 -3.57 -0.06 1.29 3.87 None
90 2.77 -0.05 1.09 3.27 None
2nd Degree linear
80 1.74 0.02 0.87 2.61 None
70 -1.39 0.02 0.73 2.19 None
60 1.12 0.01 0.59 1.77 None
100 7.07 0.01 1.67 5.01 >1
95 -3.80 -0.10 1.29 3.87 None
90 2.74 -0.11 1.11 3.33 None
2nd Degree parabolic
80 1.68 0.01 0.88 2.64 None
70 1.36 0.02 0.74 2.22 None
60 -1.11 -0.00 0.61 1.83 None
100 6.74 0.03 1.64 4.92 >1
95 -3.57 -0.06 1.29 3.87 None
90 -2.77 -0.05 1.09 3.27 None
2nd Degree polynomial
80 1.75 0.02 0.88 2.64 None
70 -1.39 0.02 0.73 2.19 None
60 1.12 0.01 0.60 1.8 None
3rd Degree 100 6.44 0.04 1.56 4.68 >1
95 -3.40 -0.03 1.21 3.63 None
71
PHILIPPINES
90 -2.64 -0.00 1.03 3.09 None

80 1.66 0.05 0.80 2.4 None
70 -1.27 0.43 0.42 1.26 None
60 0.97 0.01 0.54 1.62 None
100 6.60 0.04 1.55 4.66 >1
95 3.43 -0.03 1.20 3.60 None
4th Degree polynomial 90 2.58 -0.01 1.01 3.04 None
80 1.66 0.04 0.79 2.36 None
70 -1.21 0.03 0.64 1.91 None
60 0.97 0.02 0.53 1.58 None
100 6.13 0.06 1.51 4.54 >1
95 3.32 -0.01 1.18 3.55 None
90 2.55 0.04 1.00 3.01 None
5th Degree polynomial
80 -1.68 0.02 0.77 2.31 None
70 1.23 0.01 0.61 1.83 None
60 -0.92 0.00 1.51 4.54 None
100 5.92 0.06 1.50 4.49 >1
95 -3.38 -0.03 1.17 3.51 None
90 2.47 0.04 0.99 2.96 None
80 1.66 0.05 0.76 2.27 None
70 1.23 0.02 0.59 1.78 None
60 0.91 -0.01 0.47 1.42 None
100 4.37 -0.09 0.93 2.78 >1
95 2.06 -0.05 0.68 2.03 >1
90 1.61 -0.03 0.55 1.64 None
ANUDEM
80 -0.98 -0.02 0.38 1.13 None
70 0.62 -0.03 0.27 0.81 None
60 0.46 0.00 0.20 0.59 None
The DTM accuracy of all interpolation 700 Histogram of error dataset (TIN)
techniques, except TIN, can be compared at 80 th
percentile (table 4-7). The TIN-based 600
interpolation method couldn’t be compared

500
through this method due to extreme values in the
Frequen
right side of the histogram (figure 4-14). In the 400
80th percentile, the error values are within 3-

300
standard deviation (3- sigma) interval. According
to table 4-7, ANUDEM interpolation method has 200
produced the most accurate DTM with the

maximum error 0.98 m. The DTM produced by 100
the Ordinary Kriging with 5 m block average has

0
the second lowest error with maximum error 1.00 0.0000 5.0000 10.0000 15.0000 20.0000 25.0000 30.0000
Elevation Residual
m.
Figure 4-14: Histogram of the error dataset
produced by the TIN-based terrain
modelling
72
PHILIPPINES
The third is the Ordinary Kriging (100 m contour lines to point conversion interval) with 1.17 m
maximum error. Errors in the polynomial interpolation methods seem confined between 1.60 m and
1.70 m, and errors generated by the higher degree polynomial interpolation methods are not
significantly different with the lower degree polynomial interpolation methods. The errors produced
by this interpolation method are not spatially dependence (appendix D).
2) Spatial error distribution
The second DTM quality assessment focuses on the spatial distribution of the error datasets.
In this study, the flood modelling will be focused at a particular area of interest as illustrated in figure
4-15. Thus errors in this area should be as small as possible. This area contains large terrain changes
due to new developments. Furthermore, according to section 3.1.1, for flood modelling purpose, the
accuracy of the relatively flat area is more important than the hilly areas.
a
b
7 pts 10 pts
Area of interest
7
Error less than 1.0 m
Error more than 1.0 m
d
19 pts 21 pts
Figure 4-15: Errors distribution for TIN (a), Ordinary Kriging with 5 m block average (b), Ordinary
Kriging (100 m line to point conversion) (c) and ANUDEM (d); TIN has the least error within the
study area, followed by block ordinary kriging with 5 m, Ordinary Krigging for 50 m line to point
conversion interval and ANUDEM
The error distribution for the polynomial interpolation methods are shown in appendix B
73
PHILIPPINES
Figure 4-15 shows the distribution of errors with more than one meter (large circle) and less
than one meter (small circle). The DTM produced through the TIN-based terrain modelling has the
least errors less than 1.0 m inside the area of interest with only 7 points. This is followed by the
Ordinary Kriging with 5 m block average (10 points), Ordinary Krigging for 50 m line to point
conversion interval distance dataset (19 points),
and ANUDEM (21 points). Finally, the
polynomial interpolation methods with more than
39 points (Appendix B). Most of the errors with
more than 1.00 m are located at the meandering
river course, North-east side of the study area.
Meandering river course, commonly, exhibits
a
complex terrain with large changes in elevation in
a short distance. This area needs dense elevation
data to represent the real terrain changes or the
landscape. On the other hand, errors with more
than 1.00 m also appear in the South east part of
the Naga City. This area is quite high with more
complex landforms. Through this assessment, it is b
clearly shown that the complex terrain landform
required denser elevation points to accurately
represent the inhomogeneous terrain. However,
Kriging interpolation method is sensitive to
inhomogeneous surface. The interpolation method
tends to smoother interpolated surface. Most of
the interpolation methods performed well in c
relatively flat area.
3) Visual assessment
However, the DTM accuracy assessment

and the distribution of errors only consider local d
or individual accuracy of each single estimated
elevation point. In general, terrain surface consists
of neighbouring terrain elevations which are
Barangay Triangulo
together used in defining the landscape.
Barangay Mabulo
e
Figure 4-16: Natural terrain derived from 6th

Polynomial degree (a), TIN-based terrain
modelling (b), ANUDEM (c), Ordinary Kriging
(100 m conversion interval) (d), and Ordinary
Kriging (5 m average block) (e), visualized in
3D
74
PHILIPPINES
Therefore, the visual assessment of the DTM is crucial to assess visually, on the ability of
the DTM in representing the real landscape of the study area based on expert knowledge on the
geomorphology and experience in the fieldwork. Each DTM is visualized in 3-dimensional form,
thus helps to reveal the general landscape of the study area. The DTMs should represent the general
landscape of the flood plain area for instance, natural levee, back-swamps and etc. Figure 4-16 show
the ground terrain models of the study area in 3D perspective. Generally, the DTM produced by,
Ordinary Kriging, TIN and ANUDEM have shown the general landscape of the study area, for
instance, the natural levee, high elevation area in the upper-stream of the Naga River, and back-
swamp area. The DTMs also reveal the areas, which were elevated for bridge construction. At a
glance, Barangay Triangulo and Barangay Mabolo can be easily identified as among the lowest area
in Naga City (figure 4-16 (e)).
The depression area in Barangay Triangulo can be classified as a back-swamp area, and
protected by higher natural levee along Naga River. However the DTMs produced by the Polynomial
interpolation method, for instance the 6 th polynomial degree is very smooth and able to show only the
general landscape of the study area. Thus, the interpolated surface has lost several landscape features
of the study area. On the other hand, the DTMs produced by TIN and ANUDEM contain quite high
number of small depressions which are called “pits”. TIN and ANUDEM have successfully revealed
the drainage pattern in the higher part of the Naga City, which claims that these interpolation
methods would be the proper methods be used for rugged terrain modelling. In section 3.1.1, it was
discussed that one of the most problematic features in the DTM for any hydrological application that
is an artificial pits. The DTM produced by the Ordinary Kriging shows quite smooth terrain surface
while preserving the main landscape of the study area.
Table 4-8: The summary of the DTMs quality assessment

Number of errors
Interpolation method Accuracy assessment with more than 1.0 m Visual assessment
in the area of interest
Able to represent the real
Ordinary Kriging (5 m
1.00 m (at 80 percentile) 10 landscape of the study
elevation block average)
area
Ordinary Kriging (100 m Able to represent the
contour lines to point 1.17 m (at 80 percentile) 19 real landscape of the
conversion interval) study area
Contains large number
TIN N/A 7
of artificial pits
Contains large number
ANUDEM 0.98 m (at 80 percentile) 21
of artificial pits
Smooth interpolated
Polynomial interpolation
1.66 m to 1.74 m (at 80 surface and with very
methods (2nd Polynomial > 39
percentile) general landscape of the
to 6th Polynomial )
study area
Finally, it can be concluded that, the DTM produced by the Ordinary Kriging with 5 m
block average has a good balance result between error values at 80 percent of errors, error
distribution and
75
PHILIPPINES
visual assessment on general landscape of the study area. This DTM is used to construct the DSM of
the study area.
Man-made terrain modeling
Current situation of Naga City
The DSM will be constructed by adding the man-made terrain features onto the ground terrain
model. The present situation of Naga City includes:
1. Developments in CBDII
2. Almeda Highway.
3. Buildings
1) Roads
The contour lines of the river have been excluded during the
construction of the road terrain to prevent underestimation of road
elevation for the areas near to the river. The road elevation data were
derived through the levelling process in Naga City. However the
measurement didn’t cover the whole area of Naga City. In this case,
the road elevation for the areas without the levelling data, are
assumed have the same level with the natural terrain (figure 4-17).
The width of the roads in Naga City varies from 2 m to 12 m.
Tennakoon (2004) has identified that the road width need to be
increased at least up to √2 (pixel resolution) to ensure water can flow
continuously along the road during flood.
Figure 4-17: Road elevation

compared to the residential
area in the roadside
2) Central Business District II (CBD II) development
The construction of the terrain in this area is separated into

roads, developed areas and swampy areas. Detailed terrain models
for the roads was constructed through the elevation points and
detailed CBD II development plan. On the other hand, the terrain
model for the swampy areas was derived from the ground terrain of
the whole Naga City.
Figure 4-18: Examples

on Buildings in CBD
II
76
PHILIPPINES
There are 4 main buildings in CBD II; 1) integrated bus terminal, 2) satellite or district market,
3) Jeepneey station and 4) coliseum (figure 4-18). This area was formerly covered by swampy and
rice field areas, and acted as a temporary water retention area during heavy rainfall. The elevation of
this area was between 1.0 m and 1.5 m above mean sea level. The new developments have elevated
the original ground terrain roughly between 2.05 m and 3.24 m. In future, the empty spaces, which
are currently covered by the swampy areas, will be elevated to accommodate shops and other
commercial buildings. This area will be elevated at least to the road level.
3) Almeda highway
The Almeda highway terrain model is constructed based on

its detailed construction plan. The elevation data were mapped for
every 20 meters distance apart. According to the plan, the natural
ground has been elevated up to 2.3 meter in average for
approximately 11 m road width including road buffer sides (figure
4-19).
Figure 4-19: Almeda highway

connecting Naga City and
Almeda
4) Buildings
Certain parts of the residential areas in Barangay Sabang do not exist anymore, and the
residents have been relocated to the new residential areas. Furthermore the existing building
footprints are believed has not been updated for a long time. There are buildings that do not exist
anymore and new commercial buildings that have not been mapped in the existing building
footprints. The updating process was done through integrating development plans and field survey
done by Saut Sagala.
Future developments
The future situation in Naga City will consider:

