Professional Documents
Culture Documents
Feasibility Study of The Kisegese Irrigation Scheme
Feasibility Study of The Kisegese Irrigation Scheme
DISCLAIMER
The author’s views expressed in this publication do not necessarily reflect the views of the United
States Agency for International Development or the United States Government.
Using this Report
The feasibility study of the Kisegese irrigation scheme is documented in a main
report with a series of associated annexes covering specific components. This report
is one of those annexes. This Kisegese study has been completed in parallel with
similar studies for the Udagaji-Mgugwe and Mpanga areas in Kilombero District,
which are documented in separate reports. A single executive summary, prepared
as a standalone document, covers the three study areas, given the similarities
between them. The figure below illustrates the layout of the Kisegese feasibility
study report in relation to the other study areas.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME i
Table of Contents
Using this Report.......................................................................................................... i
Table of Contents ......................................................................................................... ii
List of Appendices ....................................................................................................... iv
List of Tables ............................................................................................................... v
List of Figures.............................................................................................................. vi
List of Acronyms and Abbreviations ........................................................................... vii
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME ii
2.7.2.3 Shallow Flow ........................................................ 2-24
2.7.2.4 Channel Flow ....................................................... 2-25
2.7.2.5 Tc for each Sub-catchment .................................. 2-25
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME iii
6 REFERENCES ................................................................................................... 6-1
List of Appendices
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME iv
List of Tables
Table 2.1: Summary of Basin-wide Rainfall Data Considered for Analysis ....... 2-2
Table 2.2: Summary of IRRIP2 Rain Gauge Data ............................................. 2-7
Table 2.3: Summary of IRRIP2 Climate Station Data ........................................ 2-7
Table 2.4: Daily Rainfall Summary Statistics from Long-Term Gauges .......... 2-13
Table 2.5: Comparison of IRRIP2 Rain Gauge Data with Long-Term Data .... 2-14
Table 2.6: Comparison of Rainfall Derivation at Ifakara and Lumemo ............ 2-19
Table 2.7: Rainfall Derivation at Ifakara .......................................................... 2-21
Table 2.8: Types of Flow in Derivation of the Time of Concentration .............. 2-24
Table 2.9: Average Velocity in Channel Watercourses ................................... 2-25
Table 2.10: Land Use and Curve Numbers for Sub-Catchment Peak Flows .... 2-27
Table 2.11: Kisegese Drainage Sub-Catchment Peak Flows ............................ 2-29
Table 3.1: Mbingu Spot Gaugings ..................................................................... 3-2
Table 3.2: Comparison of Flow Statistics at 1KB8 (Mpanga at Mpanga
Mission) .......................................................................................... 3-10
Table 3.3: Q80 Discharge at the Chiwachiwa Weir Offtake ............................ 3-13
Table 3.4: Q80 Discharge at the Ruipa Weir Offtake ...................................... 3-14
Table 3.5: Comparison of Flow Statistics at 1KB8 (Mpanga at Mpanga
Mission) and 1KB9 (Mnyera at Taveta Mission) ............................. 3-15
Table 3.6: Calculated Flood Growth Factors at 1KB8 (Mpanga at Mpanga
Mission) and 1KB9 (Mnyera at Taveta Mission) ............................. 3-15
Table 3.7: Kisegese Flood Modeling Peak Flows (m³/s) ................................. 3-16
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME v
List of Figures
Figure 1.1: Kisegese Irrigation Scheme Layout .................................................. 1-2
Figure 2.1: Rain Gauge and Climate Station Locations in the Kilombero
Catchment ........................................................................................ 2-2
Figure 2.2: Rainfall and Climate Stations Installed by IRRIP2 ............................ 2-6
Figure 2.3: Monthly Rainfall Total Comparison ................................................... 2-9
Figure 2.4: Comparison of Overlapping Records at Ifakara and Katrin ............ 2-10
Figure 2.5: Comparison of Ifakara and Katrin Monthly Average Rainfall
Totals .............................................................................................. 2-11
Figure 2.6: Monthly Rainfall Total Comparison Between Kilombero and
Great Ruaha Gauges ..................................................................... 2-12
Figure 2.7: Monthly Rainfall Total Comparison Between Kilombero
Gauges ........................................................................................... 2-13
Figure 2.8: Ifakara Monthly Peak Hourly Rainfall Intensity ............................... 2-16
Figure 2.9: Uhafiwa Monthly Peak Hourly Rainfall Intensity ............................. 2-17
Figure 2.10: Malinyi Monthly Peak Hourly Rainfall Intensity ............................... 2-18
Figure 2.11: Percentage Exceedance Rainfall from Ifakara ............................... 2-22
Figure 2.12: Kisegese Irrigation Scheme Drainage Sub-Catchments ................ 2-23
Figure 2.13: SCS Type II Synthetic Precipitation Distribution ............................. 2-26
Figure 3.1: River Gauging Stations in Kilombero Catchment ............................. 3-1
Figure 3.2: Standardized Average Monthly Flows .............................................. 3-3
Figure 3.3: Water Level Series from Gauge 1KB15 ........................................... 3-4
Figure 3.4: Level Duration Curves from Gauge 1KB15 ...................................... 3-6
Figure 3.5: Summary of Flow Data Records across the Kilombero
Catchment, 1956-2014 ..................................................................... 3-8
Figure 3.6: FDC Comparison between 1KB8 and 1KB8 after Infilling .............. 3-11
Figure 3.7: Kisegese Analysis Gauge Locations .............................................. 3-12
Figure 3.8: Hydrograph Shape Derivation at 1KB14 (Lumemo at
Kiburubutu) ..................................................................................... 3-17
Figure 4.1: Observed and Projected Temperature Anomalies ........................... 4-1
Figure 4.2: Projected Changes in Annual Precipitation in 2071-2100
Compared to Baseline (1971-2000) Conditions (mm/year) .............. 4-2
Figure 4.3: Observed and Projected Rainfall Anomalies .................................... 4-2
Figure 4.4: Projected Global Multi-model Averages of Precipitation Trends
under an A1B Emissions Scenario ................................................... 4-4
Figure 4.5: Potential Impacts of Temperature Rise on Crop Agriculture ............ 4-5
Figure 4.6: Potential Impacts of Precipitation Changes on Crop
Productivity ....................................................................................... 4-6
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME vi
List of Acronyms and Abbreviations
ADCP acoustic doppler current profiler
AMAX annual maximum
AMC antecedent moisture content
CN curve number
IDF intensity-duration-frequency
IHH Institute of Hydrology and Hydraulics (of Tanzania)
IPCC Intergovernmental Panel on Climate Change
IRRIP2 Technical Assistance to Support the Development of Irrigation and
Rural Roads Infrastructure Project
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME vii
1 INTRODUCTION
1.1 General
Under the U.S. Government’s Feed the Future (FTF) program, CDM International
Inc. (CDM Smith) is implementing the “Technical Assistance to Support the
Development of Irrigation and Rural Roads Infrastructure Project” (IRRIP2), funded
by the Tanzania Mission of the United States Agency for International Development
(USAID). Among other activities, IRRIP2 is completing feasibility studies (FS) to
support the development of four irrigation schemes in the Kilombero District of
Morogoro Region. These irrigation schemes will be for small farmers who are
currently growing one crop of rainfed rice prone to flooding and episodes of drought.
This Annex presents the results of reviews and analyses of climatic and hydrometric
data collected near the proposed Kisegese irrigation scheme and the wider
Kilombero basin as part of the FS. The data used for analysis includes data
recorded within the Kilombero catchment and surrounding areas by the Government
of Tanzania (GoT), as well as data collected from climatic stations installed by
IRRIP2 in January 2015.
The data review includes a catchment-wide assessment of available data in the
Kilombero catchment of the Rufiji Basin to identify appropriate gauging stations and
climatic stations with robust records for use in determining water resources for the
Kisegese scheme. The data has been further analyzed where anomalies were
noted, or where infilling and extension of records could improve the analysis.
The most appropriate data series was determined to calculate the water resources
available for irrigation; develop flood flows and hydrograph shapes to model and map
flood extents; and derive rainfall intensity-duration-frequency (IDF) curves for
drainage duty design within the scheme area. Potential impacts of climate change
on the hydrology were also evaluated.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 1-1
Figure 1.1: Kisegese Irrigation Scheme Layout
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 1-2
3. Derive a suite of flow series at each of the irrigation scheme weir locations to
support analysis of likely available flow volumes for irrigation.
4. Determine flood hydrology to be used as an input to hydraulic modeling and
ultimately flood mapping to assess the flood risk to the proposed scheme and
the placement of mitigation options to protect it.
5. Determine the 100-year flow for use in the weir design calculations.
6. Derive flood hydrograph shapes for hydraulic modeling and flood mapping.
To increase the robustness of the analyses, data was gathered from a wide variety of
sources across the Kilombero catchment and beyond, where data supported
calculations relevant to determining climatic and hydrometric series for use in the
Kisegese analysis.
The main sources of data are the Rufiji Basin Development Authority (RUBADA) and
the Rufiji Basin Water Office (RBWO) who provided data in digital form from a large
number of gauging stations and climatic stations. Hard copies of water yearbooks
and the Rufiji Basin Hydropower Master Plan were also provided. The data has been
critically assessed to determine its degree of quality and applicability for the study,
with only the highest quality data incorporated in the analysis.
In addition to existing gauging stations and climatic stations monitored by the GoT,
IRRIP2 installed new data gathering stations to enhance existing datasets. The new
stations were placed in locations to provide value to the IRRIP2 Feasibility Studies
and enhance ongoing understanding of the climate within the Kilombero catchment.
The analysis presented in this Report uses local data where appropriate to give best
estimates of local conditions and to help reduce uncertainty in the analyses that can
result from using low-resolution, global datasets.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 1-3
2 RAINFALL AND CLIMATE DATA ANALYSIS
2.1 Introduction
The rainfall records have been used to derive suitable estimates of IDF curves. IDF
curves were derived to calculate runoff from small catchments adjacent to the
Kisegese irrigation scheme to input into the external drainage channel design and to
calculate the internal drainage requirements. IDF data has been used to determine
runoff in conjunction with the United States Soil Conservation Service Curve Number
(SCS CN) method, taking into account catchment-specific data. Details of analyses
and calculations are presented below.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-1
Figure 2.1: Rain Gauge and Climate Station Locations in the Kilombero
Catchment
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-2
Daily
Name Catchment Rainfall Comment
The rainfall record at Ilangali is relatively
short and the gauge is located a
significant distance from the Kilombero
Ilangali Great Ruaha 1992-2011 catchment and therefore it is distant from
the Kisegese irrigation scheme and
unlikely to replicate local rainfall
conditions.
The rainfall record at Kimani is long and
reasonably complete, thought the gauge
is located a significant distance from the
Kimani Great Ruaha 1963-2014 Kilombero catchment and therefore it is
distant from the Kisegese irrigation
scheme and unlikely to replicate local
rainfall conditions.
The rainfall record at Lower Rufiji is short
and the gauge is located a significant
distance from the Kilombero catchment
Lower Rufiji Utete 2006-2015
and therefore it is distant from the
Kisegese irrigation scheme and unlikely to
replicate local rainfall conditions.
The data record at Lumemo is reasonably
long and complete and therefore valuable
Lumemo Kilombero 1969-2011
in determining long term rainfall trends
and high-intensity storm rainfall depths.
The rainfall record at Madibira is long and
reasonably complete, thought the gauge
is located a significant distance from the
Madibira Great Ruaha 1971-2014 Kilombero catchment and therefore it is
distant from the Kisegese irrigation
scheme and unlikely to replicate local
rainfall conditions.
The data record at Malinyi is a short
record and therefore of little value in
determining long-term rainfall trends and
Malinyi Kilombero 1998-2001 high intensity storm rainfall depths. In
addition to daily flow data, approximately
two and a half years of 15-minute data
has been recorded at Malinyi.
The rainfall record at Msembe is long and
reasonably complete, thought the gauge
is located a significant distance from the
Msembe Great Ruaha 1971-2014 Kilombero catchment and therefore it is
distant from the Kisegese irrigation
scheme and unlikely to replicate local
rainfall conditions.
The rainfall record at Mtandika is long and
reasonably complete, thought the gauge
is located a significant distance from the
Mtandika Great Ruaha 1982-1997 Kilombero catchment and therefore it is
distant from the Kisegese irrigation
scheme and unlikely to replicate local
rainfall conditions.
