Journal of Hydrogeology & Hydrologic EngineeringISSN: 2325-9647

Reach Us +1-209-730-0872
All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

Research Article, J Hydrogeol Hydrol Eng Vol: 8 Issue: 1

Rainfall-Runoff Modelling of Koshi River Basin Using HECHMS

Mukesh Raj Kafle*

Department of Civil Engineering, Institute of Engineering, Pulchowk Campus, Lalitpur, Nepal

*Corresponding Author : Mukesh Raj Kafle
Department of Civil Engineering, Institute of Engineering, Pulchowk Campus, Lalitpur, Nepal
Tel:
9779851041278
E-mail: [email protected]

Received: June 19, 2019 Accepted: July 30, 2019 Published: August 07, 2019

Citation: Kafle MR (2019) Rainfall-Runoff Modelling of Koshi River Basin Using HEC-HMS. J Hydrogeol Hydrol Eng 8:1.

Abstract

This paper presents the results of frequency analysis of rainfall and Rainfall-Runoff modelling of Koshi River basin. The simulation was carried out by model software HEC-HMS. The Gumbel’s method was used for the frequency analysis of 1 day, 2 day and 3-day maximum rainfall data of forty-nine stations. The mean value of PMP for 1 day,2 day and 3 day rainfall were 324 mm, 415 mm and 554 mm respectively. The calibration of model was mainly based on the observed discharge data at Chatara station (outlet of whole basin). The Nash-Sutcliffe efficiency was 83% for calibration and 77% for validation. The volume bias is +1.1% for calibration and +20% for validation. Four types of modelling results in terms of parameters namely deficit and Constant loss parameters, Clark runoff transform parameters, Monthly base flow (m3/s) parameters and Muskingum routing parameters were estimated. The minimum base flow in all sub-basins except Arun lower were resulted in the month of March whereas maximum base flow was estimated in the month of August in all sub-basins. The highest contribution of base flow was resulted in Arun up sub-basin ranging from 107 m3/s even in the month of February to 530 m3/s in the month of August. On the other hand, among sub-basins, the lowest contribution of base flow was resulted in Likhu sub-basin ranging from 11 m3/s in the month of March to 70 m3/ s in the month of August.

Keywords: Rainfall; Runoff; Gumbel; HEC-HMS; Calibration; Validation; PMP; Base flow

Introduction

Simulation and prediction of rainfall and corresponding runoff are essential for planners, stakeholders and policy makers for flood risk management. There are range of methods available to estimate stream flow from catchments, using observed data wherever possible, or using empirical and statistical techniques to estimate river discharge, more commonly known as rainfall-runoff models [1]. Among available various techniques and methods for simulation and predictions rainfall and runoff, each method is different in terms of accuracy, scope, time horizon and the cost [2]. All Rainfall-Runoff (RR) models are the simplified characterizations of the real-world system [3]. Hydrological studies are often aimed at establishing rainfallrunoff relationships [4]. Rainfall-runoff models can be categorized according to the model type [2,5] classified the hydrological models as deterministic or stochastic, global or semi-distributed, kinematic or dynamic and finally empirical or conceptual. The selection of the model depends on the watershed and the objective of the hydrological forecast. Many scientists have conducted important hydrologic studies using HEC-HMS model, which proved its ability to simulate and forecast stream flow. It is applicable in diverse geographic areas as arid environment, tropical catchment, semi-arid region; small areas catchment etc. for solving the widest possible of problems [6-12] used HEC-HMS for rainfall-runoff simulation.

Since long time, flooding in the Koshi River and its management has been challenging issues. Floods from the Koshi River in the past have created havoc in the downstream area of Nepal and India leading to loss of lives and property and causing widespread human suffering. To avoid anticipated flood disaster and timely implementation of preparedness plan the establishment of early warning system (EWS) in the Koshi River is must. A crucial component of early warning system is the flood forecasting modelling, which needs rainfall-runoff simulation.

In this study, the conceptual approach is adopted for the hydrologic modelling using semi-distributed hydrologic model of HEC-HMS in order to investigate the rainfall-runoff interactions in the Koshi River basin.