1. The new mall located next to Almeda highway,
2. Micro drainage in Barangay Triangulo and
3. The elevated future commercial areas at CBD II.
77
PHILIPPINES
1) Mall
The new mall as proposed by the LANDCO Pacific Corp., will be located next to the Almeda
Highway (figure 4-20). According to officer in Naga City Mayor
Office, the detailed plan has not yet been submitted by the
developer. Thus, it is assumed that this area will be elevated to the
same level of the Almeda Highway. This area will be elevated to
5.07 m above mean sea level (the average elevation of the Almeda
Higway, in front of the proposed area).
Figure 4-20: New mall which
is proposed to be built next to
the Almeda highway
2) Micro Drainage System at Barangay Lerma and Barangay Triangulo
A New drainage is being constructed in Barangay Triangulo to overcome the flood problem in
the urban poor residential area that located next to the CBD II. The project is called the Micro
Drainage System in Barangays Lerma and Triangulo, Naga City,
implemented by the Community Organization Enterprise Foundation
Inc. (COPE). However the residential area should be elevated to at
least the same level with the drainage. The drainage couldn’t be built
much lower due to low earth gradient in this area. The drainage
should maintain the slope to ensure that the floodwater can be
transported to a natural creek in the Bicol River. The dimension of
the drainage increases towards the creek, and the type of drainage
changes from the open drainage to the close drainage. Since now,
there is no clear plan on which part of the residential area will be
elevated. Unfortunately, for the time being, the drainage behaves like
a barrier during flooding and causes long period of inundation (figure
4-21). The floodwater should rise up to about one meter before it can
be transferred to the main drainage.
Figure 4-21: Micro drainage
project at Barangay Triangulo
3) The elevated commercial areas at CBD II
According to the development plan of the CBD II, this area has been divided into several
lots. This lot will accommodate commercial buildings in future. However, detailed information on
the buildings could not be found. In the present situation, the slots are covered by swamp. This area
will be elevated to the same level of the roads in the final DSM.
78
PHILIPPINES
Terrain model integration
The final DSM for the present and the future situations of Naga City are illustrated in figure 4-
22 and 4-23 respectively. Both DSMs are used to assess the impact of the new developments in Naga
City on flood behaviour. However, the final DSMs have been modified many times, especially for
areas which elevations were underestimated by the Ordinary Kriging interpolation. Most of the areas
are well recognized, for instance LCC Mall, Metro Ville (figure 4-22) and so forth, and the elevations
were increased to the road level. The new DSM contains about 437,932.5 m3 new elevated terrain.
Jeepney Station
Satellite market
LCC Mall Metro Ville Bus Station

Coliseum
Almeda highway
18.41 m
-4.50 m
Figure 4-22: DSM of present situation of Naga City
COPE drainag Elevated commercial

e
area
Mall
18.41 m
-4.50 m
Figure 4-23: DSM of future situation of Naga City and the elevated areas (right window)
79
PHILIPPINES
5 Flood modelling
Introduction
The SOBEK 1D 2D model is used for the flood modelling of the study area. The SOBEK 1D
2D model combines one dimensional channel flow with overland flow, which is represented by a 2D-
grid of elevation information. The overall methodology of the modelling is indicated in figure 5-1.
Figure 5-1: Overall methodology for flood modelling
80
PHILIPPINES
The purely 2D flood model requires fine resolution of DTM in representing the physical
characteristics of the river. Thus in this study, the 5 m DTM resolution is not good enough to
represent very narrow river course. Therefore the river flow will be separated from the 2D floodplain
model. The river flow will be simulated using the 1D model. The boundary conditions were derived
from the flood modelling over the lower part of the Bicol region, based on the previous research by
Bin Usamah (2005). This data is then used for the flood modelling and the flood model calibration.
The flood depths of the super typhoon Nanmadol (collected by Peters Guarin Graciela and Saut
Sagala) are used to calibrate the flood model. In This study, the flood model calibration is done step
by step. In the flood calibration the predicted flood depths are influenced by the building structures
and surface roughness. Thus, 2 sets of building structures and 3 sets of Manning coefficient are used
in the flood model calibration. The output will be the suitable building structures and Manning’s
coefficient to be used in the final flood modelling. The flood model calibration is based on the 10
years return period flood, since the super typhoon Nanmadol is defined approximately similar to this
flood event. In this study, the flood modelling only focuses on the flooding attributed by river
overflows of both the Naga River and the Bicol River. The overland flow will be represented by the
high spatial resolution terrain model of Naga City (5 m). The flood modelling focuses on 2, 5, 10 and
17.5 years return period floods.
Boundary condition
Two water discharge boundaries for a specific return period flood are located at the Naga
River and the Bicol River upstream. The Bicol River, at Naga City is affected by tidal wave that can
elevate normal water level up to 3-meter for long term recurrence event. Therefore, the observed
correlation between water level and discharge are not consistent because of the backflow effect
(figure 5-5). According to the previous literature review in section 3.4.2.1, series of water level is the
best data that can be used to represent this inter-tidal effect (figure 5-4). The discharge information
for both the Naga River (figure 5-3) and the Bicol River (figure 5-2) upstream are derived from a
study by the NIPPON KOEI Consultant that modelled the whole Bicol River basin using MIKE 11
(Bin Usamah, 2005) for the return periods of 2, 5, 10 and 25 years. Discharge values for the 2 years
return period is derived from interpolated and extrapolated values of Nippon KOEI data. Series of
water level for the Bicol River downstream boundary is derived from Bin Usamah (2005).
81
PHILIPPINES
Hydrograph for Bicol River Upstream
600
500
400
2 Years
5 Years
10 Years
Discharge
300
17.5 Years
200
100
0
12.00
37.00 62.00 87.00 112.00

Time (hour)
Figure 5-2: Boundary condition for the Bicol River Upstream
Hydrograph for Naga River Upstream
300
250
200
2 Years
5 Years
Discharge
150 10 Years
17.5 Years
100
50
0
12.00 37.00 62.00 87.00 112.00
Time (hour)
Figure 5-3: Boundary condition for the Naga River Upstream
Water Level for Bicol River Downstream
2.500
2.000
1.500
1.000
0.500
0.000 2 Years
Water level
-0.500 5 Years
-1.000 10 Years
-1.500 17.5 Years
0
8
1
6
2
4
3
2
4
0
4
8
5
6
6
4
7
2
8
0
-2.000
-2.500
Time (hour)
Figure 5-4: Boundary condition for the Bicol River Downstream
82
Discharge
Discharge and Water Level Relationship (QH) for 2 Years Return Period Flood and Water Level Relationship (QH) for 5 Years Return Period Flood
2.000
PHILIPPINES
-20. .000 -50. .000
2.000 Discharge (m3) Discharge (m3)

1.500 1.500
1.000 1.000
0.500 0.500Relationship (QH) for 17.5 Years Return Period Flood
Water Level
Water Level
Discharge and Water Level Relationship (QH) for 10 YearsDischarge
Return Period Flood
and Water Level
0.000 0.000
0-00.500 00 20.000 40.000 60.000 80.000 100.000 120 000 -0.500.000 50.000 100.000 150
.0 2.50
2.000 -1.000
-1.000
-1.500 -1.500
-2.000 -2.000
-2.500 -2.500
Water Level
Water Level
-15 .000
-40 .000
Discharge (m3) Discharge (m3)

1.500 2.00
1.000 1.50
0.500 1.00
Figure 5-5: Discharge and Water Level relationship (QH)
0.000 0.50
0.00
0.00 -100.00 -50.0-000.500.0 00 50.000 100.000 150.000 200
-100.0-0.50.0
0 0 -1.000 00 100.000 200.000 300
0 0 0 0 -1.00
-1.500
-1.50
Surface roughness -2.000
-2.500
-2.00
-2.50
Manning coefficients are derived from the transformation of the Naga landuse or landcover
information into the corresponding Manning’s values derived from the literature. Land use data is
divided into 2 categories; 1) present and 2) the future situation of Naga City (figure 5-6).
Table 5-1: Manning’s coefficient (After Tennakoon (2004)); This coefficient will be used in model
calibration
Manning’s value
Land use
Initial run 2nd run 3rd run
Agricultural (paddy
0.025 0.03 0.035
field)
Cemetery 0.025 0.03 0.035
Commercial 0.032 0.04 0.045
Dumpsite 0.025 0.03 0.035
Industrial 0.032 0.04 0.045
Institutional 0.032 0.04 0.045
Open area 0.025 0.03 0.035
Parks 0.025 0.03 0.035
83
PHILIPPINES
Residential 0.032 0.04 0.045

Swampy area 0.025 0.03 0.035
Roads and railway 0.025 0.03 0.035
Buildings 1.0 1.0 1.0
Three sets of Manning’s coefficient are used for the flood model calibration. The water
depth, velocity and flood extent are compared in the model sensitivity analysis.
a b
Figure 5-6: Landuse of Naga City in future (a), and in 2005 (b)
Building structure
Building structures can be represented as solid blocks, partially solid structure which allows
water intrusion inside the buildings during flooding or as a rough surface. In this study, building
structure is treated as both solid block and rough surface (figure 5-7). The results produced by both
approaches (flood depth, velocity and flood extent) are evaluated and compared in order to determine
the best representation of building structures for further modelling.
a b
Figure 5-7: Building structure as solid block (a), and rough surface (b)
84
PHILIPPINES
Model schematization
The network editor in SOBEK is called NETTER, which allows schematizing flood model
components on the top of the GIS layer as a reference (figure 5-8). Two edit modes offered by
NETTER, namely schematization and attribute edit modes. The components in the flood model (i.e.
river cross-section, calculation points, boundary nodes, 2D grid, connection nodes and etc) are
defined in the schematization mode. While their attributes, can be defined through the attribute
editing mode.
2D grid
Naga River upstream
Connection boundary (discharge)
point
1D river cross
section
Downstream Boundary
(Water level)
History station
Bicol River upstream boundary
(discharge)
Line Line Calculation point

boundary measurement
Figure 5-8: Model Schematization in SOBEK 1D2D
The schematization starts with 2D network, which represents the floodplain terrain. The terrain
information can be directly imported through ASCII data format. Every single terrain value in the
imported DTM, can be checked and manually changed using the grid cell editor. The model
computation time is largely influenced by the DTM resolution, in which the high resolution DTM
will require longer processing time than coarser DTM resolution. Thus, the selection on the suitable
DTM resolution is based on the application and level of detail needed in flood investigation and
analyses. Surface roughness is important in defining flow velocity, direction and flood extent. The
surface roughness value can be defined by a single value for the whole floodplain or one value for
each pixel. This function is very useful to change and modify the DTM and surface roughness value,
especially in
85
PHILIPPINES
model calibration and flood simulation for different scenarios. In this study, the flood modelling is
done through the combination of 1D and 2D flood models. In this case, the riverbed elevation is
excluded from the final DTM. The 2D schematization is further continued by placing the boundary
nodes at the edge of the model. Furthermore for selective and quick assessment on the model
performance, history boundaries are placed at the selected areas. Series of water depth, level and
water velocity can be produced by the history station for a specific pixel.
The 1D flow module is a representation of water

system which is flowing along the river or channel. The
characteristic of the flow is determined by the
characteristics of the channel (i.e. cross section, bed
level, surface level, roughness and etc.). The river
course is defined by reaches that connected together by
connection nodes. Straight line of river networks is
produced and further editing using “edit reach vector” is
needed to reshape the river course. The discharges
through the river are influenced by the shape of the
riverbed.
Figure 5-9: River cross-section input

