You are on page 1of 11

Uncorrected Proof

1 © IWA Publishing 2019 Water Supply | in press | 2019

Simulation of river discharge in ungauged catchments


by forcing GLDAS products to a hydrological model
(a case study: Polroud basin, Iran)
Hadis Pakdel Khasmakhi, Majid Vazifedoust, Safar Marofi
and Abdollah Taheri Tizro

ABSTRACT

Due to unavailability of sufficient discharge data for many rivers, appropriate approach is required to Hadis Pakdel Khasmakhi
Safar Marofi
provide accurate data for estimating discharge in ungauged watersheds. In this study, Global Land Abdollah Taheri Tizro
Department of Water Science Engineering, Faculty
Data Assimilation System (GLDAS) datasets were integrated with HEC-Hydrologic Modeling of Agriculture,
Bu-Ali Sina University,
System (HEC-HMS) to simulate the outlet river discharge in Polroud watershed located in the North of Hamedan,
Iran. Temperature and precipitation products generated by GLDAS were calibrated using regression Iran

analysis based on observation data for the period of 2004–2006. Then, river discharge was simulated Majid Vazifedoust (corresponding author)
Department of Water Science Engineering,
by using HEC-HMS based on two different datasets (GLDAS meteorological product and gauged data) Faculty of Agricultural Sciences,
University of Guilan,
on the scale of the basin for the same period. The results clearly indicated that the forcing of GLDAS Iran
data into HEC-HMS model leads to promising results with acceptable correlation with observed data. E-mail: majid.vazifedoust@gmail.com

Although, in comparison with direct GLDAS runoff products, the proposed approach improved the
accuracy of river discharge, but the problem of underestimation still reduces the expected accuracy.
Because of global accessibility, GLDAS datasets would be a good alternative in ungauged or poorly
gauged watersheds.
Key words | GLDAS datasets, HEC-HMS, river discharge, ungauged watershed

INTRODUCTION

Flooding is the most common natural disaster in the snow- hydrological parameters such as river discharge continues
dominated catchments such as Polroud river basin in to be one of the most widely measured elements.
North of Iran. Therefore, accurate data of river discharge Simulation of river discharge is not usually an easy task
(in time and space) is very important for flood early warning since the hydrological data required for simulation are not
and the proper water resources management. Precipitation, available in such basins. However, global land surface
air temperature and runoff are key determinants of the river models provide a potential alternative source of forcing
discharge and the origin of devastating floods in some cases. data for hydrological models in regions where the hydrologi-
However, ground-based approaches for data collection cal data required for simulation are not readily available.
are very costly and time consuming. In addition, these The Global Land Data Assimilation System (GLDAS)
hydrological parameters have significant gaps over remote (Rodell et al. ) is an important tool on hydrology
and ungauged regions. Notwithstanding the recent decline and water resources due to the provision of hydrological par-
in the number of active river gauges, simulation of ameters on a global scale. GLDAS was designed to provide

doi: 10.2166/ws.2019.160

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
2 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

optimal estimates of land surface fluxes and storages of water In this study, an alternative approach was introduced
and energy. Therefore, many of the problems related to the to simulate the river discharge based on forcing GLDAS
acquisition of land data would be eliminated by providing datasets into the HEC-HMS model (U.S. Army Corps of
the accurate data from this system (Fang et al. ). Engineers (USACE) ) in two cases of the continuous
The GLDAS rainfall product is available at 0.25-degree simulation using soil moisture accounting (SMA) and
spatial resolution and 3-hourly time intervals for entire single-event simulation using the soil Conservation Services
 