Study Area and Data

Basin area and tributaries

SaptaKoshi River is popularly known as Koshi in Nepal and India. After the confluence of three rivers Arun, Sun Koshi and Tamor at Tribeni in Sunsari District, Koshi zone of Nepal, the river is named as Sapta Koshi, meaning a river with seven major tributaries. Among the seven rivers: Indrawati, Sunkoshi, Tamakoshi, Likhu and, Dudh Koshi, are the major tributaries of Sunkoshi. Arun and Tamor are the remaining two major tributaries. The Koshi River is a trans-boundary river originating in Tibet (China) and flowing through Nepal and India. It is one of the largest tributaries of the Ganges River. The entire Koshi River basin up to its confluence with the Ganges in India has a catchment area of 69,300 km2, out of which, 29,400 km2 lies in China; 30,700 km2 in Nepal and 9,200 km2 in India. About 5,700 sq. km. area of the catchment is under glaciers. The Koshi River basin as a tributary of Ganges is shown in Figure 1

Figure 1: Koshi basin as a tributary of Ganges.

Hydro meteorological data

Hydrological and meteorological stations, whose data are available continuously, are selected in this study. Forty-nine rainfall stations and twelve gauging stations are considered for the analysis of precipitation characteristics and discharge. The location rainfall stations and hydrological stations are shown in the Figures 2 and 3 respectively. The figure shows that most of the rainfall stations are located in the mid and lower part of the basin. The upper high mountain part (snow covered area) is not easily accessible.

Figure 2: Location of Rainfall stations.

Figure 3: Drainage network of Koshi basin with hydrological stations.

Regarding major sub-basins’ contribution, station nos. 600.1 and 600.4 represent the flow at two locations of Arun Basin whereas station numbers 684 and 690 represent discharge at two locations of Tamor. Stations nos 647, 660 and 670 represent contribution from Tamakoshi, Likhu and Dudhkoshi respectively. Stations 610, 630, 652 and 680 represent locations of flow at Sunkoshi. The rainfall contribution of the monsoon period, which spans from June-September, is around 80%. Almost none or very little rainfall is observed in rest of the period. The spatial variation of annual rainfall ranges from 870 mm in Nepalthok to 4900 mm in Nam No records of snowfall and accumulations are available.

The hydrological data from Chatara gauging site (station no.695), located at 26° 52’ 00” N and 87° 09’ 30” E and elevation of 140 m, is taken for the study and analysis of peak flood. The drainage area of the Koshi basin at Chatra hydrological gauging station is 54100 km2. The average monthly variation of flow at this station shows that the peak discharge occurs in August. The mean annual flow of Chatara Station is 1620 m3s-1 (Figure 4). The extreme flow in the Koshi River at Chatara station was 25879 m3s-1 in 1968 followed by 24240 m3s- 1 in 1954 and 24000 m3s-1 in 1980. The fl ood flo w on 18th August 2008 at Chatara was only about 4250 m3s-1 when the flood disaster in Koshi River was initiated (Source: Department of Hydrology and Meteorology, Government of Nepal) (Figure 5).

Figure 4: Monthly average discharges at Chatara for (1977-2015).

Figure 5: Maximum instantaneous discharge at Chatara (1977-2015).

Methodology

Rainfall frequency analysis

The Gumbel’s method was used for the frequency analysis of 1 day, 2 day and 3 day maximum rainfall data of forty-nine stations. From the station data (1990-2010 period), the frequency of rainfall of different sub-basins was computed.

PMP can be estimated statistically by using the following relationship.

image

Where PMP=Probable maximum precipitation

image =Mean of annual maximum precipitation,

σ=standard deviation of annual maximum precipitation,

K=frequency factor, K is computed as follows:

image (1)

Pl=highest value of precipitation in the series,

image=mean of precipitation time series excluding highest value,

Sn=standard deviation of precipitation time series excluding highest value.