The rainfall record at Mtera is long and
Mtera Great Ruaha 1971-2014
reasonably complete, thought the gauge
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-3
Daily
Name Catchment Rainfall Comment
is located a significant distance from the
Kilombero catchment and therefore it is
distant from the Kisegese irrigation
scheme and unlikely to replicate local
rainfall conditions.
The data record at Njombe is a short
record and therefore of little value in
determining long-term rainfall trends and
Njombe Great Ruaha 1998-2002 high intensity storm rainfall depths. In
addition to daily flow data, approximately
two and a half years of 15-minute data
has been recorded at Njombe.
The data record at Uhafiwa is of moderate
length. In addition to daily flow data,
Uhafiwa Kilombero 1998-2014 approximately two and a half years of 15-
minute data has been recorded at
Uhafiwa.
The daily record at Illovo MET station is
Illovo MET Kilombero 1999-2014 relatively short though monthly rainfall
totals are recorded from 1961 – 2014.
The daily record at Illovo Msolwa station
Illovo Msolwa Kilombero 1999-2014 is relatively short though monthly rainfall
totals are recorded from 1961 – 2014.
The daily record at Illovo Ruembe station
Illovo Ruembe Kilombero 1999-2014 is relatively short though monthly rainfall
totals are recorded from 1974 – 2014.
A short daily record has been recorded at
KPL, though the majority of the record is
KPL Kilombero 2011-2015 missing and therefore of little use.
Weekly data is recorded from 2010 –
2014 though the dry months are omitted.
No daily rainfall record has been made
Kivere Mufindi Kilombero N/A available for this study. Monthly rainfall
records from 1961 – 1989 are available
Idege No daily rainfall record has been made
(Livalonge Kilombero N/A available for this study. Monthly rainfall
Estate) records from 1961 – 1989 are available
No daily rainfall record has been made
Luiga Mufindi Kilombero N/A available for this study. Monthly rainfall
records from 1961 – 1989 are available
No daily rainfall record has been made
Taveta
Kilombero N/A available for this study. Monthly rainfall
Mission
records from 1942 – 1990 are available
No daily rainfall record has been made
Ukalawa Tea Kilombero N/A available for this study. Monthly rainfall
records from 1971 – 1989 are available
No daily rainfall record has been made
Lupembe Tea
Kilombero N/A available for this study. Monthly rainfall
Office
records from 1971 – 1988 are available
No daily rainfall record has been made
Matembwe
Kilombero N/A available for this study. Monthly rainfall
Mission
records from 1951 – 1989 are available
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-4
Daily
Name Catchment Rainfall Comment
No daily rainfall record has been made
Kilolelo Farm Kilombero N/A available for this study. Monthly rainfall
records from 1975 – 1989 are available
No daily rainfall record has been made
Tanganyika
Kilombero N/A available for this study. Monthly rainfall
Wattle No. I
records from 1961 – 1989 are available
No daily rainfall record has been made
Tanganyika
Kilombero N/A available for this study. Monthly rainfall
Wattle No. II
records from 1961 – 1989 are available
No daily rainfall record has been made
Tanganyika
Kilombero N/A available for this study. Monthly rainfall
Wattle No. IV
records from 1961 – 1989 are available
No daily rainfall record has been made
Upper Spray
Kilombero N/A available for this study. Monthly rainfall
Forest Kihansi
records from 2007 – 2012 are available
Rungwe No daily rainfall record has been made
Research Kilombero N/A available for this study. Monthly rainfall
Centre Kihansi records from 2007 – 2012 are available
Mid Gorge No daily rainfall record has been made
Sus Bridge Kilombero N/A available for this study. Monthly rainfall
Kihansi records from 2007 – 2012 are available
No daily rainfall record has been made
Lower Kihansi
Kilombero N/A available for this study. Monthly rainfall
Forest
records from 2007 – 2012 are available
No daily rainfall record has been made
Kihansi ENV
Kilombero N/A available for this study. Monthly rainfall
Centre
records from 2004 – 2015 are available
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-5
Figure 2.2: Rainfall and Climate Stations Installed by IRRIP2
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-6
depth and intensity and it does not rely on staff reading the gauge on a daily (or sub-
daily) basis.
Table 2.2 presents a summary of the recorded rainfall data at each station. The
closest station to the Kisegese scheme is Mbingu, which unfortunately has suffered
from a malfunctioning logger and therefore rainfall data collected at this gauge is
considered spurious. Daily totals in excess of 2,500 mm are reported in the recorded
data, which is not considered appropriate for consideration in this study.
The data collected at the other station locations is considered to be within the bounds
of what is expected within the Kilombero basin when compared to long-term records.
Table 2.2: Summary of IRRIP2 Rain Gauge Data
Daily Daily Hourly
Station Average Max Average Hourly Max
Name Record Period (mm) (mm) (mm) (mm)
17/March/2015 –
Chisano 12.34 76 3.22 54
12/May/2015
Mbingu N/A N/A N/A N/A N/A
23/February/2015 –
Mutenga 21.09 191 5.46 31.5
11/June/2015
05/March/2015 –
Treefarms 6.7 36 2.17 17.5
12/May/2015
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-7
Chisano1 Mbingu Mutenga Treefarms
Maximum N/A 984.2 983.8 983.0
Average N/A 83.1 80.5 80.5
Humidity (%) Minimum N/A 33.7 33.8 31.0
Maximum N/A 100 99.4 100
1Chisano gauge malfunctioned during the March-June period. It has since been repaired for future
collection efforts.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-8
Figure 2.3: Monthly Rainfall Total Comparison
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-9
Two gauges have been located at Ifakara, in close proximity to one another, with a
combined, substantial record period of 88 years showing an extended history of
rainfall patterns in the Kilombero basin including the capture of a significant number
of large rainfall events.
Comparison of the two datasets was undertaken to understand if combining the
records was appropriate. The records have an overlapping period from January
1969 to July 1972. Figure 2.4 presents monthly rainfall totals over this period to
assess the similarity in the recorded rainfall records. In general, the monthly rainfall
totals are similar during the overlapping period, though during April 1968 and 1971
the totals recorded at Ifakara are significantly higher. It is unclear why these monthly
totals are so different, but could be down to an extremely localized storm event, or a
malfunction at the Katrin gauge, where a single, significant event has not been
recorded resulting in a large difference in the totals recorded.
In addition to comparing the overlapping period, the full records at each gauge have
been analyzed and compared. Figure 2.5 presents a comparison of average
monthly rainfall totals for Ifakara (1927 to 1972) and Katrin (1968 to 2014).
Percentage difference for each month is also presented on the graph. The Katrin
data shows slightly higher rainfall totals for ten months of the year. The average
percentage difference between the records is 12%.
The Katrin record may be showing a slight increase in monthly rainfall totals over the
Ifakara record due to the type of rain gauge used or potentially due to the impact of
climate change.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-10
Figure 2.5: Comparison of Ifakara and Katrin Monthly Average Rainfall Totals
The records at Ifakara have been combined for the purposes of this study given the
proximity of the rain gauges. Given the record length available at Ifakara, it is
considered the most appropriate to derive return period rainfall depths.
Data from the Lumemo gauge, which is located reasonably close to Ifakara, has
been recorded for 42 years. This record has been identified as a record that can be
included in the analysis undertaken, potentially as a check of rainfall derived using
the Ifakara and Katrin records.
2.4.2 Data Analysis
Several comparisons of the available rainfall series have been undertaken to
understand which data is of best use in further analysis and derivation of IDF curves.
Analyses include assessing the Kilombero and Great Ruaha gauges to determine if
there is a different rainfall profile between the catchments and also comparison of
daily and sub daily data within the Kilombero basin.
2.4.2.1 Comparison of Kilombero with Great Ruaha Data
A significant amount of available data is from rain gauges located outside the
Kilombero basin and therefore a comparison has been undertaken to determine if
this data is representative of rainfall depths and patterns at the Kisegese irrigation
scheme.
Figure 2.6 shows average monthly rainfall totals calculated for the gauges located
within the Kilombero basin and compared with those located in the Great Ruaha
basin. The gauges used to produce the Kilombero and Ruaha totals are as identified
in Table 2.1. During the dry period from June to September, the comparison
between the basins is similar, but during the wet season, rainfall totals are
significantly higher in the Kilombero in comparison with the Great Ruaha.
Given the rainfall data is to be used in derivation of high intensity rainfall events for
design of internal drainage and flood mitigation channels, only data from the
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-11
Kilombero basin has been used given the obviously higher wet season rainfall
compared with more distant rain gauges.
Figure 2.6: Monthly Rainfall Total Comparison Between Kilombero and Great
Ruaha Gauges
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-12
Figure 2.7: Monthly Rainfall Total Comparison Between Kilombero Gauges
The daily average rainfall totals from the long-term gauged data are much lower than
those recorded by the IRRIP2 gauges. This is to be expected, given that the short
data records are from the wet part of the year, not taking into account the dry season
like the long-term records do.
Table 2.5 presents the results of an analysis to fairly compare the daily rainfall data
at the IRRIP2 gauges with the long-term gauged records. Where the IRRIP2 gauge
has been in operation, the average daily rainfall for that period has been calculated at
each long-term record, over the entire record period (i.e., the daily average rainfall for
the period 17/03 to 12/05 over the full 88-year record at Ifakara).
The comparison shows that for the Chisano and Treefarms gauges the daily average
value is within the range of values calculated for the long-term gauges. For Mutenga,
the daily average is more than double that of the highest daily average of the long-
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-13
term gauges. The Mutenga gauge is at a higher altitude than the Ifakara, Lumemo
and Malinyi gauges, but much lower than the gauge at Uhafiwa. It is unclear why a
much larger daily average is recorded at Mutenga, but it may be the result of
capturing a very intense, localized storm.
In general, the daily totals recorded by the IRRIP2 gauges are considered
appropriate when compared to the wet seasons recorded by the long-term gauges.
Table 2.5: Comparison of IRRIP2 Rain Gauge Data with Long-Term Data
17/March to 23/February to 05/March to
Station Name
12/May (mm) 11/June (mm) 12/May (mm)
Ifakara 10.83 7.75 10.45
Lumemo 6.88 4.93 6.64
Uhafiwa 12.94 9.02 12.36
Malinyi 9.55 6.59 9.3
Chisano 12.34 - -
Mbingu1 - - -
Mutenga - 21.09 -
Treefarms - - 6.7
Notes:
IRRIP2 gauges (Chisano, Mbingu, Mutenga, and Treefarms) are identified in bold.
1 Rainfall data from Mbingu considered unreliable due to faulty gauge
The daily maximum rainfall totals from the long-term gauges are generally higher
than the IRRIP2 gauges. This is also to be expected as the chances of recording
large daily rainfall totals are much higher looking at the long-term records compared
to 3-month data records recorded at the IRRIP2 gauges. The maximum totals
recorded by the IRRIP2 gauges are considered appropriate, with the gauge at
Mutenga capturing a maximum daily rainfall of 191 mm, which is directly comparable
with the long-term gauged datasets.
2.4.2.4 Sub-Daily Data Analysis
In addition to daily rainfall totals, a very limited amount of 15-minute, 30-minute and
60-minute rainfall peak intensity measurements are available from gauges at Ifakara,
Malinyi and Uhafiwa, all located in the Kilombero basin. Figure 2.8, Figure 2.9 and
Figure 2.10 show the largest peak hourly precipitation intensity for each month at the
Ifakara, Uhafiwa, and Malinyi gauges, respectively.
It is not considered appropriate to derive 15-minute, 30-minute or 60-minute return
period rainfall totals given the short data record available. The period of record that
the data covers is not considered significant when compared to long-term climatic
variation and therefore results of any statistical analysis using the data would be held
in very low confidence. The available data does give an idea of the likely magnitude
of rainfall that is experienced in the Kilombero basin and therefore is considered
acceptable to use as a reasonableness check to other analyses undertaken. It is
noted that the sub-daily data shows a yearly trend which has been observed in the
daily rainfall datasets.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-14
2.4.2.5 Comparison of Available Sub-Daily Data with IRRIP2 Rainfall Data
Maximum hourly rainfall totals have been calculated from the IRRIP2 rainfall records
(presented in Table 2.2) for comparison with the sub-daily rainfall data recorded at
Ifakara, Uhafiwa and Malinyi.
