Abera Abdi and Ayenew Ecological Processes (2021) 10:41

https://doi.org/10.1186/s13717-021-00305-5

RESEARCH Open Access

Evaluation of the WEAP model in

simulating subbasin hydrology in the
Central Rift Valley basin, Ethiopia
Debele Abera Abdi1,2* and Tenalem Ayenew2

Abstract
Background: The subbasin hydrologic behaviors have been altered by many natural and anthropologic factors
such as climate change and land development activities. Model-based assessment can be used to simulate both
natural hydrological processes, human-induced effects, and management strategies on water resources. For the
Ketar subbasin, the WEAP (water evaluation and planning) hydrologic model was developed that aimed at (1)
evaluating the application of the WEAP model in the Ketar subbasin, (2) evaluating the demonstration of the WEAP
model using model efficiency evaluation criteria, and (3) simulating hydrological processes of the subbasin using
the WEAP model.
Methods: WEAP-based soil moisture method (rainfall-runoff) hydrology routine is comprised of a lumped, one-
dimensional, two-layer soil water accounting that uses empirical functions to designate evapotranspiration, surface
runoff, interflow, and deep percolation for a sub-unit at root zone. A catchment is considered as the smallest
hydrologic response unit. The catchment’s surface hydrological balance is typically estimated by discretizing the
catchment into multiple land uses for which water balance is estimated at root zone.
Results: The monthly measured and simulated streamflow statistics showed a positive strong relationship with R2
of 0.82, NSE of 0.80, and IA of 0.95; and with R2 of 0.91, NSE of 0.91, and IA of 0.98 for calibration and validation
periods respectively. Similarly, the mean monthly measured and simulated streamflow showed an agreement with
R2 of 0.99, NSE of 0.97, and IA of 0.99, and R2 of 0.94, NSE of 0.93, and IA of 0.93 for the periods of calibration and
validation respectively.
Conclusion: The model has demonstrated the capability to represent the hydrologic dynamics of the subbasin
both at monthly and mean monthly periods. In general, the overall model performance evaluation statistics show a
very good agreement between measured and simulated streamflow at the outlet of the subbasin.
Keywords: Hydrologic processes, Model, Soil moisture method

Background physical processes like precipitation, evapotranspiration

A proper understanding of subbasin hydrological pro- (ET), overland flow, infiltration, recharge or discharge,
cesses is essential to estimate changes in the dynamic and groundwater flow and their interactions in the atmos-
response of a hydrologic system in spatio-temporal di- phere, land surface, and sub-surface that govern water
mensions. The earth’s hydrological system encompasses movement, distribution, and change from one system to
the other (Delfs et al. 2013; Niu et al. 2014). These could
* Correspondence: debelersgis@gmail.com be largely influenced by meteorological parameters (pre-
1
Department of Natural Resources Management, Arsi University, P.O. Box 193, cipitation, temperature, solar radiation, wind speed, and
Asella, Ethiopia
2 relative humidity), land surface properties, and soil param-
Department of Earth Sciences, College of Natural Science, Addis Ababa
University, P.O. Box 1176, Addis Ababa, Ethiopia eters. To estimate hydrologic parameters, models are
© The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons
licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 2 of 14

widely used. Hydrological models are simplified mathem- et al. 2007). Presently, water resources management is
atical representation of a real-world system that is used to primarily used model-based scenario analysis to explore
enhance the understanding, prediction of hydrologic pro- different alternatives. Water resources modeling is a re-
cesses (Pacheco 2015; Santos et al. 2014). quirement, and an essential part of scenario analysis in
In the recent decade, hydrologic models play a very water resources management for which different generic
crucial role in properly characterizing subbasin proper- software platforms can be used depending on the spe-
ties by predicting target parameters based on input data. cific purpose (Ahmadi et al. 2018). In Ethiopia, several
They provide a chance to estimate some variables that hydrological studies have been developed using physic-
are difficult to measure in the field because of their in- ally based models (Desta and Lemma 2017; Legesse et al.
herent nature (may vary in spatio-temporal scale) 2003; Zeray et al. 2007). Specifically, in the CRV basin,
(Wheater et al. 2010). Moreover, well-developed, cali- some of these hydrological models have been conducted
brated, and validated models can be used for different to evaluate the hydrologic response of the basin.
applications such as impact evaluation of change in It is well known that basin hydrologic processes can be
some of the input variables (e.g., land-use change, cli- altered by changing land development and climate condi-
mate change, development activities are a few among tions. For instance, the hydrological response of the main
others), and suitable for setting scenarios and analysis. Ethiopian Rift valley basin to land use and climate change
Generally, there are two classes of modeling tech- on streamflow, as well as lakes level fluctuation was studied
niques that are intensively used to estimate time series with physically based distributed Precipitation Runoff Mod-
hydrologic parameters (Mohammadi et al. 2020a, b, c). eling System (PRMS) model (Legesse et al. 2003, 2004,
These are the physical-based models and artificial 2010). They have investigated the basin’s hydrologic pro-
intelligence (AI) models. Recently, the applications of cesses (ET, surface runoff, interflow, baseflow components,
artificial intelligence models have been widely and effect- and evaporation of lakes). Another study with Soil Water
ively used to predict hydrologic and meteorological pa- Assessment Tool (SWAT) model was used by Zeray et al.
rameters (Mohammadi et al. 2020a, b, c). The artificial (2007) to assess the impact of climate change on Ziway
intelligence models can estimate required parameters Lake water level. The same model was developed to assess
based on a series of input predictors without under- the hydrologic processes and distributed water balance of
standing the physical processes (Mohammadi et al. the Ketar subbasin (Desta and Lemma 2017). However, in-
2020a, b, c). In contrast to the artificial intelligence tegrated hydrological and water resource models have not
model, the physically based models are used to simulate been done to date. In this effect, water evaluation and plan-
time series hydrologic parameters through modeling the ning (the WEAP) model was selected to be implemented in
potential interactions among various factors (Fang et al. the CRV basin. The WEAP modeling software, developed
2019; Wang et al. 2016). These groups of models have by the Stockholm Environment Institute (SEI) was used as
been widely practiced across the world; for instance, a basis for building the hydrology component. It is water re-
SWAT (Arnold et al. 1998) and the WEAP (SEI 2007) sources modeling system that includes options to simulate
are a few among others. both the natural rainfall-runoff processes and the manage-
Currently, the Central Rift Valley (CRV) basin has ment of implemented water system (Yates et al. 2005). Pre-
been facing challenges for satisfyingever-growing viously, the WEAP has been successfully applied to
demands for water, domestic use, irrigation, and the re- agricultural and urban catchments for the simulation of cli-
quirement for environmental flow to maintain mate, land use, and population growth in many parts of the
ecosystem health (Alemayehu et al. 2006; Ayenew 2007; world (Höllermann et al. 2010; Joyce et al. 2006; Mehta
Legesse and Ayenew 2006; Pascual-Ferrer et al. 2013). et al. 2013; Purkey et al. 2008).
Consequently, competition has steadily tightened for Therefore, this study aimed at (1) evaluating the appli-
Lake Ziway water abstraction among many water users cation of the WEAP model in the Ketar river subbasin,
(Pascual-Ferrer et al. 2014). Besides, the sustainability of (2) evaluating the demonstration of the WEAP model
water resources has become more challenged under using model efficiency evaluation criteria, and (3) simu-
potential impacts of human-induced changes such as lating hydrological processes of the subbasin using the
land development and climate change. Effective water WEAP model.
resources management depends on a thorough under-
standing of the quantity and quality of available water in Methods
the subbasin (Lorenz and Ziegeweid 2015). Description of the Ketar subbasin
Therefore, a model-based assessment can be used to The Ketar River subbasin is located in the central part of
simulate both the natural hydrologic processes, human- the main Ethiopian Rift Valley basin (Fig. 1) with a drainage
induced effects, and management strategies on water re- area of 327,160 hectares. It supports more than 675,000 res-
sources (Dougherty et al. 2007; Lerner 2002; Scanlon idents’ livelihoods that mainly rely on subsistent farming.
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 3 of 14

