Change detecting land use/land cover helps assess and quantify its impact on water resources. This study aims to assess the impacts of land-use/land-cover change on water resources using the SWAT model in the Tana sub-basin, Ethiopia. The research detects and presents the changes between three LULC maps (1986, 2000, and 2014). The results suggest that over the last 28 years, the water body is the least disturbed and sub-afroalpine vegetation is the most transformed in terms of coverage. Cultivated lands gained a large area of cover from the other types. Most of the vegetation cover showed a decreased trend. Forest land and grassland decreased continuously while wetland showed a small variation compared to the other cover types. On the other hand, bush and shrubland recorded about a 1% increase in the total area and an unexpected fast decline in the second period. LULC changes would have an impact on water resources in the study area. The average annual water yield increased by 14.88 and 12.6%, baseflow increased by 18.4% and decreased by 7.16%, surface runoff increased by 12 and 16.16%, evapotranspiration decreased by 18.39 and 13.49%, for 2000 and 2014, respectively, compared to baseline 1986.

  • The impact of land-use/land-cover change on water resources is assessed in the Tana sub-basin.

  • The study analysis indicated that the changes in LULC have implicated effects on the water balance components using SWAT.

  • The approach used has accredited contributions and provides perceptible information that will allow stakeholders and decision-makers to make prominent choices regarding natural resource planning and management.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Land-use changes are altering the hydrologic system and have potentially large impacts on water resources (Wagner 2014). Ethiopia's natural resources are influenced by several interlinked factors such as agricultural expansion, population pressure, rapid urbanization, migration and resettlement, climate change, and environmental pollution (Wassie 2020). There is, therefore, a meaningful change in land use/land cover. Most of the study area population lives in rural areas and their livelihood depends on agriculture (McCartney et al. 2010; Abera et al. 2020). As cited in the study of Nile Transboundary Environmental Action Project of Nile Basin Initiative, the author explained that the land-use trend of the Tana sub-basin describes an area that is heavily cultivated and populated (Alemayhu 2006). The expansion of Bahir Dar and other cities in the Amhara Region is putting pressure on natural resources and land with resultant compromises for biodiversity conservation and food security. The status of land-cover and continued land-use changes as a result of uncontrolled land fragmentation and the intensive use of sub-land division and deforestation have long been encouraging massive soil erosion rates almost in all parts of the watershed (Bogale 2020; Tewabe & Fentahun 2020). Poor agricultural practices and improper grazing land management, overexploitation of natural resources, and climate change and variability will result in increasing pressure on the limited water resources. In the future, there is significant potential for further socio-economic development based on increased utilization of water in the catchment (McCartney et al. 2010).

Land-use/land-cover (LULC) changes have potentially large impacts on water resources (Stonestrom et al. 2009). The spatial distribution of land cover information is important for different purposes. The trend of LULC of an area can be used as one component in the determination of a planning unit for its actual and potential appraisal of predefined land utilization types. Land-use and -cover maps are frequently used as a tool for natural resources management and urban planning, and they can assist in targeting and prioritizing risk mitigation activities (Cleve et al. 2008). In general, identifying LULC and area can be used as input for viable decisions in resource allocation and sustainable use of the available resources. However, a proper plan is required to ensure that such development is sustainable and does not adversely impact those communities that depend on the natural resources of Lake Tana. Several schemes are under development and planned for the future in the study area. Hence, assessment of LULC and knowing its trend in the study area is fundamental to assessing its impact on the water resources. As Woldesenbet et al. (2017, 2018) showed the change in LULC has an impact on the hydrology components of the Lake Tana basin. It is obvious that changes in LULC such as the expansion of cultivation land, and reduction of bush and shrubland, grassland, and forest land help its response on water balance components such as increased surface runoff, water yield, and reduce evapotranspiration (ET) and baseflow. This could be due to more land being left unprotected by deforestation by natural such as fire and anthropogenic causes such as urbanization, settlement, industry, and cropland, the more it will be prone to erosion and thus increasing the runoff. The LULC of the study area was changed over time, and this could have an impact on water resources. Therefore, it is very crucial to assess and update the impact of LULC change on the water resources in the Tana sub-basin.

This research aimed to assess the impact of LULC change on water resources in the Tana sub-basin using the Soil and Water Assessment Tool (SWAT). The application of the SWAT model in the study area was tested and validated by the scholar Setegn et al. (2009). The existing LULC and its changes were assessed and the effect of LULC change on the sub-basin water resources was analyzed.

Description of the study area

The sub-basin is an elevated region in Northern Ethiopia, situated in the headwaters of the Blue Nile Basin. The geographical location extends from 36°45′ E to 38°15′ E longitude and from 10°57′ N to 12°46′ N latitude as shown in Figure 1. The mean elevation is 2,026.54 m.a.s.l., with the highest elevations at 4,112 m.a.s.l. in the eastern part of the basin around Mountain Guna, and the lowest elevation at the point of outflow into the Blue Nile at Bahir Dar is about 1,786 m.a.s.l. (Abebe et al. 2017).
Figure 1

Location of the Tana Sub-Basin from Ethiopia and Abbay Basin.

Figure 1

Location of the Tana Sub-Basin from Ethiopia and Abbay Basin.

Close modal
The sub-basin covers an area of 15,070.14 km2, and 20% of the sub-basin is the Lake Tana water body. Lake Tana is a shallow freshwater lake and is the largest surface area in Ethiopia (Woldesenbet et al. 2017; Lemma et al. 2020) and the third-largest in the Nile Basin which is fed by a large number of smaller streams (Kebede et al. 2006) and a few larger rivers with catchments over 1,000 km2: the Gilgel Abbay (1,656.35 km2) in the southern part of the Tana sub-basin, and the Ribb (1,318.01 km2) and Gumara (1,354.35 km2), both in the eastern and southeastern part of the Tana Sub-Basin, Megech (515.06 km2) in the northern part of the study area as shown in Figure 2. Gilgel Abay, Gumara, Rib, and Megech rivers that flow into the Lake Tana water body contribute about three-quarters of the inflow (Atanaw et al. 2018). As stated by the author Atanaw et al. (2018), Gilgel Abay entering the lake from the south is the largest river draining 35% of the Tana sub-basin while Gumara and Rib entering the lake from the east drain each 27% of the sub-basin. Megech river has a drainage area that covers 11% of the basin.
Figure 2

Drainage, major watersheds, and slope distribution in the Tana sub-basin. *MG: Megech, RB: Rib, GU: Gumara, AB: Abbay, and GA: Gilgel Abay.

Figure 2

Drainage, major watersheds, and slope distribution in the Tana sub-basin. *MG: Megech, RB: Rib, GU: Gumara, AB: Abbay, and GA: Gilgel Abay.

Close modal

One of the hydrological response units (HRU) in the SWAT model is the slope. The distribution of slope in watersheds will directly or indirectly affect the generated response related to water resources and the runoff derived from the catchment. Accordingly, out of the total drainage area, 63.18% has a slope of 0–8%. The remaining 36.82% of the drainage area has a slope above 8%; out of which 15.65% has a slope of 8–15% (Figure 2).

Hydrological model

The SWAT model was developed by the United States Department of Agricultural (USDA) Agriculture Research Service to model the hydrology of a given watershed (Arnold et al. 1998). SWAT is a widely used tool in the world to evaluate and assess the influences of environmental and ecological alterations and hydrological responses at different watershed scales, even with limited data (Fu et al. 2009; Liu et al. 2018). In this study to assess the impact of LULC change on water resources in the Tana sub-basin, the SWAT hydrological model was developed and used. The application of the SWAT model in the study area was tested and the result showed that the model was suitable for the analysis of hydrological response in the Tana sub-basin (Setegn et al. 2009). The SWAT tool requires a digital elevation model (DEM), LULC, and soil data for the delineation of watersheds and generating HRU. ArcGIS 10.3 tool was used to analyze the LULC change in the Tana sub-basin. The areas covered by each LULC type for the various periods were compared. The flow data in the study area were available up to the year 2013 only and used for calibration. Then the directions of the changes in each LULC type between 1986 and 2000, 2000 and 2014, and 1986 and 2014 were determined.