window
Constant interpolation
Linear interpolation
Figure 5-10: The original cross section data (a), Constant interpolation (b) and Linear interpolation (c)
86
PHILIPPINES
Thus, proper definition of the river cross-section and the amount of cross section nodes in the
model are crucial to accurately simulate the river flow. The definition of the river cross section is
made using the cross section nodes (figure 5-9). In this study, the cross section nodes are added along
the river course at 150 m distance interval. In SOBEK, the bed levels and hydraulic radius are
interpolated for each calculation point based on the cross-section data supplied by user. The
interpolation will be made either through constant interpolation or linear interpolation (figure 5-10).
The same interpolation principle is applied for the ground layer, the surface level and the cross
sectional of the river. The bed friction is a component that exerts a force on the flowing water, which
always in the opposite direction of the water flow. The friction in 1D channel flow is defined as the
interaction between flowing water with channel bed. The main factors that determine the flow
condition is the friction force and earth gravity, while other factors are less important.
Model calibration
The principle behind the flood calibration is certain flood model parameter need to be tuned to
reduce the deviation between the results of the flood modelling and the corresponding observed flood
data, for instance flood depth and flood extent. In section 3.4.2.3, it was found that the calibrated
parameters from one flood event might no be appropriate to be used especially for the larger flood
event. However, in this case, there is only one reliable flood data derived from the fieldwork, which
is the flood depth. The information on the flood extent is only available in the Metro Ville residential
area. Thus the flood calibration will only rely on this observed data. Inundation depth and flood
extent are the most common parameters used in flood model calibration. Flood depth is a local factor,
thus the calibration need to focus on the flood model parameters that have an impact on flood depth.
According to the previous literature in section 3.4.2.3, building structure and surface roughness have
quite pronounced impact on the flood depth.
In this study, proper boundary condition for the suprtyphoon Nanmadol can’t be constructed
due to lack of data. The 10 years return period flood boundary conditions produced by the NIPPON
KOEI Consultant is used in the model calibration. The model calibration process consists of two
main steps. For each step, the results of the flood modelling including water depth and flood extent
are compared to the water depth (figure 5-11) data and flood extent information derived during the
fieldwork. The first step focuses on the representation of building structure as a solid block and rough
surface. The second step emphasizes on the different set of Manning coefficients. The results derived
from several simulations will be evaluated through 3 main criteria; 1) overestimation or
underestimation of the flood depth, 2) overall absolute accuracy of water depth and 3) flood extent.
87
PHILIPPINES
Igualdad Interior Triangulo
Water depth (m)

Saba
Figure 5-11: The distribution of the flood depth samples (Saut Sagala and Peters Guarin Graciela)
The flood depths of the super typhoon Nanmadol are used to assess the accuracy of flood depth
prediction. There are 249 points of flood depths collected through interviews in Barangay Triangulo,
Barangay Sabang and Barangay Igualdad Interior (the interviews were conducted by Saut Sagala and
Peters Guarin Graciela). The distribution of the samples is depicted in figure 5-11. According to this
figure, the residential areas in Barangay Triangulo and Barangay Sabang were severely inundated
during the super typhoon. Nevertheless, the swampy areas in Barangay Triangulo were also affected
by the flood.
A common rule of thumb for normality test based on skewness and kurtosis is to divide the
skewness and kurtosis value by their standard errors. The normally distributed data should have
skewness and kurtosis values within +2 to -2. In this case, both error datasets are within this
boundary (table 5-2).
Table 5-2: Skew and kurtosis value for error datasets of flood model with rough surface and Solid Block
building structures
Rough surface Solid block
Skewness 0.24 0.164
Standard error of skew 0.154 0.154
Kurtosis -0.242 -0.268
Standard error of kurtosis 0.307 0.307
Thus, the flood depth predictions derived from both situations (rough surface and solid
building structures) neither underestimate nor overestimate the flood depth of the super typhoon
(figure 5-12). The absolute water depth prediction accuracy for both building structures is 0.35 m.
There are only 4 percent of the samples with errors between 1.0 m to 1.2 m. These errors might be
introduced by the local terrain that can’t be represented in the DSM, for instance the elevated ground
terrain to alleviate the flood problem by individual. Beside that, this model also doesn’t take into
account the rainfall run-off information, thus ignore the flood water produced by the direct rainfall.
However, since the information on water depth was derived through interview, it may contain biases.
88
PHILIPPINES
Histogram of Maximum Water Depth Error Histogram of Maximum Water Depth

(Building Structure as Rough Surface) Error (Building Structure as Solid
Block)
60 60
a b
50 50
40 40
Freque
30 30
20
20
10
10
0
0
-1.000 -0.500 0.000 0.500 1.000
-0.900 -0.600 -0.300 0.000 0.300 0.600 0.900 1.200
Error
Error
Figure 5-12: The histogram of maximum water depth prediction error for flood modelling with rough
surface (a), and solid block (b) as a building structure
The next assessment is based on the flood extent of both building structures. The flood extents
are compared to the known flood extent derived during the fieldwork. In general, the flood extent of
solid structure building is less compared to the rough surface building structure (figure 5-13). The
Metro Ville is a new residential area, which was elevated to accommodate the relocated residents
from the flood prone areas. During the recent flood, this area was not flooded.
Neuva Caceres University

Residential area in
Barangay Tinago
Metro-Ville (solid block)
Metro-Ville (rough surface)
Higher maximum water depth

(solid block building) (0.83 m
difference)

(rough surface) (1.27 m difference)
Additional flooded area for building

structure treated as a rough surface
Figure 5-13: Difference in maximum flood water depth and flood extent for 10 years flood modelling
treated building structure as solid block and rough surface
89
PHILIPPINES
In this case water velocity measurements were observed at University of Neuva Caceres (open
area) and residential area at Barangay Tinago (figure 5-13). It was found that, the water velocity
derived from the rough surface building structure is higher compared to the solid block (figure 5-14).
In this case, the momentum of the overflowed water decreases when the water-flow is obstructed by
solid building block, and decreases the ability of water to go further. On the other hand, the rough
surface building structures allows water to flow much further and increase the flood extent (figure 5-
13). The obstruction has increased the flood depth in the area as depicted in figure 5-13. In addition,
there is only slight difference in water flow pattern between the two building structures (figure 5-15).
Therefore, it was proved that by treating building structure as a rough surface, it overestimates the
flood extent, the flood velocity and underestimates the flood depth. However, this concept is reliable
for this study area, since different area has different environment setting. Buildings in Naga City is
quite dense for both residential and commercial areas, thus this factor will influence the flood
behaviour. Therefore it is decided to treat building structure as a solid block structure rather than
rough surface.
Water Velocity (m/s) Vs Time (hour) for building structure as solid block and Rough surface (UNC)
Water Velocity (m/s) Vs Time (hour) for building structure as solid block and Rough surface (Residentia
0.04
0.035
0.16
0.03
0.14
0.025
0.12
0.02
0.1
0.015
Rough surface 0.08 Rough surface
Velocity
Velocity
0.01
Solid block 0.06 Solid block
0.005
0.04
0
0.02
a 0
Time (Hour)
b
Time (Hour)
Figure 5-14: The comparison of water velocity for different representation of building structure; solid
block and rough surface at University of Neuva Caceres (open area) (a) and Residential area in Barangay
Tinago(b)
m/s a
m/s b
m/s m/s
Figure 5-15: The comparison of water velocity for different representation of building structure; rough
surface (a) and solid block (b) after 56th hour simulation (the small right windows show the inset of the
study area)
90
PHILIPPINES
Further analysis is carried on the remaining 2 sets Manning coefficient with 10 years return
period flood. The modelling revealed that the absolute error of the maximum water depth prediction
is
0.35 m, similar to the accuracy derived from the initial Manning coefficient (figure 5-17). According
to the z-test at 95% confidence interval, the error dataset of all Manning coefficient datasets are not
significantly difference and thus conclude that there is no significant difference in maximum
inundation depth of using those 3 Manning datasets. As a result, the initially used Manning
coefficient value is used for further flood modelling.
Table 5-3: Z- test between Manning coefficient datasets prediction error; the result are not significant
(less than 1.96)
Manning set 1 Manning set 2 Manning set 3

Manning set 1 N/A 0.06 0.06
Manning set 2 N/A 0.06
Manning set 3 N/A
Histogram of Maximum Water Depth Error

(Manning Coefficient Set 2) Histogram of Maximum Water Depth Error
(Manning Coefficient Set 3)
70
a 70
b
60 60
50 50
Freque
Freque
40 40
30 30
20 20
10 10
0
-1.000 -0.500 0.000 0.500 1.000 0
-1.000 -0.500 0.000 0.500 1.000
Error
Error
Figure 5-16: The histogram of maximum water depth prediction error for flood modelling with Manning
coefficient set 1 (a), and Manning coefficient set 2 (b)
Through the flood calibration process, it was found out that the solid block buildings and the
initial Manning coefficient are suitable for further flood modelling (table 5-4).
Table 5-4: Overall results of the flood model calibration
Building structure Manning Overestimated or Prediction Flood extent

coefficient underestimated accuracy (water (Metro Ville area)
depth)
Rough surface Initial set No 0.35 m Inundated
Solid block Initial set No 0.35 m Not inundated
Solid block Set 1 No 0.35 m Not inundated
Solid block Set 2 No 0.35 m Not inundated
91
PHILIPPINES
6 Development impact assessment

The development impact assessments are based on three main aspects, namely, 1) detailed
investigation on changes in flood behaviour, 2) community perception on flood risk, and finally 3)
investigation on flood hazard changes based on flood depth and flood velocity (figure 6-1). All these
analyses will be used to show the consequences of the developments in various aspects. The
assessment are made for the flood event of 2 different situations; present situation and after
development situation in Naga City. Figure 6-1 shows the general work flow in this chapter. The
assessment will be made for 5, 10 and 17.5 years return period flood, since the 2 years return period
flood event is rather small and no changes in flood behaviour observed in both situations.
Figure 6-1: General work flow of development impact assessments
Investigation on the changes in flood behaviour for the situation before and after the
developments will focus on flood depth, flood velocity, inundation duration, warning time and flood
extent. This information will be used as input for the next assessment which focuses on the changes
of flood risk perception in the residential areas. Finally the assessment ends with the investigation on
changes in flood hazard area. The flood hazard definition based on drowning hazard is adopted from
Ramsbottom et al.(2003). These assessments do not oppose any particular development in Naga City
and it rather just aims to show ideas to assess the impact of development in flood prone area. For the
purpose of discussion, figure 6-2 will be used as a reference for the specific name of the Barangays in
the study area.
92
PHILIPPINES
DAYANGDANG
CONCEPCION PEQUENA
TINAGO SAN FRANCISCO