latitudes (approximately 60 S–60 N), which is suitable for (SCS) loss method in the Polroud basin.
flood risk analysis studies (Seyyedi et al. ). GLDAS
integrates satellite and ground-based observations for
parameterizing, forcing and constraining global offline simu- MATERIALS AND METHODS
lations of advanced land surface models (LSMs). GLDAS
includes four LSMs: Mosaic (Koster & Suarez. ), the Com- Study area
munity Land Model (CLM) (Dai et al. ), the Variable
Infiltration Capacity (VIC) model (Liang et al. ) and Polroud basin with an area of 1,688 square kilometers
Noah (Koren et al. ; Ek et al. ). Noah was developed located in north latitude from 36 320 to 37 010 and east
by the National Centers for Environmental Prediction (NCEP) longitude from 49 450 to 50 330 , is the largest watershed in
through the cooperation of public and private institutions. the east of Guilan province. The drainage basin is mostly
Because of the importance of GLDAS outputs, these data mountainous, which its maximum and average height
have been applied and evaluated in many studies. relative to sea level is 3,800 and 1,928 meters, respectively.
Numerous estimations and applications have been car- Long-term average annual precipitation is equivalent to
ried out to provide a better and deeper understanding 929 mm and its average annual temperature is 15.5  C.
about GLDAS/Noah products over different regions (Yang The physiographic parameters of Polroud sub-basins
& Koike ; Zhang et al. ; Zaitchik et al. ; He consists of area, slope, height, as well as the length and
et al. ; Van Loon et al. ). slope of the main rivers, which are effective in the rainfall-
Seyyedi et al. () compared TRMM satellite rainfall runoff simulation, have been presented in Table 1.
product and the GLDAS reanalysis precipitation datasets. In the Polroud basin, there are three hydrometric
Results from that study suggest the use of downscaling and stations: ‘Samosh’, ‘Musa Kalayeh’ and Tul-lat, However,
error correction for the GLDAS reanalysis precipitation because of lack of data, just data of Tul-lat station were
dataset before implementing it for runoff simulations. used in this study to simulate the river discharge (Figure 1).
In ungauged watersheds, the accurate extraction of
physiographic characteristics has a great influence on esti-
mating flood discharge in ungauged basins; however, the GLDAS
overall accuracy of GLDAS runoff outputs are not suitable
for estimation of river discharge (Zaitchik et al. ). The GLDAS (Rodell et al. ) combined satellite and ground-
subsequent approach may be entering the precipitation and based observations with the aim of estimating terrestrial
near surface air temperature data generated by GLDAS water and energy storages. GLDAS raster outputs for the
imported into hydrological models such as Hydrologic period 2004–2006 including rainfall, near surface air
Engineering Center Hydrologic Modeling System (HEC- temperature, snow water equivalent, surface and subsurface
HMS) (U.S. Army Corps of Engineers (USACE) ). runoff were downloaded from Earthdata portal and then
HEC-HMS has been successfully used by many meteorological outputs including rainfall, near surface air
researchers to simulate runoff and river discharge in many temperature were forced to the calibrated HEC-HMS
basins worldwide (Oloche Oleyiblo & Zhi-jia. ; Kure model to simulate the river discharge in Tul-lat outlet. The
et al. ; Bhuiyan et al. ; Gao et al. ; Quedraogo GLDAS outputs implemented in this study have been
et al. ). summarized in Table 2 (Fang et al. ).

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
3 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Table 1 | The physiographic parameters of the Polroud sub-basins from WMS model

Slope (%)

Watershed Area (km)2 Mean Basin Elevation (m) Stream Basin Main Stream (km) Perimeter (km)

Sub-basin 1 1.25 1,632.10 0.28 0.4 0.3 7.44


Sub-basin 2 243.95 1,127.20 0.06 0.41 34.3 111.3
Sub-basin 3 101.02 1,463.10 0.13 0.46 18.10 59.49
Sub-basin 4 156.75 1,507.50 0.07 0.38 21.73 81.31
Sub-basin 5 476.20 1,790.40 0.02 0.24 30.77 154.96
Sub-basin 6 144.51 2,240.80 0.07 0.37 30.77 74.18
Sub-basin 7 565.26 2,327.60 0.05 0.40 48.94 159.12

Figure 1 | Location of Polroud basin with its sub-basins and the HEC-HMS hydrographic network.