The available data such as weather data (rainfall, maximum and minimum temperature, wind speed, sunshine hour, and relative humidity) were collected and converted into the usable format for the receiving convenient models to be applied in the study; SWAT Calibration and Uncertainty Problems (SWAT-CUP). The quality of time-series data and its outliers were checked. Then, time-series data were organized and converted into a usable format for the model as an input: weather data from 1987 to 2013 and flow data from 1990 to 2013 which were used to calibrate the SWAT model setup.

Data collection and analysis

DEM describes the elevation of any point in each area at a specific spatial resolution as a digital file. DEM is one of the essential inputs required by SWAT: (1) to delineate the watershed into several sub-watersheds or sub-basins; (2) to analyze the drainage pattern of the watershed, slope, stream length, the width of the channel within the watershed. The DEM was generated from 30 × 30 m Shuttle Radar Topographic Mission (SRTM) data obtained from NASA's website (http://srtm.csi.cgiar.org/). The DEM data were used to generate stream networks and slope classification. Based on the generated stream networks, 69 sub-basins were delineated. About 942, 886, and 869 HRUs were then created considering dominant soil and LULC types for 1986, 2000, and 2014 LULC, respectively. The model generated different HRUs due to the change in LULC (Figures 3 and 4; Table 1). For this study, 10% threshold values for LULC, soil, and slope layers were considered; therefore, the land units below the threshold size were not considered. The overall water resources of the study area will be influenced by the HRU. The soil data sourced from the Ministry of Water, Irrigation and Energy (MoWIE) Ethiopia were reclassified to match the SWAT database requirement.
Table 1

The summary of LULC assessment: the years 1986, 2000, and 2014

Land coverChange assessment
1986
2000
2014
Area (km2)%Area (km2)%Area (km2)%
Afroalpine and sub-Afroalpine vegetation 121.37 0.81 10.23 0.07 37.75 0.25 
Built-up area 148.60 0.99 42.40 0.28 43.82 0.29 
Bush and shrubland 3,619.32 24.06 3,773.61 25.09 2,688.23 17.87 
Cultivated land 5,511.20 36.64 6,122.91 40.71 7,343.69 48.82 
Forest land 656.51 4.36 351.09 2.33 272.75 1.81 
Grassland 1,824.30 12.13 1,577.75 10.49 1,433.93 9.53 
Waterbody 3,051.42 20.29 3,063.50 20.37 3,101.32 20.62 
Wetland 108.80 0.72 100.03 0.67 120.03 0.80 
Land coverChange assessment
1986
2000
2014
Area (km2)%Area (km2)%Area (km2)%
Afroalpine and sub-Afroalpine vegetation 121.37 0.81 10.23 0.07 37.75 0.25 
Built-up area 148.60 0.99 42.40 0.28 43.82 0.29 
Bush and shrubland 3,619.32 24.06 3,773.61 25.09 2,688.23 17.87 
Cultivated land 5,511.20 36.64 6,122.91 40.71 7,343.69 48.82 
Forest land 656.51 4.36 351.09 2.33 272.75 1.81 
Grassland 1,824.30 12.13 1,577.75 10.49 1,433.93 9.53 
Waterbody 3,051.42 20.29 3,063.50 20.37 3,101.32 20.62 
Wetland 108.80 0.72 100.03 0.67 120.03 0.80 
Figure 3

Summary of transformed versus unchanged land use land cover.

Figure 3

Summary of transformed versus unchanged land use land cover.

Close modal
Figure 4

Transformed versus unchanged LULC classes: 1986–2000, 2000–2014, and 1986–2014.

Figure 4

Transformed versus unchanged LULC classes: 1986–2000, 2000–2014, and 1986–2014.

Close modal

The slope data that were derived from the DEM were also reclassified to correspond with the SWAT database requirement and adopted from Food and Agricultural Organization (FAO) system (Yimer et al. 2006; Dagnachew et al. 2020). Accordingly, the study area slope was calculated from the DEM and classified into flat to very gently sloping (<3%), gently to sloppy sloping (3–8%), strongly sloping (8–15%), moderately steep (15–30%), and steep to extremely steep (>30%).

LULC changes are altering the hydrologic system and have potentially large impacts on water resources (Wagner 2014; Woldesenbet et al. 2017, 2018). The study evaluated the impacts of LULC changes on water resources. To assess these, the LULC map of the entire Tana sub-basin and beyond the boundary was sourced from the Amhara Design and Supervision Work Enterprise (ADSWE), Ethiopia. Then, the LULC map of the Tana sub-basin area was extracted using the study area boundary from the shapefile obtained from ADSWE which bound the entire Tana sub-basin. The maps obtained would be evaluated before being used, and adapted to the study area, and the assessed LULC maps were LULC maps in 1986, 2000, and 2014. The LULC maps, areal estimates, and percentage of each LULC class are present in Figure 5 and Table 1. The LC was categorized into eight namely Cultivation land, Forest, Shrub/Bushland, Water bodies, Afroalpine, Grassland, and Settlement/Built-up area.
Figure 5

LULC change of four major watersheds in the Tana Sub-basin.

Figure 5

LULC change of four major watersheds in the Tana Sub-basin.

Close modal

Measured weather data for the period from 1987 to 2013 was used. The full data set of daily time-series measured rainfall, maximum and minimum temperature, relative humidity, sunshine hour, and wind speed data were available from Amhara Metrological Agency for six weather stations. The rest of the stations were used to correct the bias of Climate Forecast System Reanalysis (CFSR) rainfall data. Since we have limited meteorological station spatial distribution and a short period of records, this study used additional meteorological data from the CFSR (https://globalweather.tamu.edu/) sourced from National Centers for Environmental Prediction (NCEP). This study, therefore, includes 42 rainfall grid points covering the extent of the Tana sub-basin (one point contains 6 meteorological variables including rainfall, minimum temperature, maximum temperature, relative humidity, wind speed, and solar radiation). These CFSR data were used after bias correcting it using ground observations. Data analysis was done, filling in missed data, data quality check, and interpretation of hydrology and meteorology data were done. This study used the method of arithmetic mean, normal ratio, and linear regression to estimate the missing observation of the station, and these methods were tested and applied in this area by the scholar Mesfin et al. (2021). The study adopted arithmetic mean techniques to estimate the missing observation data if the annual rainfall data at surrounding gauges are within the range of 10% of the annual rainfall of the considered station, while it exceeds 10% of the normal ratio method applied (Caldera et al. 2016). The linear regression method was applied if the correlation of the annual rainfall of data missing stations with an annual rainfall of the same years at nearby stations is good enough to estimate the missing observation of the station (Caldera et al. 2016). The missed data were computed from records of several stations at the same time. Screening the precipitation data requires careful examination of large time-series files, graphical tools, and standard reports of precipitation data were applied to facilitate this process.

The corrected CFSR data with nearby observed data were tested by performance evaluation criteria such as Nash Sutcliffe efficiency (NSE), coefficient of determination (R2), and Percent Bias (PBIAS). NSE indicates how well the simulation matches the observation and it ranges from the negative infinitive to 1. The higher the NSE value, the more reliable the model is in comparison to the mean (Nash & Sutcliffe 1970). PBIAS value shows the average tendency of the simulated to their observed data counterparts (Gupta et al. 1999). Positive values indicate an overestimation of observation, while negative values indicate an underestimation. The optimal value of PBIAS is 0.0, with low-magnitude values indicating accurate model simulations. MAE demonstrates the average model prediction error with less sensitivity to large errors. R2 indicates how much the observed and corrected data fit (Van Liew et al. 2003).

Three stations' rainfall data were used to represent the climate variability of the study area. The daily maximum and minimum air temperature, solar radiation, wind speed, and 1-h rainfall (1-h rainfall is one-third of the daily rainfall of the station) were available for the representative three stations. The statistical parameters for precipitation were computed using the tools of weather generator maker and precipitation (WGNmaker and PCP stat) and then used by the weather generator of the SWAT model. Daily dewpoint was computed using the formula expressed below and for verification, Dew02.exe: was used to calculate it using minimum and maximum daily temperature and the average daily humidity. WGEN user were developed using daily rainfall, daily maximum and minimum temperature, 1-h rainfall, daily solar radiation, daily wind speed, and daily dew point temperature. Bahir Dar, Dangla, and Gonder weather stations in the study area were used to generate weather generator (WGEN) user.