STA. CRUZ
LERMA
DINAGA CBD II
ABELLA
IGUALDAD
TRIANGULO
SABANG
TABUCO
MABULO
Figure 6-2: Barangays located inside the study areas
93
PHILIPPINES
Detailed changes in flood characteristics
The investigation on the detailed flood behaviour changes is a starting point for any further
impact assessments. The assessment will emphasize on the changes in flood velocity, flood depth,
warning time and inundation period. Warning time in this study is defined as the time when the water
first comes to each pixel representing the floodplain. Simple change detection method is used to
show the changes of flood behaviour for present and the after development situations. The change
map will be produced to show the amount of changes including the areas involved. Figure 6-3 shows
the general work flow of this assessment. The results of the flood modelling (5, 10 and 17.5 years
return period flood) for present situation are shown in appendix G to I.
characteristics
The flood modelling for 17.5 years return period flood was successfully made only for 60
hours. The modelling took 6 days to complete. Nevertheless the changes in flood behaviour can still
be detected as illustrated in figure 6-4. According to figure 6-4 (a), changes in maximum water depth
are observed in Barangays Triangulo, Tabuco and Mabulo. The highest changes in maximum flood
depth are located in the area behind the Almeda Highway and the elevated commercial area in
Barangay Triangulo. On the other hand, changes in maximum water velocity are also observed in the
same areas including Barangay Sabang (figure 6-4 (b)). After 60 hour flood simulation, some parts of
the elevated commercial areas in Barangay Triangulo still not be inundated, however at the same
time the flood water had already overflowed to the areas behind the Almeda Highway. However,
the full
94
PHILIPPINES
flood modelling will be 120 hours, thus, more pronounced changes are expected if the modelling
could be finished successfully.
Higher maximum water Higher maximum water

depth after development (+ velocity after development (+
1.55 m) 1.67 m/s)
Lower maximum water Lower maximum water

a b velocity after development (-
depth after development (-
1.51 m) 1.65 m/s)
Shorter warning time

after development (+ 7
hours)
c Longer warning time

after development (- 8
hours)
Dry area after development

Newly inundated area after development
Negative change
Positive change
Figure 6-4Changes in maximum water depth (a), changes in maximum water velocity (b) changes in
warning time (c) and changes in inundation period for 17.5 years return period flood
In general the changes of flood behaviour in the 10 years return period flood is less
compared to the 17.5 years return period flood. The impact of the development seems more localized
to the vicinity areas. Changes in maximum flood depth are observed mostly in the Barangay
Triangulo and some part of the Panganiban Drive area (figure 6-5 (a)). The maximum different in
maximum flood depth is 0.75 m (maximum) and it is located in the area along the Panganiban Drive.
On the other hand, the maximum water depth has decreased in the elevated commercial areas in the
95
PHILIPPINES
CBD II. According to figure 6-5 (b), changes in maximum flood velocity were also located mostly in
Barangay Triangulo. Pronounced different in maximum flood velocity can be observed at the edge
part of the
96
PHILIPPINES
elevated area in CBD II and in the residential areas in Barangay Triangulo. Areas with shorter
warning time are observed in Barangays Triangulo, Sta. Cruz and Sabang (figure 6-5 (c)). While
some parts along the Panganiban Drive, Barangays Sabang and Sta. Cruz suffers long inundation
period as illustrated in figure 6-5 (c). Most of the elevated commercial areas in the CBD II were not
inundated, and it seems that the flood problem in this area was transferred to another area.
Panganiban Drive

after development (+ 0.75 m) Higher maximum water velocity
after development (+ 1.25 m/s)
a Lower maximum water depth

Lower maximum water velocity
after development (- 1.16 m) b after development (- 1.15 m/s)
Shorter warning time after development Longer inundation duration after

(+ 22 hours) development (+ 68 hours)
c Longer warning time after

d Shorter inundation duration after
development (- 20 hours) development (- 68 hours)
Newly inundated area after Dry area after development
development
Figure 6-5: Changes in maximum water depth (a), changes in maximum water velocity (b) changes in
warning time (c) and changes in inundation period for 10 years return period flood
97
PHILIPPINES
The next observation is for the 5 years return period flood. In general the changes of the
flood behaviour are more localized compared to 17.5 and 10 years return period flood (figure 6-6).
The changes are mostly occurred in the residential area in Barangay Triangulo. In this flood event,
the impact of the small structure, for instance drainage system in Barangay Triangulo is quite
pronounced, especially in the lower part of the residential area. The construction of the drainage
system had eased the flood problem in this area.
Higher maximum water depth Higher maximum water velocity

after development (+ 0.58 m) after development (+ 0.39 m/s)
a Lower maximum water depth Lower maximum water velocity

b
after development (- 0.83 m) after development (- 0.59 m/s)
Shorter warning time after development

(+ 11 hours) Longer inundation duration after
development (+ 60 hours)
c Longer warning time after
development (- 14 hours) d Shorter inundation duration after
Dry area after development development (- 62 hours)
Newly inundated area after
development
Figure 6-6: Changes in maximum water depth (a), Changes in maximum water velocity (b) Changes in
warning time (c) and changes in inundation period for 5 years return period flood
98
PHILIPPINES
Maximum Water Velocity Difference (17.5 years return period flood

Maximum Water Depth Difference (17.5 years return period flood)
182
195
132
145
82
95
Area
Area
32
45
-18
-5
50
25
00
75
50
65
39
13
88
62
36
0.
0.
1.
1.
1.
0.
1.
0.
0.
0.
1.
1.
Water Velocity Difference (m/s)
-
-
Water Depth Difference (m)
Warning time Difference (17.5 years return period flood)
7.4
6.4
5.4
4.4
3.4
2.4
Area
1.4
0.4
-0.6
-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7
Warning Time Difference (hour)
Maximum Water Depth Difference (10 years return

Maximum
period flood)
Water Velocity Difference (10 years return period flood)
85 240
75 190
65 140
55 90
45 40
35 -10
25
Area
Area
15
5
-5 Water Velocity Difference (m/s)
Water Depth Difference (m)
Warning time Difference (10 years return period flood) Inundation Period Difference (10 years return period flood)
16.5 22
14.5
12.5
10.5 17
8.5
6.5
4.5 12
2.5
Area
Area
0.5
-1.5 7
2
6
8
5
8
4
8
3
7
2
6
1
6
6
0
8
2
0
1
6
1
0
1
2
-
-
-3
Inundation Period Difference (hour)
99
PHILIPPINES
Maximum Water Depth Difference (5 years returnMaximum

period flood)
Water Velocity Difference (5 years return period flood)
46 606
36 506
26 406
16 306
6 206
-4 106
6
Area
-94 -0.49 -0.36 -0.23 -0.10 0.03 0.16 0.29 0.42
Area
Water Depth Difference (m) Water Velocity Difference (m/s)
Warning Time Difference (5 years return period flood)

Inundation Period Difference (5 years return period flood)
1.3 4.755
1.1
0.9
0.7 3.755
0.5
0.3 2.755
0.1
Area
-0.1
Area
1.755
0.755
-0.245 6
7
5
9
5
3
4
6
4
0
3
4
-
-
Inundation Period Difference (hour)
Figure 6-7: Histogram of changes in flood characteristics
10
PHILIPPINES
Community-based flood risk perception
The community flood risk perceptions were only derived for the residential areas in
Barangays Mabulo and Triangulo through the interview sessions conducted by the ongoing PhD
research by Peters Guarin Graciela. (table 6-1 and 6-2). Therefore, another flood risk perception
should be generated in order to represent the flood risk perception in another residential area.
Figure 6-8: Flow chart for development impact assessment based on community-based flood hazard
perception
10
PHILIPPINES
In section 3.6, it was found that more than 50 percent of the residents in Barangays Triangulo
and Mabulo perceived flooding as a nuisance rather than a catastrophe. Furthermore, they had chosen
to live there for the sake of economic livelihood and education of their children. Barangays Mabulo
and Triangulo are the most severely affected areas by flooding in Naga City. In general, the flood
risk perception of other residential areas, with less experience of severe flood would be quite
sensitive to even a small magnitude of flooding. Thus, the new flood risk perception is created based
on the worst case of the original flood risk perception in Barangays Mabulo and Triangulo (table 6-
3).
Table 6-1: Flood risk perception for the residential area in Barangay Triangulo (ongoing PhD research
by Peters Guarin Graciela)
Flood Risk Perception (Barangay Triangulo)

Depth/ Duration 1 day (24 hrs) 1 to 2 days (24hrs – More than 2 days (more
Ankle (20 cm) Normal Normal Normal
Knee (40 cm) Manageable Disturbing Unmanageable
Waist (80- 100 cm) Manageable Highly disturbing Disastrous
Breast (140 cm) Unmanageable Disastrous Disastrous
Breast + high wind Disastrous Disastrous Disastrous
Table 6-2: Flood risk perception for the residential area in Barangay Mabulo (on going PhD research by
Peters Guarin Graciela)
Flood Risk Perception (Barangay Mabulo)

Depth/ Duration 1 day 1 to 2 days (24hrs – More than 2 days (more than
48 hrs) 48 hrs)
Ankle (20 cm) Normal Normal Disturbing (skin diseases)
Knee (40 cm) Manageable Highly disturbing Disastrous
Waist (80- 100 cm) Unmanageable Disastrous Disastrous
Breast and strong wind Disastrous (< 5 hours) Disastrous Disastrous
Table 6-3: Probable flood risk perception for other residential areas in Naga City
Probable Risk Perception for other residential areas

Depth/ Duration 1 day 1 to 2 days (24hrs – More than 2 days (more
Ankle (20 cm) Normal Normal Disturbing
Knee (40 cm) Manageable Highly disturbing Disastrous
Waist (80- 100 cm) Unmanageable Disastrous Disastrous
Breast and strong wind Disastrous Disastrous Disastrous
This assessment aims at investigating the impact of the developments in Naga City on the
surrounding residential areas. In specific, this assessment will define the amount of changes
quantitatively and also the corresponding areas for different flood recurrence intervals (5, 10 and
17.5
10
PHILIPPINES
years return period flood). The general methodology for this assessment is shown in figure 6-8. Flood
depth information from the flood modelling of both scenarios (present situation and after
development situation of Naga City) are classified into flood depth classes as defined in the flood
risk perception (ankle, knee, waist and chest level). Then, all the classified maps are used to produce
binary maps for each flood depth class. Finally, the flood duration information is derived by adding
all the binary maps for each flood depth class. The flood depth and the flood duration data are then
used as input to derive flood risk perception for both flood events; present and after development
situations. All the classified maps are then combined by selecting the worst case perception for each
pixel. Cross map is the easiest change detection approach for nominal dataset and able to show the
changes from one to another situation in pixel by pixel basis. The changes in flood perception will be
classified into negative and positive impact. The negative impact represents flood risk perception
changes from better flood risk perception to worse perception. On the other hand, the changes from
worse perception to better perception are classified as a positive impact. The map that will be used to
show this information will be called as impact map.
10 years return period flood
Figure 6-9 shows the flood risk perception maps of 10 years return period flood for present
and after development situations. According to this figure, the most affected residential areas are the
residential areas in Barangays Triangulo, Mabulo and also the lower part of Barangay Sabang. These
areas are either located near to the riverbank or in the depression areas for instance, in Barangay
Triangulo. According to the flood modelling results, Barangay Triangulo collects flood water from
the neighbour Barangays. Most of these Barangays are located in the northern part of Naga City
where the river overflows initiated. Instead of that, the inundation in Barangay Triangulo was also
caused by water overflows from the Bicol River.
Disastrous
Unmanageable Disastrous
Highly disturbing Unmanageable
Disturbing Highly disturbing
Manageable Disturbing
a Normal Manageable
b Normal
Figure 6-9: Flood risk perception maps of 10 years return period flood for present situation (a) and after
development situations (b) in Naga City
10
PHILIPPINES
In general, for both flood situations, there are slight changes in all flood risk perception
categories. The highest changes is in Disturbing area, which has increased by 1.07 ha (from 14.27 ha
to 15.34 ha), while the lowest changes is in Unmanageable area which has decreased by 0.01 ha
(from
20.34 ha to 20.33 ha) (figure 6-10). On the other hand, Highly Disturbing, Disturbing, Manageable
and Normal flood hazard perception have increased by 0.29 ha, 1.07 ha, 0.22 ha and 0.67 ha
respectively. Meanwhile, the Disastrous area has declined from 18.26 ha to 17.83 ha.
Area Classified into Community-based Flood Hazard

Perception for Present Situation (10 years return period
Perception for After Development Situation (10 years
flood)
return period flood)
Normal
Normal Disastrous
13.91 ha Disastrous
14.58 ha 17.83 ha
18.26 ha
Manageable
7.24 ha Manageable
7.46 ha
Unmanageable
Disturbing Unmanageable Disturbing 20.33 ha
14.27 ha Highly disturbing 20.34 ha 15.34 ha Highly disturbing
0.21 ha 0.50 ha
a b
Figure 6-10: Areas classified into flood risk perception for present (a) and after development situation
(b) in Naga City (10 years return period flood)
Changes Between 2 Situations (Present Situation and

