Professional Documents
Culture Documents
erosion management
Kumar Manish2*, Hassan Kazi1, Kumar Nikhil1, Kumar Mani1 Kumar Anil2, and Jaiswal Ritesh1
1 Mathematical Modelling Centre, FMISC, WRD, PATNA, India
2 Flood Management Improvement Support Centre, WRD, Bihar
*Corresponding author, e-mail: manishkr14@gmail.com
ABSTRACT
There are uncertainties in flood level prediction and sediment load prediction relating to fluvial
sediment transport. The present practices of flood risk assessment, river behaviour management
modelling around the world are mostly not sediment driven, they are purely based upon non-
erodible bed hydrodynamic modelling. Legislations (e.g., EU Water Framework Directive (EC
2000/60/EC)) lists engineering activities regulated with respect to sediment management; there
is increased awareness that sediment controls to floodwater conveyance are important for flood
risk management. In future climate change scenarios, long-term channel conveyance may change
significantly in morphologically active rivers, particularly in the major rivers in Bihar; this may
increase flood risk in some reaches due to massive siltation (Pender et al. 2016). In view of the
above, two-dimensional mobile bed and bank morphological model of the Ganga in Bihar has
been developed. The model domain is about 600 km long, from Buxar up to Farakka. MIKE21C
modelling software of DHI (DHI, 2016) has been used. MIKE 21C is a generalised mathematical
modelling system for the simulation of hydrodynamics of vertically homogeneous flows, and for
the simulation of sediment transports. This is one among the best 2-D morphological modelling
software (Langendoen, 2001). Key elements in MIKE21C modelling tool are: i) curvilinear grid
(movable adaptive grid to account bank erosion, sandbar erosion), ii) hydrodynamics, iii) helical
flow-analytical model, iv) bed and suspended load, iv) alluvial resistance, v) river bed
morphology, vi) bank erosion and vii) river planform. The model can generate advance
knowledge, in short term (one-year) to medium term (3 to 5 years) on river bank erosion and
morphological developments (Enggorb and Tjerry, 1999). Detail depth, flow distribution, bed
shear stress and sediment concentration at a particular bend (Bindtoli) have been discussed.
Depth and flow distribution maps (Figure 1) identify the deeper main channel and areas of higher
flow velocities, and thus higher erosion vulnerabilities. Shear stress maps identify areas of
erosion, and erosion-deposition map quantifies their extent and magnitude. The maps, at Bindtoli
area, are from a coarse grid 600 km long 2-D model and thus, do not have fine enough resolution
of representing very finer details of local structures (e.g. spurs or revetment) in this model. For
that purpose, a very fine resolution, sub-scale 2-D model, should have to be derived from the
larger model for planning and design of erosion protection works. This 2-D model will remain
operational for design of new erosion protection works and for detail evaluation of similar
existing structures. Similar local scale detailed model could be derived from this large model at
any areas of interest within the domain of this model. This paper presents the results of flood
forecasting and erosion prediction models for the river system of Bihar.
Keywords: 2-D modelling, Erosion Prediction, MIKE21C, Bank Erosion, Bed Erosion and
Secondary Flow
1.0 INTRODUCTION
Gangetic Basin
The Ganga originates from the Gangotri glacier in the Himalayas at an elevation of about
7,010 m in the Uttarkashi district of Uttarakhand. After confluence of streams Bhagirathi and
Alakhnanda at Devprayag, the river is known as Ganga. The total length of river Ganga
(measured along the Bhagirathi and the Hooghly) up to its outfall into Bay of Bengal is 2,525
km. The principal tributaries joining the river from right are the Yamuna and the Sone. The
Ramganga, the Ghaghra, the Gandak, the Kosi and the Mahananda join the river from left. Ganga
river basin (GRBEMP Interim Report, 2013) is the largest river basin in India in terms of
catchment area, constituting 26% of the country's land mass (8,61,404 Sq. km). The Ganga basin
mainly consists of alluvial sediments, which has been collected over most of the Quaternary
period building one of the largest alluvial plains in the world. The basin covers 11 States of India,
viz., Uttarakhand, Uttar Pradesh, Madhya Pradesh, Rajasthan, Haryana, Himachal Pradesh,
Chhattisgarh, Jharkhand, Bihar, West Bengal and Delhi. The Ganges River system has wide
diurnal, seasonal and annual variations in the sediment-carrying capacity and it varies from 403
to 660 × 106 tonnes/year (Subramanian, V. et al, 1984).