The saturation vapor pressure (es) was computed using daily minimum and maximum air temperature values, then average actual saturation pressure (ea) was computed using saturation vapor pressure (es) and average humidity data. The study computed the value of es and ea according to Allen et al. (1998) for es and Hackle (1999) for ea as indicated below:
formula
(1)
formula
(2)
The daily dew point temperature dew was calculated by the following equation
formula
(3)
where es is the saturation vapor pressure (hPa), ea is the actual vapor pressure (hPa), RF is the relative humidity (%), T is the air temperature (°C), and Dew is the dew point temperature.

The flow data were collected from Abbay Basin Development Office and used for calibration, and flow data generation for missed values and ungauged stations. Available flow data were used for Gilegl Abbay near Merawi, Megech near Azezo, Rib near Addis Zemen, and Gumara near Bahir Dar from 1987 to 2013. In this study, the flow data were used for model simulation for sensitivity analysis, calibration, and validation.

Three model setups were developed for the years 1986, 2000, and 2014 LULC maps. For the three model setups, all data were the same except for the LULC. The SWAT database files were adapted for local conditions. The model has been implemented in the 69 watersheds/sub-basins. The simulation covered 27 years (from 1987 to 2013) where the first three years (1987–1989) were used as model warm-up periods, 16 years (1990–2005) for calibration, and the last 8 years (2006–2013) for validation. The model was simulated and calibrated at a monthly time scale against observed discharge series at the four stations of the major rivers of the Tana sub-basin namely Gilgel Abbay, Megech, Rib, and Gumara.

Model performance indices/statistical analysis

To evaluate the performance of models for water balance components, statistical parameters such as Nash Sutcliff efficiency (NSE), coefficient of determination (R2), percent bias (PBIAS), and ratio of the root mean square error to the standard deviation of measured data (RSR) has been used before further application and analysis were done as recommended by Krause et al. (2005) and Moriasi et al. (2007). These statistical parameters are widely used in evaluating the performance of the hydrological models (Arnold et al. 1999; Gassman et al. 2007). The parameters are first computed by maximizing the NSE (Nash & Sutcliffe 1970), then R2 (Van Liew et al. 2003; Lemma et al. 2020), and finally minimizing the magnitude of PBIAS (Gupta et al. 1999) and RSR (Moriasi et al. 2007). The statistical parameters of the model result showed a good fit most of the time between the simulated and measured streamflow and underperformed in some high or low streamflow events and extremes, and the result aligns with the study done by Shrestha et al. (2017). The goodness of fit between the simulated and measured data was evaluated by the four statistical parameters and calculated by the equations as shown below.
formula
(4)
formula
(5)
formula
(6)
formula
(7)
formula
(8)
where Qm,i and Qs,i are observed and simulated variable/flow at time i, respectively, and and are average observed and simulated variables, respectively, and STDV is the standard deviation of measured data.

The model performance is satisfactory and applied for further analysis if the NSE > 0.5 (Nash & Sutcliffe 1970), PBIAS < 25% (Gupta et al. 1999), and RSR ≤ 0.7 (Moriasi et al. 2007) for a monthly time step variable/flow. The values of R2 greater than 0.5 are considered acceptable based on the previous criteria reported by Santhi et al. (2001) and Van Liew et al. (2003) while Setegn et al. (2009) also stated on the model performance can be judged as satisfactory if R2 is greater than 0.6 and NSE is greater than 0.5. The value NSE shows the level of reliability of the model in comparison to the mean and the value of R2 indicates how much the observed and simulated streamflow fit. The comparison between the observed and simulated streamflow indicated that there is a good agreement between the simulated and observed discharge which was verified by lower values of RSR and PBIAS and higher values of R2 and NSE. RSR ranges from zero to a large positive value. The lower the RSR, the lower the RMSE and the better the model simulation performance.

LULC changes: 1986, 2000, and 2014

The LULC map for 1986, 2000, and 2014 is shown in Figure 5. In the year 1986, cultivated land accounted for about 36.64% of the study area while water bodies, grassland, and bush/shrubland collectively represented about 56.48%. The built-up area, forest land, Afroalpine, and wetland collectively covered only 6.88% as shown in Table 1. In the year 2000, cultivated land covered 40.71% of the study area while water body, grassland, and bush/shrubland collectively represented about 55.95% as shown in Table 1. The built-up area, forest land, Afroalpine, and wetland covered only 3.35% of the Tana sub-basin. For the year 2014, cultivated land accounted 48.82% of the study area while water body, grassland, and bush/shrubland collectively represented about 48.02%. The built-up area, forest land, Afroalpine, and wetland covered only 3.15% of the Tana sub-basin as shown in Table 1.

Change detection

The LULC of the study area for the last 28 years (1986, 2000, and 2014) were assessed (Table 1). Most of the vegetation covers showed a decrease in the last 28 years. Forestland and grassland have decreased continuously in these years. Sub-afroalpine vegetation showed a dramatic decrease in the second period of assessment. On the other hand, bushes and shrubs recorded about a 1% increase in the total area and an unexpectedly fast decline in the second period. The forest land showed a continuous reduction while water bodies and wetlands showed a small variation as compared to the other cover types in the study area. Waterbody occupies 20.29, 20.37, and 20.62% of the Tana sub-basin in the years 1986, 2000, and 2014, respectively. From the total area of the sub-basin, wetlands decreased to 0.67 from 0.72% in the first assessment period and increased to 0.80% in the second period.

Another good expectation was cultivated land increment. It progressed from 36.64 to 40.71% in the first and then, to 48.82% in the second period of assessment, considering the total area of the sub-basin. The study area is one of the areas in Ethiopia influenced by population pressure dominantly engaged in agriculture (Abera et al. 2020). Therefore, there was continuous pressure on the surrounding vegetation cover in need of expanding the cultivated lands.

Transformed versus unchanged LULC: 1986–2000, 2000–2014, and 1986–2014

Due to the complex influx and the heavy agrarian population pressure in Tana sub-basin, the amount of LULC change was more than 25%. As indicated in Figure 3, the amount of land cover remain unchanged in the first period (1986–2000) was 73.47% and it was 71.54% in the next period (2000–2014). However, the overall (1986–2014) changes recorded increased a little bit more than 30%, which pushed the unchanged down to 66.06%. In the first period (1986–2000), the water body keeps by far the highest percentage of survival coverage, having 99.7% unchanged. There is also no significant cover transformation on cultivated land. More than half of bushes, shrubs, and grasslands remain unchanged in this period. On the other hand, sub-afroalpine vegetation (4.09%) followed by built-up areas (8.24%) have the lowest unchanged coverage. In other words, these land covers are the most disturbing cover types and strongly transformed into other cover types in the last 28 years.

In the second period (2000–2014), still water bodies preserved 99.7% of coverage in the first period, 99.47% in the second, and 99.87% in the whole assessment years. Similarly, built-up areas were also the least maintained (9.24%) cover type in the period. Wetlands were the second least preserved; only maintain 24.65% of their previous wetland areas (Figure 4).

Despite the increased pressure of cultivated land expansion, the Tana sub-basin still has a large area of grassland (grazing land) coverage. Grassland is the fourth largest in the amount of coverage with 1,824.30 km2 followed by cultivation land, water body, and bush, respectively. The recent conservation and rehabilitation program of the government resulted in significant coverage of bushes and shrubs inland, and the current socio-economic pressures may influence it for the last year of assessments. The built-up areas of the study area are highly transformed and converted to other land cover types (Figure 4). However, the transition statistics of the built-up area are not a good indicator of the change expected actually on the ground.

Over the last 28 years, the waterbody, including Lake Tana, has been the least disturbed in terms of coverage. On the other hand, sub-afroalpine vegetation is the most transformed cover. Cultivated lands gained a large area of cover from the other types.

LULC change of four major watersheds

As indicated in Figure 5 and Table 2, the LULC of the major watershed of the Tana sub-basin changed in the last 28 years (1986–2014). The highest land use/land cover change was recorded in Rib (72.04%) and Gumara watersheds (47.41%) from 1986–2000 and 2000–2014, respectively. Gilgel Abbay watershed has shown high transformed LULC (57.93%) for the last 28 years. The highest amount of land cover remains unchanged in Megech (70.82%, 66.55%) from 1986 to 2000 and 2000 to 2014, respectively. However, the overall (1986–2014) unchanged LULC recorded in the Rib watershed is 57.19%.