After Development Situation) for 10 Years Retun Period
Flood
Negative impact
4.20 ha
Unchanged Positive impact
69.43 ha 2.61 ha
Negative
Unchanged
Positive impact
Figure 6-11: Flood impact map for 10 years return period flood
Impact map as illustrated in figure 6-11 shows the distribution of the overall changes in flood
risk perception for 10 years return period flood. This figure shows that the impacts have been
10
PHILIPPINES
transferred from one area to another, though they are located quite a distance from the main
development areas. In this case, there are about 4.20 ha areas were classified into area with the
negative impact, 2.61 ha areas with the positive impact and 69.43 ha areas remain unchanged.
The next analysis focuses on the impact of the developments in 5 years return period flood.
In the 5 years return period flood, most of the flood water come from the Bicol River. In this event,
the residential areas near to the riverbank and the depression area in Barangay Triangulo were
inundated by flood. Therefore, most of these areas were classified as Disastrous. While only a few
parts of the residential areas located in the upper stream of Naga River were inundated and they were
just classified into Normal flood perception. Figure 6-12 shows the flood risk perception maps for
present and after development situations.
Disastrous
Unmanageable Disastrous
Highly disturbing Unmanageable
Disturbing Highly disturbing
Manageable Disturbing
a
Normal Manageable
b
Normal
Figure 6-12: Flood risk perception maps of 5 years return period flood for present (a) and after
development situations (b) in Naga City
According to figure 6-13, the highest changes are observed in Highly Disturbing group
which has decreased by 1.05 ha after the developments (3.19 ha to 2.13 ha). On the other hand, the
Unmanageable group has the lowest changes from 3.99 ha to 4.14 ha. In this flood event, areas
classified into Disastrous and Normal classes have increased by 0.47 ha (7.54 ha to 8.00 ha) and 0.46
ha (3.55 ha to 4.01 ha) respectively. The Disturbing and Manageable classes have decreased by 0.27
ha (1.72 ha to 1.45 ha) and 0.46 ha (1.81 to 1.35) areas respectively. Figure 6-12 shows that most of
the changes have taken place in the residential area in Barangay Triangulo.
10
PHILIPPINES

Perception for Present Situation (5 years return period Area Classified into Community-based Flood Hazard
flood) Perception for After Development Situation (5
years return period flood)
Normal
3.55 ha Normal
Disastrous 4.01 ha
Manageable 7.54 ha Disastrous
1.81 ha Manageable 8.00 ha
1.35 ha
Disturbing
Disturbing
1.72 ha
1.45 ha
Highly Highly
disturbing Unmanageable disturbing Unmanageable
3.19 ha 2.13 ha
3.99 ha 4.14 ha
a b
Figure 6-13: Areas classified into flood risk perception for present (a) and after development situation (b)
in Naga City (5 years return period flood)
Area Changes Between 2 Situations (Present

Situation and After Development Situation) for 5
Years Retun Period Flood
Negative impact
Unchanged
1.42 ha
18.07 ha
Positive impact
2.35 ha
Negative impact
Unchanged
Positive impact
Figure 6-14: Flood impact distribution (negative and positive impact) represents changes in
community flood hazard perception (5 years return period flood)
The flood impact map as depicted in figure 6-14, suggested that the impact is more localized
in the Residential area of Barangay Triangulo. In fact, the impact also more localized compared to
the 10 years return period flood. In the 5 years return period flood, the negative impact accounts
about
1.42 ha residential areas, 2.35 ha received the positve impact, while 18.07 ha areas remain unchanged.
10
PHILIPPINES
Flood hazard
The community-based flood risk perception assessment was based on the flood duration and
flood depth. However, flood hazard can also be expressed by other combination of flood
characteristics, for instance, flood depth and flood velocity based on research by Ramsbottom et al.
(2003). Hazard here specifically refers to the wading hazard of adults, children and also selected
vehicles during flood. In this assessment, the flood hazard maps for each simulation hour are created
and integrated based on the worst case of hazard level at each pixel. The comparison on the flood
hazard between present and after development situations is made by crossing both maps.
Adult wading limit
Children wading limit
Safe to use 4WDs Safe to use small cars
Extreme hazard
High hazard
Medium hazard
Low hazard
a
characteristics (a) and hazard level (after Ramsbottom et al. (2003))
10
PHILIPPINES
The changes in flood hazard for both flood situations will be classified into negative and
positive impacts. The negative impact represents flood hazard changes from better flood hazard level
to worse level. On the other hand, the changes from worse hazard level to better hazard level are
classified as a positive impact. The map used to show this information will be called as impact map.
The general workflow for this analysis is shown in figure 6-15 (a).
17.5 years return period flood
Figure 6-16 shows the flood hazard map for present and after development situations in Naga
City. In general, most of the areas located near the riverbank and in the depression areas were
classified into Extreme and High hazard area. According to figure 6-16, obvious changes occurred in
the elevated commercial area in the CBD II. This area was classified into Extreme hazard in present
situation. However, the hazard level was changed to Low hazard for the situation after development.
Furthermore, flood hazard level has also changed in Barangay Mabulo and in the area next to the
Almeda Highway.
Extreme hazard High hazard Medium hazard Low hazard Extreme hazard High hazard Medium hazard L
a b
hazard for 17.5 years return period flood
10
PHILIPPINES
Area Classified into Flood Hazard for Before

Development Situation (17.5 years return period Area Classified into Flood Hazard for After
flood) Development Situation (17.5 years return period flood)
Low hazard
69.05 ha Low hazard
Extreme hazard 80.16 ha Extreme hazard
87.93 ha 97.05 ha
Medium hazard
38.04 ha
Medium hazard
37.32 ha
High hazard
High hazard
58.50 ha
a
76.98 ha b
Figure 6-17: Areas classified into flood hazard classes for present (a) and after development situation (b)
in Naga City (17.5 years return period flood)
According to figure 6-17, the highest changes was in High hazard class, which has decreased
from 76.98 ha to 58.50 ha (18.48 ha). On the other hand, the Medium hazard area has decreased from
38.04 ha to 37.32 ha (0.72 ha) and appears as the lowest changes hazard class. However, the areas
classified into Extreme hazard have increased from 87.93 ha to 97.05 ha (9.12 ha). Finally the Low
hazard area has increased by 11.11 ha (69.05 ha to 80.16 ha).
Figure 6-18 shows that the negative impact of the development spreads to quite a large extent,
especially to Barangays Triangulo, Mabulo, Tabuco and Sabang. It is clear that areas with the
positive impact are in the elevated commercial areas in CBDII. In this case, the negative impact
accounts about
33.28 ha areas and on the other hand, 18.76 ha areas received the positive impact of the development.
While 229.58 ha remains unchanged.
Area Changes Between 2 Situations (Present Situation

and After Development Situation) for 17.5 Years Return
Period Flood
Negative impact
33.28 ha
Unchanged
229.58 ha Positive impact
18.76 ha
Figure 6-18: Flood impact map for 17.5 years return period flood
10
PHILIPPINES
According to figure 6-19, most of the areas located near to the riverbank (where the river
overflow had initiated) and in depression areas are in general classified into Extreme and High
hazard. In the 10 years return period flood, the elevated commercial areas in CBD II were not
inundated, and the changes in flood hazard seems smaller compared to the changes in 17.5 years
flood.
Extreme hazard
High hazard Extreme hazard
a Medium hazard High hazard
Low hazard b Medium hazard
Low hazard
Figure 6-19:Flood hazard map for present (a) and the situation after development (b); the worst case
hazard for 10 years return period flood
Area Classified into Flood Hazard for Present

Area Classified into Flood Hazard for After
Situation (10 years return period flood)
Development Situation (10 years return period flood)
Extreme hazard Extreme hazard

18.97 19.23 ha
Low hazard Low hazard

86.34 89.18 ha
High hazard High hazard
86.52 82.34 ha
Medium hazard Medium hazard

68.09 a 65.70 ha b
In this flood event, areas classified into Extreme and Low hazard has increased by 0.26 ha
(18.97 ha to 19.23 ha), and 2.84 ha (86.93 to 89.19 ha) respectively. On the other hand, areas
classified into other type of hazard classes have decreased by 4.18 ha (86.52 ha to 82.34 ha) for High
11
PHILIPPINES
hazard and 2.39 ha (68.09 ha to 65.70 ha) for Medium hazard (figure 6-20).
11
PHILIPPINES
Area Changes Between 2 Situations (Present Situation

and After Development Situation) for 10 Years Return
Period Flood
Negative impact
Unchanged
9.27 ha
244.30 ha
Positive impact
10.98 ha
Negative impact
Unchanged
Positive impact
Most of the positive changes are observed in the CBD II area, where the former swampy
areas were elevated for the commercial purpose (figure 6-21). The negative impact areas are
observed in Barangays Triangulo, Sabang, Sta. Cruz and Panganiban Drive. Approximately it is
about 9.27 ha areas in Naga City suffer the negative impact of the development. On the other hand,
10.98 ha areas receive the positive impact, and the rest remains unchanged.
The 5 years return period flood is less severe compared to 10 years and 17.5 return period
floods. The overflow from the Naga River is rather small and most of the flooding caused by the
overflowing Bicol River. Obvious changes are observed in Barangay Triangulo (figure 6-22). The
Extreme and Low hazard areas have increased by 0.02 ha (8.77 ha to 8.79 ha) and 0.22 ha (37.04 ha
37.26 ha) respectively. On the other hand, the High hazard area has decreased by 1.16 ha (55.96 to
54.80 ha), and 1.8 ha (45.30 to 41.35 ha) for Medium hazard (figure 6-23).

High hazard High hazard
Medium hazard Medium hazard
a
Low hazard b
Low hazard
hazard for 5 years return period flood
11
PHILIPPINES
Area Classified into Flood Hazard for Present

Area Classified into Flood Hazard for After
Situation (5 years return period flood)
Development Situation (5 years return period
flood)

8.77 ha 8.79 ha
Low hazard Low hazard
37.04 ha 37.26 ha
High hazard
High hazard
55.96 ha
54.80 ha
Medium hazard
Medium hazard
43.50 ha
a 41.35 ha
b
For the 5 years return flood in Naga City, it is predicted that only 0.61 ha areas will suffer
the negative impact of the development, while for 5.34 ha areas still receives the positive impact with
decreasing in hazard level (figure 6-24). Through this analysis, it is also revealed that, most of the
changes occurred in Barangay Triangulo. In addition, the small scale construction, for instance the
drainage system in Barangay Triangulo has a great impact on the flood behaviour. In this situation,
the drainage system has reduced the flood problem in the lower part of the Barangay Triangulo
residential areas.
Area Changes Between 2 Situations (Present