Simulation of river discharge and WMS functions. Runoff processes were simulated on
each sub-basin system from the upstream to the watershed
To simulate the river discharge during the period of 2004– outlet throughout the streamflow network.
2006, both the gauge data and calibrated GLDAS data was In this study, the constant monthly base flow method
separately entered to the HEC-HMS model. The watershed was used to simulate the base discharge. The constant
was first divided into homogeneous sub-basins (Figure 1) monthly base flow method allows the specification of a
using WMS software. Then, the hydrological modelling constant base flow for each month of the year (HEC
was developed in HEC-HMS for the watershed using the 2000). Therefore, three-year average of river baseflow for
sub-basin geometric data (Table 1). The model takes into each 12 months were entered to the model as constant
account the influences of physical parameters of the water- monthly baseflow data. Simulations was performed using
shed such as climatic, topography, landuse and soil data both the SMA loss method (for the continuous simulation)
representing the boundary condition over the watershed to and SCS loss method (for the single-event simulation).
simulate runoff. The hydrology parameters needed in the The SCS-CN loss method is used in runoff estimation to
rainfall-runoff modelling were generated using GIS layers specify the amount of infiltration rates of soils. The method

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
4 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Figure 2 | Schematic of soil moisture accounting algorithm (HEC 2000).

Table 2 | Main characteristic of GLDAS outputs implemented for river discharge simulation

Parameter Unit Format Spatial resolution Temporal resolution Period

average surface temperature, K GRIB 0.25 3-hourly 2004–2013


2 
Snow water equivalent kgm GRIB 0.25 3-hourly 2004–2013
Rainfall rate, kgsm2 GRIB 0.25 3-hourly 2004–2013
2 
Surface runoff, Subsurface runoff kgm GRIB 0.25 3-hourly 2004–2013

uses an integration of landuse and soil data to determine CN The initial abstraction (Iα) for each sub-basin was con-
values of the watershed. The CN values were adopted from sidered as 20 percent of basin storage (Tassew et al. ):
Technical Release 55 (USDA-NRCS 1986) based on hydro-
logic soil groups (HSGs). The surface storage coefficient Ia ¼ 0:2S (2)
(S) was calculated for each sub-basin via CN value based
on Equation (1) as follow:
HMS includes also SMA method to define the rainfall,
runoff, storage and losses relationships and simulate the
1000 continuous simulation. As shown in Figure 1, in SMA algor-
S¼  10 (1)
CN ithm five storage zones are considered. For the simulation of

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
5 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
water movement through the various storage zones, 12 par- u n
uX (Qsim  Qobs )2
ameters including the maximum capacity (maximum depth) RMSE ¼ t (5)
1
n
of each storage zone, initial storage condition in terms of
percentage of the filled portion of each zone, and the trans- P
n
fer rates, such as the maximum infiltration rate, are required (Qsim  Qobs )
1
Bias ¼ (6)
(Fleming & Neary ). The whole 12 parameters needed n
for the SMA were taken into consideration in this simu- 0 P n
2
1
1 (Qobs Qsim )
lation. The maximum infiltration rate and the maximum B 1 C
B C
soil depth as well as the percolation rates and groundwater E¼1B P n
C (7)
@ n
1
2
A (Qobs Qobs )
components had significant influence on the simulated flow 1
n
discharges. The remaining parameters were also adjusted to
match the simulated and observed peak flows, volumes, time (Qsim  Qobs )
PEP ¼ (8)
to peaks and hydrograph shape. Qobs
The expression used in SCS method for estimating
runoff may be presented as: where, Qobs and Qsim show the observed and simulated dis-
charges, respectively; n is the number of observations; and
Qobs indicates the average observed discharge.
(P  Ia )2
Q¼ (3)
((P  Ia ) þ S)

where, Q is accumulated storm runoff in m and P is accumu- RESULTS AND DISCUSSION