Figure 2.8, Figure 2.9 and Figure 2.10 show a comparison of the sub-daily data at
Ifakara, Uhafiwa and Malinyi in comparison with maximum hourly rainfall totals
recorded at the IRRIP2 rain gauges.
The maximum hourly rainfall total from Chisano is higher than records at Uhafiwa
and Malinyi and higher than all but two of the recorded months at Ifakara. It is
unclear why a higher maximum hourly rainfall total has been recorded at Chisano in
comparison with Mutenga and Treefarms gauges. Given the maximum total from
Chisano has been exceeded twice in the gauged record from Ifakara it is still
considered a plausible rainfall depth to fall in one hour in the Kilombero basin.
The maximum hourly rainfall totals recorded at the Mutenga and Treefarms gauges
have been exceeded a number of times in the sub daily records from Ifakara,
Uhafiwa and Malinyi and are therefore also considered plausible.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-15
Figure 2.8: Ifakara Monthly Peak Hourly Rainfall Intensity
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-16
Figure 2.9: Uhafiwa Monthly Peak Hourly Rainfall Intensity
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-17
Figure 2.10: Malinyi Monthly Peak Hourly Rainfall Intensity
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-18
2.4.3 Conclusions
Observation and analysis of the available rainfall data results in the following:
Given the long-term data record available at the Ifakara gauge, it will be used to
derive the IDF curves to be used in the internal and external drainage analysis.
Data from the gauge at Lumemo will be used as a check of this analysis.
Short duration gauged data from Ifakara will be used as a check on the derived
IDF data.
The gauge at Ifakara will also be used to derive required rainfall datasets for
use in CROPWAT models of the scheme.
2.6 Outputs
2.6.1 Intensity-Duration-Frequency Analysis
The annual maximum daily rainfall totals recorded at Ifakara and Lumemo were
selected from the available data records and a generalized logistic distribution fitted
to the data points to derive daily rainfall depths at 5-year, 10-year, 25-year and 50-
year return periods. Table 2.6 summarizes the outputs from the analysis.
Table 2.6: Comparison of Rainfall Derivation at Ifakara and Lumemo
Return Period (years) Ifakara (mm/day) Lumemo (mm/day)
5 122 112
10 140 137
25 166 169
50 188 192
The derived daily rainfall totals at Ifakara correspond closely to those at Lumemo,
which increases confidence in the analysis. The rainfall intensities derived from the
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-19
Ifakara record are robust given the record length and the close comparison with the
Lumemo data.
Appendix A.1 presents the annual maximum (AMAX) record from Ifakara used in
derivation of return period rainfall. Appendix A.2 presents distributions fitted to the
AMAX data.
The data from Ifakara has been taken forward to use in CROPWAT calculations and
scheme drainage design. The data from Ifakara is preferred over the data from
Lumemo primarily due to the long record length available, which increases
confidence in deriving return period rainfall.
2.6.2 Rainfall Records for CROPWAT Calculations
The Ifakara gauged record has again been used to determine rainfall depths as
inputs to CROPWAT calculations given the record length and its representativeness
of rainfall characteristics in the Kilombero Basin. Both the average and a variety of
percentage likelihood rainfall series have been derived.
The data has been derived in 36 decads across the year. A decad would normally
be 10 days in length, but for ease of calculations and organization of data, it is
assumed that each month in the year has 31 days and therefore includes two decads
of 10 days and a third of 11 days.
Table 2.7 summarizes the average and 80% likely rainfall depths for the Kisegese
scheme. The 80% likely rainfall depths are calculated from the Ifakara record as the
depth exceeded 80% of the time. Figure 2.11 presents a range of decadal rainfall
series derived over a range of percentage exceedance values.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-20
Table 2.7: Rainfall Derivation at Ifakara
Ifakara Ifakara
Average Ifakara 80% Average Ifakara 80%
Decad (mm) Likely (mm) Decad (mm) Likely (mm)
1 66.8 10.64 19 4.4 0.00
2 68.5 15.20 20 3.2 0.00
3 62.5 17.44 21 2.4 0.00
4 64.5 6.96 22 2.3 0.00
5 69.5 15.92 23 2.4 0.00
6 50.4 1.68 24 1.2 0.00
7 80.4 25.82 25 2.5 0.00
8 92.0 27.44 26 2.0 0.00
9 132.5 57.32 27 2.4 0.00
10 133.9 65.38 28 3.2 0.00
11 137.3 57.04 29 4.8 0.00
12 107.5 39.34 30 7.3 0.00
13 66.7 16.24 31 12.3 0.00
14 33.8 4.92 32 21.1 0.00
15 24.7 0.00 33 26.5 0.00
16 8.7 0.00 34 34.2 0.00
17 6.0 0.00 35 46.6 5.98
18 3.9 0.00 36 61.2 8.12
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-21
Figure 2.11: Percentage Exceedance Rainfall from Ifakara
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-22
Figure 2.12: Kisegese Irrigation Scheme Drainage Sub-Catchments
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-23
Table 2.8: Types of Flow in Derivation of the Time of Concentration
Flow Typical Area (km2) Method of Calculation
Sheet < 0.3 Kinematic (Tc1 )
Shallow 0.5 to 1 Kirpich (Tc2 )
Channel >1 Channel (Tc3 )
Depending on the size of the sub-catchment, all three components may contribute to
the Tc for the catchment. The contributions from each component are derived in
accordance with the procedures outlined in the following sections.
2.7.2.2 Sheet Flow
Time of concentration for sheet flow component, Tc1, is given by the Kinematic
equation (Hoggan 1989):
where:
Tc1 = Time of concentration for sheet flow component (hr)
n = Manning’s coefficient
L1 = Longest flow path (m) up to 100 meters
S = Slope
I = Rainfall intensity (mm)
Sheet flow is considered applicable to the first 100 m of the flow path, after which the
shallow flow process is the controlling mechanism. A conservative estimate of
Manning’s n was used in the calculation (0.05). Sensitivity analysis was undertaken
which showed results were relatively insensitive to the Manning’s roughness
coefficient.
2.7.2.3 Shallow Flow
One of the most common empirical equations to calculate time of concentration is
Kirpich’s equation (Smedema and Rycroft 1983). The derivation is:
.
.
0.01953
60
where:
Tc2 = Time of concentration for shallow flow component (hr)
L2 = Length of the longest channel ( = LT - 100) (m) up to 400 meters
LT = Longest total flow path (m)
S = The overall catchment slope in m/m = H/L
H = The difference in elevation between the most remote point in the
catchment and the outlet
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-24
Shallow flow is considered applicable to any sub-catchment with a flow path greater
than 100 m. The effective length of channel is taken as the total flow path length less
100 m (the first 100 m is attributable to the sheet flow component) up to a maximum
channel length of 400 m. For those sub-catchments with a flow path longer than 500
m, the channel flow process is the controlling mechanism for the remaining section.
2.7.2.4 Channel Flow
Time of concentration in open channels has been determined from channel
conveyance characteristics for different reaches. Since flow in a channel is confined
within a limited cross sectional area, it is faster than sheet flow or shallow flow. An
indicative velocity guide for channel watercourses is provided in Design of Small
Dams (USBR 1987) as shown in Table 2.9.
Table 2.9: Average Velocity in Channel Watercourses
Average Slope of Channel from
Farthest Point to Outlet (%) Average Velocity (m/s)
1-2 0.6
2-4 0.9
4-6 1.2
6 - 10 1.5
Channel flow is considered applicable to any sub-catchment with a flow path greater
than 500 m. The effective length of channel is taken as the total flow path length less
500 m (the first 500 m is covered by the sheet flow and shallow flow components).
The time of concentration for the channel flow component is determined as:
3600
where:
Tc3 = Time of concentration for channel flow component (hr)
L3 = Stream length ( = LT - 500) (m)
v = Velocity (dependent on slope as Table 2.9) (m/s)
where:
Tc1, Tc2, and Tc3 = Time of concentration for sheet flow, shallow flow and
channel flow.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-25
The SCS synthetic rainfall distributions were to aid in the calculation of peak
discharges for use in the design of engineering projects across the United States.
The authors used available precipitation gauge data to compute and model the
intensity for various sub-daily precipitation rates. Four different SCS synthetic rainfall
distributions were developed to cover all regions in the United States, where the peak
intensity changes depending on the dominant climate in the region.
For this study, the Type II synthetic rainfall distribution was chosen, which is the most
intense SCS distribution and is appropriate to represent potential runoff from large,
convective thunderstorms that typically occur in the study area. Figure 2.13 shows
the 24-hour SCS Type II hyetograph. The 24-hour SCS Type II hyetograph was
applied to the computed 10- and 25-year precipitation depths.
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00
Duration (hours)
where:
R = accumulated direct runoff (mm)
P = accumulated rainfall (mm)
Ia = initial abstraction including surface storage and infiltration prior
to runoff (mm)
S = potential maximum retention (mm)
I = Rainfall intensity (mm)
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-26
0.2
The value of S is determined from the SCS curve number, CN, as:
25400 254
For the design of the cross-drainage structures, the peak discharge is required. To
compute the peak discharge for each sub-catchment, the SCS Dimensionless Unit
Hydrograph technique was implemented in the HEC-HMS modeling software using
the excess precipitation computed using the SCS CN method for each sub-
catchment. The SCS Dimensionless Unit Hydrograph uses the time of concentration
and runoff calculations described in the SCS Curve Number methodology to develop
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-27
a hydrograph shape for runoff produced by the SCS CN calculations. The storm
hydrograph given by the SCS Type II synthetic storm is developed by adding up
scaled individual unit hydrographs computed for each interval of precipitation
occurring across the 24-hour distribution. For this analysis, the peak discharge was
computed using the HEC-HMS modeling software at a 15-minute time interval, which
means that the 24-hour storm total was broken up into 96 intervals using the SCS
Type II synthetic hyetograph, with each interval producing an individual unit
hydrograph with the total volume equivalent to the runoff during that interval. The
peak discharge was calculated by overlapping all 96 hydrographs to produce a
composite hydrograph for the entire 24-hour period. The parameters and peak runoff
for the 10- and 25-year events are summarized in Table 2.11.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-28
Table 2.11: Kisegese Drainage Sub-Catchment Peak Flows
Length of Time of Concentration (hr)1 Peak Discharge (m3/s)
Sub- Drainage Area Flow Path AMC III
catchment (ha) (m) Slope (%) CN Tc1 Tc2 Tc3 Tc 10-year Storm 25-year Storm
1 134.2 1,887 0.8 75 1.55 0.21 0.89 2.65 6.9 9.1
2 126.6 1,755 1.2 75 1.37 0.18 0.66 2.21 7.5 9.8
3 355 4,091 0.9 75 1.5 0.2 2.18 3.87 13.9 18.3
4 102.9 1,691 1.1 75 1.41 0.19 0.65 2.25 6.0 7.9
5 47.6 1,055 0.8 85 1.55 0.21 0.36 2.12 3.9 4.9
6 165.9 2,299 0.2 85 2.35 0.36 1.81 4.52 7.9 9.9
7 95.8 1,728 0.3 85 2.08 0.31 1.12 3.51 5.5 6.9
8 150.7 1,643 0.3 84 2.08 0.31 1.05 3.43 8.6 10.8
9 123.3 2,199 0.3 84 2.08 0.31 1.55 3.94 6.3 7.9
10 68.8 1,621 0.4 84 1.91 0.27 0.94 3.12 4.2 5.3
11 129 2,217 0.3 85 2.08 0.31 1.57 3.96 6.8 8.5
12 420 4,081 0.3 85 2.08 0.31 3.28 5.66 16.9 21.1
13 456.5 4,970 0.2 85 2.35 0.36 4.49 7.2 15.3 19.1
14 1,304.0 6,862 0.3 85 2.08 0.31 5.82 8.21 39.4 49.3
15 634 8,291 0.2 85 2.35 0.36 7.89 10.54 15.8 19.7
1Tc values calculated based on 10-year storm depth.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 2-29
3 HYDROMETRIC DATA ANALYSIS
3.1 Introduction
All hydrometric data made available for the study has been considered in relation to
deriving suitable estimates of long-term flow series and flood flows for the Kisegese
irrigation scheme.
Derivation of long-term flow series has been undertaken to understand the water
resource available for the irrigation scheme, specifically the average and 80% likely
flows.