Fig. 1 Geographical location of the Ketar subbasin

Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 4 of 14

The Ketar upstream subbasin is one of the main feeders of level. It is a lumped continuous model that is based
water to the Lake Ziway. Eventually, the overflow from on rainfall-runoff method (soil moisture method).
Lake Ziway has a major share in sustaining Lake Abijata. This takes into account a one-dimensional, 2-layer
The subbasin can be divided into three physiographic re- (or “bucket”) soil moisture dynamic accounting sys-
gions: the rift floor (<1750 masl), the transitional escarp- tem that uses empirical functions to partition water
ment (1750–2000 masl), and the highlands (>2000 masl). into ET, surface runoff, sub-surface runoff (i.e.,
Similarly, the climatic conditions characterizing the rift interflow), and deep percolation for a subbasin unit
floor, the escarpment, and the highlands differ greatly. at the root zone (Eq. 1) and shown in Fig. 2 (SEI
Mean annual rainfall over the entire basin is around 900 2007). One of the most important input data is the
mm, showing high spatial variations as no more than 700 climate dataset as precipitation, temperature, relative
mm/year and as high as 1200 mm/year in the rift floor and humidity, and wind speed. The subbasin has a
highlands respectively. There is also temporal variability of unique climate dataset that is uniformly distributed
total rainfall around 59% occurs during the rainy season across subbasin.
(June–September), the remaining 28%, and 13% fall during This method allows for the characterization of land
months of Match–May and October–February respectively use, climate, and/or soil type impacts to these pro-
(Pascual-Ferrer et al. 2014). cesses. The Ketar subbasin is fractionally subdivided
The mean annual temperature is about 15°C in the into a unique set of independent land use/land cover
highlands and around 20ºC in the rift and actual ET var- classes (j) which sums up 100% of the sub basin’s
ies from around 900 mm/year in the highlands to area. To reflect the lumped hydrologic response,
650 mm/year in the rift (Ayenew 2003). The climate is LULC values from each fractional area within the
humid to sub-humid in the highlands and semi-arid in subbasin are then summed, with the surface runoff,
the rift floor. The hydrology of the region is character- interflow, and base flow being linked to a river fea-
ized by seasonal variations. Since precipitation is the ture and ET being lost from the system. For each
only input of water that is available for ET, surface run- fractional area, j of N, a water balance is computed.
off, recharge to groundwater, and interflow, and in turn, When proper linking is made between the subbasin
it determines stream flow that mostly feeds the lakes. unit node and a groundwater node, the deep percola-
tion within the subbasin unit can be transmitted to a
Hydrologic model development surface water body as baseflow or directly to ground-
The WEAP hydrologic model was applied to evaluate water storage and finally, the water balance empirical
the hydrologic behavior at the subbasin or land-use equation was computed as follow (SEI 2007).

Fig. 2 Conceptual diagram and equations incorporated in the soil moisture model after (Sieber and Purkey 2015)
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 5 of 14

XN  
!
dz1; j 5Z 1; j −2Z 21; j R¼ j¼1
A j 1− f j K s; j Z 21; j Þ ð4Þ
Rd j ¼ P e ðt Þ−ET 0 ðt ÞK c; j ðt Þ ð1Þ
dt 3
  For applications where no return flow link is created
RRF
−P e ðt ÞZ 1; j j −f j k s; j Z 21; j − 1−f j K s; j Z 21; j from a catchment to a groundwater node baseflow em-
anating from the second bucket is computed as
Where Z1, j ϵ⌊0, 1⌋ is relative soil water storage, a frac- dz2 XN   
tion of the total effective water storage in the root zone S max ¼ 1− f j K s; j Z 2
1; j −K s;2 Z 22; j ð5Þ
dt j¼1
layer land use j [dimensionless], j is land use and land
cover unit (e.g., cultivated land, forest land, shrub) and Where the inflow to this storage, Smax is the deep per-
each subbasin has N fraction of land use and land cover colation from the upper storage given in Eq. (1) and Ks,2
units, Rdj is soil water holding capacity of land use j is the saturated conductivity of the lower storage (mm/
[mm], Pe is effective precipitation [mm], ET0(t) is refer- time), which is given as a single value for the subbasin,
ence evapotranspiration [mm/day], Kc,j is crop coeffi- and therefore, does not include a subscript, j.
cient land use j; RRFj is runoff resistance factor of land Actual evapotranspiration (ET) is also estimated using
use and land cover j, Pe ðtÞZ 1; j
RRF j
is the surface runoff, f reference ET, crop coefficient (Kc), and soil water level
j
in the modeling unit root zone given by Eq. (6).
k z; j Z 21; j is interflow from the first layer land use j, fj is
 