lated storm rainfall in mm.
Routing of the runoff process from the upstream to the Simulation of river discharge based on meteorological
watershed outlet throughout the streamflow network was data from station
conducted using lag time value calculated for each sub-
basin as follow: To improve the accuracy of simulation, internal parameters
in the both functions of SCS and SMA were calibrated
L0:8 (S þ 1) 0:7 manually using gauge data during the period of 2004–
Tl ¼ (4)
1900y0:5 2006. The calibrated parameters for the loss methods includ-
ing SCS with seven sub-basins and SMA for whole basin
where Tl shows the basin lag time, the interval between pre- have been demonstrated in the Tables 3 and 4, respectively.
cipitation center to peak point of hydrograph (hours), L The results of river discharge simulation using HEC-
indicates the length of the main river (feet), y is the average HMS model based on SMA loss method and meteorological
slope of the basin (percentage).
The accuracy assessments were performed based on Table 3 | The calibrated parameters for the SCS loss method with seven sub-basins

statistical criteria including the coefficient of determination


Watershed Curve Lag Time Initial Abstraction
(R2), Nash-Sutcliffe coefficient (E), Bias, root mean square Sub-basin (km2) Number (min) (Ia)

error (RMSE), as follows: Sub-basin 1 1.25 40 438 76.2


Sub-basin 2 223.9 40 3,888 76.2
0 n 12
P Sub-basin 3 101.02 38 2,526 82
1
B 1 (Qobs  Q obs )(Qsim  Qsim ) C
B C Sub-basin 4 136.75 40 2,958 76.2
B C
R ¼B n C
2

B C (4) Sub-basin 5 421.20 48 3,744 50.8


B Qobs × Qsim C
@ A Sub-basin 6 134.3 48 2,196 50.8
Sub-basin 7 516.2 48 4,230 50.8

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
6 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Table 4 | The calibrated parameters for the SMA loss method for the whole basin was implemented between the simulated and the observed
data values. Table 6 presents the error rate for the simulation
SMA Parameter Value
of 44 flooding cases in the Polroud basin. The values of
Canopy storage (mm) 1.2
minimum, maximum and average error are 2%, 35% and
Surface storage (mm) 4
15.6%, respectively. The results indicate that the accuracy
Max rate of infiltration (mm/hr) 12
of the SCS loss method for simulation of flooding events
Soil storage (mm) 120
were favorable and acceptable.
Tension storage (mm) 110
Soil Percolation (mm/hr) 2
Simulation of river discharge based on calibrated
Groundwater 1 storage (mm) 250
GLDAS data
Groundwater 1 percolation (mm/hr) 1.2
Groundwater 1 coefficient (hr) 120
To see the possibility of river discharge simulation in the
Groundwater 2 storage (mm) 200
ungauged catchments using global dataset, calibrated
Groundwater 2 percolation (mm/hr) 0.8
meteorological GLDAS data was entered to the HEC-
Groundwater 2 coefficient (hr) 150
HMS model to simulate the river discharge during the
period of 2004–2006. Meteorological GLDAS datasets
data from Tul-lat station have been presented in the Table 5 were first calibrated using bias correction based on the
as well as Figure 3. The simulation was performed from 1 gauge measurements from 2004 to 2006. We studied the
January 2004 to 30 December 2006 (36 months). Period correlation between GLDAS products and gauge records,
of 2004–2005 was used for the calibration process and the performed the regression analysis and applied the bias
period of 2006 was assigned for model verification. The corrections to the GLDAS dataset.
results demonstrated that simulated discharge agree signifi- Spatial distribution of both annual accumulated precipi-
cantly well with the gauge observation. The correlation tation and annual averaged air temperature as well as their
between simulated discharge and gauged discharge scatter plot with gauged data has been indicated in Figure 4.
station