Table 2

Percentage of unchanged and transformed LULC for major Tana sub-basin watershed

YearTotal unchanged LULC (%)
Total transformed LULC (%)
GumaraRibMegechGilgel AbbayGumaraRibMegechGilgel Abbay
1986–2000 64.49 27.96 70.82 49.39 35.51 72.04 29.18 50.61 
2000–2014 52.59 60.99 66.55 56.55 47.41 39.01 33.45 43.45 
1986–2014 52.90 57.19 56.73 42.07 47.10 42.81 43.27 57.93 
YearTotal unchanged LULC (%)
Total transformed LULC (%)
GumaraRibMegechGilgel AbbayGumaraRibMegechGilgel Abbay
1986–2000 64.49 27.96 70.82 49.39 35.51 72.04 29.18 50.61 
2000–2014 52.59 60.99 66.55 56.55 47.41 39.01 33.45 43.45 
1986–2014 52.90 57.19 56.73 42.07 47.10 42.81 43.27 57.93 

Afroalpine, wetland, and forest land were the most highly disturbed cover types and strongly transformed to other cover types in the last 28 years and decreased their coverage whereas cultivated land increased over time in all watersheds. Bush and shrubland coverage decreased within 28 years for the major watersheds while it increased for Gilgel Abbay. Grassland increased for Gumara and Rib watersheds and decreased for Gilgel Abbay and Megech watersheds (1986–2000). There was also no significant cover conversion observed on the water body as all eight major LULC types (Figure 5). This transformed LULC shows a significant impact on the water balance components (Figures 8,91011).

Sensitivity analysis, calibration, and validation

SWAT Calibration and Uncertainty Programs were developed for calibration, validation, and uncertainty analysis of the SWAT model, and were also used to optimize the SWAT model parameters (Abbaspour 2013). The sensitivity analysis was undertaken and performed using the data period 1990–2013 for the streamflow of four major rivers in the Tana sub-basin by using a built-in tool in SWAT-CUP that uses the global sensitivity design method. Twenty-four hydrological parameters were tested for sensitivity analysis for the simulation of the streamflow in the study area. The list of the most 10 sensitive parameters for streamflow simulation was used out of 24 hydrological parameters based on their t-stat and p-value of parameter sensitivity order for model calibration and validation (Table 3). The parameter with the high absolute value of ‘t-stat’ and low ‘p-value’ is the more sensitive parameters for streamflow sensitivity analysis in the model. The results showed that among the selected parameters, the curve number for moisture condition II (CN2) was the most sensitive parameter for Gilgel Abay and Megech rivers while groundwater ‘revap’ coefficient (GW_REVAP) and groundwater delay (GW_DELAY) were the highest sensitivity parameter for Rib and Gumara river, respectively. In most of the watersheds, the threshold depth of water in the shallow aquifer required for return flow to occur (GWQMN), baseflow alpha factor (ALPHA_BF), deep aquifer percolation fraction (RCHRG_DP), and depth from the soil surface to bottom of layer (Sol_Z) were found to have high sensitivity order as shown in Table 3. The result suggests that parameters play key roles in the overall hydrological process in the Tana sub-basin. In other words, for better streamflow simulations, accurate estimation of these parameters in the sub-basin is required.

Table 3

SWAT streamflow sensitive parameters and fitted values after calibration for four rivers

Gumara river
Rib river
Megech river
Gilgel Abbay river
Sensitivity parametersFitted valueSensitivity parametersFitted valueSensitivity parametersFitted valueSensitivity parametersFitted value
A_GW_DELAY −38.06 V_GW_REVAP 0.25 R_CN2 1.68% R__CN2 13.58% 
R_CN2 −3.05% V_ALPHA_BF 0.02 A_GW_DELAY −66.11 R__SOL_Z 15.81% 
A_GWQMN −257.03 V_RCHRG_DP 0.09 A_GWQMN 917.52 A__GWQMN −186.62 
V_GW_REVAP 0.02 A_GW_DELAY −17.16 V_GW_REVAP 0.09 R__SOL_ALB 30.88% 
V_RCHRG_DP 0.21 V_CH_EROD 0.48 V_RCHRG_DP 0.52 V__CH_EROD  0.58 
R_SOL_AWC −10.74% R_CN2 −3.33% R_SLSUBBSN −6.16% R__SOL_AWC −7.38% 
V_CH_EROD 0.40 A_REVAPMN −129.70 A_EPCO 0.51 A__GW_DELAY 2.19 
A_EPCO 0.38 R_SOL_ALB 5.10% V_SURLAG 4.78 V__SURLAG 10.46 
V_BIOMIX 1.00 R_SOL_Z 0.73% V_CH_EROD 0.53 V__CH_COV 0.85 
10 V_CH_N2 0.42 R_OV_N −5.89% R_OV_N 5.86% V__ALPHA_BNK 0.19 
Gumara river
Rib river
Megech river
Gilgel Abbay river
Sensitivity parametersFitted valueSensitivity parametersFitted valueSensitivity parametersFitted valueSensitivity parametersFitted value
A_GW_DELAY −38.06 V_GW_REVAP 0.25 R_CN2 1.68% R__CN2 13.58% 
R_CN2 −3.05% V_ALPHA_BF 0.02 A_GW_DELAY −66.11 R__SOL_Z 15.81% 
A_GWQMN −257.03 V_RCHRG_DP 0.09 A_GWQMN 917.52 A__GWQMN −186.62 
V_GW_REVAP 0.02 A_GW_DELAY −17.16 V_GW_REVAP 0.09 R__SOL_ALB 30.88% 
V_RCHRG_DP 0.21 V_CH_EROD 0.48 V_RCHRG_DP 0.52 V__CH_EROD  0.58 
R_SOL_AWC −10.74% R_CN2 −3.33% R_SLSUBBSN −6.16% R__SOL_AWC −7.38% 
V_CH_EROD 0.40 A_REVAPMN −129.70 A_EPCO 0.51 A__GW_DELAY 2.19 
A_EPCO 0.38 R_SOL_ALB 5.10% V_SURLAG 4.78 V__SURLAG 10.46 
V_BIOMIX 1.00 R_SOL_Z 0.73% V_CH_EROD 0.53 V__CH_COV 0.85 
10 V_CH_N2 0.42 R_OV_N −5.89% R_OV_N 5.86% V__ALPHA_BNK 0.19 

The symbol R indicates multiple the existing values, A add on existing values, and V replaces the existing values.

SWAT model performed well in all the watersheds for the calibration and validation periods. The Nash Sutcliff efficiency (NSE) value during the calibration ranged from 0.72 to 0.88 and validation from 0.57 to 0.89. The performance of the best simulation of streamflow result that used these fitted parameter values for calibration and validation is shown in Figure 6. The model performed well in streamflow simulation. The efficiency values greater than or equal to 0.50 are considered adequate for SWAT model application as stated by Santhi et al. (2001). Setegn et al. (2009) also stated on the model performance can be judged as satisfactory if R2 is greater than 0.6 and NSE is greater than 0.5. Hence, it is observed that SWAT exhibited strong performance in representing the hydrological conditions of the Tana sub-basin which most of the R2 and NSE were greater than 0.7 (see Table 4). The flow hydrograph of the simulation well replicates with observation except for some pick and baseflows.
Table 4

Performance of simulated versus observed flow for calibration and validation periods

Sub-basinMajor watershedCalibration
Validation
R2NSPBIASRSRR2NSPBIASRSR
Tana Gumara 0.84 0.83 16.20 0.41 0.83 0.80 24.60 0.45 
 Megech 0.72 0.72 −9.4 0.53 0.74 0.57 44.1 0.65 
 Gilgel Abbay 0.88 0.88 2.90 0.35 0.71 0.65 −4.3 0.93 
 Rib 0.83 0.82 4.10 0.42 0.90 0.89 11.60 0.33 
Sub-basinMajor watershedCalibration
Validation
R2NSPBIASRSRR2NSPBIASRSR
Tana Gumara 0.84 0.83 16.20 0.41 0.83 0.80 24.60 0.45 
 Megech 0.72 0.72 −9.4 0.53 0.74 0.57 44.1 0.65 
 Gilgel Abbay 0.88 0.88 2.90 0.35 0.71 0.65 −4.3 0.93 
 Rib 0.83 0.82 4.10 0.42 0.90 0.89 11.60 0.33 
Figure 6

Streamflow hydrographs of watersheds in the Tana sub-basin during calibration and validation: (a) Rib (b) Gumara, (c) Megech, and (d) Gilgel Abay.