Rainfall - Runoff (R-R) model

The study was conducted with numerical simulation with model software Hydrologic Modelling System [13]. It is open source computer software developed by U.S. Army Corps of Engineering´s Hydrologic Engineering Center that helps in simulating the hydrologic cycle (precipitation, evapotranspiration, infiltration, surface runoff and base flow) of a catchment by describing its physical and meteorological properties. A simple schematic representation of runoff process replicated in HEC-HMS is shown in Figure 6. Wide options of mathematical models for all the hydrological components that conceptually represent watershed behavior are incorporated in this program. The program uses separate model to represent each component of the runoff process l ike model t o c ompute r unoff volume, model of direct runoff/base fl ow/channel flo w as wel l as alternative models to account for the cumulative losses for e.g.: SCS CN loss model. Then, it computes runoff volume by subtracting losses (infiltration, storage, interception, evaporation etc.) from precipitation. HEC-HMS 4.2.1 was used during this project [14].

Figure 6: System diagram of runoff process (Feldman, 2000).

Depending upon the soil type, ground cover, antecedent moisture and other watershed properties, a portion of precipitation infiltrates into the soil. The infiltrated water moves horizontally as interflow just beneath the surface, or it percolates vertically to the groundwater aquifer beneath the watershed. The interflow eventually moves into the stream channel. Water in the aquifer moves slowly, but eventually, some returns to the channels as baseflow.

Water that does not infiltrate moves by overland flow to a stream channel. The stream channel is the combination point for the overland flow, the precipitation that falls directly on water bodies in the watershed, and the interflow and base flow. Thus, resultant stream flow is the total watershed outflow.

HEC-HMS uses a separate model to represent each component of the runoff.

a) Runoff volume model (Loss model)

b) Direct runoff model (transform model)

c) Base flow model d. Routing model.

Loss model represents the volume of water lost due to interception, infiltration, storage, evaporation, and transpiration. Excess precipitation computed after subtracting losses is transformed into direct runoff by using runoff transform model. Base flow model represents the contribution of groundwater. Routing model computes a downstream hydrograph, given an upstream hydrograph as a boundary condition.

Selected components for the study:

Deficit and constant loss model: HEC-HMS includes a quasicontinuous model of precipitation losses; this is known as the deficit and constant-rate loss model. The interception and depression loss are treated as initial loss. This model is similar to the initial and constant-rate loss model, but the initial loss can recover after a prolonged period of no rainfall.

Clark’s model for direct runoff: Clark’s model derives a watershed unit hydrograph by explicitly representing two critical processes in the transformation of excess precipitation to runoff. First: translation or movement of the excess from its origin throughout the drainage to the watershed outlet: based on synthetic time area curve and time of concentration. Second: attenuation or reduction of the magnitude of the discharge as the excess is stored throughout the watershed: modelled with linear reservoir.

Constant, monthly-varying base flow: This is the simplest base flow model in HEC-HMS. It represents base flow as a constant flow; this may vary monthly. This user-specified flow is added to the direct runoff computed from rainfall for each time step of the simulation.

Muskingum model for routing: Muskingum model of flow routing is based on continuity equation and momentum equation approximated by storage equation. Continuity equation is given by:

image (2)

Where, I is inflow, O is outflow and S is storage.

The storage equation of Muskingum model is given by

Where K is a proportionality coefficient

image (3)

X is a weighting factor (0 ≤X ≤0.5)

Solving continuity and storage equation, the equation relating to inflow and outflow is given as:

image (4)

Where C1, C2 and C3 are constants (functions of parameters K and X).

Model parameters of the selected modules: Parameters of Deficit and Constant loss model - Initial loss, Constant loss rate

Parameters of Clark model -Time of concentration (tc), Storage coefficient (R)

Parameters of Muskingum model - K: Time constant, X: Weighting factor

Parameters of Base flow model- Constant, monthly varying base flow

Model setup

To represent the heterogeneity of the hydrological characteristics, the basin is divided into several sub-basins. The sub-basin division is based on the available discharge gauging stations. The schematization of the basin for running HEC-HMS model is shown in Figure 7. The schematic consists of 10 sub-basins, 12 reaches and 12 junctions (Table 1).