partitioning coefficient related to the land cover type, ET ¼ ET 0  K c 5Z 1 −2Z 1 2 =3 ð6Þ
soil, and topography for the area j, that divides flow into
horizontal fj and vertical (1−fj) flows, and Ks, j is satu- ET0 is commonly known to be the amount of water
rated hydraulic conductivity of the root zone layer of from a land surface that would be lost to the atmosphere
land use j [mm/time]. where water is adequate to meet the demand for atmos-
In Eq. (1), the second term represents the reference pheric evaporation from the reference surface. ET0 esti-
ET; it is estimated using the Penman-Monteith equation mation uses standard climatological records of solar
modified for a standardized crop of grass, 0.12 m in radiation (sunshine), air temperature, humidity, and
height and with a surface resistance of 69 s/m. Continu- wind speed above an extensive surface of green grass,
ing with Eq. (1), the kc,j is the crop/plant coefficient for shading the ground, and not short of water (Allen et al.
each fractional land cover. The third term represents 1998). The Penman-Monteith method to estimate ET0 is
surface runoff, where RRFj is the runoff resistance factor expressed as:
of the land cover. Higher values of RRFj lead to less sur- 900
face runoff. The fourth and fifth terms are the interflow 0:408ΔðRn −GÞ þ γ u2 ðes −ea Þ
ET 0 ¼ T þ 273 ð7Þ
and deep percolation terms, respectively, where the par- Δ þ γ ð1 þ 0:34u2 Þ
ameter ks,j is an estimate of the root zone saturated con-
ductivity (mm/time) and fj is a partitioning coefficient Where, ET0 is the reference evapotranspiration (mm/
related to soil, land cover type, and topography that frac- day), Rn is the net radiation at the crop surface (MJ/
tionally partitions water both horizontally and vertically. m2day), G is soil heat flux density (MJ/m2day), T is
Thus, total surface runoff (RT) and interflow, from mean daily air temperature at 2 m height (°C), u2 is the
each sub-catchment at time t is given as Eq. (2). wind speed at 2 m height (m/s), es is the saturation
vapor pressure (kPa), ea is the actual vapor pressure
XN RRF (kPa), es − ea is saturation vapor pressure deficit (kPa), Δ
RT ðt Þ ¼ A j P e ðt ÞZ 1; j j Þ ð2Þ
j¼1 is slope vapor pressure curve (kPa /°C), and γ is the psy-
chrometric constant (kPa/°C).
The total interflow (IF), for each sub-catchment at
time t, is given as Eq. (3). The WEAP model calibration and validation
Model calibration (parameter estimation) primarily
X
N aimed at obtaining a set of parameters applicable to the
IFðt Þ ¼ A j f j k s; j Z 21; j ð3Þ subbasin to reasonably represent the hydrology of the
j¼1 Ketar River at the Abura gauging station. It involves the
automatic (Doherty 2010) and/or manual adjustment of
When an alluvial aquifer is introduced into the model model parameters to minimize the difference between
and a runoff/infiltration link is established between the observed and simulated values by tuning the model par-
watershed unit and the groundwater node and ground- ameter values (Ingol-Blanco and McKinney 2013). In
water recharge (R) (volume/time) to the aquifer is com- this study, the manual approach (trial-and-error) was
puted as follows: used to fit as closely as possible; the simulated and
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 6 of 14

measured streamflow data. The WEAP embedded soil performance evaluation metrics have successfully been
moisture method involves seven soil and land use-related implemented with the WEAP model (Barbaro and
parameters (Ingol-Blanco and McKinney 2013; Sieber and Zarriello 2006; Ingol-Blanco and McKinney 2013;
Purkey 2015) that can be used to re-calibrate the hydro- Mccabe and Legates 1999) for measured and simu-
logic model. These are crop coefficient (Kc), soil water lated flows over the calibration and validation periods.
capacity (Sw), deep water capacity (Dw), runoff resistance
factor (RRF), the conductivity of root zone (Ks), conductiv-
ity of deep zone (Kd), preferred flow direction (f) and The WEAP model input data
initial storage fraction at the beginning of simulation of To simulate the subbasin monthly hydrological
upper soil layer (Z1) and initial storage fraction at the be- processes with the WEAP hydrological model, me-
ginning of simulation of lower soil layer (Z2). teorological data are required including monthly
On the other hand, the main importance of imple- precipitation, temperature, wind speed, and relative
menting hydrologic models is to model subbasin hydro- humidity in addition to land use and soil layer
logic conditions. In this effect, the model validation depth, and geographic latitude data. Fifteen years
process is essential to assess the validity of a model to (1991–2005) of monthly precipitation data were col-
simulate the hydrologic response of subbasin for condi- lected from 5 meteorological stations and used both
tions unlike that used during the calibration period model calibration and validation periods. To estimate
(Legesse et al. 2003). the average areal precipitation (depth) across the
subbasin, the Thiessen polygon method was used,
The WEAP model performance evaluation measures and the station weight percentage is presented in
The model performance was assessed using various Table 1. The average monthly temperature, relative
techniques: (1) joint plots of the monthly and mean humidity, and wind velocity data were obtained for
monthly simulated and observed hydrographs, (2) com- the years 1991–2005 from National Meteorological
monly used statistical methods. The coefficient of deter- Agency (NMA). The long-term average monthly pre-
mination (R2) (Krause et al. 2005); Nash-Sutcliffe cipitation, ET0, discharge, and temperature were
coefficient of efficiency (NSE) (Nash and Sutcliff 1970), shown in Fig. 3.
index of agreement (IA) (Willmott 1981), root mean Besides climatic data, the land use and land cover
square error (RMSE), and RMSE-observations standard (LULC) have also a significant role in determining the
deviation ratio (RSR) (Singh et al. 2005) were used to hydrologic processes for the subbasin. The land use
measure the goodness-of-fit of a model as below: and land cover maps were developed by MOWIE and
P  2 the LULC types and their percentage is described in
X i −X Y i −Y Table 2 and Fig. 4. The daily discharge data were ob-
R ¼ P
2
2 P  2 ð8Þ tained from the MOWIE for the Ketar River at Abura
X i −X Y i− Y
for the period of 1991–2005 and later summed to
Pn 2 monthly time steps and pictorially presented in Fig. 3.
ðX i −Y i Þ
NSE ¼ 1−Pni¼1 2 ð9Þ Of which the monthly data from 1991 to 2000 and
i¼1 X i −X 2001 to 2005 were used for calibration and validation
Pn respectively. The missing data were filled by the linear
ðX i −Y i Þ2
IA ¼ 1− Pn  i¼1    2 ð10Þ interpolation method. Another important data input
   