Situation and After Development Situation) for 5 Years
Return Period Flood
Unchanged
Negative impact
139.52 ha
0.61 ha
Positive impact
5.34 ha
Negative impact
Unchanged
Positive impact
11
PHILIPPINES
Analysis on the relationship between flood risk perception and flood

hazard
Flood risk perception in the residential areas takes into account several aspects of flood
impacts for instance, physical damages, loss of lives, economical and social disruption. The flood
hazard mapping in this study has in specific focused in drowning hazard. Thus the flood hazard
aspect is one of the aspects in the flood risk perception. This analysis aims at investigating the
“agreement” and “disagreement” between the flood risk perception maps (before and after the
developments) and the hazard maps (before and after the developments). The analysis is based on the
10 years return period flood. Figure 6-25 shows the overall methodology of the analysis. The flood
risk perception maps and flood hazard maps are crossed and the comparisons are made on the flood
hazard classes and the flood risk perception classes. The comparison will find out the matched,
underestimation and overestimation cases between flood risk perceptions and flood hazard classes
(table 6-4).
Figure 6-25: Overall methodology of the relationship analysis between flood risk perception and flood
hazard
Table 6-4: The comparison between flood hazard classes and flood risk perception
Extreme hazard High hazard Medium hazard Low hazard
Disastrous Matched Matched O O
Unmanageable Matched Matched O O
Highly disturbing Matched Matched Matched O
Disturbing U U Matched O
Manageable U U Matched Matched
Normal U U Matched Matched
O = Residents overestimate the flood hazard
U = Residents underestimate the flood hazard
The matched cases are defined when the residential areas with high level of flood risk
perception classes are also classified in high level of flood hazard classes. On the other hand, the
11
PHILIPPINES
matched cases also defined when the residential areas with low level flood risk perception are also
classified in the low level of flood hazard classes. The overestimation of the flood risk perception
could be happened when the residential areas with low level flood hazard classes are classified in the
high level of flood risk perception classes. This scenario is possible, since the flood hazard only takes
into account the drowning hazard, which is only one of the several parts of the flood risk perception.
However, the underestimation of the flood risk perception on the flood hazard is dangerous, where
people might underestimate the flood hazard in their area (table 6-4).
a b
Figure 6-26: The comparison results between flood risk perception and flood hazard for present situation
(a) and after development situation (b); 10 years return period flood
Figure 6-26 (a) shows that for the situation before the development, the match cases account
about more than 50 percent of the residential areas, while 41 percent of the residential areas
overestimates the flood hazard. Only 0.3 percent of the residential areas have underestimated the
flood hazard. On the other hand, for after development situation, figure 6-26 (b) shows that about
59.34 percent of the residential areas have been classified into matched case, 40.32 percent of the
residential areas overestimate the flood hazard and only 0.3 percent of the residential areas
underestimate the flood hazard. Thus we can conclude that in the 10 years return period flood, the
flood risk perception derived from the interview sessions are related with the flood hazard map.
11
PHILIPPINES
7 Conclusion and Recommendation
DTM and DSM generation based on multi sources elevation data
Elevation data was derived from various sources, which characteristics differ in data scale,
vertical accuracy and method of derivations. All datasets should be integrated and interpolated to
produce the final DTM and DSM for flood modelling. However, each dataset covers different areas
and contains different type of landforms. In this case, complex terrain features for instance, rugged
terrain and meandering river course will exhibit high variation in terrain elevation in a short distance.
In this study, the step-by-step selection and integration of the elevation dataset had reduced the
disagreement and overlapping dataset. Further than that, since the selection of the elevation dataset
had based on the specific criteria, for instance the nominal accuracy, the best elevation value might
be selected from all the datasets to represent the real terrain information. Nevertheless, disagreements
between datasets could also exhibit high elevation variation in a short distance. Thus, reducing the
“nugget effect” from the integrated dataset had consequently reduced the complexity of the terrain
and also the disagreement between datasets. This is the trade-off that had been faced in this study;
losing the detailed information while reducing the disagreement between the datasets. The next step
was relied on the capability of commonly used terrain interpolation methods in generating the terrain
model. Finally, the Ordinary Kriging had come out as the best terrain model, which had full-filled the
requirements for flood modelling purpose. The other interpolation methods, for instance TIN and
ANUDEM were quite sensitive with errors or noise in the integrated elevation dataset. Therefore, the
final products of these interpolators contain artificial “pits” which is one of the big problems in
hydrological applications. On the other hand, the DTMs produced by the polynomial interpolation
methods lost the detailed landscape of the study areas. Besides, these DTMs have large error
compared to other interpolation methods and contain large number of error point, with accuracy more
than 1.0 m in the area of interest.
The 5 m resolution DTM is selected based on the nominal accuracy of the datasets and also
the requirement in flood modelling for urban area. The 5 m DTM resolution is good enough to model
the flood behaviour for individual structure and reclamation of small lands. Finally, the final DSM
was generated by adding the man-made terrain onto the DTM.
This is one of the cheapest approaches in constructing terrain model without precise and
expensive elevation data acquisition techniques for instance, LiDAR, Radar Interferometry and
Aerial Photos. However, the integration of all the available elevation datasets should be made very
carefully. In this case, the terrain updating process can be made through the conventional levelling
and the compilation of recent and future development plans to be added onto the natural ground
terrain model.
11
PHILIPPINES
Flood model calibration
In the flood model calibration it was revealed that by treating building structure as a solid
block and rough surface have shown discrepancies in water depth, flow velocity and flood extent.
Building structure as a solid block, in general has reduced the flood extent since the flood velocity
was reduced after hitting the solid structure. This has lead to an increase in flood depth, especially in
area with high building density. It was also observed that the maximum difference in maximum
water depth is about 1.27 m. Clear effect in water depth can be observed in the upper stream of Naga
River, especially in the residential areas near to the location where the river overflows initiated
(section 5.6, figure 5-13). In general there was only a small difference observed in flow pattern.
Furthermore, by treating building structure as rough surface had overestimated the flood extent. The
Metro Ville residential area located in Barangay Sabang was not inundated during the recent flooding
triggered by the super typhoon Nanmadol, but this area was inundated in flood modelling with
building structure as rough surface. The difference in flood extent between both building structures
can be seen in Barangays Sta. Cruz and Sabang. However, according to the accuracy assessment
based on the maximum flood depths in Barangays Sabang and Triangulo, both cases neither
underestimate nor overestimate the flood depth. The absolute accuracy of the maximum flood depth
prediction of both cases is 0.35 m. Further analysis was focussed on the effect of the Manning
coefficient value to the maximum water depth during the 10 years return period flood. According to
the significant test (z- test) on the error dataset, it was found that the differences in maximum
inundation depth were not significance for the flood model with those Manning coefficients.
Development impact assessment
The development impact assessment doesn’t intend to oppose any developments in Naga
City. The assessment aims at giving ideas on how flood behaviour would change after several major
developments were carried out especially in flood prone area. Instead of that, this research has also
given several simple methods in quickly detecting and assessing the detailed impact on flood
behaviour. On the other hand, it gives information about the impact on the surrounding areas by
considering the perception of the residents to a particular magnitude of flood (for instance, flood
depth and duration). In general the spatial extent of the impact is larger for a larger flood magnitude
(figure 7-1 (a) and (b)).
11
PHILIPPINES
The impact of the development for The impact of the development for
each flood reccurence interval each flood reccurence interval (Flood
(Hazard area) risk perception)
40
5
30 17.5 years 10 years
4 10 years
5 years
20 3 5 years
Area
Area
10 2
1
0
Positive impact Negative impact 0
Positive impact Negative impact
Im pact
a Impact b
Figure 7-1: The impact of the developments on hazard area (a) and flood risk perception (b)
Figure 7-1 shows the impact of the development based on the flood hazard mapping.
According to figure 7-1(a), for 17.5 years return period flood, area with the negative impact is higher
compared to the positive impact. Apart from that, areas with the positive impact for 10 years and 5
years return period floods are higher compared to the negative impact. Figure 7-1 (b), shows the
impact of the developments in flood risk perception aspect. It shows that areas with the negative
impact for 10 years return period flood are higher compared to the positive impact. However, in 5
years return period flood, areas with positive impact is higher than the negative impact.
According to these results, further analysis is carried out to investigate the detailed changes
of flood hazard area with respect to the landuse or landcover in Naga City. Figure 7-2 shows the
detailed changes of flood hazard areas in Naga City in 17.5 and 10 years return period flood.
According to figure 7-2 (b), in 10 years return period flood, it is about 92 percent of the areas with
the positive impact are in the commercial area. Meanwhile, areas with the negative impact are mostly
in the residential and commercial areas (figure 7-2 (a)). On the other hand, for 17.5 years return
period flood (figure 7-2 (d)), commercial areas account about 96 percent from the total areas with
positive impact. According to figure 7-2 (c), it is about 34 percent of the agriculture areas, 21 percent
of the residential areas and 13.88 percent of the commercial areas were classified in the area with
negative impact.
11
PHILIPPINES
Areas classified into Negative Impact (10 Years Return Period Flood ) classified into Positive Impact (10 Years Return Period Flood )
Areas
3.5 12
3 10
2.5 8
2 6
Area
Area
1.5 4
1 2
0.5 0
0
Open_ar
Agricult
Commerc
Industr
Institutio
Resident
Swam
Transportatio
Agricult
Commerc
Resident
Swam
Transportatio
a Landus e or Landcove r b Landuse or Landcover
Areas)classified into Positive Impact (17.5 Years Return Period Flood )