Flood flows have been derived to undertake flood modeling for the irrigation scheme
area to determine flood extents and the need for flood mitigation.
Details of analysis and calculations are presented below.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-1
3.3 IRRIP2 Hydrometric Data Stations
3.3.1 Overview
Spot gauging of river flows upstream of the proposed irrigation schemes in the
Kilombero basin has been undertaken to increase the understanding of potential
water resources available. The gauging used an Acoustic Doppler Current Profiler
(ADCP), which is a hydro-acoustic current meter that measures flow velocities and
channel cross section using the Doppler effect of sound waves reflected by particles
in the water. This allows for a standard velocity-area calculation to be undertaken to
give discharge. Figure 3.1 presents locations where spot gauging has taken place
within the catchment, including at the Mbingu gauging station on the Ruipa River.
At the time of writing, eight spot gaugings had been recorded at the Mbingu gauge –
four historical recordings and four undertaken by the IRRIP2 field teams. Table 3.1
summarizes the spot gaugings available for analysis, with the IRRIP2 measurements
highlighted with bold text. In addition to these measurements, limited spot gaugings
have been undertaken upstream of the Mbingu gauge.
Table 3.1: Mbingu Spot Gaugings
Date Stage (m) Discharge (m³/s)
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-2
Gauging of the Ruipa River at the Mbingu gauge is unlikely to enhance the analysis
undertaken for the Kisegese scheme in the future unless the water level
measurement at the gauge is extended. Spot gauging at the scheme intake location
is considered useful and should continue to build a reasonable dataset from which
comparison with flows derived from this study can be undertaken as a validation.
3.3.3 Continuing Hydrometric Monitoring
Spot gauging will continue for the remaining duration of IRRIP2. This will enable
increasingly detailed analysis to be undertaken and used for any future detailed
design requirements using the Ruipa River as a source of water.
A notable feature of a number of the available water level records is a step change
occurring, normally in the early 1970s, where the range of water levels and maximum
water levels recorded are much reduced. This is sometimes reflected in the flow
data and at other times not. Figure 3.3 shows an example of this occurring at gauge
1KB15, where in 1972 there is an obvious step change in the water level series.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-3
Examination of the records suggests this is potentially a scaling error, where an
assumption has been made that the water levels have been recorded in feet rather
than meters, resulting in the step change. No periods of overlap were available to
definitively prove this is the case so an approach considering level duration curves
has been applied to correct the stepped level record where possible.
After initial assessment of the gauged data available for analysis, a red, amber and
green scoring approach was used to classify gauges into categories according to the
quality of the data available and its relevance to the study at the Kisegese irrigation
scheme. The classifications are included in Appendix C, with gauges classified as
red excluded from further analysis.
The application of the red, amber and green classification is applied in the context of
the study and the available data. Gauges highlighted amber are considered to be of
questionable quality. Therefore, the data is held in lower confidence and should only
be used for verification purposes, if required. The study relies on data from gauges
highlighted green, where some degree of confidence is held in the data record due to
the length of record, completeness of record, lack of step changes and checks of flow
statistics. In general, there is a significant amount of data missing from the available
records, with limited stage-discharge data provided. It was therefore difficult to
definitively assess the quality of the data. Best use of available datasets has been
made, with a level of trust applied to the records provided.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-4
to determine which flow series were of sufficient quality and duration to take forward
for infilling in order to derive long records for analysis.
The second stage of the process included exploring the use of appropriate infilling
techniques to ensure long duration, consistent and accurate flow records were
available for use in deriving available water resources in the Kisegese scheme inflow
sub-catchments and also to derive appropriate flood flows for hydraulic modeling and
flood risk assessment.
3.4.2.2 Methodology
The methodology used to derive long-term flow records across the catchment utilizes
recorded flow data as much as possible while considering data quality. It is noted
that flow records have decreased in quality over the recent period and therefore there
is a limited data set from which to derive accurate flows from the last 20 years.
Long-term climate data with sufficient spatial coverage was not available during the
timescale of the project and therefore development of a rainfall runoff model has not
been attempted. Given there is a reasonable coverage of recorded flow series it was
preferred to use this rather than any climatic data sets available for rainfall runoff
modeling. There are benefits to using a rainfall-driven hydrological model to
determine long-term flow series though it was considered more beneficial to
incorporate gauged data in this analysis. It is noted that undertaking rainfall-driven
hydrological modeling can result in flow series that are unrepresentative unless
detailed calibration is undertaken. This can produce unrealistic low-flow series and
overestimated flood peaks, both of which are of paramount importance to the study.
The chosen methodology was applied as follows:
Derive robust flow series at the gauges identified as being appropriate for use,
using available rating curves and other analyses to extend the duration of
datasets where possible.
Assess the coverage of flow series across the catchment to determine which
gauges have long-term flow series and therefore require minimal infilling and
the gauges to be used for infilling.
Infill the data series from similar gauges through assessment of relationships
between gauged records, flow statistics and catchment flow duration curves
(FDCs).
Where a long-term data series is not available for a specific catchment of
interest (i.e., those feeding the irrigation scheme), scale the derived long-term
series to the target catchment considering catchment-specific characteristics
such as area, mean flow and hydrograph shape.
Use short-term records local to the scheme to ensure the derived flow series is
fitted as closely as possible to catchment characteristics, therefore mimicking
local climatic characteristics.
It is noted that certain assumptions must be made when scaling a flow series from
one catchment to the other, but by using short-term local gauged data, the scaled
flow series can be tuned to the target catchment, or one close by, through scaling to
the FDC shape. This method ensures the flow series is representative of the
catchment runoff characteristics and local rainfall patterns.
3.4.2.3 Gauge Data Analysis
As noted above, a significant number of water level records include a step change in
the early 1970s. It is unclear why this occurs, but given it is prevalent in a large
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-5
number of records across the catchment a method was devised to produce an
estimate of a consistent record at these locations. A method was derived where level
duration curves were used to remove the step in the data to increase record lengths
where appropriate.
No examples were found where an overlap between a stepped level series and a
non-stepped level series existed at the same location and therefore the derivation of
a direct scaling factor was not possible. To ensure a consistent scaling factor could
be used with the stepped data, a method was devised which takes into account water
levels across the entire gauged range to restrict any bias of low or high flows.
First, the record was split into pre- and post-step series and level duration curves
derived from each series. A level duration curve shows the percentage of time a
level is exceeded in a given record. Figure 3.4 shows the comparison of pre- and
post-step level duration curves at gauge 1KB15. The mean factor between level
percentiles across the range of percentiles was calculated and applied to the post
step level record. This process was applied to level data series wherever records
were deemed appropriate for use.
Where appropriate, relationships between water levels and flows have been derived
to extend or test record periods. This has enabled the data series provided to be
used to enhance the analyses undertaken. For example, where a concurrent level
and flow series was available, but the level series is longer, the relationship between
level and flow during the concurrent period has been applied to the extended level
record, allowing the flow record also to be extended. The assumption is made that
the relationship between level and flow is applicable to the full level record.
In addition to this, several occurrences were noted where rating curves were
available or derived and used with water level data series to extend the available flow
series record.
Appendix D provides a summary of analyses undertaken across all the gauges
deemed suitable for use.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-6
3.4.3 Conclusions
Analysis of the available hydrometric data results in the following observations:
Several robust, long-term gauged records are available from within the
Kilombero basin from which representative long-term flow series and flood
flows can be derived.
Several of the gauged records available within the Kilombero basin are of poor
quality and therefore have been omitted from further analysis.
Several gauges are available throughout the catchment that can be used to
tune long-term flow series to catchments with short records.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-7
Figure 3.5: Summary of Flow Data Records across the Kilombero Catchment, 1956-2014
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-8
3.5.2.1 Gauged Data Relationships
Examination of the flow series available shows a reasonable similarity in flow
patterns across the catchment, with defined wet and dry seasons, and therefore
potential relationships between flows at different gauges exist across the catchment.
The relationship between flows at each of the gauges was tested by plotting the daily
flow series at each location against each other to form scatter plots. If a significant
relationship is present, the scatter plot should show a reasonably tight cluster of
points along a line, to which a mathematical relationship could be applied. The
goodness of fit has been tested using best-fit relationships and r-squared values.
The majority of the relationships show low r-squared values, less than 0.5, with only
four comparisons showing r-squared values greater than 0.5. The highest r-squared
value comes from the comparison of flows at gauges 1KB17 (Kilombero at Swero)
and 1KB27 (Ruipa at Mbingu), with a value of 0.64. This is still considered a poor fit.
A reasonable comparison would only be used if the gauged records showed a good
fit, with an r-squared value greater than 0.8.
The results of the analysis show that robust relationships between gauged flows are
difficult to achieve across the catchment due to its size. Therefore, the method has
not been taken forward further to derive long-term flow series. It is considered that
comparison of FDCs is a more appropriate way to compare catchment flows in this
case as their shape shows how the catchment responds to rainfall, and therefore
similar FDC shapes show similar catchment responses.
Full details of the analyses are presented in Appendix B.
3.5.2.2 Flow Scaling and Infilling
Examination of available flow records was undertaken to determine which record
series could be used for infilling to produce a consistent long-term flow record.
Long-term records are noted at gauges 1KB8 (Mpanga at Mpanga Mission), 1KB9
(Mnyera at Taveta Mission), 1KB14 (Lumemo at Kiburubutu), 1KB18 (Ruhudji at
Njombe) and 1KB28 (Kihansi at Lugoda). Further analysis of these gauges was
undertaken to determine how and where long-term flow series could be derived.
Gauges 1KB18 (Ruhudji at Njombe), 1KB14 (Lumemo at Kiburubutu) and 1KB28
(Kihansi at Lugoda) were all considered for use in the infilling process, but
examination of available records and catchment characteristics ruled them out of the
process:
Gauge 1KB18 (Ruhudji at Njombe) was rejected based on its FDC shape which
is considered to be abnormal in comparison with other gauges in the Kilombero
basin.
Gauge 1KB28 (Kihansi at Lugoda) is located downstream of the Kihansi
hydropower scheme where releases are regulated and therefore no longer
natural. Infilling with the record was not considered appropriate, though using it
as a check record may be useful.
The record at gauge 1KB14 (Lumemo at Kiburubutu) is considered to be
robust, but is not as long as the records at 1KB8 (Mpanga at Mpanga Mission)
and 1KB9 (Mnyera at Taveta Mission). Gauge 1KB14 also has a much smaller
catchment area.
Examination of available data records shows gauge 1KB8 (Mpanga at Mpanga
Mission) and 1KB9 (Mnyera at Taveta Mission) both have reasonably long records,
with gauge 1KB9 (Mnyera at Taveta Mission) covering data gaps in the record at
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-9
1KB8 (Mpanga at Mpanga Mission). The two catchments are reasonably closely
matched in terms of scale and their FDCs show a reasonably similar profile (i.e., the
gradient of the FDCs is similar which suggests a similar response to rainfall within
each catchment). The catchments are also adjacent to one another so they are likely
to experience similar rainfall patterns.
It was noted that neither gauge 1KB8 (Mpanga at Mpanga Mission) or 1KB9 (Mnyera
at Taveta Mission) has a record beyond 2006. Gauge 1KB32 (Kihansi at Lutaki) has
reasonable data from this period and therefore flow from this location was scaled to
the 1KB8 (Mpanga at Mpanga Mission) catchment to increase the record up to the
end of 2014.
Table 3.2 and Figure 3.6 present a comparison of the original 1KB8 (Mpanga at
Mpanga Mission) series with the infilled series. It is noted that there is a slight
difference in flows between the two series during mid flows, which is manifested in a
slightly higher mean flow from the infilled series. The low and high flows match well
and therefore the series is considered appropriate for use.
The long-term and complete flow series which has been derived at gauge 1KB8
(Mpanga at Mpanga Mission) can be tuned to short-term gauged records which are
more representative of inflows to the Kisegese irrigation scheme through scaling to
their FDC shape. The process is described below.
Full details of the infilling process are presented in Appendix B.2.