was in the range of 0.75 and 0.86 with RMSE of 6.9–10.3 The results showed that GLDAS near surface air tempera-
respectively. As shown in the Table 5, 2005 (calibration ture in daily time steps have a good correlation with gauge
period) was identified as the best performance with R2, records, however, calibrated GLDAS precipitation indicates
Nash and RMSE equal to 0.81, 0.80 and 6.9, respectively. a rather low accuracy and a range of underestimation com-
In order to evaluate the performance of simulation pared to the similar gauged data. In the Figure 4(a), the areas
based on SCS loss method, the rainfall-runoff model was with the highest temperature are colored light brown and
implemented for single flooding events. Therefore, analyzing the areas with the lowest temperature are colored dark
the errors for such cases performed independently without brown. The scatter plot in this figure indicates high corre-
considering the base flow. Based on gauge records, flooding lation between GLDAS and gauged temperature after
events were recognized in each year separately and flooding calibration. Spatial distribution of precipitation provides a
was simulated based on SCS loss method. For accuracy better insight into how precipitation is distributed over the
assessment, index of Percentage Error Parameter (PEP) catchment. As clearly demonstrated in Figure 4(b), the
highest precipitation occurs in the coastline area of Caspian
Table 5 | The statistical assessment of simulated river discharge station using HEC-HMS Sea with dark blue color. The scatter plot between GLDAS
based on SMA loss method in Polroud basin and gauged precipitation indicates rather low correlation
between these two datasets.
Period R2 Nash Bias RMSE
Figure 5, compares the results of river discharge
2004 0.75 0.64 6.3 10.3
simulation derived by running HEC-HMS using calibrated
2005 0.81 0.80 0.12 6.9
GLDAS datasets with GLDAS runoff product and gauged
2006 0.86 0.85 1.69 8.0
discharge data.

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
7 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Figure 3 | Comparisons between daily simulated discharge station (hollow circles for 2006, black circles for 2005 and gray circles for 2004), gauged discharge (black line) and gauged
precipitation (gray line) in the Polroud basin for the period of 2004–2006.

Although GLDAS datasets have simplified the way we GLDAS dataset (temperature, rainfall) into the simulation
access data in ungauged catchments, there is evidence that process led to promising results comparing to the obser-
we may not able to use them without data calibration. The vation data. The results confirmed more than 50%
primary result indicated that in case of direct use of improvement in accuracy of simulation after calibration pro-
GLDAS runoff products without initial calibration will cess of forcing GLDAS dataset, but the problem of
lead to low accuracy (R2 ¼ 0.14–0.3). The comparison underestimation still exists.
results shown in the Figure 5 indicates that forcing Table 7 indicates the statistical assessment of simulated
river discharge derived by forcing GLDAS data into the
HEC-HMS model instead of using gauge data. The corre-
Table 6 | The error rate for the simulation of 44 flooding cases (period of 2004–2006)
using SCS loss method in the Polroud basin lation between simulated and observation data was in the
range of 0.36 and 0.66 with RMSE of 12–14 respectively.
Year 2004 2005 2006 period of 2004–2006
Based on Table 7, the best performance was observed in
Number 14 15 15 44 the year 2006 (R2 ¼ 0.66, Nash ¼ 0.63, Bias ¼ 1.6 and
Minimum 2 3 6.9 2 RMES ¼ 13).
PEP (%) Maximum 35 25.6 21.6 35 As shown in Figure 5, both the GLDAS runoff products
Average 22 10.9 14.1 15.6
or discharge simulated by forcing GLDAS dataset into the

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
8 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Figure 4 | (a) the spatial distribution of annual near surface air temperature ( C) from GLDAS product averaged over period of 2004–2006, Scatter plot of GLDAS temperature and gauge
temperature for the years of 2004–2006, (b) the spatial distribution of annual accumulated GLDAS precipitation product (mm) averaged over period of 2004–2006, scatter plot of
GLDAS precipitation and gauge rainfall for the years of 2004–2006. GLDAS dataset has been calibrated with gauged data.

HEC-HMS have underestimation problem, compared to for hydrological modeling as a substitute approach in a
gauge data specially from January to June because of snow used as ungauged catchment.
dominated. From July to December, the difference was less In this study, the SMA and SCS loss method in
marked and it can ensure the accuracy and usefulness HEC-HMS model was calibrated using available gauge
GLDAS runoff products. observations in the Polroud basin. The SCS loss method
was used to simulate 44 flooding events during the period
of 2004–2006 and SMA was implemented for continuous
CONCLUSIONS simulation of river discharge in the same period. The results
indicated that the simulated river discharge agreed well with
The accurate data with proper resolution (both spatial and gauge records.
temporal) are essential for river discharge studies. However, To evaluate the application of GLDAS products directly
due to lack of data in many basin, GLDAS datasets pro- and indirectly for estimation of river discharge in ungauged
duced by global land surface models can be implemented catchment, GLDAS data including average near surface air

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
9 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Figure 5 | Comparison of daily discharge from GLDAS runoff product (gray line and gray circles) with observed discharge (black line), simulated discharge based on forcing GLDAS data
(dotted line and black circles) and simulated discharge based on station data (hollow circles) for the years of 2004 (a), 2005 (b) and 2006 (c).