i¼1 Y i −X − X i −X is the soil layer depth and hence the soil depth con-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn ffi sidered is 500 mm and 250 mm for the top and bot-
2
i¼1 ð X i −Y i Þ tom layers respectively.
RSR ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn  2 ð11Þ
i¼1 X i −Y
Table 1 Precipitation percentage weight of stations by Thiessen
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 2 polygon method in the Ketar upstream
i¼1 ðX i −Y i Þ
RMSE ¼ ð12Þ Station Weighted percentage
n
Asella 13.7
Where Xi, Yi ; X , Y ; and n denote the ith measured Bekoji 35.73
monthly discharge data, the ith simulated monthly Degaga 4.41
discharge data, the mean of the measured monthly Ketar genet 34.63
discharge data, the mean of the simulated monthly Kulumsa 11.5
discharge data, and the total number of observation
Sum 100
data respectively. These methods of model
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 7 of 14

Fig. 3 Mean monthly temperature (Temp), evapotranspiration (ET0), discharge, and precipitation (Prec)

Results and discussion validation period, there was a reasonably very good
Model: calibration and validation agreement between the measured and simulated stream-
The monthly based climatic data such as precipitation, flow during this period with R2 of 0.91, NSE of 0.91, and
average temperature, relative humidity, sunshine hour, IA of 0.98. On the other hand, the mean monthly mea-
wind speed, land use, and soil parameters were used to sured and simulated streamflow as shown in Fig. 7a and
simulate streamflow outputs. The model was calibrated Table 6 exhibited a strong agreement with R2 of 0.99,
to estimate land use and soil-related parameters using NSE of 0.97, and IA of 0.99 for the calibration period.
the manual (trial-and-error) method until a good fit is Figure 7b showed that for the mean monthly measured
observed between the measured and simulated stream- and simulated streamflow for the validation period, the
flow. Figure 5 shows a schematic diagram of flow chart achieved performance was with R2 of 0.94, NSE of 0.93,
for input and output, and modeling processes using the and IA of 0.93 for the mean monthly period.
WEAP hydrologic model.
Table 3 shows a range of values for the estimated or Model performance evaluation
calibrated model parameters. Figure 6a and b shows the The efficiency of the model performance was tested
observed and simulated monthly runoff for the Ketar using statistics from the model simulated output and
River basin for the calibration and validation periods re- measured streamflow data. According to Santhi et al.
spectively. The monthly hydrographs of measured and (2001), Van Liew et al. (2007), the values of R2 that are
simulated streamflow statistics presented in Table 4 greater than 0.5 are acceptable with higher values indi-
showed a stronger agreement with R2 of 0.82, NSE of cating less error variance. For monthly hydrographical
0.80, and IA of 0.95 for the calibration period. For the data, NSE values range between −∞ and 1.0 and values
between 0.75<NSE≤1 and 0<=RSR≤0.5 rated as very
Table 2 Land use and land cover data of Ketar subbasin in good; 0.65<NSE≤0.75 and 0.5<RSR≤0.6 as good; 0.5<
2003 NSE≤0.65 and 0.6<RSR≤0.7 as satisfactory; and NSE≤0.5
Land use/land Ketar subbasin area and RSR>0.7 rated as unsatisfactory for monthly data
cover
Hectare Percentage (Moriasi et al. 2007). Another model performance
Cultivated land 210,841.5 64.4
evaluation is the IA that can be used as a standard-
ized measure of the degree of model prediction error
Grassland 1628.7 4
and with values range from 0 to 1; the value of 1 in-
Forestland 11,564.1 20 dicates a perfect agreement and 0 indicates no agree-
Shrubland 96,150.5 9.4 ment at all between the measured and simulated
Barren land 6732.2 2.1 values (Willmott 1981).
Waterbody 190.0 0.1 For the calibration period, the model performance was
Wetland 54.3 0.0
achieved to simulate streamflow with R2, NSE, IA, and
RSR values of 0.82, 0.8, 0.95, and 0.44 respectively. Ac-
Total area (hectare) 327,161 100
cording to Moriasi et al. (2007), this result has shown a
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 8 of 14

Fig. 4 Land use land cover map of the Ketar subbasin

very good agreement between monthly measured and agreement in reproducing the overall streamflow
simulated streamflow in the Ketar River subbasin. Like- characteristics.
wise, for the validation period, there was a reasonably Similarly, previous studies have confirmed that the
good agreement between the measured and simulated capability of the WEAP hydrologic model in reproducing
flows with R2 of 0.91, NSE of 0.91, IA of 0.98, and RSR catchment hydrology processes in a different part of the
of 0.3 during this period. On the other hand, the mean world (Table 5). Among these, Asghar et al. (2019) re-
monthly measured and simulated streamflow (Table 4) ported the WEAP hydrologic model to attain the NSE
shows a stronger agreement with R2 of 0.99, NSE of and R2 values of 0.85, 0.86, 0.89, and 0.87 for the
0.97, IA of 0.99, and RSR of 0.15 for the calibration monthly calibration and validation periods between the
period. For the validation period, the model performance measured and simulated streamflow in the central Indus
agreement indices were R2 of 0.94, NSE of 0.93, IA of basin, respectively. From five gauging stations in the
0.93, and RSR of 0.25 for the mean monthly period. In Western Algeria watersheds, ranging values of NSE=
general, the model was able to maintain a very good 0.23–0.88, and R2=0.74–1.0 were achieved between
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 9 of 14