S.N Sub basins Area (km2) Rainfall stations Discharge stations
1  Sunkoshi up 4920 1006, 1008, 1017, 1018, 1020, 1023, 1025, 1027, 1028, 1036, 1058, 1063 610, 630
2  Sunkoshi mid 2380 1049, 1103, 1104, 1115 652
3   Sunkoshi lower 2920 1206, 1207, 1210, 1211, 1222 680
4   Tamakoshi 2750 1101, 1102 647
5    Likhu 820 1224 660
6   Dudhkoshi 3900 1202, 1203, 1204, 1219 670
7     Arun 28200 (including Chinese part) 1301, 1317, 1325 600.1, 604.5
8    Tamor 5890 1307, 1308, 1403, 1404, 1405, 1406, 1419 684, 690
9   Arun low 2980 1303, 1305, 1306, 1309, 1321, 1322, 1324  
10   Saptakoshi low 2030 1311, 1316, 1320, 1226 695

Table 1: Sub-basins for HEC-HMS.

Figure 7: HEC-HMS schematization.

Results and Discussion

Rainfall frequency and PMP

The output of 1 day, 2 day and 3 day maximum rainfall for different return periods is presented in Table 2. The comparison of output results for an example, the result of 10-year return period, 1 day maximum rainfall ranges from 119-235 mm, 2 day maximum rainfall from 164-304 mm, and 3 day maximum rainfall from 193-344 mm respectively. The 1 day maximum rainfall with 5-year return period is higher than 100 mm. The data shows that Sunkoshi upper part and Arun upper part have higher magnitude of rainfall (excluding Saptakaoshi lower part). This result is consistent with the percentage contribution presented in Table 3 as these part contribute to higher amount of runoff to Chatara.

  1 day 2 days 3 days
Sub-basins 100 yr 50 yr 25 yr 10 yr 5 yr 100 yr 50 yr 25 yr 10 yr 5 yr 100 yr 50 yr 25 yr 10 yr 5 yr
Sunkoshi up 191 175 158 135 116 258 237 215 185 162 314 289 263 227 200
Sunkoshi mid 175 158 142 119 101 268 243 219 186 160 325 293 261 218 183
Sunkoshi lower 264 235 206 167 136 341 305 268 219 180 421 375 329 266 217
Tamakoshi 198 177 157 129 107 269 242 215 178 149 288 262 235 199 171
Likhu 187 168 149 122 101 243 219 196 164 139 310 280 249 208 175
Dudhkoshi 230 205 181 148 122 283 256 228 191 162 307 281 254 218 189
Arun up 347 310 274 224 185 416 376 336 282 239 477 433 388 328 280
Tamor 186 168 151 127 108 234 214 193 166 144 275 252 228 197 172
Arun low 180 163 145 122 103 263 236 209 172 143 291 262 232 193 161
Saptakoshi lower 361 323 286 235 194 471 421 371 304 250 522 469 416 344 287

Table 2: Frequency analysis for 1 day, 2 days and 3 days maximum rainfall in mm.

S.N. Station no. River Location % Contribution
1 600.1 Arun River Uwagaon 16
2 604.5 ArunRiver Turkeghat 28
3 610 BhoteKoshi Barabise 5
4 630 SunKoshi Pachuwarghat 14
5 647 Tamakoshi Busti 9
6 652 Sunkoshi Khurkot 31
7 660 Likhu Sangutar 3
8 670 DudhKoshi Rabuwa-Bazar 13
9 680 SunKoshi Kampughat 53
10 684 Tamur Majhitar 16
11 690 Tamur Mulghat 22

Table 3: Percentage contribution of tributaries to mean annual flow at Chatara.