Table 7 | Statistics analysis of simulated daily discharge (a) using forcing GLDAS datasets into the HEC-HMS (b) direct GLDAS runoff product in the Polroud river basin during the period of
2004–2006

Simulated Discharge with forcing GLDAS dataset GLDAS runoff product

Period R2 Nash RMSE Error % R2 Nash RMSE Error %

2004 0.38 0.04 14.1 40 0.19 1.1 21 77


2005 0.36 0.34 12.9 20 0.33 0.064 15.48 53
2006 0.66 0.63 13 19 0.14 0.16 23.5 58

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
10 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

temperature and rainfall was calibrated using available northwestern China. Sciences in Cold and Arid Regions.
gauge observations in the basin. Then GLDAS data was 7 (2), 0157–0169. doi:10.3724/SP.J.1226.2015.00157.
Koren, V., Schaake, J., Mitchell, K., Duan, Q.-Y., Chen, F. &
forced into the HEC-HMS to simulate continuously the Baker, J. M.  A parameterization of snowpack and
river discharge during the period of 2004–2006 based on frozen ground intended for NCEP weather and climate
SMA loss method. models. Journal of Geophysical Research 104 (D16),
19,569–19,585.
The results clearly indicated that the forcing of GLDAS
Koster, R. D. & Suarez, M. J.  Modeling the land surface
data into HEC-HMS model leads to promising results with boundary in climate models as a composite of independent
acceptable correlation with observed data. Although, in vegetation stands. Journal of Geophysical Research 97 (D3),
comparison with direct GLDAS runoff products, the pro- 2697–2715. doi:10.1029/91JD01696.
Kure, S., Jang, S., Ohara, N., Kavvas, M. L. & Chen, Z. Q. 
posed approach improved the accuracy of river discharge,
Hydrologic impact of regional climate change for the
but the problem of underestimation still reduces the snowfed and glacierfed watersheds in the Republic of
expected accuracy. GLDAS underestimates most of the par- Tajikistan. Hydrological response of flow to climate change.
ameters (rainfall, runoff and temperature) (Zaitchik et al. Hydrological Processes 27, 4057–4070. doi:10.1002/hyp.
9535.
), while this model is able to predict the process of Liang, X., Lettenmaier, D., Wood, E. F. & Burges, S. J. 
changes in parameters over the time correctly. A simple hydrologically based model of land surface water
and energy fluxes for GSMs. Journal of Geophysical Research
99 (D7), 14,415–14,428. doi:10.1029/94JD00483.
Oloche Oleyiblo, J. & Zhi-jia, L. I.  Application of HEC-HMS
REFERENCES
for flood forecasting in Misai and Wan’an catchments in
China. Water Science and Engineering 3 (1), 14–22. doi:10.
Bhuiyan, H. A. K. M., McNairn, H., Powers, J. & Merzouki, A. 3882/j.issn.1674-2370.2010.01.002.
 Application of HEC-HMS in a cold region watershed Quedraogo, W. A. A., Messo Raude, J. & Mwangi Gathenya, J.
and use of RADARSAT-2 soil moisture in initializing the  Continuous modeling of the Mkurumudzi river
model. Journal of Hydrology 4 (9), 1–19. doi:10.3390/ catchment in Kenya using the HEC-HMS conceptual model:
hydrology4010009. calibration, validation, model performance evaluation and
Dai, Y., Zeng, X., Dickinson, R., Baker, I., Bonan, G. B., sensitivity analysis. Hydrology 5 (44), 1–18. doi:10.3390.
Bosilovich, M. G., Denning, A. S., Dirmeyer, P. A., Houser, Rodell, M., Houser, P. R., Jambor, U., Gottschalck, J., Mitchell, K.,
P. R., Niu, G., Oleson, K. W., Schlosser, C. A. & Yang, Z. Meng, C. J., Arsenault, K., Cosgrove, B., Radakovich, J.,