Fig. 5 The schematic diagram for WEAP hydrologic model procedure

measured and simulated average monthly flows (Hamlat NSE=0.78 and R2=0.83 at calibration and validation pe-
et al. 2013). To balance future water availability and de- riods between measured and simulated flows data
mand, the WEAP model was applied in Benin (Höller- respectively.
mann et al. 2010) obtained values NSE=0.91, R2=0.92, Finally, Ingol-Blanco and McKinney (2013) developed
and NSE=0.78 and R2=0.83 for calibration and validation the WEAP model to assess the hydrologic processes in
periods, respectively. To assess the potential impacts of Rio Conchos Basin, Mexico. Six gauging stations were
climate and land-use change on irrigation water supply used in the basin for model performance evaluation.
in the USA, the WEAP hydrologic model was developed Values of NSE=0.65–0.87; R2=0.92-0.97, and NSE=0.60–
by Mehta et al. (2013) and the model has demonstrated 0.88 and R2=0.92–0.97 pertinence between measured
the capability with values of NSE=0.91, R2=0.92, and and simulated flows were found respectively.

Table 3 Range of values for the estimated parameters used in the WEAP model for the Ketar River subbasin
Parameters Definition Range in values
Kc Crop coefficient 0.34–1.05
SWC Effective soil water capacity of the upper layer (mm) 500
DWC Effective soil water capacity of the lower layer (mm) 250
RRF Runoff resistance factor 2.5–405
RZC Root zone rate of conductivity at full saturation (mm/month) 0.8–-179
DC Rate of conductivity of deep layer at full saturation (mm/month) 4–24
PF Preferred flow direction 0.65
Z1 Initial storage fraction of upper layer at the beginning of the simulation (%) 28–50
Z2 Initial storage fraction of lower layer at the beginning of the simulation (%) 70
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 10 of 14

Fig. 6 Monthly hydrograph of measured and simulated streamflow of the Ketar River at Abura gauging station and monthly precipitation of the
subbasin: a the calibration period and b the validation period

Hydrological processes (2007) which indicated that 88% of precipitation is

Precipitation is the only source of water input for the consumed by ET in the CRV basin of Ethiopia. In
subbasin and it amounts to 930 mm annually in-depth. some months, ET exceeded the amount of precipita-
Over 60% of global precipitation is consumed by ET and tion in January, October, November, and December.
it is tough to get accurate methods to determine it in On the contrary, in months ranging from March to
less time and cost-effective (Mohammadi and Mehdiza- August ET was less than the precipitation amounts
deh 2020; Wang et al. 2019), since it is the major com- and nearly equal in February and September. On the
ponent of the hydrologic cycle affecting water resource other hand, the remaining 14% of precipitation is the
availability and important in water resource develop- Ketar streamflow (runoff) which is comprised of 4.5%,
ment and management (Abtew and Melesse 2013). 8%, and 1.3% baseflow, interflow, and surface runoff
Long-term monthly values hydrological components of respectively.
the Ketar subbasin are presented in Table 6. This study Another study by Desta and Lemma (2017) using the
indicated that the ET loss was estimated to 799 mm/ SWAT model shows lower values as compared to the
year. The remaining is the streamflow (runoff) compo- current study that ranges from 64–69% of precipitation
nents in order of their contribution are the interflow, lost to the atmosphere through ET. The relative propor-
baseflow, and surface runoff account 8 mm/year, 4.5 tion of the three components of the runoff (baseflow,
mm/year, and 1.3 mm/year respectively. ET loss ac- interflow, and surface runoff) depends on the physical
counts for 86% of overall precipitation available water. characteristics of the watershed, the land use, soil, top-
This result is consistent with the study by Jansen et al. ography, geomorphology, and geology characteristics.
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 11 of 14

Table 4 Summary of model performance statistics for measured and simulated the Ketar monthly and mean monthly streamflow at
Abura gauging station
Statistics Monthly Mean monthly
Calibration period 1991–2000
Drainage area (ha) 327,161
Number of months 120
Mean measured flow (m3/s) 13 12.52
3
Mean simulated flow (m /s) 12.5 12.98
Standard deviation (SD) measured 17.8 16
Standard deviation (SD) simulated 17.2 18.2
Coefficient of determination (R2) 0.82 0.99
Nash-Sutcliffe coefficient (NSE) 0.80 0.97
Index of agreement (IA) 0.95 0.99
Root mean square error (RMSE) 7.62 6.56
RMSE-observations standard deviation ratio (RSR) 0.44 0.15
Validation period 2001–2005
Number of months 60
3
Mean measured flow (m /s) 10.4 10.4
Mean simulated flow (m3/s) 9.6 9.6
Standard deviation (SD) measured 14.5 13.1
Standard deviation (SD) simulated 14.4 13.3
2
Coefficient of determination (R ) 0.91 0.94
Nash-Sutcliffe coefficient (NSE) 0.91 0.93
Index of agreement (IA) 0.98 0.98
Root mean square error (RMSE) 4.3 3.3
RMSE-observations standard deviation ratio (RSR) 0.3 0.25

