2.1. Description of the Study Area
The Upper Gelana watershed is part of the Abaya-Chamo subbasin within the Rift Valley Lakes basin. The Abaya-Chamo subbasin lies in Southern Ethiopia, in the middle of the Main Ethiopian Rift (MER) Valley. The watershed is located between 37°50′–38°20′ east longitude and 5°25′–6°18′ north latitude [
29]. The watershed elevation ranges from 1295 m to 3062 m above sea level, and it covers an area of 1238 km
2 (
Figure 1). The topography of the watershed area is characterized by hills and ridges that produce an undulating land surface with a tiny and narrow plateau at some ephemeral stream sides. The watershed predominantly drains towards the south from the northeast peak to the southwest outlet.
The climate of the Gelana River basin varies from semi-arid on the rift floor to humid in the escarpment mountains [
29]. The mean annual precipitation ranges from 842.94 mm in the lowland rift floors up to 1387.13 mm in the highland areas. Precipitation in this region has a bimodal pattern, with two hyetograph peaks in April–May during the “little rainy season” and in September–October during the “major rainy season”. The warm seasons of the year are from June to August and from November to March, and the annual average temperature is about 21.2 °C. The mean monthly maximum temperature varies from 21.2 to 28.15 °C, and the mean monthly minimum temperature varies from 8.97 to 13.32 °C.
The geology of the watershed (
Figure 2) is extraordinarily intricate and faulted, as numerous earlier studies have described [
30,
31]. Since the study area lies at the escarpment of the Main Ethiopia Rift Valley, local structural features are observed following the main regional structure in the north–south direction. The geological unit of the study area is classified as Lower Basalt: aphyric to porphyritic basalt (Pgl), Alluvium (Qa), Eluvium (Qe), Shale Ignimbrite: rhyolitic ignimber (Pgs), Hornblende-biotite-quartz-feldspar gneiss (Phbg), Magnetite-quartz-feldspar gnesiss (Pmfg), and Rhyolites and trachytes were separated from TV1(Try).
2.2. WetSpass Model
WetSpass model is a physically based and spatially distributed water balance model [
32,
33] that resolves
water and
energy
transfer among
soil,
plants, and
atmosphere in quasi-
steady-
state conditions. The model name, WetSpass, was derived from the first underlined letters of its dominant functions. The WetSpass model is commonly used to estimate the long-term mean spatial patterns of actual evapotranspiration, surface runoff, and groundwater recharge of watersheds at wide ranges of spatial scales. A special version of the model, the WetSpass-M model, is applied for the estimation of spatial groundwater recharge on a monthly, seasonal, and yearly basis for the Upper Gelana watershed. The model categorizes the land use land cover of the watershed into four major classes: vegetated, bare soil, open water, and impervious surfaces. The water balance components of the four predominant land use land cover classes are used to calculate the total water balance components of a raster cell with the following equations [
32]:
where, ET
raster, S
raster, and R
raster are total evapotranspiration, surface runoff, and groundwater recharge of a grid cell, respectively. The water balance components of the four land use land cover (LULC) classes have respective subscripts of (
v) vegetated, (
s) bare soil, (
o) open water, and (
i) impervious area. The terms
av,
as,
ao, and
ai are the fraction area of vegetated, bare soil, open water, and impervious area, respectively.
The transpiration component of the evapotranspiration process can only occur in the vegetated land cover class of the watershed. The total evapotranspiration (ET) is the sum of actual evapotranspiration (AET) and evaporation of water intercepted by vegetation (I). The actual evapotranspiration is often computed from the potential evaporation (PET) of open water with a proper application of vegetation coefficient and water content function [
32].
The WetSpass model assigns interception percentages for different land use land cover classes for both summer and winter seasons. The monthly interception amount for each grid cell is calculated by multiplying the respective interception ratio with the precipitation amount of the grid cell. The remainder amount of the precipitation contributes to the runoff amount at the grid cell with the proper application of a surface runoff coefficient and a soil moisture accounting coefficient for Hortonian runoff. The Hortonian runoff coefficient varies with soil texture and seasons. The detailed mathematical description of the monthly water balance computation of the WetSpass model is widely available in the literature [
32].
2.5. Input Data for WetSpass
The WetSpass model requires two major types of input data: geo-spatially referenced grid maps and parameter tables [
33]. The model input grid maps were prepared for the following biophysical variables; slope angle, land use, soil texture, groundwater depth, and average meteorological maps of precipitation, potential evapotranspiration, temperature, and wind speed on monthly, seasonal and annual temporal scales. Initially, the model input grid maps were prepared at the spatial resolution of the coarser biophysical factor, 12.5 × 12.5 m. The hydro-meteorological observations in the watershed were very sparse, and their grid maps at 12.5 m resolution could not portray clear spatial variability. Therefore, the model input grid maps were further resampled to a spatial resolution of 100 m, and this spatial scale was found to produce coherent outputs during model simulation.