 The common land model (CLM). American Bosilovich, M., Entin, J. K., Walker, J. P., Lohmann, D. &
Meteorological Society 84 (8), 1013–1023. doi:10.1175/ Toll, D.  The global land data assimilation system.
BAMS-84-8-1013. Bulletin of the American Meteorological Society 85 (3),
Ek, M. B., Mitchell, K. E., Lin, Y., Rogers, E., Grunmann, P., 381–394.
Koren, V., Gayno, G. & Tarpley, J. D.  Implementation of Roy, D., Begam, S., Ghosh, S. & Jana, S.  Calibration and
Noah land surface model advances in the national centers for validation of HEC-HMS model for a watershed in eastern
environmental prediction operational mesoscale Eta model. India. ARPN Journal of Engineering and Applied Sciences
Journal of Geophysical Research 108 (D22), 8851. doi:10. 8 (1), 40–56.
1029/2002JD003296. Seyyedi, H., Anagnostou, E. N., Beighley, E. & McCollum, J. 
Fang, H., Beaudoing, H., Rodell, M., Tengl, W. & Vollmer, B.  Hydrologic evaluation of satellite and re-analysis
Global land data assimilation (GLDAS) products, services precipitation datasets over mid-Latitude basins. Atmospheric
and application from NASA hydrology data and information Research 164–165, 37–48.
services center (HDISC). In: ASPRS 2009 Annual Tassew, B. G., Belete, M. A. & Miegel, K.  Application of
Conference Baltimore, March 8–13, Maryland. HEC-HMS model for flow simulation in the Lake Tana
Fleming, M. & Neary, V.  Continuous hydrologic modeling Basin: the case of Gilgel Abay Catchment, Upper Blue Nile
study with the hydrologic modeling system. Journal of Basin, Ethiopia. Hydrology Open Access Journal. Hydrology
Hydrologic Engineering 9, 175–183. 6, 21. doi:10.3390/hydrology6010021.
Gao, P., Carbone, G. J. & Lu, J.  Flood simulation in South United States Army Corps of Engineers  HEC-HMS Technical
Carolina watersheds using different precipitation inputs. Reference Manual. Hydro. Eng. Cent., Davis, Calif.
Advances in Meteorology Journal. Article ID 4085463. Van Loon, A. F., Kumar, R. & Mishra, V.  The European 2015
https://doi.org/10.1155/2018/4085463. drought from a groundwater perspective: estimation in
He, S. W., Nan, Z. T. & Hou, Y. T.  Accuracy evaluation of two absence of observed groundwater data. Hydrology and Earth
precipitation datasets over upper reach of Heihe River Basin, System Science doi:10.5194/hess-2016-561.

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user
Uncorrected Proof
11 H. P. Khasmakhi et al. | Simulation of discharge in ungauged catchment by forcing GLDAS Water Supply | in press | 2019

Yang, K. & Koike, T.  Satellite monitoring of the surface water Resource Research 46, W06507. doi:10.1029/
and energy budget in the Central Tibetan Plateau. Advances 2009WR007811.
in Atmospheric Science 25, 974–985. doi:10.1007/s00376- Zhang, J., Wang, W.-C. & Wei, J.  Assessing land-atmosphere
008-0974-8. coupling using soil moisture from the Global Land Data
Zaitchik, B. F., Rodell, M. & Olivera, F.  Evaluation of the Assimilation System and observational precipitation. Journal
Global Land Data Assimilation System using global river oF Geophysical Research 113, D17119. doi:10.1029/
discharge data and a source to sink routing scheme. Water 2008JD009807.

First received 2 March 2019; accepted in revised form 19 October 2019. Available online 5 November 2019

Downloaded from https://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2019.160/615421/ws2019160.pdf


by UNIVERSITY OF MASSACHUSETTS user

You might also like