Table 3.2: Comparison of Flow Statistics at 1KB8 (Mpanga at Mpanga
Mission)
Flow Statistic 1KB8 1KB8 Infilled
Qmin 8.86 8.86
Qmax 345.27 345.27
Qmean 48.77 52.36
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-10
Figure 3.6: FDC Comparison between 1KB8 and 1KB8 after Infilling
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-11
Figure 3.7: Kisegese Analysis Gauge Locations
To determine which of the gauged catchments show the closest similarity to the
target catchment, FDC comparison was undertaken using flows standardized by
Qmean to compare how the catchments react to rainfall. In addition to this,
catchment area, Qmean per unit area, and record length were also considered along
with any catchment specific information available.
The comparison of data from each of the gauges shows that 1KB14 (Lumemo at
Kiburubutu) is the most appropriate gauge to provide an accurate representation of
the FDC shape at the Kisegese scheme.
It is noted that gauge 1KB27 (Ruipa at Mbingu) is adjacent to the Kisegese scheme
and includes the catchments of the Chiwachiwa and Ruipa rivers. Given its location,
the FDC from gauge 1KB27 (Ruipa at Mbingu) was immediately considered to use in
scaling the long-term flow series derived at 1KB8 (Mpanga at Mpanga Mission) to a
catchment-specific flow regime at the Kisegese scheme. In addition to this, ADCP
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-12
gauging has been undertaken at this location. It is noted however, that the flow
series at this gauge is based on a rating curve that has been derived using eight data
points only. Therefore, there is uncertainty in the flow series, particularly the high
flow end. This is manifested in the catchment FDC where the shape at the high flow
end of the FDC is an abnormal shape. It is also noted that the Qmean per unit area
for the gauged record is around half that of other similarly scaled catchments in the
Kilombero basin. Given the uncertainty in the rating curve and consequently the flow
record at this gauge, it is considered that the FDC is underestimating mid to low flows
in the catchment and is not representing high flows accurately. This gauge has not
been considered further in the analysis.
Full details of the selection process are presented in Appendix B.3.
3.5.4 Calculation of Irrigation Scheme Flows
To derive a reliable flow series at the Kisegese weir offtake locations the infilled
series derived at gauge 1KB8 (Mpanga at Mpanga Mission) has been scaled to the
FDC shape at gauge 1KB14 (Lumemo at Kiburubutu), then scaled by area to each of
the intake catchments. The Kisegese scheme includes two weirs as shown on
Figure 1.1.
The available discharge is calculated as the flow that is exceeded 80 percent of the
time throughout the year based on the historical record, or the 80 percent reliable
flow (Q80). This means that each month is broken down into three 10-day periods
(decads), or two 10-day periods and the remaining days left in the month. The 80-
percent likely flow was calculated for each decad at each weir location. Table 3.3
and Table 3.4 summarize the calculated flows.
Table 3.3: Q80 Discharge at the Chiwachiwa Weir Offtake
Month Decad Flow (m³/s) Month Decad Flow
1 2.319 19 1.915
Jan 2 2.308 Jul 20 1.783
3 2.113 21 1.681
4 2.615 22 1.618
Feb 5 2.680 Aug 23 1.543
6 2.710 24 1.433
7 3.107 25 1.411
Mar 8 3.160 Sep 26 1.354
9 3.871 27 1.353
10 4.412 28 1.327
Apr 11 5.681 Oct 29 1.270
12 5.853 30 1.248
13 4.954 31 1.201
May 14 3.614 Nov 32 1.184
15 2.970 33 1.271
16 2.570 34 1.346
Jun 17 2.251 Dec 35 1.546
18 1.939 36 1.914
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-13
Table 3.4: Q80 Discharge at the Ruipa Weir Offtake
Month Decad Flow (m³/s) Month Decad Flow
1 8.262 19 6.823
Jan 2 8.223 Jul 20 6.351
3 7.528 21 5.988
4 9.315 22 5.764
Feb 5 9.549 Aug 23 5.498
6 9.654 24 5.106
7 11.069 25 5.028
Mar 8 11.257 Sep 26 4.824
9 13.791 27 4.822
10 15.720 28 4.729
Apr 11 20.238 Oct 29 4.524
12 20.851 30 4.448
13 17.652 31 4.279
May 14 12.878 Nov 32 4.219
15 10.583 33 4.527
16 9.157 34 4.797
Jun 17 8.021 Dec 35 5.508
18 6.906 36 6.818
where:
AREA = Catchment area (km2)
MAF = Mean annual flood (m3/s)
Gauges selected to derive flows were chosen based on record length, since a
reasonable record length is required to ensure the process of deriving flood flows is
statistically significant.
As discussed above, several of the gauged records available have been rejected
from the analysis given they are held in low confidence, are of short record duration,
or come from catchments which are small in comparison to the catchments of
interest to the flood study, or are impacted by hydropower generation.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-14
Gauges 1KB8 (Mpanga at Mpanga Mission) and 1KB9 (Mnyera at Taveta Mission)
have been used to derive flood flows given their record length and that significant
confidence is held in the records available here. It is noted that the infilled record at
1KB8 has not been used in derivation of flood flows given that the scaling undertaken
is not likely to represent the very highest flows accurately. Relying on the gauged
record only reduces uncertainty in the analysis.
Table 3.5 compares the mean annual flood flows calculated from the IHH equation
and the gauged AMAX series at Mpanga Mission and Taveta Mission. Table 3.6
includes growth factors calculated from the gauged data.
The data has been derived to allow calculation of flood flows for ungauged
catchments depending on whether they show similar characteristics to the Mpanga
Mission catchment or the Taveta Mission catchment. To calculate flood flows the
selected growth factors are applied to the MAF for the ungauged catchment
calculated using the IHH equation. Where the Mpanga Mission growth factors are
favored, a 27% increase is applied to the calculated MAF given the difference
between it and the MAF derived from the AMAX series at the gauge presented in
Table 3.5.
Table 3.5: Comparison of Flow Statistics at 1KB8 (Mpanga at Mpanga
Mission) and 1KB9 (Mnyera at Taveta Mission)
1KB8 1KB9
(m³/s) (m³/s)
MAF IHH equation 173.52 249.75
MAF gauged record 220.73 248.28
It is noted that at gauge 1KB8 (Mpanga at Mpanga Mission), the mean annual flood
calculated from the data record is 27% larger than that calculated using the IHH
equation. At gauge 1KB9 (Mnyera at Taveta Mission), the values are almost
identical. Given the discrepancy, where the 1KB8 (Mpanga at Mpanga Mission)
growth factors have been applied, a 27% increase to the mean annual flood value
calculated using the IHH equation has been applied. Where the 1KB9 (Mnyera at
Taveta Mission) growth factors have been applied no adjustment has been made.
The growth factors derived from gauge 1KB8 (Mpanga at Mpanga Mission) are larger
than those derived from gauge 1KB9 (Mnyera at Taveta Mission) which is due to
gauge 1KB8 (Mpanga at Mpanga Mission) having a large area located on the
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-15
escarpment whereas 1KB9 is much flatter. The inflows to the Kisegese flood model
are more similar to the 1KB8 (Mpanga at Mpanga Mission) catchment and therefore
all inflows use the growth factors derived from it.
Use of the IHH empirical equation to derive MAF flows for ungauged catchments
contains some inherent uncertainty given it is a generalized equation used for various
types and sizes of catchments. The comparison presented above shows that for the
Taveta Mission gauge the MAF is predicted very accurately by the empirical formula,
whereas at Mpanga Mission it is significantly underestimating the MAF. Best effort
has been made to match ungauged catchments used in modeling to either the
Taveta Mission catchment or the Mpanga Mission catchment, and applying the uplift
in MAF where appropriate.
For the Kisegese scheme, the flood model has been schematized with three inflow
locations. Catchment areas have been measured to the confluence of the
Chiwachiwa and Ruipa rivers to ensure the flows entering the model represent the
whole area under consideration rather than that upstream of the inflow points. Table
3.7 presents peak flows derived for a range of return periods that were considered for
flood modeling.
Table 3.7: Kisegese Flood Modeling Peak Flows (m³/s)
Chiwachiwa Ruipa Londo
Inflow (Inflow 1) (Inflow 2) (Inflow 3)
Catchment Area (km²) 439 761 192
Calculated MAF 62.98 86.55 39.05
Scaled MAF 79.98 109.92 49.59
Q2 75.31 103.50 46.69
Q5 99.35 136.54 61.60
Q10 115.27 158.42 71.47
Q25 135.38 186.06 83.94
Q50 150.30 206.56 93.19
Q100 165.11 226.92 102.37
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-16
The events were standardized by their peak flow and phase-shifted so their peak
occurred concurrently. The events were then averaged to give a site-specific
hydrograph shape that could be scaled by peak flow for any event and used in the
flood modeling process. Figure 3.8 shows an example of the average hydrograph
shape (red dashed line) derived from a number of selected events (grey). This
average hydrograph allows the flood modeling result to be representative of the
flooding patterns at the proposed irrigation scheme.
For the Kisegese scheme, all the inflow catchments are relatively small compared to
gauged catchments 1KB8 and 1KB9 and therefore the hydrograph shape derived
from gauge 1KB14 (Lumemo at Kiburubutu) has been used. The hydrograph is 10
days long, with the peak occurring after 5 days.
3.6.2 Observations
The critical element of deriving a long-term flow record and flood flows for the
watercourses near the Kisegese scheme is the reliability of gauged records used in
the analysis.
The basis of the analysis has been undertaken using long-term gauged records from
1KB8 (Mpanga at Mpanga Mission) and 1KB9 (Mnyera at Taveta) where the data is
considered to be robust and relatively complete compared to other locations across
the catchment. The infilling process relies on gauge 1KB9 (Mnyera at Taveta), but
given it is a reasonably similar catchment to 1KB8 (Mpanga at Mpanga Mission) the
infilling is held in reasonably high confidence. Comparison of baseline and infilled
FDCs at 1BK8 (Mpanga at Mpanga Mission) further reinforces that the infilled record
is reasonable.
Derivation of flood flows has been undertaken using the standard Tanzanian IHH
methodology to derive catchment MAF values and is therefore considered to be
robust and appropriate for the Kilombero catchment. Growth curves to be applied to
the MAF have been derived using a standard methodology of fitting an appropriate
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-17
distribution to AMAX record. This again relies heavily on gauge 1KB8 (Mpanga at
Mpanga Mission) to provide a long-term and reliable AMAX series from which to
derive the growth curve factors.
The derivation of hydrograph shapes for use in flood modeling was a difficult
procedure given the general trend in flows over a year in the Kilombero catchment.
The peak flows experienced in a year tend to build over a period of months, building
on past events, so choosing single events on which to derive hydrograph shapes was
challenging. Best efforts have been made to choose individual events from the
available hydrograph records, but it is recognized that there may be some uncertainty
in their shape and duration.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 3-18
4 CLIMATE CHANGE ASSESSMENT
This section presents an overview of trends and projections of climate change with
respect to the water resources available for the proposed irrigation scheme.
Potential impacts of such changes are presented. FS Annex J: Engineering provides
additional discussion on the application of climate change to the irrigation designs,
including mitigation measures and adaptation strategies to mitigate the impacts.
4.1 Temperature
In Tanzania, mean annual temperature has increased by 1.0 °C since 1960, with an
average rate of 0.23 °C per decade. The mean annual temperature is projected to
increase by 1.0 to 2.7 °C by the 2060s, and 1.5 to 4.5 °C by the 2090s depending on
the emissions scenario as shown in Figure 4.1. The figure shows recent trends in
annual and seasonal mean temperature for the recent past and projected future. All
values shown are anomalies, relative to the 1970-1999 mean climate. The colored
lines from 2006 onwards show the median (solid line) and range (shading) of several
global climate model (GCMs) projections of climate under three different emission
(A2, A1B, B1) scenarios (McSweeney et al. 2010)1. The A2 emissions scenario is
the continuation of current rising emissions trends. The A1B is the gradual reduction
from current emissions trends beginning around mid-century. The B1 emissions
scenario has substantial emissions reductions.
Similar increasing trends in temperature were reported in a recent report for the
Tanzanian Ministry for Agriculture, Food Security, and Cooperatives (MAFC). The
Intergovernmental Panel on Climate Change (IPCC) report for Africa also predicts a
warming trend across Africa, and a likely mean annual temperature rise of over 2 °C
by 2100 (MAFC 2014; IPCC 2014). Studies also agree that the rise in temperature
will be greater during cooler months (June to August) than warmer ones (December
to February) (MAFC 2014; Wambura et al. 2014).