The biophysical parameters of land use, soil, and runoff have to be provided in four lookup tables for the seamless application of the WetSpass model. The attribute tables contain model parameters for different land use classes, soil types, and the two dominant seasons. Relevant literature reviews and the model user guide were carefully used to modify and estimate parameter values of the watershed features. Previous studies in the region, for example [
34], made certain parameter adjustments to the leaf area index, root depth, and bareness for the Geba basin in Ethiopia. Accordingly, through professional consultation and frequent field observations, the relative differences between the Geba basin and the environmental setup of the Upper Gelana watershed are taken into account in parameter estimation for the watershed. The parameter lookup tables are listed in
Table 2,
Table 3 and
Table 4.
The land use map of the watershed was derived from cloud-free Copernicus open access hub Sentinel-2A satellite images on 23 February 2019, using the standard ERDAS IMAGINE supervised image classification method. The six land use classes identified in the land use map (
Figure 4a) of the watershed are agriculture (54.00%), shrub and bare land (21.91%), forest (15.12%), and urban and grassland (8.98%).
The soil map of the watershed was extracted from the soil map data of the Rift Valley Lake basin master plan (scale 1:250,000), which was obtained from the Ministry of Agriculture of Ethiopia. The soil code system used in WetSpass is based on the soil texture triangle developed by the United States Department of Agriculture (USDA), which is characterized by its percentage of clay, silt, and sand, ranging from the fine textures (clay), through the intermediate textures (loam), to the coarser textures (sand). The percentage of the topsoil textures (coarse, medium, and fine) was used to identify the soil type from the universal soil texture triangle. The soil map of the Upper Gelana watershed in
Figure 4b revealed four types of soil texture classes: sandy loam (58.53%), loam (40.98%), loamy sand (0.24%), and clay (0.25%).
The watershed’s digital elevation model (DEM) was obtained from ALOS-PALSAR. It was downloaded from the Alaska Satellite Facility (ASF) website (
https://www.asf.alaska.edu (accessed on 2 November 2018)) at a spatial resolution of 12.5 m. The DEM is used to create a topographic elevation map, a slope map (
Figure 4c), and a river network map (
Figure 1) for the Upper Gelana watershed. The slope angle of the watershed ranges from 0° to 55° with a mean slope angle of 13°. Groundwater depth data were collected from the Regional Water Resources Office. The groundwater level elevation contours were interpolated from the static water levels of 85 boreholes and springs found within the watershed as shown in
Figure 4d.
The meteorological data of the watershed, such as rainfall amount (mm), temperature (°C), sunshine hour, relative humidity (%), and wind speed (m/s), were kindly provided by the National Meteorological Agency (NMA) of Ethiopia. Based on the availability of data and station locations, a total of six meteorological stations in the proximity of the study area were selected. While the Chlelektu, Fesehagenete, Gedebe, Hageremariam, and Yiregachefe weather stations are located within the watershed, the Dilla weather station is situated outside the watershed (
Figure 1). The record length of these daily meteorological variables spans from 1988 to 2019. The Upper Gelana watershed’s hydro-meteorological characteristics and other biophysical features need to be accurately captured by the spatially distributed WetSpass model. The annual and seasonal meteorological grid maps of precipitation, temperature, potential evapotranspiration, and wind speed were prepared from the available meteorological stations in the format of the model input requirements.
In the Ethiopian context, there are three major seasons: July–August–September–October (JASO), November–December–January–February (NDJF), and March–April–May–June (MAMJ) are, respectively, referred to as the Kiremt, Bega, and Belg seasons. The Kiremt and Belg seasons are commonly known as wet seasons, whereas the Bega season is typically dry.
Streamflow data were collected from the Ministry of Water, Irrigation, and Energy (MoWIE) of Ethiopia for the Tore gauging station located near the Tore town. The daily streamflow data cover the period from 1990 to 2019. The streamflow data were used to validate the flow simulation of the WetSpass model, which comprises both surface runoff and groundwater recharge. In addition, the baseflow estimates and results of various previous regional and fragmented studies were used as a supplementary confirmation approach for validation of the total simulated flow [
35,
36,
37]. The basic assumption of this validation exercise is that the stream baseflow is equivalent to the groundwater discharge in that watershed, which implies that groundwater discharge to rivers is equivalent to groundwater recharge [
36]. The BFI+3 baseflow separation program was used to separate the gauging river flow records into the surface runoff and baseflow components. Using a numerical digital filter, the BFI+3 tool can partition the rainfall-runoff discharge into its components; overland surface flow and subsurface baseflow flow.