The weather in the Ganga basin is characterized by a distinct wet season during the period of
southwest monsoon (June to October). The mean annual rainfall of the Ganga River Basin is
about 1,170 mm. About 88% of the annual rainfall is received during the period of June to
October. The bulk of the remaining 12% occurs mostly in the periods of March to May and
November to December. The Ganga-Brahmaputra is the world highest Sediment carrying river.
(Sediment loads of major world rivers (after Knighton, 1998: Figure 3.2), Degens et al (1991),
Meybeck (1976) and Milliman and Meade (1983))
Surface
Bottom
Figure 1 Deviation in directions of flow velocity and shear stress due to helical flow in a river bend
The direction of the bed shear stress in a curved flow plays a crucial role in a bed
topography model for river bends. The bed shear stress direction is specified as (Rozowskii,
1957, Engelund, 1974, Struiksma et. al, 1985):
h
tan s = -
Rs
Where h is water depth, Rs is radius of curvature of flow stream lines and δs is angle
between bed shear stress depth average shear stress
is defined as:
2 g
= (1 -
)
2
C
Where κ is von karman’s constant (0.4), g is acceleration due to gravity, C is chezy
number and α is calibration constant.
The approximate value of is 10. In regions of changing curvature of the streamlines,
the secondary flow will adapt gradually. The adaptation of the secondary flow profile is
considerably faster near the bottom (where the bed shear stresses act) than further up in the water
column.
Bed load is calculated on the basis of empirical sediment transport formulas. In contrast
to suspended load, it is assumed for the bed load that its responses immediately to changes in
local hydraulic conditions without any time and space lag. Two important effects are taken into
account:
1. The deviation of the direction of the bed shear stress from the mean flow direction due
to helical flow, and
2. Effect of a sloping river bed.
Since the pioneering work of van Bendegom, 1947, a large number of models for the
prediction of the transverse bed slope has been proposed. The following formula (Olesen, 1987,
Talmon, 1992) is implemented in the present modelling system:
-a z
*
S n = tan s - G S bl
n
Where G and a are transverse bed slope factor and exponent (approximate value of G is
0.6)
Based on the calculated suspended load and the bed load, the bed level changes are found
from solving the sediment continuity equation:
z S x S y
(1 - n) + + = Se
t x y
C = a hb
With b=1/6, the equation equals the Manning formula. Calibration experiences of the
braided River like Ganga suggest that the b exponent must be higher (around 0.5) to reproduce the
flow and riverbed development.
Where E is yearly bank erosion (m), W is channel width (m), R is radius of curvature of
the curved channel in meter, C is Chezy coefficient (m1/2/s) and B is overall bank resistance
coefficient
The 1st and the 2nd methods have already applied in in different studies in various rivers of Indian
Subcontinent. The 4th method i.e., stochastic model is yet to computerized. The 3rd method is
used in the mathematical 2-D model (MIKE21C) of Ganga for short term (one or two monsoons)
prediction of erosion rates.
2.2 Methodology
The fluvial modelling tools utilized in the present study are NAM, MIKE11 and
MIKE21C
NAM
It is DHI’s hydrological modelling software, which simulates rainfall-run-off processes
occurring at catchment scale. NAM forms the rainfall-runoff module (RR module) of MIKE11.
The RR module can either be developed and used independently or be used directly with the
MIKE11 model to provide lateral inflows (run-off) from the catchment to the river. NAM is a
deterministic, lumped, conceptual model. NAM (DHI, 2016) comprises of a set of linked
mathematical statement describing, in a simplified quantitative form, the land phase of the
hydrological cycle. NAM mainly simulates surface, sub-surface run-off and baseflow.