Figure 6

Streamflow hydrographs of watersheds in the Tana sub-basin during calibration and validation: (a) Rib (b) Gumara, (c) Megech, and (d) Gilgel Abay.

Close modal

In the calibration period, the monthly flow hydrograph of observed and simulated streamflow for the Tana sub-basin major watersheds showed that the simulated streamflow replicated the observed streamflow (Figure 6). In the streamflow hydrographs, the model overestimated the observed streamflow for the calibration period except for the Megech watershed. In addition to this, peaks and baseflow of the hydrographs were not well predicted compared to the rising and falling limb. The different trend observed during the calibration and validation period for the Megech watershed indicates that there are uncertainties in simulated flow due to errors in input data such as temperature and rainfall, errors in the type of soil and the corresponding soil characteristics such as infiltration capacity, and/or other unknown activities in the watershed. As the model does not simulate certain input data, the predictions can be uncertain. This result is more or less similar to other hydrological studies done by Setegn et al. (2009).

As compared to the calibration period, the observed streamflow was not replicated by the simulated flow for the validation period. However, the result was satisfactory to use the calibrated model for further analysis. The hydrograph of the Megech and Gilgel Abbay watershed during the validation period showed that the observed streamflow was poorly replicated by the simulated one. This is similar to the previous studies done on the Megech watershed by Halefom et al. (2018) and on the Gilgel Abay watershed (Worqlul et al. 2018).

Impacts of LULC change on water resources

The LULC changes during the past 28 years (1986, 2000, and 2014) in the Tana sub-basins are shown in Table 1. The assessment of LULC maps for the years 1986, 2000, and 2014 indicates that the most significant changes occurred in LULC classes in the Tana sub-basins, namely built-up area, grassland, Afroalpine, bush and shrubland, cultivated land, and forest land. Cultivation land expanded continuously while forest land and grassland declined throughout the study period of the study area. This study shows the change in land use/land cover has an impact on the water resources in the study area. The study considered baseflow (BF) as the sum of lateral flow and groundwater contribution to the streamflow for further analysis.

Figure 7 and Table 5 show the changes in the eight LULC classes and their corresponding average annual water balance components differences (ET, surface runoff, water yield, and baseflow) from 1986 to 2013 simulated for each LULC map. The average annual water yield in the Tana sub-basin was 93.57 and 79.26 mm above in 2000 and 2014, respectively, as compared to the baseline year of 1986 (i.e. increased by 14.88 and 12.6%, respectively, compared to 1986). As with water yield, the average annual baseflow with LULC increased gradually from 291.5 in 1986 to 345.16 mm in 2000 (increased by 18.4%), from 345.16 in 2000 to 320.45 mm in 2014 (decreased by 7.16%) and increased by 9.93% for the whole study period (1986–2014). The average annual surface runoff also increased by 12 and 16.16% in 2000 and 2014, respectively. In contrast to this, the average annual ET for LULC was 168.03 and 123.31 mm lower in 2000 and 2014, respectively, than that in 1986; i.e. it decreased by 18.39%, and 13.49%, respectively, in these years. The average annual surface runoff and ET increased whereas the average annual water yield and baseflow decreased from scenario 2000 to 2014 (Table 5).
Table 5

Proportional LULC extent, changes of LULCs, the annual average value of hydrological components, and changes in hydrological components for the Tana sub-basin

LULC (%)
Water resources component (mm)
PeriodAfroalpine and sub-Afroalpine vegetationBuilt-up areaBush and shrublandCultivated landForest landGrass landWater bodyWetlandET SURQWYLDBF
1986 0.81  0.99 24.06 36.64 4.36 12.13 20.29  0.72 913.65 233.61 628.88 291.50 
2000 0.07  0.28 25.09 40.71 2.33 10.49 20.37  0.67 745.62 261.65 722.45 345.16 
2014 0.25  0.29 17.87 48.82 1.81  9.53 20.62  0.80 790.34 271.36 708.14 320.45 
2000–1986 −0.74 −0.71  1.03  4.07 −2.03   − 1.64 0.08 −0.06 −168.03  28.04  93.57  53.65 
2014–2000 0.18  0.01 −7.22  8.12 −0.52   − 0.96 0.25  0.13 44.72  9.71   − 14.31 −24.71 
2014–1986 −0.56 −0.70 −6.19 12.18 −2.55   − 2.60 0.33  0.07 −123.31  37.75  79.26  28.95 
LULC (%)
Water resources component (mm)
PeriodAfroalpine and sub-Afroalpine vegetationBuilt-up areaBush and shrublandCultivated landForest landGrass landWater bodyWetlandET SURQWYLDBF
1986 0.81  0.99 24.06 36.64 4.36 12.13 20.29  0.72 913.65 233.61 628.88 291.50 
2000 0.07  0.28 25.09 40.71 2.33 10.49 20.37  0.67 745.62 261.65 722.45 345.16 
2014 0.25  0.29 17.87 48.82 1.81  9.53 20.62  0.80 790.34 271.36 708.14 320.45 
2000–1986 −0.74 −0.71  1.03  4.07 −2.03   − 1.64 0.08 −0.06 −168.03  28.04  93.57  53.65 
2014–2000 0.18  0.01 −7.22  8.12 −0.52   − 0.96 0.25  0.13 44.72  9.71   − 14.31 −24.71 
2014–1986 −0.56 −0.70 −6.19 12.18 −2.55   − 2.60 0.33  0.07 −123.31  37.75  79.26  28.95 

ET, Evapotranspiration; SURF, Surface runoff; WYLD, Water yield and BF, Baseflow.

Figure 7

Monthly average response of evapotranspiration, surface runoff, water yield, and baseflow to the different LULC scenarios and the change between each scenario.

Figure 7

Monthly average response of evapotranspiration, surface runoff, water yield, and baseflow to the different LULC scenarios and the change between each scenario.

Close modal
Figure 8

Monthly average response of evapotranspiration to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Figure 8

Monthly average response of evapotranspiration to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Close modal
Figure 9

Monthly average response of surface runoff to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Figure 9

Monthly average response of surface runoff to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Close modal
Figure 10

Monthly average response of water yield to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Figure 10

Monthly average response of water yield to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Close modal
Figure 11

The monthly average response of baseflow to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Figure 11

The monthly average response of baseflow to different LULC scenarios (a) Megech, (b) Rib, (c) Gumara, and (d) Gilgel Abbay.

Close modal

The change in the average monthly ET, SURQ, WYLD, and BF of the study area could be mainly ascribed to the LULC changes from 1986 to 2014 (Table 1; Figure 7). It is obvious that the expansion of cultivation land and reduction of bush and shrubland, grassland, and forest help increase surface runoff, and water yield, and reduce ET and baseflow. This could be because more and more land is left unprotected by deforestation by natural such as fire and anthropogenic causes such as urbanization, settlement, industry, and cropland, the more it will be prone to erosion and thus increasing the runoff. As Woldesenbet et al. (2017) stated very strong positive correlation with the Pearson correlation factor was observed between the proportional extent of cultivation land and surface runoff components. On the other hand, strong negative correlations were found between shrub and surface runoff components. In contrast to the surface runoff component, the groundwater component is strongly negatively correlated with the expansion of cultivation and is strongly positively correlated with the percentage of bush and shrubland (Woldesenbet et al. 2017). This study also showed similar increments in average monthly and annual surface runoff due to the expansion of cultivation land and reduction of shrubland within 27 years whereas baseflow declined on average annual values from 1986 to 2000 and increment from 2000 to 2014. Increased average monthly/annual sub-basin surface runoff associated with expanding cultivation areas probably is due to the decrease in the infiltration of soil (Franczyk & Chang 2009). The increase in surface runoff and water yield in the study area corresponds to sub-basins with a reduction in forest cover and shows an effect on ET. High forest cover will respond to a high rate of transpiration, and this will increase the value of ET while more or less undisturbed water bodies of the sub-basin do not show significant changes in evaporation. The rate of transpiration within the forest will decline due to the reduction of forest cover in the study area. Due to decreasing transpiration rates related to forest cover reduction and unchanged water bodies, ET values will decrease over time in the study area, Tana sub-basin. This was affirmed by the high negative correlation between open forests, surface runoff, and water yield (Awotwi et al. 2019). An association between the decrease in ET and cultivation expansion from 1986 to 2014 can be inferred from the comparison between the variations of average monthly/annual sub-basin ET and changes in LULC from 1986 to 2014. As shown in Table 5, ET decreased from 1986 to 2014, and baseflow increased from 1986 to 2014 and decreased from 2000 to 2014.