The annual discharge (runoff) proportion of the Ketar Conclusion

River was estimated to 14% of annual precipitation. The model performance evaluation statistics shows a
Subsequently, it is composed of 58% interflow predom- very good agreement between measured and simulated
inantly from June to October and followed by 32% base- monthly streamflow at the outlet for both the calibration
flow which sustained throughout the year, and 10% and validation periods. The WEAP model outcome re-
surface runoff that mostly occurs in July, August, and vealed that the lion’s share of available water returns to
September. These results are consistent with the study the atmosphere via ET in the subbasin. The annual
by Legesse et al. (2003) using the PRSM model in the stream discharge is mainly composed of interflow that
same subbasin. The same authors revealed that of the contributes a major part of flow during the wet season
total annual flow, the interflow was about 60% and the and is followed by the baseflow that is critical to sustain-
most important component during the rainy season, ing the streamflow during the dry season. While consid-
while the baseflow was 30% that maintains the river run- ering water availability in a temporal context, more than
off during the dry season. The same study explains an 80% of the annual total runoff is concentrated in 4
abundance of interflow and baseflow as compared to months (July–October) and the remaining months
surface runoff associated with a presence of huge pro- largely depend on the baseflow contribution to the
portion of cultivated/grazing land in the subbasin that streamflow. This is because of the seasonality of rainfall
enhances infiltration in the soil zone and thereby lateral in the catchment. However, in recent years, water re-
subsurface flow along subsurface channels, macro-pores, quirement for irrigation particularly is increasing in the
and fractures. Contrary to this, the less contribution of dry periods as a result of little precipitation in these
surface runoff to the streamflow is explained by the ab- months.
sence of significant impervious surfaces in the subbasin Therefore, the sustainable water availability of the
(Legesse et al. 2003). Ketar stream throughout the year is largely dependent
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 12 of 14

Fig. 7 Mean monthly hydrograph of measured, simulated flows of Ketar River at Abura gauging station for a the calibration and b the validation
period; black line and green line error bars indicate the standard error of monthly measured and simulated flows respectively

Table 5 Comparison of WEAP hydrologic model performance evaluation with other similar studies elsewhere
Country Model accuracy statistics Modeling Reference
of origin periods
R2
NSE
Ethiopia 0.82 0.8 Calibration Present study
0.91 0.91 Validation
Pakistan 0.96 0.85 Calibration (Asghar et al. 2019)
0.87 0.89 Validation
Algeria 0.74–1.0 0.23–0.88 (Hamlat et al. 2013)
Benin 0.92 0.91 Calibration (Höllermann et al. 2010)
0.83 0.78 Validation
USA 0.92 0.91 Calibration (Mehta et al. 2013)
0.83 0.78 Validation
Mexico 0.92–0.97 0.65–0.87 Calibration (Ingol-Blanco and McKinney 2013)
0.92–0.97 0.60–0.88 Validation
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 13 of 14

Table 6 Average monthly estimated depth of hydrologic Declarations

components in the Ketar subbasin (in mm)
Ethics approval and consent to participate
Month Prec ET Baseflow Interflow Surface runoff Runoff Not applicable.
Jan 23.7 29.7 1.7 0.1 0.0 1.8
Feb 33.5 30.0 1.6 0.1 0.0 1.7 Consent for publication
Mar 64.5 39.4 1.4 0.3 0.0 1.7 All authors agreed and approved the manuscript for publication in
Ecological Processes.
Apr 91.7 57.8 1.8 0.9 0.0 2.7
May 88.3 72.4 2.2 1.0 0.0 3.2 Competing interests
Jun 112.6 83.9 2.5 1.6 0.2 4.3 There is no conflict of interest.