MIKE11
It is a professional engineering software for simulation of flows, sediment transport and
water quality in rivers, estuaries, channels, lakes, reservoirs and other water bodies. MIKE11
(DHI, 2016) is a fully dynamic, one-dimensional modelling software; it solves vertically
integrated Saint Venant one-dimensional flow equations for conservation of mass and
momentum. The tool is used for detailed analysis, design, management and operation of simple
and complex river/channel systems. The hydrodynamic (HD) module forms the nucleus of the
software, and forms the basis for rest of the modules, e.g., flood forecasting, advection-
dispersion, sediment transport and water quality.
MIKE21C
MIKE 21C is a generalized mathematical modelling system for the simulation of
hydrodynamics of vertically homogeneous flows, and for the simulation of sediment transports.
MIKE21C solves 2-D Saint-Venant’s flow equations in curvilinear; applies 2nd order time
integration to enhance accuracy and allowing larger time step for efficient computation; capable
of wetting and drying; uses spatiotemporal variation on flow roughness; compute
erosion/deposition by sediment continuity equation and allows bank erosion in fixed and in
moveable grid. The modelling system is composed of a number of modules like hydrodynamic
module (HD), Advection-dispersion module (AD), Sediment transport module (ST) & large-
scale morphological module.
An approximately 510 km long 2D model of Ganga river is built on 106,400 computational cells (Figure
4) in curvilinear orthogonal grid system of MIKE21C modelling technology. The model covers full active
width of the Ganga River; at some of the reaches, the width of the active braid belt is approximately
18km. Computational grids have been generated in multi-block grid generation approach. The resolution
across the width is more important, due to bend scour and bank erosion at numerous bends along the
course of the river. Finer cell resolution across the width of the river is necessary for assessing bend scour,
obstruction scour and bank erosion.
Figure 4: Curvilinear grid of Ganga 2D model, zoomed at the reach in Munger to demonstrate grid
resolution distinctly
Bathymetry for the 2D model has been generated from Jaxa satellite imagery of 30m resolution (Figure -
5).
Buxar is the upstream boundary and Farakka is the downstream boundary of the model.
Discharge at Buxar and water level at Farakka has been used as boundary condition for the 2D model;
2016 monsoon flow condition has been chosen. There are ten lateral inflows to the 2D model from the
tributaries on the north and south bank of the Ganges. All boundary data has been taken from the
calibrated MIKE11 1d model of the regional network model
Figure -5: Ganga 2D model bathymetry
Chezy’s C has been used as roughness parameter for calibration of the model. Spatially varying depth
dependent roughness values have been generated for model calibration (Figure -7). Based on the
assumption that deeper areas will have lower friction and shallower areas will have higher friction, a
depth varying roughness values (map) were generated and used for calibration of both hydrodynamic and
morphological model. The minimum and maximum Chezy values are 24.2 and 60.5 respectively. The
relation between Manning’s M (1/n) and Chezy C is C = Mh1/6.
Figure 7: Ganga 2D model Chezy’s roughness map (depth dependent)
The present 2D model is for hydrodynamics and sediment transport modelling; as such bed material
size is an input to the 2D model. In the present model, uniform grain size of 0.15mm has been used.
4.0 RESULTS
The hydrodynamic calibration has been done against observed water level at eight CWC gauging stations
in the Ganga. Calibration is excellent in terms of peak water level and shape of the hydrograph at
Gandhighat, Hatidah, Munger, Bhagalpur, Kahalgaon and Sahebganj (Figure 6.8 and 6.9); calibration
requires improvement at Buxar and Dighaghat; difference at peak water level between modelled and
observed data is considerably high, nearly a metre (Figure 6.7). Thus, this requires investigation; probable
reasons may be 2d water level has been extracted from a different location than where the observed CWC
station is and also may be effect of roughness parameter. There is a bit of issue with the initial condition
in the 2d model, where modelled and observed level are a bit different; however, this would be easy to
fix, just have to set the appropriate initial condition. We should mention that the results presented here
are from initial trial runs for calibration. Further fine tuning of the model is being done and it is expected
that this will improve model results considerably.