In the Tana sub-basin, the expansion of cultivation land replaced the grassland and/or bush and shrubland. Cultivation land decreases soil infiltration rate/percolation/baseflow and increases surface runoff compared to grassland and shrubland. As an expansion of cultivation land, the change in grassland and bush and shrub can also change the water balance component of the basin. Further comparison between changes in water yield and changes in LULCs in the Tana sub-basin (Table 5) indicates that the increase in water yield from 1986 to 2014 is due to the gradual increase in cultivation land and a simultaneous decrease in the bush and shrubland. This study shows that the LULC change has significant impacts on infiltration rates, runoff production, total simulation flow, interflow, base flow, water yield, ET, and water retention capacity of the soil or change in storage of the soil; hence, it affects the water balance of the study area.

Impact of LULC change on ET of major watersheds

The average monthly values of ET simulated from each LULC map and its changes between each scenario for four major watersheds of the Tana sub-basin are shown in Figure 8. ET values increased with time in each watershed except for Megech. The average monthly ET values were high from June to September (wet season) of the year. The average annual ET was 9.6 and 24.45 mm in 2000 and 2014, respectively (i.e. decreased by 5.42 and 2.13%) for Megech, 18.41 mm lower in 2000, 367.9 mm above in 2014 (i.e. decreased by 3.39% and increased by 67.76%) for Rib, 90.67 and 123.36 mm above in 2000 and 2014, respectively (i.e. increased by 16.71 and 22.73%) for Gumara, and 39.39 and 46.86 mm above in 2000 and 2014, respectively (i.e. increased by 5.71 and 6.79%) for Gilgel Abbay as compared to the baseline year of 1986. This study shows as the variation of LULC changes has a significant impact on the ET of the study area as shown in Figure 8. In the watersheds, significant expansion of cultivation land and reduction of forest and shrub land coverage, and relatively unchanged water bodies were observed and could cause decreasing the amount of ET because of transpiration declines. There will be a change in the leaf area index (LAI). Therefore, changes in LAI can lead to changes in ET, which controls the soil moisture content. Additionally, vegetation growth with taller canopy height and more abundant leaves becomes possible when sufficient soil moisture is available (Wang & Qu 2009).

Impact of LULC change on surface runoff of major watersheds

Average monthly values of surface runoff simulated from each LULC map and its changes within the three scenarios for the four major watersheds of the Tana sub-basin are shown in Figure 9. Surface runoff increased with time for four major watersheds and the response was significantly high in the wet season (June to September) of the year while in the dry season, there was an increment, but the values were small (Figure 9). The average annual surface runoff was 17.13 mm lower in 2000, 31.76 mm above in 2014 (i.e. decreased by 9.63% and increased by 15.99%), 41.22 mm above in 2000, 239.82 mm above in 2014 (i.e. increased by 16.08 and 93.54%), 38.98 mm above in 2000, 49.32 mm above in 2014 (i.e. increased by 8.74 and 11.05%), and 132.42 mm above in 2000, 180.91 mm above in 2014 (i.e. increased by 15.14 and 20.69%) as compared to the baseline year of 1986 for Megech, Rib, Gumara, and Gilgel Abbay, respectively. Continuous increment of surface runoff was observed in all scenarios except for Megech. From 1986 to 2000, there was a reduction in surface runoff value observed and for the whole period, the value averagely increased. This may be uncertainties due to errors in input data such as temperature and rainfall, errors in the type of soil and the corresponding soil characteristics such as infiltration capacity, and/or other unknown activities in the watershed. In the Tana sub-basin, the average annual surface runoff also increased by 12 and 16.16% in 2000 and 2014, respectively. As Figure 5 shows, cultivation lands were expanding, and shrub and forest land were reducing over time. The surface runoff response was positively correlated with the expansion of cultivation land and negatively correlated with the reduction of forest and shrubland coverage in the study area. Settlement/urbanization creates more impervious surfaces that do not allow the percolation of water down through the soil to the aquifer which leads to increased surface runoff. The LULC change in a watershed determines to what degree water infiltrates, accumulates, or flows over the land, and influences the runoff characteristics to a significant extent, which in turn, affects the surface and groundwater availability of the area. As a result, these LULC changes showed an effect on surface runoff responses for the major watersheds of the study (Figure 9). The study outcomes are consistent with the conclusions from studies by Woldesenbet et al. (2017) and Woldesenbet et al. (2018) in the Upper Blue Nile Basin, Ethiopia, where an increase in surface runoff was identified as the result of increasing settlement/built-up and agriculture/cultivation land at the expense of closed and open forests.

Impact of LULC change on water yield of major watersheds

As Figure 10 shows, the average monthly water yield simulated from each LULC map and its changes within the three scenarios for the four major watersheds of the Tana sub-basin. Water yield values increased with time for four major watersheds and the value of seasonal changes was significantly high in the wet season (June to September) of the year while in the dry season, there was an increment but small (Figure 10). The average annual water yield was 229.24 and 56.13 mm lower in 2000 and 2014, respectively (i.e. decreased by 46.63 and 11.42%) for Megech, 14.8 and 288.13 mm above in 2000 and 2014, respectively (i.e. increased by 4.2 and 81.68%) for Rib, 204.25 and 232.01 mm above in 2000 and 2014, respectively (i.e. increased by 27.09 and 30.79%) for Gumara and 253.58 and 153.62 mm above in 2000 and 2014, respectively (i.e. increased by 23.43 and 14.19%) for Gilgel Abbay as compared to the baseline year of 1986. A continuous increment of water yield was observed in all scenarios while it declined in the Megech watershed.

In the Tana sub-basin, the average annual water yield also increased by 14.88 and 12.6% in 2000 and 2014, respectively. In Figure 7, the water yield of the sub-basin was shown an increased trend monthly. This was due to the expansion of cultivation lands and reduction of shrub and forest land coverage over time and the result was similar to the previous studies by Awotwi et al. (2019) in Pra River Basin, Ghana, and Sead et al. (2010) in Upper Blue Nile River. The response of water yield was observed for LULC variability in the study area, Tana sub-basin.