4.2 Precipitation
From a global analysis using three GCMs, Hagemann et al. (2013) projects an
increase in precipitation of 200 mm per year in the area of the Rufiji Basin by the end
of the 21st century when compared to annual average precipitation from 1971 to 2000
(Figure 4.2).
1 The methods used to determine observed and projected temperature and precipitation trends are outlined in the
United Nations Development Program Climate Change Country Profile Documentation by McSweeney et al. (2010),
accessible at: http://www.geog.ox.ac.uk/research/climate/projects/undp-cp/UNDPCCCP_documentation.pdf
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-1
Figure 4.2: Projected Changes in Annual Precipitation in 2071-2100
Compared to Baseline (1971-2000) Conditions (mm/year)
In Tanzania, national average annual rainfall varies from below 500 mm to 2,500
mm, depending on altitude and climatic zone. Precipitation, in general, is highly
variable for the region. Observations of rainfall over Tanzania have shown
statistically significant decreasing trends for mean annual rainfall at an average rate
of 2.8 mm per month (3.3%) per decade (McSweeney et al. 2010).
Converse to observed trends, GCM projections of future mean rainfall indicate an
increase in annual rainfall for the country as a whole. The study by McSweeney
(2010) found GCM ensemble precipitation projections of -4 to +30% by the 2090s
relative to the 1970-1999 mean precipitation, with variability based on the emissions
scenario as shown in Figure 4.3 (McSweeney et al. 2010).
The Kilombero District experiences a single rainfall pattern (unimodal rainfall regime)
from December to April unlike other parts of Tanzania (MAFC 2014; Wambura et al.
2014). In contrast to projections for the country, the Kilombero District could
experience annual rainfall decreases of 5 to 15%. This range of decrease in
precipitation is also likely to occur in other parts of Tanzania including the central,
western, southern, southwestern, and eastern zones (MAFC 2014; Matari 2008).
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-2
Determining the impact of climate change patterns on rainfall patterns is highly
uncertain – projections vary widely between seasons, regions, and rainfall regimes
across the country and the Kilombero District. Changes in rainfall patterns will vary
depending on current climate and geography; not all climactic zones will experience
the same changes.
4.3 Streamflow
In Africa, longer dry seasons and more uncertain rainfall are attributed to climate
change. It is anticipated that there will be a 10-20% decrease in river flows over
some regions at mid-latitudes and in the dry tropics, some of which are presently
water stressed areas (IPCC 2007; Kashaigili et al. 2007). For the Great Ruaha
River, a tributary to the Rufiji Basin, Kashaigili et al. performed a trend analysis on
mean annual runoff. The analysis did not reveal any significant trend at the 95%
level of confidence, but a declining trend in low flows was detected. The modification
of land use and land cover has resulted in changes to flow regimes in the catchment
(Kashaigili 2008).
There are limited studies that evaluate and quantify observed and projected trends in
hydrologic or streamflow responses for the region of interest.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-3
Figure 4.4: Projected Global Multi-model Averages of Precipitation Trends
under an A1B Emissions Scenario
a) Number of days with precipitation
greater than 10 mm
Note: Stippled regions correspond to areas where at least five of the nine averaged GCMs concur in
determining that the projected change is statistically significant. Source: Tebaldi, 2006
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-4
4.5 Impacts to Irrigation Schemes
An increase in temperature will affect the proposed irrigation scheme. Table 1 of the
MAFC report summarizes the potential impacts of an increase in temperature on crop
production and agriculture (reproduced as Figure 4.5). Stakeholders provided input
to the most significant impacts, which were primarily water availability, proliferation of
pests and disease, as well as evaporative impacts to soil moisture and soil fertility,
especially for food crops, such as wheat, maize, sorghum, and rice. Irrigation layouts
and/or types of crop may have to change in order to adapt to changing temperature
conditions (MAFC 2014).
Even though uncertainty exists as to the direction, extent and location of changes in
future precipitation, it can affect proposed irrigation schemes. Similarly, the MAFC
report provides a summary of the response that stakeholders provided given three
potential rainfall patterns (Table 2, reproduced as Figure 4.6), applicable to the
design of irrigations schemes. An increase in rainfall may cause flooding, damage to
infrastructure and crops, and degradation of soil. A decrease or variability in rainfall
may have a significant impact on crop productivity, and may also impact water
availability both annually and seasonally. Although an increase in total annual rainfall
may seem to improve crop productivity, total rainfall could become more
concentrated as less frequent, more extreme events, which could damage crops,
cause erosion, etc.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-5
Figure 4.6: Potential Impacts of Precipitation Changes on Crop Productivity
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 4-6
5 CONCLUSIONS AND RECOMMENDATIONS
These analyses have used a large number of datasets over a wide area of the
Kilombero catchment and beyond.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 5-1
Data from gauge 1KB32 was required to infill the series from 2006 to 2014, as
no other reliable data series were available. Data from the 1KB32 gauge is
thought to be reasonably appropriate for infilling purposes.
The analysis undertaken has produced an unbroken daily flow series from 1958
to 2014 with analysis showing the infilled series is reasonably similar to the
gauged data series at 1KB8 before infilling and therefore considered robust.
Chiwachiwa and Kisegese intake-specific flow series have been derived by
scaling the long-term daily flow series from gauge 1KB8 to the FDC derived
from the gauged record at gauge 1KB14, then scaling to each catchment by
catchment area. Gauge 1KB14 is considered an appropriate analogue for the
Kisegese catchments based on catchment area and topography, with the upper
catchment located on the upper escarpment.
Scaling to gauge 1KB14 FDC is appropriate and required, as the smaller
catchments which include a significant portion of their area on the escarpment
show a different response to rainfall (read ‘flashier’ or steeper gradient FDC)
than the larger catchments in the Kilombero basin.
The Tanzanian IHH equation has been used to calculate MAF values for use in
flood modeling based on catchment area. Use of the Tanzanian-specific
equation increases confidence in the analysis.
The flood flows derived for hydraulic modeling and flood risk assessment at
Kisegese use the long-term gauged data from gauge 1KB8. Gauge 1KB8 has
a long record in comparison with many of the gauges within the catchment,
which increases confidence in derivation of growth curves that were applied to
calculated MAFs to determine flood flows.
Flood hydrographs have been derived using observed data records rather than
a theoretical shape, which adds confidence to flood modeling as gauged
records capture the long duration events observed within the Kilombero basin,
ensuring large volume events are included in the flood models.
Hydrograph shapes have been derived at three gauges at different scales and
topography to ensure appropriate shapes are fitted to inflow catchments.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 5-2
5.4 Recommendations
Recommendations for future work and detailed design of the Kisegese irrigation
scheme are as follows:
The climatic stations installed as part of the IRRIP2 project should be
maintained for as long as possible but at least for 1 year to ensure data is
gathered from the wet season and can be used to undertake corroboration of
rainfall analyses presented in this report.
Further efforts should be made to ensure full rainfall data records from across
the catchment are accessible from all sources, rather than 5-year blocks. This
will enable detailed assessment of rainfall profiles across the Kilombero basin
and local to the Kisegese scheme.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 5-3
6 REFERENCES
Reference Title
CDM Smith IRRIP2 Preliminary Hydrology Report - Kilombero Valley Irrigation
2014 Schemes. Engineering Services for the Tanzania Irrigation and Rural
Roads Infrastructure Project (IRRIP2). CDM Smith for USAID. March
2014
CDM Smith Environmental Flows in Rufiji River Basin Assessed from the Perspective
2015 of Planned Development in Kilombero and Lower Rufiji Sub-Basins.
Technical Assistance to Support the Development of Irrigation and Rural
Roads Infrastructure Project (IRRIP2). CDM Smith for USAID. March
2016.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 6-1
Reference Title
MAFC 2014 Ministry of Agriculture, Food Security and Cooperatives (2014).
Tanzania Agriculture Climate Resilience Plan, 2014-2019. Accessed
September 30, 2015
http://www.kilimo.go.tz/publications/english%20docs/ACRP_TANZANIA_
ENDORSED.pdf
Matari et al. Matari E. R., Chang’a L. B., Chikojo G. E., Hyera T. (2008). Climate
2008 Change scenario development for Second National Communication –
Tanzania. TMA Research Journal, Vol.1, pp. 40-54
McSweeney McSweeney, C., New, M. & Lizcano, G. 2010. UNDP Climate Change
et al. 2010 Country Profiles: Tanzania. Accessed September 30, 2015
http://country-profiles.geog.ox.ac.uk/
USBR 1987 Design of Small Dams. United States Bureau of Reclamation. 3rd Ed
1987
WREM 2012 Rufiji Integrated Water Resources Management and Development Plan:
Interim Report. United Republic of Tanzania Ministry of Water. WREM
International Inc. January 2012
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME 6-2
APPENDICES
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME
A.1 Annual Maximum Rainfall Series
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-1
A.2 Extreme Rainfall Distributions
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-2
Figure A.1: Ifakara Extreme Rainfall Distribution
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-3
A.3 SCS Method Curve Numbers
Crop residue
cover (CR) Poor 77 86 91 94
Good 67 78 85 89
SR + CR Poor 71 80 87 90
Good 64 75 82 85
Good 65 75 82 86
C + CR Poor 69 78 83 87
Good 69 78 83 87
Contoured & 82
Terraced (C & T) Poor 66 74 80
Good 62 71 78 81
C & T + CR Poor 65 73 79 81
Good 61 70 77 80
Small SR Poor 65 76 84 88
Good 63 75 83 87
SR + CR Poor 64 75 83 86
Good 60 72 80 84
C Poor 63 74 82 85
Good 61 73 81 84
C + CR Poor 62 73 81 84
Good 60 72 80 83
C&T Poor 61 72 79 82
Good 59 70 78 81
C & T + CR Poor 60 71 78 81
Good 58 69 77 80
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-4
Table A.3 SCS Method Curve Numbers
Curve numbers for hydrological soil group
Hydrological
Cover Type Treatment Condition A B C D
Closed-
seeded
legumes or
rotation
meadow SR Poor 66 77 85 89
Good 58 72 81 85
C Poor 64 75 83 85
Good 55 69 78 83
C&T Poor 63 73 80 83
Good 51 67 76 80
Continuous
Pasture or forage for
range grazing Poor 47 67 81 88
Fair 25 59 75 83
Good 6 35 70 79
Continuous
Meadow grass - 30 58 71 78
Brush weed
Brush grass mix Poor 48 67 77 83
Fair 35 56 70 77
Good 30 48 65 73
Woods Poor 45 66 77 83
Fair 36 60 73 79
Good 25 55 70 77
Buildings,
Farmsteads driveways etc. - 59 74 82 86
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-5
Table A.4 SCS Method Curve Numbers
Curve numbers for hydrological soil group
Hydrological
Cover Type Treatment Condition A B C D
Brush, weed,
Herbaceous grass Poor - 80 87 93
Fair - 71 81 89
Good - 62 74 85
Oak-aspen Poor - 66 74 79
Fair - 48 57 63
Good - 30 41 48
Pinyon-
juniper Poor - 75 85 89
Fair - 58 73 80
Good - 41 61 71
Sagebrush Poor - 67 80 85
Fair - 51 63 70
Good - 35 47 55
Fair 55 72 81 86
Good 49 68 79 84
Grass cover
>50% Fair 49 69 79 84
Good 39 61 74 80
Impervious
Areas 98 98 98 98
Paved;
Streets and curbs/storm
road sewers 98 98 98 98
Paved; open
ditches 83 89 92 93
Gravel 76 85 89 91
Dirt 72 82 87 89
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-6
Table A.5 SCS Method Curve Numbers
Curve numbers for hydrological soil group
Hydrological
Cover Type Treatment Condition A B C D
Western Natural
Desert landscaping 63 77 85 88
Artificial desert
landscaping 96 96 96 96
Impervious
Commercial
Urban and business
districts industrial 85% 89 92 94 95
72% 81 88 91 93
1/8 acre or
Residential less 65% 77 85 90 92
1 acre 20% 51 68 79 84
2 acre 12% 46 65 77 82
Pervious, no
Newly graded vegetation 77 86 91 94
Use
appropriate
values from
“other
agricultural
Idle lands lands”
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-7
Table A.6 SCS Curve Numbers for Different AMC Classes
CN Corresponding Values
Class II Class I Class III
95 87 98
90 78 96
85 70 94
80 63 91
75 57 88
70 51 85
65 45 82
60 40 78
55 35 74
50 31 70
45 26 65
40 22 60
35 18 55
30 15 50
25 12 43
20 9 37
15 6 30
10 4 22
5 2 13
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME A-8
B.1 Hydrometric Data
Gauged data records available have been compared to determine if relationships
between their flows exist to an extent where mathematical formula could be fitted to
the data and used for infilling where data gaps are present.