Jul 154.3 83.9 3.4 8.2 2.7 14.3 Received: 6 November 2020 Accepted: 27 April 2021
Aug 159.0 99.6 7.5 30.6 7.8 45.8
Sep 105.8 96.8 10.2 22.6 1.9 34.7
References
Oct 65.0 105.1 5.4 8.2 0.0 13.6
Abtew W, Melesse A (2013) Evaporation and evapotranspiration: measurements
Nov 18.9 66.0 2.6 1.1 0.0 3.7 and estimations. Springer, New York. https://doi.org/10.1007/978-94-007-473
7-1
Dec 12.9 34.6 1.8 0.1 0.0 2.0 Ahmadi A, Ohab-Yazdi SA, Zadehvakili N, Safavi HR (2018) A dynamic model of
Sum 930.2 799.3 42.2 74.7 12.5 129.5 water resources management using the scenario analysis technique
downstream of the Zayandehroud basin. Int J River Basin Manag 17(4):451–
Min 12.9 29.7 1.4 0.1 0.0 1.7 63. https://doi.org/10.1080/15715124.2018.1505734
Max 105.1 105.1 10.2 30.6 7.8 45.8 Alemayehu T, Ayenew T, Kebede S (2006) Hydrogeochemical and lake-level
changes in the Ethiopian Rift. J Hydrol 316(1–4):290–300. https://doi.org/10.1
Mean 77.5 66.6 3.5 6.2 1.0 10.8 016/j.jhydrol.2005.04.024
% 100.0 86 4.5 8.0 1.3 14 Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotranspiration: guidelines
for computing crop water requirements. FAO irrigation and drainage paper
no. 56, Rome
on the baseflow, and anything that alters the hydrologic Arnold J, Srinivasan R, Muttiah R, Williams J (1998) Large area hydrologic
modeling and assessment part I: model development. Am Water Resour
behavior of the subbasin may impact the amount and Assoc 34(1):73–89. https://doi.org/10.1111/j.1752-1688.1998.tb05961.x
the sustainability of streamflow. Later, it may affect se- Asghar A, Iqbal J, Amin A, Ribbe L (2019) Integrated hydrological modeling for
quentially the downstream water users and water level assessment of water demand and supply under socio-economic and IPCC
climate change scenarios using WEAP in Central Indus Basin. J Water Serv
of Lake Ziway and lastly Lake Abijata. The model can be Res Technol 68(2):136–148. https://doi.org/10.2166/aqua.2019.106
used for further related to the impacts of climate change Ayenew T (2003) Evapotranspiration estimation using thematic mapper spectral
and land use/land cover dynamics on the subbasin hy- satellite data in the Ethiopian rift and adjacent highlands. J Hydrol 279(1–4):
83–93. https://doi.org/10.1016/S0022-1694(03)00173-2
drology that in turn affects the basin water availability. Ayenew T (2007) Water management problems in the Ethiopian rift: challenges
for development. J Afri Earth Sci 48(2–3):222–236. https://doi.org/10.1016/j.ja
Abbreviations frearsci.2006.05.010
WEAP: Water evaluation and planning system tool; SWAT: Soil and Water Barbaro BJR, Zarriello PJ (2006) A precipitation-runoff model for the Blackstone
Analysis Tool; PRSM: Precipitation and Runoff Simulation Model; River Basin, Massachusetts, and Rhode Island, U.S. Geological Survey Scientific
Prec: Precipitation; ET: Evapotranspiration; masl: Meter above sea level; Investigations Report 2006–5213, p 95
LULC: Land use and land cover Delfs JO, Wang W, Kalbacher T, Singh AK, Kolditz O (2013) A coupled surface/
subsurface flow model accounting for air entrapment and air pressure
Acknowledgements counterflow. Environ Earth Sci 69(2):395–414. https://doi.org/10.1007/s12665-
We want to express our greatest appreciation to the National Meteorological 013-2420-1
Agencies (NMA) and the Ministry of Water, Irrigation, and Electricity of Desta H, Lemma B (2017) Journal of hydrology: regional studies SWAT based
Ethiopia (MOWIE) for providing the necessary data. The opinions expressed hydrological assessment and characterization of lake. J Hydrol Reg Stud 13:
herein are the authors’ own and do not necessarily express the view of NMA 122–137. https://doi.org/10.1016/j.ejrh.2017.08.002
and MOWIE. Doherty, J. (2010) PEST, Model-Independent Parameter Estimation—User Manual.
5th Edition, with Slight Additions, Watermark Numerical Computing,
Authors’ contributions Brisbane.
The first author is currently conducting his Ph.D. research on model-based Dougherty M, Dymond RL, Grizzard TJ, Godrej AN, Zipper CE, Randolph J (2007)
evaluation of water resources in the Central Rift Valley of Ethiopia. This work Quantifying long-term hydrologic response in an urbanizing basin. J Hydrol
is a major part of the research in the completion of one of the objectives. Eng 12(1):33–41. https://doi.org/10.1061/(ASCE)1084-0699(2007)12:1(33)
Hence, he has made a significant role in all aspects of the research. He is Fang W, Huang S, Ren K, Huang Q, Huang G, Cheng G, Li K (2019) Examining the
researching under the very close supervision of the second author. The latter applicability of different sampling techniques in the development of
has involved from conception and design to analysis and interpretation of decomposition-based streamflow forecasting models. J Hydrol 568:534–550.
data. The author(s) read and approved the final manuscript. https://doi.org/10.1016/j.jhydrol.2018.11.020
Hamlat A, Errih M, Guidoum A (2013) Simulation of water resources management
Funding scenarios in western Algeria watersheds using WEAP model. Arab J Geosci
No funding has been received for this study. 6(7):2225–2236. https://doi.org/10.1007/s12517-012-0539-0
Höllermann B, Giertz S, Diekkrüger B (2010) Benin 2025-balancing future water
Availability of data and materials availability and demand using the WEAP “Water Evaluation and Planning”
The datasets used and/or analyzed during the current study are available system. Water Resour Manag 24(13):3591–3613. https://doi.org/10.1007/s112
from the corresponding author on reasonable request. 69-010-9622-z
Abera Abdi and Ayenew Ecological Processes (2021) 10:41 Page 14 of 14