The groundwater level measurements were obtained from the South Region Water Works & Construction Enterprise (SWWCE). The monthly groundwater level data for the wells in
Figure 4d span from 2017 to 2018. The groundwater level at the wells in the watershed is interpolated to produce a grid input map for the WetSpass model (
Figure 4d).
Table 2.
Lookup parameters for summer land use land cover.
Table 2.
Lookup parameters for summer land use land cover.
No | Land Use Type | Runoff Vegetation | Runoff Class | Impervious Runoff Class | Vegetated Area | Bare Area | Impervious Area | Open Water Area | Root Depth | LAI | Min Stomatal Open | Interception Percentage | Vegetation Height |
---|
21 | Agriculture | Crop | 1 | 0 | 0.8 | 0.2 | 0 | 0 | 0.4 | 4 | 180 | 15 | 0.6 |
7 | Bare land | Bare soil | 4 | 0 | 0 | 1 | 0 | 0 | 0.05 | 0 | 110 | 0 | 0.01 |
33 | Forest | Forest | 3 | 0 | 1 | 0 | 0 | 0 | 2 | 5 | 375 | 35 | 16 |
23 | Grassland | Grass | 2 | 0 | 1 | 0 | 0 | 0 | 0.3 | 2 | 100 | 10 | 0.2 |
2 | Built-up | Grass | 2 | 1 | 0.2 | 0 | 0.8 | 0 | 0.3 | 2 | 100 | 10 | 0.12 |
36 | Shrub land | Grass | 2 | 0 | 1 | 0 | 0 | 0 | 0.6 | 0.6 | 110 | 15 | 2 |
Table 3.
Lookup parameters for winter land use land cover.
Table 3.
Lookup parameters for winter land use land cover.
No | Land Use Type | Runoff Vegetation | Runoff Class | Impervious Runoff Class | Vegetated Area | Bare Area | Impervious Area | Open Water Area | Root Depth | LAI | Min Stomatal Open | Interception Percentage | Vegetation Height |
---|
21 | Agriculture | Crop | 1 | 0 | 0 | 1 | 0 | 0 | 0.35 | 0 | 180 | 0 | 0.6 |
7 | Bare land | Bare soil | 4 | 0 | 0 | 1 | 0 | 0 | 0.05 | 0 | 110 | 0 | 0.01 |
33 | Forest | Forest | 3 | 0 | 0.5 | 0.5 | 0 | 0 | 2 | 4.5 | 500 | 38 | 15 |
23 | Grassland | Grass | 2 | 0 | 1 | 0 | 0 | 0 | 0.3 | 2 | 100 | 10 | 0.2 |
1 | Built-up | Grass | 2 | 1 | 0.2 | 0 | 0.8 | 0 | 0.3 | 2 | 100 | 10 | 0.12 |
36 | Shrub land | Grass | 2 | 0 | 0.2 | 0.8 | 0 | 0 | 0.6 | 0 | 110 | 5 | 2 |
Where LAI is Leaf Area Index.
Table 4.
Soil parameter attribute table.
Table 4.
Soil parameter attribute table.
Soil Type Number | Soil Texture | Field Capacity | Wilting Point | PWA | Residual Water Content | AI | Bare Soil ET | Tension Saturated Height | Pf_Sum | Pf_Win |
---|
12 | Clay | 0.46 | 0.33 | 0.13 | 0.09 | 0.21 | 0.05 | 0.37 | 0.95 | 0.85 |
5 | Loam | 0.25 | 0.12 | 0.13 | 0.027 | 0.37 | 0.05 | 0.11 | 0.15 | 0.02 |
2 | Loamy sand | 0.15 | 0.07 | 0.08 | 0.035 | 0.47 | 0.05 | 0.09 | 0.09 | 0.01 |
3 | Sandy loam | 0.21 | 0.09 | 0.12 | 0.041 | 0.44 | 0.05 | 0.15 | 0.09 | 0.01 |
Where PAW is plant available water content; AI is the calibration parameter dependent on the sand content of the soil; Pf_Sum is the fraction of summer precipitation contributing to Hortonian runoff; and Pf_Win is the fraction of winter precipitation contributing to Hortonian runoff.