Areas classified into Negative Impact (17.5 Years Return Period Flood
12 20
10 18
8 16
14
6 12
4 10
2 8
0 6
4
2
0
c Landuse or Landcove r
d Landuse or Landcove r
Figure 7-2: Areas classified into negative and positive impact (based on the hazard map) for 10 years
return period flood (a) and (b) and 17.5 years return period flood (c) and (d)
In general, for 10 and 17.5 years return period floods, the commercial areas received the
positive impact of the developments. However, the negative impact area covered quite a large portion
of the residential areas. In addition, according to figure 7-1(b), based on the flood risk perception, the
residential areas with the negative impact in 10 years return period flood is higher compared to the
areas with the positive impact. This might because of most of the flood problem in 10 years return
period flood was transferred to the residential areas as illustrated in figure 7-2 (a). However, in 5
years return period flood, the areas with the positive impact are higher compared to the negative
impact areas (figure 7-1 (a) and (b)). As indicated in section 6.4, there relationship between the flood
risk perception and flood hazard is quite good and it is clearly shown in figure 7-1 (a) and (b).
Most of the changes are concentrated in the residential areas of Barangay Triangulo. This
area is the nearest area to the main developments in the CBD II. The elevated commercial area in the
CBD II in most cases received the positive impact of the development.
Through this study, it was proved that, by elevating the ground terrain can only solve the
flood problem in a particular area, but unfortunately this problem is transferred to another area. Small
scale development, for instance the Micro Drainage Project in Barangay Triangulo had quite a
pronounced role in changing the flood behaviour, especially in the small magnitude flood. The 17.5
and 10 years return period flood are quite large, thus the contribution of such small terrain changes
can be neglected. In 2, 5, 10 and 17.5 years return period flood, the new Mall proposed to be
11
PHILIPPINES
constructed next to the Almeda Highway was not inundated by water. Furthermore, the Almeda
Highway acted as a man-made levee or barrier and had efficiently obstructed the flood water,
especially from the Bicol River.
Figure 7-3: Risk management cycle; the results of this study can be used in HAZARD mapping and
MITIGATION process. Source: Smith (2001)
Development is necessary to provide present necessities and the needs of the future
generations. However, sustainable development may not be achieved without any effort in reducing
the intensity of hazard as a result of the developments. In this case, Naga City had taken many wise
steps in managing hazards, for instance flood hazard mapping, wind hazard mapping and damage
assessment, which are crucial for assessing potential loss and further actions to reduce it. In this
study, it was proved that, the 1D2D flood modelling can be used to investigate the behaviour of
floods and the flood impact after developments. Detailed investigation on the changes of flood
characteristics could in fact reduce future loss and damage. On the other hand, it is also a wise initial
step to reduce future investment on flood mitigation. Furthermore, several flood scenarios can be
created and simulated in investigating the effectiveness of physical flood mitigation measures to
reduce the flood problem. Figure 7-3 shows that this study basically can be used in flood hazard
mapping, flood risk assessment and mitigation phases.
Recommendation for Naga City Municipality
The landfill is considered among the cheapest ways to mitigate the flood problem. Thus, this
approach is widely applied, especially in the developing countries. However, proper planning of
landfill should be made properly to prevent creating flood problem in another place. In this study,
simple landfill approach is proposed to make sure that the landfill process will not increase the flood
level. Figure 7-4 (c) shows general approach of landfill, in which the earth supposes to be dredged
from the areas near to the river bank and this technique is known as “internal fill”. The “external fill”
approach, which earth dredged from high land area, would increase the flood level (figure 7-4 (b)).
12
PHILIPPINES
Figure 7-4: Landfill approaches; 1) External fill (b) and 2) Internal fill (c). Source: Lecture notes by
Kingma (2006)
Recommendations for future study
Recommendations for future study are as follow:
(a) Further study is needed to incorporate rainfall-runoff information into the flood model. The
computation time needed in flood modelling for this study was quite long, and even longer
for higher magnitude of flood. Therefore, the rainfall-run off information was not included
into the model. With a better performance computer, the computation time can be reduced
and give a better chance to include many others information in flood modelling, for instance
rainfall-runoff information.
(b) Landuse changes in Naga City were also obviously seen in Barangay Mabulo, for instance
areas located near to the main road were elevated for commercial activities. However, due to
limitation of time during fieldwork, it was not possible to do elevation measurement in this
area. Thus, for future study it’s suggested that detailed elevation measurement should be
carried out in this area.
12
PHILIPPINES
(c) The role of drainage to ease the flood problem in Naga City needs to be investigated,
especially for the lower magnitude of flood. Thus, further study is needed to include drainage
network schematization to define the capacity of the drainages in reducing the flood problem.
(d) The flood risk perceptions in the residential areas were only derived for Barangay Mabulo
and Barangay Triangulo. The perceptions on flood risk in these areas should be different
from another residential area, which have different experience of flood and different in
economical background. Thus, detailed investigation of their perceptions on flood will
provide complete dataset of flood risk perceptions in the residential areas. This information is
useful for detailed investigation on development impact assessment as shown by this study.
(e) Development impact assessment can be investigated in many ways. Those introduced in this
study are just the examples of evaluating the impacts through flood hazard aspect and flood
risk perception. Inevitably, potential damages and losses due to the changes of flood
behaviour is one of the important aspects. The combination of the flood risk perception,
hazard indicator and potential loss perhaps will serve much comprehensive approach in
evaluating the impact of the developments.
(f) Uncertainty in DTM should be quantified for detailed investigation of its impact on the flood
modelling results. One of the common approaches is through the uncertainty simulation.
Uncertainty simulation, for instance “Monte Carlo Simulation” produces several DTM or
DSM realizations. These realizations will be used in flood modelling, and the results will be
compared from to the results produced by the unaltered DTM. The comparison will give a
better understanding on the impact of the DTM uncertainty in flood modelling.
12
PHILIPPINES
8 References
Alkema, D., and H. Middelkoop. 2005. The Influence of Floodplain Compartmentalization on
Flood Risk within the Rhine-Meuse Delta. Natural Hazards 36:125-145.
C. A. Brebbia (ed.) 2004. 4th international conference on computer simulation in risk analysis and
hazard mitigation, Greece. September 2004.
Anderson, E.S., J.A. Thompson, D.A. Crouse, and R.E. Austin. Horizontal resolution and data density
effects on remotely sensed LIDAR-based DEM. Geoderma In Press, Corrected Proof.
Bin Usamah, M. 2005. Modelling flood events in the lower Bicol floodplain. MSc, International
Institute for Geoinformation Science and Earth Observation, Enschede.
(ed.) 2001. Conference on Hydraulics in CivilEngineering, Australia.
Blomgren, S. 1999. A digital elevation model for estimating flooding scenarios at the Falsterbo
Peninsula. Environmental Modelling & Software 14:579-587.
Chiles, J.P., and P. Delfiner. 1999. Geostatistics : modeling spatial uncertainty Wiley & Sons, New
York etc.
Cobby, D.M., D.C. Mason, and I.J. Davenport. 2001. Image processing of airborne scanning
laser altimetry data for improved river flood modelling. ISPRS Journal of
Photogrammetry and Remote Sensing 56:121-138.
Demirhan, M., A. Ozpinar, and L. Ozdamar. 2003. Performance evaluation of spatial
interpolation methods in presence of noise. International Journal of Remote Sensing
24:1237-1258.
Dowman, I. 2002. Improving the quality of digital elevation models from optical satellite data for
terrain evolution studies. International Archives of the Photogrammetry, Remote Sensing and
Spatial Information Sciences 34:617-621.
El-Ashmawy, N.T.H. 2003. Refined Modelling for Flood Extent Predictions Using Laser Scanning.
Master, International Institute for Geo-information Science and Earth Observation, Enschede.
Emergency Management Australia. 1999. Australian Emergency Manuals Series, Part III, Emergency
Management Practice. Manual. Emergency Management Australia.
Federal Emergency Management Agency (FEMA). 2006. Geospatial Accuracy Standard [Online]
http://www.fema.gov/pdf/nfip/alt_elevations/elevations_appb.pdf (verified 11 January 2006).
Geodata and Geoscience Australia. 2002. Data Use Guide; A digital elevation model of Australia
with a grid spacing of nine seconds in Longitude and Latitute [Online]
http://chapters.marssociety.org/canada/expedition-
mars.org/ExpeditionTwo/files/maps/DEM_userguide.pdf (verified 22/11/2005).
Hengl, T., G.B.M. Geuvelink, and A. Stein. 2003. Comparison of kriging with external drift and
regression-kriging [Online]. Available by International Institute for Geoinformation
Science and Earth Observation (ITC) http://www.itc.nl/library/Academic_output/ (verified
05/11/2005).
Horritt, M.S., and P.D. Bates. 2002. Evaluation of 1D and 2D numerical models for predicting river
flood inundation. Journal of Hydrology 268:87-99.
Hsu, M.H., S.H. Chen, and T.J. Chang. 2000. Inundation simulation for urban drainage basin with
storm sewer system. Journal of Hydrology 234:21-37.
Isaaks, E., and R. Srivastava. 1989. Applied geostatistics Oxford University Press, New York.
Kenny, F., and B. Matthews. 2005. A methodology for aligning raster flow direction data with
photogrammetrically mapped hydrology. Computers & Geosciences 31:768-779.
Kidner, D.B. 2003. Higher-order interpolation of regular grid digital elevation models. International
Journal of Remote Sensing 24:2981 - 2987.
Laud, J.M.S., Dizon, D.R. 2001. Quadrangle Mapping Report; Iriga, Naga and Sibono quadrangles.
Mines and Geosciences Bureau, Region V, Department of Environment and Natural
Resources, Philippines, Legaspi.
Lodwick, G.D. 1983. Digital terrain modelling The University of Calgary, Calgary.
12
PHILIPPINES
Mark, O., S. Weesakul, C. Apirumanekul, S.B. Aroonnet, and S. Djordjevic. 2004. Potential and
limitations of 1D modelling of urban flooding. Journal of Hydrology 299:284-299.
Maune, D.F. 2001. Digital elevation model technologies and applications : the DEM users manual
The American Society for Photogrammetry and Remote Sensing (ASPRS), Bethesda.
Meijerink, A.M.J., J.A.M. de Brouwer, C.M. Mannaerts, and C.R. Valenzuela. 1994. Introduction
to the use of geographic information systems for practical hydrology : IHP-IV M 2.3 ITC,
Enschede.
Naga City Government Philippines Business for Social Progress. 2001. Naga City Disaster
Mitigation Plan. Working paper. Asian Disaster Preparedness Center, Bangkok.
National Statistics Office Republic of Philippines. 2002. CAMARINES SUR: POPULATION TO
DOUBLE AFTER FOUR DECADES [Online]. Available by National Statistics Office
Republic of Philippines http://www.census.gov.ph/data/pressrelease/2002/pr02140tx.html
(posted 11/09/2002; verified 30/10/2005).
OIDC, O.I.D.C., INC. 1999. DISASTER MITIGATION PLAN (MAIN REPORT), PHILIPPINE
CITIES DISASTER MITIGATION PROGRAM, Naga City Disaster Mitigation Project.
NAGA AGA CITY ITY GOVERNMENT, Naga.
Palmiano-Reganit, M. 2005. Analysis Of Community’s Coping Mechanisms In Relation To Floods:
A Case Study In Naga City, Philippines. MSc, International Institute for Geoinformation
Science and Earth Observation, Enschede.
Pappenberger, F., K. Beven, M. Horritt, and S. Blazkova. 2005. Uncertainty in the calibration of
effective roughness parameters in HEC-RAS using inundation and downstream level
observations. Journal of Hydrology 302:46-69.
Peter, P.S., G. Pierre, H. I-Kuai, and M.B. Vaughn. 2005. Predicting ordinary kriging roughness and
dissectivity. Earth Surface Processes and Landforms 30:601-612.
Pike, R.J. 2002. A Bibliography of Terrain Modelling (Geomorphometry), the Quantitative
Representation of Topography. U.S. Geological Survey, California.
(ed.) 2002. 5th AGILE Conference on Geographical Information Science, Balearic Island, Spain.
25- 27 April 2002.
Priestnall, G., J. Jaafar, and A. Duncan. 2000. Extracting urban features from LiDAR digital surface
models. Computers, Environment and Urban Systems 24:65-78.
Raaflaub, L.D., and M.J. Collins. 2005. The effect of error in gridded digital elevation models on the
estimation of topographic parameters. Environmental Modelling & Software.
Ramsbottom, D., P. Floyd, and E. Penning-Rowsell. 2003. Defra / Environment Agency, Flood and
Coastal Defence R&D Programme. Defra – Flood Management Division, London.
(ed.) 2003. 28th International Hydrology and Water Resources Symposium, Wollongong, NSW.
10- 14 November 2003. The Institute of Engineer Australia.
Schmitt, T.G., M. Thomas, and N. Ettrich. 2004. Analysis and modeling of flooding in urban drainage
systems. Journal of Hydrology 299:300-311.
Sellin, R., T. Bryan, and J. Loveless. 2003. An improved method for roughening floodplains on
physical river models. Journal of Hydraulic Research 41.
Shi, W., M.F. Goodchild, and P.F. Fisher. 2003. Proceedings of the 2nd international symposium
on spatial data quality 2003 : 19 - 20 March 2003, Hong Kong The Hong Kong Polytechnic
University, Hong Kong.
Sinha, J. 2000. Creating DEMs using TOPOGRID [Online]. Available by Department of
Geography, University Of British Columbia
http://www.geog.ubc.ca/courses/geog516/talks_2000/topogrid.html (verified 3 December
2005).
Smith, K. 2001. Environmental hazards : assessing risk and reducing disaster. Third edition ed.
Routledge, London.
Sztruhar, D., M. Sokac, A. Holiencin, and A. Markovic. 2002. Comprehensive assessment of
combined sewer overflows in Slovakia. Urban Water 4:237-243.
Taylor, J.R. 1997. An Introduction to Error Analysis: The Study of Uncertainties in Physical
Measurements University Science Book, Sausalito, CA.
12
PHILIPPINES
Tennakoon, K.B.M. 2004. Parameterisation of 2D hydrodynamic models and flood hazard mapping
for Naga city, Philippines. MSc Thesis, International Institute for Geoinforamtion Science
and Earth Observation, Enschede.
The Geological Society of Australia. 2005. Case Study - Johnstone River Flooding (Flood Hazard
Catagories) [Online]. Available by The Geological Society of Australia,
http://earthsci.org/J_Flood04/index.htm (posted 3 December 2002; verified 10 August 2005).
Tomaž Podobnikar. 2005. Suitable digital elevation model [Online]. Available by GIS Development
http://www.gisdevelopment.net/technology/tm/tm003pf.htm (verified 01/11/2005).
Tomislav, H., G. Stephan, and P.S. Dhruba. 2003. Digital Terrain Analysis in ILWIS. International
Institute for Geoinforamtion Science and Earth Observation, Enschede.
(ed.) 2001. Conference on Hydraulics in Civil Engineering, Australia. 28-30 November 2001. The
Institution of Engineers, Australia.
Wechsler, S.P. 2003. Perception of Digital elevation Model Uncertainty by DEM Users. Urban and
Regional Information Systems Association 15.
Wechsler, S.P. 2005. Digital Elevation Model (DEM) Uncertainty: Evaluation and Effect on
Topographic Parameters [Online]
http://www.esri.com/library/userconf/proc99/proceed/papers/pap262/p262.htm (verified
19/11/2005).
Werner, M.G.F. 2001. Impact of grid size in GIS based flood extent mapping using a 1D flow model.
Physics and Chemistry of the Earth, Part B: Hydrology, Oceans and Atmosphere 26:517-522.
Werner, M.G.F., N.M. Hunter, and P.D. Bates. 2005. Identifiability of distributed floodplain
roughness values in flood extent estimation. Journal of Hydrology 314:139-157.
(ed.) 2003. 7th International Conference on GeoComputation, University of Southampton, United
Kingdom. 8 - 10 September 2003. GeoComputation.
Wood, J. 1996. The Geomorphological Characterisation of Digital Elevation Models. PhD, Leicester
University, Leicester.
12
PHILIPPINES
9 Appendices
Appendix A (Box plot for each error dataset of the interpolation