The method includes comparing each available data series against each other for the
period where overlapping data is available and creating a scatter plot. A linear
relationship is then fitted to the scatter plot and an r squared value calculated to
determine similarities in the records. If there is a good relationship between two
series a tight cluster of points will be presented, with and r squared value close to
unity.
Figure B.1, Figure B.2 and Figure B.3 show the results of the exercise undertaken
using the raw data series only, with no outliers removed or data filtering undertaken.
The results show very poor inter gauge relationships when comparing the raw flow
data series. This is perhaps not unsurprising given the distances between the
gauges across the catchment and therefore the chance of different rainfall patterns
and therefore different flow peak timings occurring.
The majority of the relationships show low r squared values, less than 0.5, with only
four comparisons showing r squared values greater than 0.5. The highest r squared
value comes from the comparison of flows at gauges 1KB17 and 1KB27, with a value
of 0.64. This is still considered a poor fit. A reasonable comparison would only be
used if the gauged records showed a good fit, with r squared value greater than 0.8.
Tests were undertaken to filter the data series to improve the relationships between
them taking into account outlier data points and the difference in the timing of flood
peaks. The testing undertaken was unable to improve the analysis to give significant
confidence in applying a relationship between two records for infilling.
Direct comparison of gauged flow records does not increase confidence in using
available gauged records for infilling. Given the geographical spread of gauging
stations across a wide area, derivation of a meaningful relationship between two
gauged records is unlikely due to widespread catchments of different sizes
experiencing rainfall at different times. To overcome this issue flow record
comparison has been undertaken using catchment FDCs rather than gauged
records. FDCs can be compared without taking into account the exact timing of a
specific flow. The shape of a FDC reflects the catchment from which it has been
derived and therefore two flow records converted to FDCs can be compared without
the need to heavily filter the available data series. This approach has been adopted
for the remainder of the study.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-1
Figure B.1: Flow Record Relationships
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-2
Figure B.2 Ifakara Extreme Rainfall Distribution
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-3
Figure B.3 Ifakara Extreme Rainfall Distribution
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-4
B.2 Flow Record Infilling
Long-term record series are noted at gauges 1KB8, 1KB9, 1KB14, 1KB18 and
1KB28 and have therefore been chosen as potential records from which to build and
infill a long-term data series. Table B.1 shows a summary of each gauge considered
and Figure B.4 shows a comparison of FDCs from each gauged, standardized by
mean flow to clearly demonstrate differences in their shape.
Table B.1. Gauged Statistics
1KB8 1KB9 1KB14 1KB18 1KB28
The catchment detail immediately highlight that several gauged records may not be
suitable for use in infilling procedures.
Gauge 1KB18 has the smallest catchment area of the gauges considered and
despite having the longest record available is located a significant distance from the
proposed irrigation scheme and the other gauges. The FDC derived from the 1KB18
data record drops off significantly for flows above the 80th percentile which means if
the record was used for infilling it would likely underestimate low flows. The low flow
end of the gauged record is important in calculating available flows for the irrigation
scheme and therefore the data from 1KB18 has not been considered further for
infilling.
Gauge 1KB28 is located downstream of a hydropower scheme in the Kihansi
catchment and although the gauged record did not show any obvious impacts from
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-5
the scheme, the FDC shown is very flat. This is symptomatic upstream attenuation
that the hydropower scheme provides, where the high flow peaks are reduced and
low flows are supported by environmental flow releases and reservoir spill. The
Qmean per unit area of 0.28 m³/s is higher than any of the other gauges considered
by approximately one third.
Therefore, the data record from gauge 1KB28 is not considered appropriate to use in
derivation of a long-term flow series.
Gauge records at 1KB8, 1KB9 and 1KB14 are considered appropriate for infilling to
create a long-term data series. Given that gauge 1KB14 has a small catchment area
compared to 1KB8 and 1KB9 it has been omitted from the infilling analysis, but it is
noted that the record here is robust and should be considered when determining
scheme specific flows as it represents flows from the escarpment are well.
Examination of available data records shows gauge 1KB8 and 1KB9 both have
reasonably long records, with gauge 1KB9 covering data gaps in the record at 1KB8.
Figure B.5 shows the standardized comparison of the FDCs from 1KB8 and 1KB9.
The two catchments are reasonably closely matched in terms of scale and their
FDCs show a reasonably similar profile i.e. the gradient of the FDCs is similar which
suggests a similar response to rainfall within each catchment. The catchments are
also neighbors so are likely to experience similar rainfall patterns. It is noted that the
FDC from 1KB8 is slightly flatter than 1KB9, with 1KB9 showing slightly lower low
flows, and slightly higher mid flows.
Figure B.5 Standardized FDC Comparison between 1KB8 and 1KB9
The record at 1KB8 was infilled using the flow series from gauge 1KB9 scaled by
mean flow to the 1KB8 catchment. This resulted in a robust data record, though only
to approximately 2007. The data record has been extended from 2007 to 2014 using
gauged data from 1KB32.
Short periods of missing data were initially infilled assuming a linear relationship
across the missing data period. These periods have been examined and adjusted
based on the available data series to fit observed hydrograph shapes.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-6
The year 2013 has not been able to be infilled from a gauged record and therefore
the average peak flow across the record has been calculated and a hydrograph
shape fitted to reach this peak based on observed data.
The infilled series stretches from 1956 to 2014. Comparison of the baseline record at
1KB8 with the infilled record has been undertaken, comparing FDCs (see Figure
B.6), Qmean, Qmax and Qmin (Table B.2).
The FDCs match very closely between approximately Q70 and Q99, which shows
the low flow characteristics of the catchment have been maintained despite the
infilling procedure. Mid to high flows are slightly higher for the infilled series than the
original record, but the comparison is close again at the very highest flow peaks, Q1
to Q5. The minimum and maximum flows in the record do not change with a slight
difference in Qmean.
The comparison shows only a slight difference between the baseline and infilled data
and therefore the infilling is considered to be delivering a reasonable representation
of flows within the1KB8 catchment.
Table B.2. Comparison of Flow Statistics at 1KB8
Project 1KB8 1KB8 Infilled
Figure B.6 FDC from Gauge 1KB8 vs FDC from 1KB8 Infilled Record
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-7
B.3 Kisegese Inflow Derivation
To ensure that the inflow series used in calculation of available flows for the
Kisegese scheme is specific to its surrounding catchments, analysis has been
undertaken to determine which gauged FDC should have the long term data record
from 1KB8 scaled to it.
Four gauges have been selected as being potentially representative of the
Chiwachiwa and Ruipa catchments that will provide water for the Kisegese Irrigation
Scheme. The selection is based mainly on catchment size and location to represent
the fact that the Chiwachiwa and Ruipa catchments have a significant proportion of
their catchments on the northern escarpment and therefore will show a flow regime
based on these specific catchment characteristics.
Four gauges were selected which show similar catchment topography to the
Chiwachiwa and Ruipa catchments namely 1KB14, 1KB15, 1KB27 and 1KB28.
Figure B.7 shows the location of these catchments and Table B.3 summaries
catchment details.
Figure B.7 Kisegese Analysis Gauge Locations
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-8
Table B.3. Comparison of Flow Statistics
1KB14 1KB15 1KB27 1KB28
To determine which of the gauged catchment show the closest similarity to the target
catchment, flow duration curve comparison was undertaken using flows standardized
by Qmean to compare how the catchments react to rainfall. In addition to this
catchment area, Qmean per unit area and record length were also considered along
with any catchment specific information. Figure B.8 shows the FDC comparison.
Figure B.8 Standardized FDC Comparison
Gauge 1KB27 is located just upstream of the Kisegese scheme and includes the
catchments of the Chiwachiwa and Ruipa rivers. Given its location, the FDC from
gauge 1KB27 was immediately considered to use in scaling the long-term flow series
derived at 1KB8 to catchment specific flow regime at the Kisegese scheme. In
addition to this, ADCP gauging has been undertaken at this location. It is noted
however, that the flow series at this gauge is based on a rating curve that has been
derived using eight data points and therefore there is uncertainty in the flow series,
particularly the high flow end. This is manifested in the FDC where the shape at the
high flow end does not look appropriate. It is also noted that the Qmean per unit
area for the gauged record is around half that of other similarly scaled catchments in
the Kilombero basin. Given the uncertainty in the rating curve and consequently the
flow record at this gauge, it is considered that the FDC is underestimating mid to low
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-9
flows in the catchment and is not representing high flows accurately and therefore
has not been considered further in the analysis.
Gauge 1KB28 is located on the Kihansi River; some 67km form the Kisegese
scheme. It is noted that a hydropower scheme is located in the upper Kihansi
catchment, which is potentially impacting the flow regime, and this is shown in the
FDC derived from the gauged record. The curve is much flatter than at gauges
1KB14 and 1KB15, suggesting attenuation of high flows, and higher mid to low flows
supported by environmental flow releases. The FDC is not considered
representative of the Chiwachiwa and Ruipa catchments due to the influence of the
hydropower scheme and is therefore not considered further in analysis.
Gauge 1KB15 is located on the Mnyera River, 28 km southwest of the Kisegese
scheme and has a catchment located on the escarpment. The FDC derived using
the flow series at this location is considered robust, though the Qmean per unit area
value is approximately three times higher than other similarly scaled catchments in
the Kilombero basin and therefore the data series is questionable and has not been
considered further in analysis.
Gauge 1KB14 is located on the Lumemo River, 43 km northeast of the Kisegese
scheme and has a catchment located on the escarpment. The FDC derived for this
catchment is considered robust, using a reasonable gauged record period and shows
a reasonable Qmean per unit area value. The gauged data here is considered to
give a reasonable estimate of the flow regime in the Chiwachiwa and Ruipa
catchments and it is therefore considered appropriate to scale the long-term daily
flow series from the Mpanga gauge to the flow duration curve shape and scale this to
intake weir catchment areas.
The comparison of data from each of the gauges shows that 1KB14 is the most
appropriate gauge to provide an accurate representation of the FDC shape at the
Kisegese scheme. It is noted that gauge 1KB27 is located just upstream of the
Kisegese scheme, but demonstrated flows and a FDC shape which are not
appropriate to represent the catchment and therefore it was omitted from further
analysis.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-10
B.4 Kisegese Flood Frequency Fitting
Figure B.9 Flood Frequency Fitting at Gauge 1KB8
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME B-11
ANNEX C HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME
Reference
Number Name Water Level Flow Comment
Reasonable record length though the gauging station is located a significant
Rufiji at
1K3 1976-1980 1956-1986 distance downstream from the Kilombero catchment and therefore is likely of limited
Stiegler’s Gorge
use in the study.
Water level and flow records are reasonable until a step change and large reduction
in range during 1971. The data record post 1975 is sporadic and not considered to
1K4A Rufiji at Utete 1978-2014 1978-1991 be robust for use in further analysis. The gauging station is located a significant
distance downstream from the Kilombero catchment and therefore is likely of limited
use in the study.
The water Level series shows a step in 1971 with peaks reduced from approximately
20m to 5m. The range of water levels recorded is reduced from approximately 15m
Kilombero at
1KB2A 1966-2003 to less than 5m. The data series is also missing significant periods of record. No
Ifakara Ferry
rating curve is available to derive flow data and therefore data is of little value to the
study.
The water level series looks reasonable until step change and large reduction in
range during 1971. The data record post 1975 is sporadic and not considered to be
Kilombero at
1KB4A 1955-1983 robust for use in further analysis. Spot gauged data is available to derive a rating
Ifwema
curve, but data is from 1960 to 1976 and therefore questionable to apply the derived
rating across the full level series.
The water level series shows a step change in 1973 where the range of levels
Mpanga at recorded is reduced and water levels peaks are reduced from 14 m to approximately
1KB8A Mpanga 1956-2014 1956-2007 4 m.
Mission
3 flow series are available with 2 series showing the reduction in peaks stated in the
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME C-1
Reference
Number Name Water Level Flow Comment
water level data. One series is consistent, where peaks and range are consistent to
1991.