Ingol-Blanco E, McKinney DC (2013) Development of a hydrological model for Pascual-Ferrer J, Pérez-foguet A, Codony J, Raventós E (2014) Assessment of
the Rio Conchos Basin. J Hydrol Eng 18(3):340–351. https://doi.org/10.1061/ water resources management in the Ethiopian Central Rift Valley:
(ASCE)he.1943-5584.0000607 environmental conservation and poverty reduction. Int J Water Resourc
Jansen H, Hengsdijk H, Dagnachew L, Tenalem A, Hellegers P, Spliethoff P Develop 30(3):572–587. https://doi.org/10.1080/07900627.2013.843410
(2007) Land and water resources assessment in the Ethiopian Central Rift Purkey DR, Joyce B, Vicuna S, Hanemann MW, Dale LL, Yates D, Dracup JA (2008)
Valley, p 81 Robust analysis of future climate change impacts on water for agriculture
Joyce B, Vicuña S, Dale L, Dracup J, Hanemann M, Purkey D, Yates D (2006) and other sectors: a case study in the Sacramento Valley. Climatic Change
Climate change impacts on water for agriculture in California: a case study in 87(S1):109–122. https://doi.org/10.1007/s10584-007-9375-8
the Sacramento Valley, p 85 Santhi C, Arnold JG, Williams JR, Dugas WA, Srinivasan R, Hauck LM (2001)
Krause P, Boyle DP, Bäse F (2005) Comparison of different efficiency criteria for Validation of the SWAT model on a large river basin with point and
hydrological model assessment. Adv Geosci 5:89–97. https://doi.org/10.51 nonpoint sources. J Am Water Resour Assoc 37(5):1169–1188. https://doi.
94/adgeo-5-89-2005 org/10.1111/j.1752-1688.2001.tb03630.x
Legesse D, Abiye TA, Vallet-Coulomb C, Abate H (2010) Streamflow sensitivity to Santos RMB, Sanches Fernandes LF, Moura JP, Pereira MG, Pacheco FAL (2014)
climate and land cover changes Meki River, Ethiopia. Hydrol Earth Syst Sci The impact of climate change, human interference, scale and modeling
14(11):2277–2287. https://doi.org/10.5194/hess-14-2277-2010 uncertainties on the estimation of aquifer properties and river flow
Legesse D, Ayenew T (2006) Effect of improper water and land resource components. J Hydrol 519:1297–1314
utilization on the central Main Ethiopian Rift lakes. Quat Int 148(1):8–18. Scanlon B, Keese K, Flint A, Flint L, Gaye C, Edmunds M, Simmers I (2007) Global
https://doi.org/10.1016/j.quaint.2005.11.003 synthesis of groundwater recharge in semiarid and arid regions. Hydrol
Legesse D, Vallet-Coulomb C, Gasse F (2003) Hydrological response of a Process 20:3335–3370. https://doi.org/10.1002/hyp.6335
catchment to climate and land-use changes in Tropical Africa: case study Sieber J, Purkey D (2015) Water evaluation and planning system: user guide.
south-central Ethiopia. J Hydrol 275(1–2):67–85. https://doi.org/10.1016/ Stockholm Environment Institute, U.S. Center, p 400 Retrieved from http://
S0022-1694(03)00019-2 www.weap21.org/WebHelp/index.html
Legesse D, Vallet-Coulomb C, Gasse F (2004) Analysis of the hydrological Singh J, Knapp H, Demissie M (2005) Hydrologic modeling of the Iroquois River
response of a tropical terminal lake, Lake Abiyata (main Ethiopian Rift Valley) watershed using HSPF and SWAT. ISWS CR 2004-08. Retrieved from http://
to changes in climate and human activities. Hydrol Process 18(3):487–504. scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Hydrologic+
https://doi.org/10.1002/hyp.1334 Modeling+of+the+Iroquois+River+Watershed+Using+HSPF+and+SWAT#3
Lerner DN (2002) Identifying and quantifying urban recharge: a review. Stockholm Environment Institute (SEI) (2007) Water evaluation and planning
Hydrogeol J 10(1):143–152. https://doi.org/10.1007/s10040-001-0177-1 system, WEAP, Stockholm Environmental Institute: Boston
Lorenz DL, Ziegeweid JR (2015) Methods to estimate historical daily streamflow Van Liew MW, Veith TL, Bosch DD, Arnold JG (2007) Suitability of SWAT for the
for ungaged stream locations in Minnesota. In: U.S. geological survey conservation effects assessment project: comparison on USDA agricultural
scientific investigations report, (2015–5181), p 18 Retrieved from http://pubs. research service watersheds. J Hydrol Eng 12(2):173–189. https://doi.org/10.1
usgs.gov/sir/2015/5181/sir20155181.pdf 061/(ASCE)1084-0699(2007)12:2(173)
Mccabe GJ, Legates DR (1999) Evaluating the use of “goodness-of-fit” measures Wang S, Huang G, Baetz B, Huang W (2016) Probabilistic inference coupled with
in hydrologic and hydroclimatic model validation. Water Resour Res 35(1): possibilistic reasoning for robust estimation of hydrologic parameters and
233–241 piecewise characterization of interactive uncertainties. J Hydrometeorol 17(4):
Mehta VK, Haden VR, Joyce BA, Purkey DR, Jackson LE (2013) Irrigation demand 1243–1260. https://doi.org/10.1175/JHM-D-15-0131.1
and supply, given projections of climate and land-use change, in Yolo Wang S, Lian J, Peng Y, Hu B, Chen H (2019) Generalized reference
County, California. Agric Water Manag 117:70–82. https://doi.org/10.1016/j.a evapotranspiration models with limited climatic data based on random
gwat.2012.10.021 forest and gene expression programming in Guangxi, China. J Hydrol 221:
Mohammadi B, Ahmadi F, Mehdizadeh S, Guan Y (2020a) Developing novel 220–230
robust models to improve the accuracy of daily streamflow modeling. Water Wheater HS, Mathias SA, Li X (2010) Groundwater modeling in arid and semi-arid
Resour Manag 34(10):3387–3409. areas. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/
CBO9780511760280
Mohammadi B, Guan Y, Moazenzadeh R, Jafar M, Safari S (2020b) Catena
Willmott CJ (1981) On the validation of models. Phys Geogr 2(2):184–194. https://
implementation of hybrid particle swarm optimization-differential evolution
doi.org/10.1080/02723646.1981.10642213
algorithms coupled with multi-layer perceptron for suspended sediment
Yates D, Sieber J, Purkey D, Huber-Lee A (2005) WEAP21 - a demand-, priority-,
load estimation. Catena 198:105024. https://doi.org/10.1016/j.catena.2020.1
and preference-driven water planning model. Part 1: model characteristics.
05024
Water Int 30(4):487–500. https://doi.org/10.1080/02508060508691893
Mohammadi B, Mehdizadeh S (2020) Modeling daily reference evapotranspiration
Zeray L, Roehrig J, Alamirew D (2007) Climate change impact on Lake Ziway
via a novel approach based on support vector regression coupled with the
watershed water availability, Ethiopia. In: Paper presented at the conference
whale optimization algorithm. Agric Water Manag 237:106145. https://doi.
on international agriculture research for development. The University of
org/10.1016/j.agwat.2020.106145
Bonn: Germany, pp 1–25
Mohammadi B, Mehdizadeh S, Ahmadi F, Thi N, Lien T, Thi N, Linh T (2020c)
Developing hybrid time series and artificial intelligence models for
estimating air temperatures. Stoch Environ Res Risk Assess 7. https://doi.org/1 Publisher’s Note
0.1007/s00477-020-01898-7 Springer Nature remains neutral with regard to jurisdictional claims in
Moriasi DN, Arnold JG, Van Liew MW, Bingner RL, Harmel RD, Veith TL (2007) published maps and institutional affiliations.
Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Trans ASABE 50(3):885–900
Nash J, Sutcliff J (1970) Riverflow forecasting through conceptual models, part
I—a discussion of principles. J Hydrol 10(2):282–290. https://doi.org/10.1080/
00750770109555783
Niu GY, Paniconi C, Troch PA, Scott RL, Durcik M, Zeng X, Huxman T, Goodrich DC
(2014) An integrated modeling framework of catchment-scale ecohydrological
processes: 1. Model description and tests over an energy-limited watershed.
Ecohydrology 7(2):427–439. https://doi.org/10.1002/eco.1362
Pacheco FAL (2015) Regional groundwater flow in hard rocks. Sci Total Environ
506–507:182–195
Pascual-Ferrer J, Candela L, Pérez-foguet A (2013) Ethiopian Central Rift Valley
basin hydrologic modeling using HEC-HMS and ArcSWAT. Geophysical Res
Abst 15(1):10563