Impact of LULC change on baseflow of major watersheds

Average baseflow values simulated monthly from each LULC map and their changes within the three scenarios for the four major watersheds of the Tana sub-basin are shown in Figure 11. As indicated in the above graph, the variation of the flow was higher in the wet season than in the dry season. Increasing monthly values of baseflow observed in Rib and Gumara whereas for Megech and Gilgel Abbay, the values were declined with time. In Rib and Gumara, both cultivation and grassland increased, and shrubland was reduced whereas in Megech cultivation land increased and grassland and shrubland reduced. For the Gilgel Abbay watershed, the coverage of cultivation and shrublands increased but grassland declined (Figure 5). As Woldesenbet et al. (2017) stated that the cultivation land expansion will decrease soil infiltration rate/percolation/baseflow, and increases surface runoff compared to grass and shrubland. But the changes LULC in each cover category will have a response on hydrology components with different scales. This study also shows the variation of land use/land cover has an impact on water balance components at different proportions in the study area. The average annual baseflow was 24.55 mm lower in 2000 and 27.55 mm lower in 2014 (i.e. decreased by 36.59 and 41.06%), 0.44 mm lower in 2000 and 40.66 mm above in 2014 (i.e. decreased by 1.27% and increased by 116.97%), 90.72 mm above in 2000 and 105.40 mm above in 2014 (i.e. increased by 31.12 and 36.15%), and 121.51 mm above in 2000 and 26.44 mm lower in 2014 (i.e. increased by 63.15% and decreased by 13.74%) as compared to the baseline year of 1986 for Megech, Rib, Gumara, and Gilgel Abbay, respectively. In the Tana sub-basin, the average annual baseflow also increased by 18.4%, and 9.93% in 2000 and 2014, respectively. In Figure 7, the baseflow show continuously increasing trends between scenario 2000 and 1986 and 2014 and 1986 while it is declining between 2014 and 2000. This will be the result of variation of LULC changes and its proportional changes of LULC types. As Woldesenbet et al. (2017) stated, the groundwater component of water balance is strongly negatively correlated with the expansion of cultivation and is strongly positively correlated with the percentage of bush and shrubland. The LULC change in a watershed determines to what degree water infiltrates, accumulates, or flows over the land, and influences the baseflow characteristics of a drainage basin, soil infiltration, and percolation to a significant extent, which in turn, affects the surface and groundwater availability of the area. Averagely, a high deviation of baseflow in each scenario was observed starting from the month of June to November due to high rainfall intensity and variability, and agricultural practice or expansion of cultivation land. As a result, these LULC changes showed an impact on the magnitude of annual and monthly baseflow for the major watersheds of the study (Figure 11).

Change in LULC is one of the factors responsible for changing the water balance of the sub-basin by altering the magnitude of surface runoff, base flow, water yield, and ET. The SWAT tool was used in this study to assess the impact of LULC change on water resources in the Tana sub-basin. The study included SWAT model evaluation, detection of LULC changes, and impact assessment. In this study, the impact of LULC changes on water resources was successfully assessed within the available data. The model has generated 942, 886, and 869 HRUs for 1986, 2000, and 2014 scenarios and 69 watersheds/sub-basin for all. The overall performance of the model was satisfactory. In four calibration stations where observational data were available, the values of NSE were above 0.72, which is above the minimum requirement to employ the model for further analysis. The study was simulated with the available flow data and lacks to calibrate with recent data to show the current impact of LULC changes on water balance components. Model uncertainty may also be the factor that affects the result of the research.

This study analysis indicated that the changes in LULC have an impact on the water resources of the study area. LULC changes have implicated effects on hydrological response and will continue to have consequences on natural resources management and development. The amount of land cover that remained unchanged in the first period (1986–2000) was 73.47% and it was 71.54% in the next period (2000–2014). However, the overall changes, for the period from 1986 to 2014, were recorded as 33.94%, which pushed the unchanged land cover down to 66.06%.

The water balance components of the study area showed positive and negative responses due to LULC changes in the Tana sub-basin. The expected reduction of surface runoff during the dry season may affect agriculture/irrigation and water-oriented activities while its increments during the wet/rainy season may lead to flooding. The rise in soil erosion and sedimentation of water resources structures and Lake Tana is also likely to occur due to the possible surface runoff increments. Overall, this could affect the livelihood in the study area. Further analysis using the latest flow dataset and LULC are required to show the overall effect of LULC change on the water resources in the study area.

Immediate measures must be taken to mitigate this by ensuring and enforcing land use plan preparation and implementation, and this may minimize illegal expansion trends of cultivation land and prevent forest cover from continuous reduction resulting in the normalization of water balance components. The approach used in this study has accredited contributions of changes in LULCs to water resources, providing perceptible information that will allow stakeholders and decision-makers to make prominent choices regarding natural resource planning and management. This approach could be applied to a variety of river basins to predict the consequences of LULC changes on water resources. The research methods used in this study can serve as a guide for other similar studies aiming at evaluating and computing hydrological responses to LULC variations.

The authors are thankful to the Ministry of Water, Irrigation and Energy, Abbay Basin Development Office, Amhara Design and Supervision Work Enterprise and Amhara Meteorology Agency for providing valuable data and information used in this study. The BRICS multilateral R&D project (BRICS2017-144) Team is sincerely acknowledged. The Durban University of Technology is sincerely acknowledged for hosting the grant. University of Limpopo and South Africa's Agricultural Research Council-Natural Resources & Engineering are most sincerely acknowledged as co-investigators in this NRF-BRICS research project.