PMP of sub-basins (Table 4) presents results of probable maximum precipitation (PMP) for 1 day, 2 day and 3 days for each sub-basins of Koshi River basin. For 1 day PMP, the lowest rainfall was estimated as 250 mm in Sunkoshi mid sub-basin and the highest value of 457 mm was estimated for Saptakoshi sub-basin. For 2 day PMP, the lowest rainfall was estimated as 342 mm in Likhu sub-basin whereas the highest value of 617 mm was estimated for Saptakoshi sub-basin. In addition, for 3 days PMP, the lowest rainfall of 449 mm was estimated in Tamakoshi sub-basin and the highest rainfall of 790 mm was estimated in Saptakoshi sub-basin.

S. No. Sub-basins Rainfall stations PMP PMP PMP
(1 day) (2 days) (3 days)
 1 Sunkoshi up 1006, 1008, 1017, 1018, 1020, 1023, 1025, 1027, 1028, 1036, 1058, 1063 274 361 492
 2 Sunkoshi mid 1049, 1103, 1104, 1115 250 352 521
 3 Sunkoshi lower 1206, 1207, 1210, 1211, 1222 406 508 742
 4 Tamakoshi 1101, 1102 291 381 449
 5 Likhu 1224 274 342 499
 6 Dudhkoshi 1202, 1203, 1204, 1219 404 436 522
 7 Arun up 1301, 1317, 1325 437 519 684
 8 Tamor 1307, 1308, 1403, 1404, 1405, 1406, 1419 305 364 484
 9 Arun low 1303, 1305, 1306, 1309, 1321, 1322, 1324 259 379 481
 10 Saptakoshi 1311, 1316, 1320, 1226 457 617 790

Table 4: PMP of sub-basins in mm.

HEC-HMS modelling results

Performance evaluation of Model: The performance evaluation of model was done by calibration and validation of modelling results with observed data. Data of 1991-1996 periods is used for calibration and 1997-1999 for validation. The calibration is mainly based on the observed discharge data at Chatara station (outlet of whole basin). The hydrograph of observed and predicted discharge at Chatara for calibration and validation is shown in Figures 8 and 9 respectively. The figures show that the model has captured the overall shape of the hydrograph. The low flow discharge is predicted well. The magnitude of high flow is also predicted satisfactorily. There is some error in the time to peak in some cases. The Nash-Sutcliffe efficiency is 83% for calibration and 77% for validation. The volume bias is +1.1% for calibration and +20% for validation.

Figure 8: Hydrograph of calibration at Chatara.

Figure 9: Hydrograph of validation at Chatara.

Model parameters

Four types of modeling results in terms of parameters namely deficit and Constant loss parameters, Clark runoff transform parameters, Monthly base flow (m3/s) parameters and Muskingum routing parameters are presented in tabular forms below (Tables 5-8). The minimum constant loss rate was found 0.01 mm/hr for two sub-basins namely Sunkoshi mid and Tamor. The maximum constant loss rate was estimated 1.3 mm/hr in Arun up sub-basin. The minimum time of concentrations were resulted as 10 hours in Likhu and Koshi lower sub-basins whereas maximum time of concentration was estimated 25 hours for Arun up sub-basins. In addition, the minimum storage constant was resulted 10 hours in Likhu subbasin and maximum storage constant was resulted 50 hours in Arun up sub-basin. The minimum base flow in all sub-basins except Arun lower were resulted in the month March whereas maximum base flow was estimated in the month August in all sub-basins. In the case of Arun lower minimum base of 24 m3/s was resulted in the month February. The highest contribution of base flow was resulted in Arun up sub-basin ranging from 107 m3/s in the even in the month of February to 530 m3/s in the month of August [15]. On the other hand, among sub-basins, the lowest contribution of base flow was resulted in Likhu sub-basin ranging from 11 m3/s in the month of March to 70-m3/s in the month of August.

S.N. Sub-basin Initial deficit Maximum deficit Constant loss rate (mm/hr)
1 Sunkoshi up 6 60 0.3
2 Sunkoshi mid 6 50 0.01
3 Sunkoshilower 5 60 0.3
4 Tamakoshi 5 60 0.1
5 Likhu 4 40 0.1
6 Dudhkoshi 6 60 0.1
7 Arun up 6 60 1.3
8 Arun lower 5 70 0.2
9 Tamor 5 70 0.01
10 Koshi lower 5 60 0.2

Table 5: Deficit and Constant loss parameters.