Mnyera at Water level series looks reasonable up to 1995 but after that, data deteriorates and
1KB9A Taveta Mission 1956-2009 1957-1987 shows a large step change. Flow series looks reasonable to 1987
u/s
Mnyera at Water level series looks reasonable with potential to support analysis at the
1KB9B Taveta Mission 1979-1995 upstream gauge.
d/s
A large step in the water level data is noted during 1973 where the range recorded is
Ruhudji at reduced from between 6m and 20 m to between 2 and 6m. A rating curve at this
1KB10 1960-1988
Mwayamulungu gauge is noted within the Rufiji Basin Hydropower Master Plan so flow series can be
derived here.
Water level series shows a large step during 1977 and there is a significant period of
1KB12A Mchilipa at Itete 1973-2014 missing data between 1990 and 2005. The data is questionable and level only data
with no means of converting to a flow series.
The level series shows a step during 1966 where the range of levels reduces
Lumemo at
1KB14A 1958-2010 substantially with maximum level also reduced substantially. A rating curve is
Kiburubutu
available to apply to the level data to derive a flows series.
Mnyera Mgeta The water level series shows a step during 1972 with the range and peak levels
1KB15 at u/s Mchombe 1960-1990 1960-1990 reduced. Available flow series do not show a step in flows. The flow peaks increase
Mission after 1977.
The water level series shows a step in 1971 that is mirrored by the flow series,
Kilombero at therefore suggesting a single rating has been used across the level series to derive
1KB17A 1957-1984 1957-1981
Swero the flow series.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME C-2
Reference
Number Name Water Level Flow Comment
The flow series derived after the data step shows flows that are far too low. A DMF
at this gauge has been calculated for the Rufiji Basin Hydropower Master Plan -
494.8 m³/s, which is approximately equal to the peak flows shown after 1971 in the
data series and therefore analysis of level data required to give robust flow record.
The water level series shows a step during 1971 where the range and peak levels
Ruhudji at d/s
1KB18B 1958-2014 1976-2011 after this data are reduced. Three flow series are available which are all similar,
Kifunga'a Falls
showing consistent flows.
The water level series shows a step during 1971, which is not reflected in the flow
Hagafiro at series. No rating curves are available to corroborate the flow series. The record is
1KB19A 1961-2013 1961-1978
Hagafiro short and is not considered reliable. Potential to derive a relationship between level
and flow from the early record to derive a longer flow series.
Only a short water level record - 10 years - though no gaps. Corroborates level data
1KB20 Ijunilo at Kibena 1960-1970 at 1KB18B but does not offer an infilling opportunity and is therefore not considered
useful to the study.
The water level data shows step in 1971 where the range of levels is slightly
1KB23 Sonjo at Sonjo 1961-2010 reduced. Virtually no data 85-00 with no rating curve to derive a flow series. Not
considered useful to the study.
The water level series shows a step in 1971 and in 1999, which suggests the data
1KB24 Sanje at Sanje 1962-2012
are unreliable with limited scope for adjustment.
Two water level series are available at this location - one named upstream and one
downstream. Upstream levels are lower than downstream levels, which reduces
Kiberege at
1KB26B 1966-2013 confidence in data, though may only be due to how the stage board is positioned.
Kiberege u/s
The data also shows a step during 1971. No rating curve is available to convert the
level series to a flow series so the data is of limited use to the study.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME C-3
Reference
Number Name Water Level Flow Comment
No steps shown in water levels though large gap in data set (1979-1988). No flow
Ruipa at series, though some spot gaugings covering limited range, are available. ADCP
1KB27 1974-1990
Mbingu gauging has been undertaken at the gauge as part of this study, which covers a
reasonable range of water levels and flows.
Water level record 1974-2007 is mostly complete with no large data gaps. The flow
record from 1998 to 2007 shows higher flows than the previous record period. Spot
Kihansi at gauged flows are available to derive a rating curve. Analysis has also been
1KB28A 1974-2007
Lugoda undertaken in the Rufiji Hydropower Master Plan for use as comparison. It is noted
that a hydropower scheme is located in the upper catchment and therefore flow
series here is not natural.
Only a short level series is available but is consistent with short flow record. The
Kihansi at
1KB32 2006-2014 1984-2014 flow series available has a long period of missing data from 1989-2006 though
Lutaki
period up to 2015 considered useful for infilling.
The water level series is constant bar sporadic spikes in level and it is assumed the
Kihansi below environmental flow from the dam is recorded with limited periods of time when the
1KB33 2004-2014
Kihansi dam is spilling. The record is of limited use to the study given the impacts of the
reservoir.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME C-4
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME
Reference
Number Name Comment
Rufiji at Stiegler’s NO ANALYSIS UNDERTAKEN.
1K3
Gorge
1K4A Rufiji at Utete NO ANALYSIS UNDERTAKEN.
Kilombero at NO ANALYSIS UNDERTAKEN.
1KB2A
Ifakara Ferry
Adjustment of level data to scale to pre-1971 range through comparison of level duration curves was considered.
Kilombero at
1KB4A Given the poor post-1975 record, this has not been undertaken. A flow series has been derived from water level
Ifwema
data with rating derived from spot gaugings. Pre-1971 series is considered robust.
The flow series provided is robust and considered appropriate for use. It is noted that the level series available is of
Mpanga at
1KB8A longer duration than the flow series. The relationship between level and flow for the period of overlap has been
Mpanga Mission
derived and applied to the most recent level series to extend the flow series available.
The record at the upstream gauge has been infilled using the relationship between the upstream and downstream
Mnyera at water level series. The relationship between the level series is consistent allowing the upstream series to be
1KB9A Taveta Mission extended.
u/s
A stage discharge relationship has been derived between the overlapping stage and flow series and applied to the
level series to derive an extended flow series. The flow series is robust up to 1995, though after this the data
record is dubious and not considered appropriate for use.
Mnyera at
1KB9B Taveta Mission A FDC for the gauge is presented in the Rufiji Basin Hydropower Master Plan, which compares well with the
d/s derived record. Comparison of FDCs shows a similar profile, though high flows are slightly reduced from
Hydropower MP document FDC.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY FOR UDAGAJI-MGUGWE IRRIGATION SCHEME D-1
Reference
Number Name Comment
Removed the step change in water level data through comparison of the level duration curves pre and post step
change. The mean factor across the level percentiles was applied to the level record post step to scale to be
consistent with the pre step record.
Ruhudji at
1KB10
Mwayamulungu
A rating presented in the Rufiji Hydropower Master Plan was applied to derive a flow series. It is noted that the
derived flow series flattens off at high flows and therefore is viewed with caution. This is likely to be due to
extrapolation of the rating curve beyond its range.
1KB12A Mchilipa at Itete NO ANALYSIS UNDERTAKEN.
Level duration curves derived pre and post step to calculate a factor to raise the post step levels to be consistent
with the pre step series. A large factor was derived (approximately 20) with low confidence held in its application.
Lumemo at The available rating curve was applied to the level record. Post step levels give flows equivalent to those in the
1KB14A
Kiburubutu digital flow record, but the pre step levels deliver extremely large flows (10000s of cumecs) which are not consistent
with a catchment of this size and shows the rating curve is not applicable to the early level series. A comparison
with 1KB15, which is of similar scale, shows these flows are too large and therefore the flow record provided is
assumed representative.
Level duration curves have been derived for the pre and post step levels, with a factor applied to the post step
levels to ensure they are consistent with the early record. Two rating curves have been supplied and applied to the
level series, but the flow series calculated is not in line with the supplied flows series. The supplied flow series
Mnyera Mgeta at shows a large increase in peak flows from 1976 (100 m³/s to 600 m³/s) onwards and is not considered to be
1KB15 u/s Mchombe representative of the catchment when compared with others of similar scale.
Mission
A flow series has been derived from the adjusted level series and the rating curve, which shows a flow series with
consistent peak flows. It is noted that the extended period shows much lower low flows than the initial period, and
therefore should be viewed with caution.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY FOR UDAGAJI-MGUGWE IRRIGATION SCHEME D-2
Reference
Number Name Comment
Level duration analysis for the pre and post data step periods has been undertaken to derive a scaling factor to
Kilombero at produce a consistent level series. A stage discharge relationship was derived from the pre step record period and
1KB17A
Swero applied to the adjusted level series. The average flow from the new derived series is 477.35 m³/s, which compares
well with the average flow at the gauge from the Rufiji Basin Hydropower Master Plan.
Level duration analysis has been undertaken between the pre and post step water levels to raise the later record in
Ruhudji at line with the early record. The available rating curve has been applied to the adjusted levels but produces flows
1KB18A
Njombe well in excess of what is expected of the catchment size. The supplied flow series is considered most appropriate
for use.
Level duration analysis has been undertaken to remove the step in the water level series. This results in a more
Hagafiro at consistent record, but the period from 1980 to 1990 shows a much reduced level range and therefore the series is
1KB19A
Hagafiro questionable. Flow series derived by determining a rating for overlapping period of level and flow series, but due to
reservations about level series the flow series is not considered reliable.
1KB20 Ijunilo at Kibena NO ANALYSIS UNDERTAKEN.
1KB23 Sonjo at Sonjo NO ANALYSIS UNDERTAKEN.
1KB24 Sanje at Sanje NO ANALYSIS UNDERTAKEN.
Kiberege at NO ANALYSIS UNDERTAKEN.
1KKB26B
Kiberege u/s
A rating curve has been derived from historic spot gaugings and recent ADCP gaugings and applied to the level
series to give a flow series. There is limited data here so of limited use to the study. The derived rating curve
1KB27 Ruipa at Mbingu covers a large range of recorded water levels, though it is noted the relationship is linear and based on limited data
points. The derived flow series is therefore held in low confidence. It is also noted that the flow series shows a
Qmean per unit area an order of magnitude less than other gauged catchments considered.
ANNEX C: HYDROLOGY
FEASIBILITY STUDY FOR UDAGAJI-MGUGWE IRRIGATION SCHEME D-3
Reference
Number Name Comment
The flow series derived using the rating curve has an average flow of 18.6 m³/s, which compares closely to the
value quoted in the Rufiji Basin Hydropower Master Plan, 17 m³/s. It is noted that the latter part of the record
Kihansi at
1KB28A shows elevated levels and flows from 1998 onwards and is held in low confidence. The earlier record from 1974 –
Lugoda
1998 is consistent considered to be robust. There is no visible impact from the hydropower scheme in the
catchment.
1KB32 Kihansi at Lutaki NO ANALYSIS REQUIRED.
Kihansi below NO ANALYSIS UNDERTAKEN.
1KB33
Kihansi
Kihansi at NO ANALYSIS UNDERTAKEN.
1KB34 Kihansi Dam
Road Bridge
Kihansi at NO ANALYSIS UNDERTAKEN.
1KB35
Tazara Bridge
ANNEX C: HYDROLOGY
FEASIBILITY STUDY FOR UDAGAJI-MGUGWE IRRIGATION SCHEME D-4
Reference
Number Name
1K3 Rufiji at Stiegler’s Gorge
1K4A Rufiji at Utete
1KB2A Kilombero at Ifakara Ferry
1KB4A Kilombero at Ifwema
1KB8A Mpanga at Mpanga Mission
1KB9A Mnyera at Taveta Mission u/s
1KB9B Mnyera at Taveta Mission d/s
1KB10 Ruhudji at Mwayamulungu
1KB12A Mchilipa at Itete
1KB14A Lumemo at Kiburubutu
1KB15 Mnyera Mgeta at u/s Mchombe Mission
1KB17A Kilombero at Swero
1KB18A Ruhudji at Njombe
1KB19A Hagafiro at Hagafiro
1KB20 Ijunilo at Kibena
1KB23 Sonjo at Sonjo
1KB24 Sanje at Sanje
1KKB26B Kiberege at Kiberege u/s
1KB27 Ruipa at Mbingu
1KB28A Kihansi at Lugoda
1KB32 Kihansi at Lutaki
1KB33 Kihansi below Kihansi
1KB34 Kihansi at Kihansi Dam Road Bridge
1KB35 Kihansi at Tazara Bridge
ANNEX C: HYDROLOGY
FEASIBILITY STUDY OF THE KISEGESE IRRIGATION SCHEME E-1
Technical Assistance to Support the Development of
Irrigation and Rural Roads Infrastructure Project (IRRIP2)