methods)
6.0000 277
184
4.0000
90
4.0000 177 142
139
90 87
74 1739
128 221
211 741
204
2.0000 131
2.0000 247 104
413
591
0.0000 0.0000
470
309
-2.0000 3127
8584
132 -2.0000 81
687
183 269
123 425
70 656
-4.0000 149
70
194 -4.0000 67
141
-6.0000
477
Elevation Residual Elevation Residual
Ordinary Kriging with 5- meter Ordinary Kriging for 100 m line to

block
point conversion distance interval
average
30.0000
748
755
4.0000 752
747
739 25.0000
731
719
2.0000
706
680 20.0000
663
637
746
0.0000 15.0000
745
744
662 741
685 10.0000 737
708 732
-2.0000 7237
728
733
738
743 720
744 5.0000 714
751 709
699
-4.0000 682
6614
631
754 0.0000
ANUDEM Triangulated Irregular Network (TIN)
8.00 8.0000
753
753
752
6.00 6.0000 747
748
7510
741
736 741
734
728
4.00 719 4.0000 725
706 719
2.00 2.0000
0.00 0.0000
-2.00 -2.0000
718
718
-4.00 -4.0000 731
732 739
748
751
-6.00 746 -6.0000
Elevation Residual
2nd Degree parabolic
12
PHILIPPINES
Elevation
Residual
2nd
Deg
ree
poly
nom
ial
12
PHILIPPINES
8.00 7.5000
753
753
749
752
6.00
747 746
5.0000
739
741 735
734 728
4.00 725 717
719 704
2.5000
2.00
0.0000
0.00
-2.5000
-2.00
715
726
734
718 748
-4.00 731 -5.0000
739
752
748
751
-6.00
-7.5000
Elevation Residual
Elevation Residual
2 Degree polynomial
nd
3rd Degree polynomial
7.5000 7.5000
753
749 753
748 749
5.0000 5.0000 744
741
736 737
733 732
724 725
711
2.5000 2.5000
710
0.0000 0.0000
-2.5000 -2.5000
703
719 718
728 729
735 735
747 7468
-5.0000 750 -5.0000 750
752 752
-7.5000 -7.5000
4 Degree polynomial
th
5 Degree polynomial
th
6.0000
751
747
744
740
4.0000 736
733
726
711
705
2.0000
0.0000
-2.0000
707
717
734
-4.0000
738
746
750
753
-6.0000
Elevation Residual
12
PHILIPPINES
Appendix B (Error distribution for each DTM interpolation method)
nd
2 Degree Linear 2nd Degree Parabolic 2nd Degree
3rd Degree 4th Degree 5th Degree
6th Degree
12
PHILIPPINES
Appendix C (Semi-variograms for 15 m and 25 m lags)
15 m Lag interval
Model:
Gaussian
Nugget: 2.2
Sill: 7.5
25 m Lag interval
Model: Gaussian
Nugget: 2.2
Sill: 16.0
Range: 2500
13
PHILIPPINES
Appendix D (Semi-variograms of the polynomial interpolation methods)
Semivariance Vs Lag (2nd Polynomial degree)

Semivariance Vs Lag (2nd Polynomial degree (Linear))
5.0 5.0
4.5 4.5
Semivarianc
4.0 4.0
Semivarianc
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Lag Lag
Semivariance Vs Lag (2nd Polynomial degree (Parabolic)) Semivariance Vs Lag (3rd Polynomial degree)
4.5 4.5
4.0 4.0
3.5 3.5
3.0
3.0
Semivarianc
Semivarian
2.5
2.5
2.0
2.0
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0.0
Lag 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Lag
Semivariance Vs Lag (5th Polynomial degree)
4.5
4.0 3.5
3.5
3.0
3.0
Semivarianc
Semivarian
2.5
2.5
2.0
1.5 2.0
1.0
0.5 1.5
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Lag
Lag
3.5
3.0
2.5
Semivarianc
2.0
1.5
1.0
0.5
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Lag
13
PHILIPPINES
Appendix E (Minimum-maximum plot for Elevation value against

Landuse/Landcover)
1 - Agriculture (paddy field)

4 – Dumpsite 7 – Open 10 – Riverbank
2 - Cemetery
5 – Industrial area 8 – 11 – Swampy area
3 – Commercial
6 - Institutional Parks 12 - Transportation utilities (railway)
9 - Residential
13
PHILIPPINES
Appendix F (Fieldwork planning for data collection)
Muhammad Zulkarnain Bin Abd Rahman

Hp no: 09154453005
Rahman10562@itc.nl
Data requirement (primary and secondary data)
A. Data collecting for recent flood, development plans and elevation
1. Land use planning office

 Recent or future land use plans (development plans)
 Central Business District I and II (CBD I and II) development plans
 Other proposed development plans in Triangulo
 Other new or proposed land-use development plans in another area
2. City Engineer office

 Recent or future flood hazard mitigation measures
 New drainage construction in Triangulo (detailed plan)
 Existing drainage network in Naga City (plan and profile)
 Other proposed flood mitigation measures in another area
3. City engineer office (levelling equipments and staff for assistant)

 Elevation point measurements for areas as stated below.
 Evacuation centers and other related areas or buildings for emergency planning
in each Barangay
 Routes used for evacuation in each Barangay
 Area with less contour data and elevation point measurement
 New development or proposed development areas in CBD I and CBD II
4. Public work office

 Road elevation from public work office
5. PAGASA
 Rainfall data for upstream Bicol River and Naga River. Rating curve at
downstream Bicol River for recent flood on December 2004.
6. Mr. Ernesto Elcamel and Graciela

 Flood depth data in Triangulo, areas covered by Saut and other areas which
were inundated during recent December 2004 flood (for 5, 10 and 20 years
return period flood). Flood depth and time of measurements are needed.
13
PHILIPPINES
B. Data collecting for flood emergency plans
1. Naga city disaster mitigation center
 Existing flood emergency plans

 Hazard indications for each element at risk which crucial in emergency
planning (alarm levels issue by Municipality which defines 3 alarm levels).
 Is there any different alarm level issued for different land use classes e.g.
residential and commercial area with house at the top of the building?.
 Is there any monitoring on the element at risk listed in table below?
Element at risk Vulnerability aspect

Evacuation routes People wading capability and the
suitable type of vehicle for evacuation
Open area People wading capability and the
suitable type of vehicle for evacuation
Residential and commercial Critical water level
with residential areas
Evacuation centers Operational aspect
Other centers that crucial Operational aspect
for emergency planning
(police station, fire brigade,
School,
Hospital and etc.)
 Which routes used for evacuation

 Flood characteristics that essential in flood emergency plans (e.g. water
depth, velocity, rate of rise and etc.)
 The effective flood emergency (evacuation plan) monitoring time interval
 The location of evacuation centres and other centres necessary in
emergency planning
▪ Municipality office (city engineer office and etc)
▪ Fire brigade station
▪ Public activity hall
▪ Barangay hall
▪ Other centers or buildings that related to flood emergency planning
and evacuation
2. Barangay captain
 Alarm level defines by Barangay level, which areas that preferred by local
resident in each Barangay to go during flooding and what is the coordination
made by them during flooding.
13
PHILIPPINES
Appendix G (5 years return period flood (present situation))

Maximum water depth Maximum water velocity
1.54 m 11.03 m/s
0.01 m 0.01 m/s
Warning time Inundation period
45th hour 75 hrs

68th hour 1 hr
Appendix H (10 years return period flood (present situation))

Maximum water depth
Maximum water velocity
2.24 m 13.84 m/s
0.01 m 0.01 m/s
Warning time Inundation period
45th hour 75 hours

83rd hour 1 hour
13
PHILIPPINES
Appendix I (17.5 years return period flood (present situation))
Maximum water depth Maximum water velocity
5.63 m 17.64 m/s

0.01 m 0.01 m/s
Warning time
44th hour
60th hour
13

Rahman

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Rahman

Uploaded by

Copyright:

Available Formats

Digital Surface Model (DSM) Construction and Flood Hazard

Simulation for Development Plans in Naga City, Philippines

Muhammad Zulkarnain Bin Abdul Rahman

Muhammad Zulkarnain Bin Abdul Rahman

Thesis Assessment Board

INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION

This document describes work undertaken as part of a programme of study at the

WWy h h blcss Ws∫

Climate and weather 11

DIGITAL ELEVATION MODEL (DEM)...................................................................................................................................18

4 DTM AND DSM GENERATIONS...........................................................................................50

6 DEVELOPMENT IMPACT ASSESSMENT............................................................................89

DETAILED CHANGES IN FLOOD CHARACTERISTICS..................................................................................................................................... 91

7 CONCLUSION AND RECOMMENDATION........................................................................111

APPENDIX I (17.5 YEARS RETURN PERIOD FLOOD (PRESENT SITUATION)).........................................................................130

Limitation of the study

Figure 1-1: Overall research methodology

This thesis contains 7 chapters.

1: 500,000 (approximate scale) 1: 2,000,000 (approximate scale)

1: 30,000 (approximate scale)

Climate and weather

Average Monthly Rainfall in Naga City

Figure 2-2: Average monthly rainfall in Naga City (www.naga.gov.ph)

Data need analysis

The collected data can be divided into 6 major groups as follow:

1. Elevation or topographical data

 CBD II development plan – This dataset contains detailed information on building

i. Area with new developments (i.e. Metro Ville and CBD I)

Landuse or Landcover and development plans

The landuse or landcover information was

 Detailed CBD II development plan

Flood emergency plan

Rainfall and floodwater depth

Flood risk perception

Digital GIS data

Digital Elevation Model (DEM)

Digital Terrain Model (DTM) for flood modelling

Applications 5m DTM 7.5m DTM 10m DTM

Flood hazard mapping in urban areas Adequate

Detailed studies related to velocity measurements,

Fixed-size grid is a common approach in any Geographical Information System (GIS)

DEM interpolation method

Three interpolation techniques will be

Figure 3-3: Cross section of Spline interpolation surface (Maune, 2001)

Kriging interpolation in terrain modelling

A Geostatistical interpolation or kriging interpolation method is similar to a probabilistic

(h) = ?2  [(Z (xi + h) – Z (x)) Equation 3-1

(h) = semi-variance at distance h

m = number of point pairs within h

distance Z (x) = elevation value at position

Z (xi + h) = value at h distance from i

Krigging interpolation is based on “a

1. The prediction is based on “linear combination of known data values”

4. Clustered measurements resulted by over-sampling in a small area, in general can

6. “The prediction error should be as small as possible”

Zo = The estimated value

λi = Weights for measurements participated in the estimation

3. Compute the semi-variogram of the residual dataset

Higher order terrain interpolation

The deterministic interpolation approach tries to fit a mathematical function to a set of

2nd degree linear Z = a + bx + cy + dxy 4

2nd degree parabolic Z = a + bx + cy + ex2 + fy2 5

2nd degree Z = a + bx + cy + dxy + ex2 + fy2 6

3rd degree Z = a +…..+ gx + hx2y + ixy2 + jy3 10