S.N. Sub-basin Time of concentration (hr) Storage constant (hr)
1 Sunkoshi up 20 45
2 Sunkoshi mid 15 40
3 Sunkoshi lower 15 40
4 Tamakoshi 20 40
5 Likhu 10 20
6 Dudhkoshi 15 35
7 Arun up 25 50
8 Arun lower 20 35
9 Tamor 20 40
10 Koshi lower 10 35

Table 6: Clark runoff transforms parameters.

S.N. Sub-basin Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1 Sunkoshi up 57 50 48 50 80 150 250 300 200 180 100 73
2 Sunkoshi mid 28 24 23 24 39 72 120 145 97 87 48 35
3 Sunkoshi lower 34 30 29 30 48 90 150 180 120 105 60 43
4 Tamakoshi 28 24 24 28 50 100 150 220 140 100 60 40
5 Likhu 15 12 11 12 18 35 55 70 55 40 30 20
6 Dudhkoshi 45 40 35 40 70 110 180 220 180 110 90 60
7 Arun up 108 107 126 158 268 380 430 530 400 280 200 135
8 Arun lower 25 24 30 35 60 80 90 110 90 60 45 35
9 Tamor 70 60 55 80 180 220 380 400 300 200 150 95
10 Koshi lower 24 20 20 20 35 60 105 125 85 75 40 30

Table 7: Monthly base flow (m3/s) parameters.

S.N. Reach K X
1 Reach-1 16.9 0.19
2 Reach-2 16.9 0.19
3 Reach-3 13.5 0.13
4 Reach-4 6.8 0.19
5 Reach-5 6.8 0.09
6 Reach-6 1.2 0.19
7 Reach-7 4.5 0.19
8 Reach-8 4.5 0.19
9 Reach-9 11.3 0.19
10 Reach-10 1.3 0.19
11 Reach-11 0.4 0.09
12 Reach-12 2 0.19

Table 8: Muskingum routing parameters.

Conclusion

In the Koshi River basin, precipitation data from forty-nine rainfall stations and discharge data from 12 gauging stations were used for hydro meteorological analysis. The Gumbel’s method was used for the frequency analysis of 1 day, 2 day and 3 day maximum rainfall data of 49 stations. The mean value of PMP for 1 day, 2 day and 3 day rainfall were 324 mm, 415 mm and 554 mm respectively. The rainfall runoff modelling of the basin was carried out by HECHMS model. The schematization of the basin for HEC-HMS model consisted of 10 sub-basins, 12 reaches and 12 junctions. To represent the heterogeneity of the hydrological characteristics, the basin was divided into 10 sub-basins based on the available dischargeobservation stations.

Data of 1991-1996 periods was used for calibration and 1997- 1999 for validation. The calibration was mainly based on the observed discharge data at Chatara station (outlet of whole basin). The hydrograph of observed and predicted discharge at Chatara for calibration and validation show that the model has captured the overall shape of the hydrograph. The low flow discharge is redirected well. The magnitude of high flow is also predicted satisfactorily. As the performance of the model is good statistically, the model has been found applicable for forecasting purpose and has been recommended.

Four types of modelling results in terms of parameters namely deficit and Constant loss parameters, Clark runoff transform parameters, Monthly base flow (m3/s) parameters and Muskingum routing parameters were estimated. Among those parameters, monthly base flow was considered as an important one from discharge estimation perspective at Chatara. The minimum base flow in all sub-basins except Arun lower were resulted in the month March whereas maximum base flow was estimated in the month August in all sub-basins. In the case of Arun lower minimum base of 24 m3/s was resulted in the month February. The highest contribution of base flow was resulted in Arun up sub-basin ranging from 107 m3/s in the even in the month of February to 530 m3/s in the month of August. On the other hand, among sub-basins, the lowest contribution of base flow was resulted in Likhu sub-basin ranging from 11 m3/s in the month of March to 70-m3/s in the month of August.

References

Track Your Manuscript

Share This Page