The financial assistance of the South Africa National Research Foundation (NRF) is hereby acknowledged. Opinions expressed and conclusions arrived at, are those of the authors and are not necessarily to be attributed to the NRF. The research is under the grant BRICS multilateral R&D project (BRICS2017-144), the NRF UID number 116021 and the Durban University of Technology UCDG Water Research Focus Area grant.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Abbaspour
K. C.
2013
SWAT-CUP 2012: SWAT Calibration and Uncertainty Programs – A User Manual
.
Eawag
,
Dübendorf
,
Switzerland
, p.
103
.
Abebe
W.
,
G/Michael
T.
,
Sime
E.
,
Beyene
S.
&
Nigate
F.
2017
Climate of Lake Tana Basin
, pp.
51
58
.
https://doi.org/10.1007/978-3-319-45755-0_5.
Abera
A.
,
Verhoest
N. E. C.
,
Tilahun
S.
,
Inyang
H.
&
Nyssen
J.
2020
Assessment of irrigation expansion and implications for water resources by using RS and GIS techniques in the Lake Tana Basin of Ethiopia
.
Environmental Monitoring and Assessment
193
(
1
),
13
.
https://doi.org/10.1007/s10661-020-08778-1.
Alemayhu
2006
GIS Applications in Suitability Modelling for Livestock Production in Tana Subbasin-Blue Nile River Basin, Ethiopia
.
Master's Thesis
,
Addis Ababa
,
Ethiopia
.
Allen
R. G.
,
Pereira
L. S.
,
Raes
D.
&
Smith
M. J.
1998
Computing crop water requirement: FAO Irrigation and drainage paper No. 56. Food Agriculture Organization of the United Nations, Rome, 56(97), e156
.
Arnold
J. G. S. R.
,
Muttiah
R. S.
&
Williams
J. R.
1998
Large area hydrologic modeling and assessment part I: model development
.
JAWRA Journal of the American Water Resources Association
34
,
73
89
.
Arnold
J. G.
,
Srinivasan
R.
,
Muttiah
R. S.
&
Allen
P. M.
1999
Continental scale simulation of the hydrologic balance1
.
JAWRA Journal of the American Water Resources Association
35
(
5
),
1037
1051
.
https://doi.org/10.1111/j.1752-1688.1999.tb04192.x
.
Atanaw
F.
,
Moges
M.
,
Alemu
M.
,
Ayana
E.
,
Demissie
S.
,
Tilahun
S.
&
Steenhuis
T.
2018
Budgeting suspended sediment fluxes in tropical monsoonal watersheds with limited data: the Lake Tana basin
.
Journal of Hydrology and Hydromechanics
66
,
65
78
.
https://doi.org/10.1515/johh-2017-0039
.
Awotwi
A.
,
Anornu
G. K.
,
Quaye-Ballard
J. A.
,
Annor
T.
,
Forkuo
E. K.
,
Harris
E.
,
Agyekum
J.
&
Terlabie
J. L.
2019
Water balance responses to land-use/land-cover changes in the Pra River Basin of Ghana, 1986–2025
.
CATENA
182
,
104129
.
https://doi.org/10.1016/j.catena.2019.104129
.
Bogale
A.
2020
Review, impact of land use/cover change on soil erosion in the Lake Tana Basin, Upper Blue Nile, Ethiopia
.
Applied Water Science
10
(
12
),
235
.
https://doi.org/10.1007/s13201-020-01325-w
.
Caldera
H. P. G. M.
,
Piyathisse
V.
&
Nandalal
K. D. W.
2016
A comparison of methods of estimating missing daily rainfall data
.
Engineer: Journal of the Institution of Engineers, Sri Lanka
49
,
1
.
https://doi.org/10.4038/engineer.v49i4.7232
.
Cleve
C.
,
Kelly
M.
,
Kearns
F.
&
Moritz
M.
2008
Classification of the wildland[ndash]urban interface: A comparison of pixel- and object-based classifications using high-resolution aerial photography
.
Computers, Environment and Urban Systems
32
(
4
),
317
326
.
https://doi.org/10.1016/j.compenvurbsys.2007.10.001.
Dagnachew
M.
,
Moges
A.
,
Kebede
A.
&
Abebe
A.
2020
Effects of soil and water conservation measures on soil quality indicators: the case of Geshy Subcatchment, Gojeb River Catchment, Ethiopia
.
Applied and Environmental Soil Science
2020
,
1868792
.
https://doi.org/10.1155/2020/1868792
.
Fu
G.
,
Charles
S. P.
,
Yu
J.
&
Liu
C.
2009
Decadal climatic variability, trends, and future scenarios for the North China Plain
.
Journal of Climate
22
(
8
),
2111
2123
.
https://doi.org/10.1175/2008jcli2605.1
.
Gassman
P. W.
,
Reyes
M. R.
,
Green
C. H.
&
Arnold
J. G.
2007
The soil and water assessment tool: historical development, applications, and future research directions
.
Transactions of the ASABE
50
(
4
),
1211
1250
.
https://doi.org/10.13031/2013.23637.
Gupta
H. V.
,
Sorooshian
S.
&
Yapo
P. O.
1999
Status of automatic calibration for hydrologic models: comparison with multilevel expert calibration
.
Journal of Hydrologic Engineering
4
(
2
),
135
143
.
https://doi.org/10.1061/(ASCE)1084-0699(1999)4:2(135)
.
Hackle
H.
1999
Meteorologie. Ulmer, Stuttgart
.
Halefom
A.
,
Teshome
A.
,
Sisay
E.
&
Ahmad
I.
2018
Dynamics of land use and land cover change using remote sensing and GIS: A case study of Debre Tabor Town, South Gondar, Ethiopia
.
Journal of Geographic Information System
10
,
165
174
.
https://doi.org/10.4236/jgis.2018.102008
.
Kebede
S.
,
Travi
Y.
,
Alemayehu
T.
&
Marc
V.
2006
Water balance of lake Tana and its sensitivity to fluctuations in rainfall, Blue Nile basin, Ethiopia
.
Journal of Hydrology
316
(
1
),
233
247
.
https://doi.org/10.1016/j.jhydrol.2005.05.011
.
Krause
P.
,
Boyle
D. P.
&
Bäse
F.
2005
Comparison of different efficiency criteria for hydrological model assessment
.
Advances in Geosciences
5
,
89
97
.
https://doi.org/10.5194/adgeo-5-89-2005
.
Lemma
H.
,
Frankl
A.
,
Dessie
M.
,
Poesen
J.
,
Adgo
E.
&
Nyssen
J.
2020
Consolidated sediment budget of Lake Tana, Ethiopia (2012–2016)
.
Geomorphology
371
,
107434
.
https://doi.org/10.1016/j.geomorph.2020.107434
.
McCartney
M.
,
Alemayehu
T.
,
Shiferaw
A.
&
Awlachew
S.
2010
Evaluation of Current and Future Water Resources Development in the Lake Tana Basin, Ethiopia
.
International Water Management Institute
,
Colombo
,
Sri Lanka
.
Mesfin
S.
,
Adem
A. A.
,
Mullu
A.
,
Melesse
A. M.
,
2021
Historical Trend Analysis of Rainfall in Amhara National Regional State
. In:
Nile and Grand Ethiopian Renaissance Dam: Past, Present and Future
(
Melesse
A. M.
,
Abtew
W.
&
Moges
S. A.
, eds).
Springer International Publishing
, pp.
475
491
.
https://doi.org/10.1007/978-3-030-76437-1_25
Moriasi
D. N.
,
Arnold
J. G.
,
Van Liew
M. W.
,
Bingner
R. L.
,
Harmel
R. D.
&
Veith
T. L.
2007
Model evaluation guidelines for systematic quantification of accuracy in watershed simulations
.
Transactions of the ASABE
50
(
3
),
885
900
.
Nash
J. E.
&
Sutcliffe
J. V.
1970
River flow forecasting through conceptual models part I – a discussion of principles
.
Journal of Hydrology
10
(
3
),
282
290
.
https://doi.org/10.1016/0022-1694(70)90255-6
.
Santhi
C.
,
Arnold
J.
,
Williams
J. R.
,
Dugas
W. A.
,
Srinivasan
R.
&
Hauck
L. M.
2001
Validation of the SWAT model on a large river basin with point and nonpoint sources
.
JAWRA Journal of the American Water Resources Association
37
,
1169
1188
.
https://doi.org/10.1111/j.1752-1688.2001.tb03630.x
.
Sead
A.
,
Bauwens
W.
&
Marwa
A.
2010
Impact of land-use change and climate change on the flows in the Blue Nile River using swat. M.S. thesis, Water Resources Engineering, Katholieke Universiteit Leuven
.
Setegn
S.
,
Srinivasan
R.
,
Melesse
A.
&
Dargahi
B.
2009
SWAT model application and prediction uncertainty analysis in the Lake Tana Basin, Ethiopia
.
Hydrological Processes
24
,
357
367
.
https://doi.org/10.1002/hyp.7457
.
Shrestha
S.
,
Shrestha
M.
&
Shrestha
P. K.
2017
Evaluation of the SWAT model performance for simulating river discharge in the Himalayan and tropical basins of Asia
.
Hydrology Research
49
(
3
),
846
860
.
https://doi.org/10.2166/nh.2017.189
.
Stonestrom
D. A.
,
Scanlon
B. R.
&
Zhang
L.
2009
Introduction to special section on impacts of land use change on water resources
.
Water Resources Research
45
(
7
).
https://doi.org/10.1029/2009WR007937.
Tewabe
D.
&
Fentahun
T.
2020
Assessing land use and land cover change detection using remote sensing in the Lake Tana Basin, Northwest Ethiopia
.
Cogent Environmental Science
6
(
1
),
1778998
.
https://doi.org/10.1080/23311843.2020.1778998.
Van Liew
M. W.
,
Arnold
J. G.
&
Garbrecht
J. D.
2003
Hydrologic simulation on agricultural watersheds: choosing between two models
.
Transactions of the ASAE
46
(
6
),
1539
1551
.
https://doi.org/10.13031/2013.15643.
Wagner
P.
2014
Possible future land use change impacts on water resources in India
.
IGCS Bulletin
3
,
8
11
.
Wang
L.
&
Qu
J. J.
2009
Satellite remote sensing applications for surface soil moisture monitoring: a review
.
Frontiers of Earth Science in China
3
(
2
),
237
247
.
https://doi.org/10.1007/s11707-009-0023-7
.
Wassie
S. B.
2020
Natural resource degradation tendencies in Ethiopia: a review
.
Environmental Systems Research
9
(
1
),
33
.
https://doi.org/10.1186/s40068-020-00194-1
.
Woldesenbet
T. A.
,
Elagib
N. A.
,
Ribbe
L.
&
Heinrich
J.
2017
Hydrological responses to land use/cover changes in the source region of the Upper Blue Nile Basin, Ethiopia
.
Science of the Total Environment
575
,
724
741
.
https://doi.org/10.1016/j.scitotenv.2016.09.124
.
Woldesenbet
T. A.
,
Elagib
N. A.
,
Ribbe
L.
&
Heinrich
J.
2018
Catchment response to climate and land use changes in the Upper Blue Nile sub-basins, Ethiopia
.
Science of the Total Environment
644
,
193
206
.
https://doi.org/10.1016/j.scitotenv.2018.06.198
.
Worqlul
A. W.
,
Ayana
E. K.
,
Maathuis
B. H. P.
,
MacAlister
C.
,
Philpot
W. D.
,
Osorio Leyton
J. M.
&
Steenhuis
T. S.
2018
Performance of bias corrected MPEG rainfall estimate for rainfall-runoff simulation in the upper Blue Nile Basin, Ethiopia
.
Journal of Hydrology
556
,
1182
1191
.
https://doi.org/10.1016/j.jhydrol.2017.01.058
.
Yimer
F.
,
Ledin
S.
&
Abdelkadir
A.
2006
Soil property variations in relation to topographic aspect and vegetation community in the south-eastern highlands of Ethiopia
.
Forest Ecology and Management
232
(
1–3
),
90
99
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).