Figure 8: Calibrated water levels from 2d model and observed CWC water levels at Buxar, Dighaghat
and Gandhighat
Figure 9: Calibrated water levels from 2d model and observed CWC water levels at Hatidah, Munger and
Bhagalpur.
Figure 10: Calibrated water levels from 2d model and observed CWC water levels at Kahalgaon and Sahebganj
Figure 11: Water surface elevation and water Depth as on 25 August 2016
Van Rijn sediment transport formula for bed and suspended load (Van Rijn, 1984) for uniform sediment
has been used in the model computation. Median grain size, D50, was used as 0.15mm. Sediment
continuity equation has been applied to carry out erosion and deposition in river bed. The model includes
secondary current and thus provides 3d effect to the 2d model. The sediment model has not been calibrated
yet; calibrated will be carried out indirectly against observed sediment load and sediment concentration
data available in literature (Subramania, 1996).
Sediment rating curves, based on 2D model results, have been generated at Gandhighat, Kahalgaon and
Sahebganj. The rating curves demonstrate that the river has efficient sediment carrying capacity at
Gandhighat and Kahalgaon. However, at Sahebganj, the sediment carrying capacity considerably drops
indicating aggradation in the river reach surrounding Sahebganj.
Figure 13: Ganga 2D model sediment rating curves at Gandhighat, Kahalgaon and Sahebganj
5.0 CONCLUSION
Bihar being India’s one of the most flood prone state which not only face recurrent flood
devastation but also the problems of frequent bank shift, bank erosion, aggradation-degradation
of river channels. The problem of frequent bank shifting is being depicted through the satellite
image comparison of November 2013 and April 2018 which clearly shows the bank shifting of
river Ganga near Ismailpur Bindtoli (Figure – 14).
Bank line in
November, 2013
Bank line in
April, 2018
In order to tackle above mentioned problems, 2-D modelling has emerged as a very useful tool.
The 2D model of various rivers in Bihar is being developed, to develop the flow field and to
carry out the flood studies of river channels and adjacent flood plains or computing the flow
characteristics like maximum flood level, flow velocity and flow direction etc. The 2-D model
is also a very useful tool for forecasting morphological changes in near future in river banks,
which would ultimately be useful for planning and execution of river training works. 2-D models
are proving to be very efficient in depicting variation of flow velocities at a cross section,
predicting flow depth and scour depth at a location, and rate of bank line shifting. 2-D sediment
transport models can be used to study the sediment deposition rate and erosion rate particularly
at the meandering portion, bifurcations and confluences.
Acknowledgment
River hydrological data and other information used in this model has been provided by Flood
Management Improvement Support Centre (FMISC), Water Resources Department (WRD),
Bihar, India. The authors gratefully acknowledge the support from the Joint Director of FMISC.
Reference:
[1] DHI (2016). “MIKE11/MIKE21C Reference Manual”.
[2] Enggrob, H. G. and S. Tjerry, (1999). “Simulation of morphological characteristics of a braided river.” In Proc.
RCEM 1999, vol. I, pages 585–594.
[3] Langendoen, E. J. (2001), ‘‘Evaluation of the effectiveness of selected computer models of depth averaged free
surface flow and sediment transport to predict the effects of hydraulic structures on river morphology.’’ Rep.,
October 2001, WES Vicksburg, National Sedimentation Laboratory, Agricultural Research Service, U.S.
Department of Agriculture
[4] Pender, D., Patidar, S., Hassan, K., and Haynes, H. (2016), “Method for Incorporating Morphological
Sensitivity into Flood Inundation Modeling “, J. Hydraul. Eng., 10.1061/(ASCE) HY.1943-7900.0001127,
04016008, 2016.
[5] Abbas, N. and Subramanian, V., 1984. Erosion and Sediment transport in the Ganges river basin (India). J.
Hydrol. 69, 173–182.