Assessing the impacts of land use/land cover changes on hydrological processes in Southern Ethiopia: The SWAT model approach

Abstract

The change in land use and land cover (LULC) due to human activities is a major cause of hydrological process changes in watersheds. This study assesses the effects of LULC changes on the hydrological processes in the Bilate catchment, located in southern Ethiopia. Landsat images of 1986, 2002, and 2018 were used to classify LULC through supervised classification. The Soil and Water Assessment Tool (SWAT) evaluated the hydrological responses to LULC changes. Results indicate an increase in built-up areas (0.70%), water bodies (0.19%), cultivated land (10.54%), and barren land (1.265%) from 1986 to 2018. Conversely, the forest (8.52%) and grazing (3.18%) areas declined. An increase in water leaving the root zone (962.02 mm/year), potential evapotranspiration (593.85 mm/year), actual evapotranspiration (201.71 mm/year), and surface runoff (514.86 mm/year) was observed, while soil water content (1423.25 mm/year) decreased at the catchment level. The declining soil water content and increasing actual and potential evapotranspiration could result in a water shortage for crop production. The impact of LULC changes on hydrology provides vital knowledge for integrated LULC and water resource management. Water resource development planning must consider LULC changes to achieve sustainable development in the catchment.

Keywords:

• Bilate catchment
• LULC change
• water resource management
• soil water
• surface runoff
• SWAT model

1. Introduction

Land use and land cover (LULC) change is a primary driver of environmental challenges in many developing countries (Matlhodi et al., 2019). The expansion of agriculture, human settlements, urban and industrial areas are among the main factors accelerating LULC changes (Mariye et al., 2022; Seyam et al., 2022). The conversion of forest and woodland areas is particularly intense and accounts for a large portion of LULC changes. Deforestation has resulted in the loss of approximately 420 million hectares of forest since 1990 worldwide (FAO, 2020). Ibrahim et al. (2022) found that the conversion of natural and forest-rich ecosystems for agriculture in less-developed areas of Nigeria threatens environmental sustainability. Similarly, Elias et al. (2019) reported that the Central Rift Valley in Ethiopia, known for its diverse ecosystem services and biodiversity, faces tremendous pressure due to rapid population growth, unsustainable developmental activities, unplanned urbanization, aggressive agricultural expansion, and associated LULC changes.

Hydrological processes in a catchment are influenced by a complex interplay of various factors, including land use land covers (LULC), climate, topography, and soil characteristics. Alterations to LULC, particularly the conversion of one form to another, can have significant impacts on hydrological processes that are dynamic in space and time. Several studies have shown that LULC changes can substantially affect catchment hydrology. For instance, in different parts of Ethiopia, LULC changes have been reported to have affected hydrological processes significantly. A study conducted at the Didessa River Basin in the Blue Nile Basin, for instance, reported an increase in average monthly flow by 4.9, 5.7, and 10.6 m³/sec between 1986 to 2001, 2001 to 2015, and 1986 to 2015, respectively (Chimdessa et al., 2018). Similarly, Belihu et al. (2020) reported that changes in land use led to an increase in surface runoff and evapotranspiration by 9.2% and 1.7%, respectively. The impact of LULC changes on catchment hydrology has also been documented in other studies, including Legesse et al. (2003), Rientjes et al. (2011), Koch et al. (2012), and WoldeYohannes et al. (2018).

The accurate modelling of hydrological processes within catchments is essential for gaining a comprehensive understanding of these phenomena and how changes in the catchment may impact them. This knowledge is crucial for implementing effective management strategies aimed at minimizing undesired effects. Chhabra et al. (2006) highlight the importance of understanding the relationships among the hydrologic processes occurring within catchments in order to compare watersheds that are sensitive to change. Hydrologic models are also crucial for improving the adaptive capacity of river basins (Krysanova et al., 2015).

In Ethiopia, the Soil and Water Assessment Tool (SWAT) model has been utilized in various catchments to simulate stream flows and sediment loads in water bodies. For example, A., . A. Shawul et al. (2013) employed SWAT to model the hydrology of the Shaya Watershed in a mountainous area, achieving acceptable statistical model efficiency criteria as recommended by SWAT developers. Similarly, Setegn et al. (2008) utilized SWAT to predict stream flow in the Lake Tana Basin and demonstrated its feasibility and performance. The replication of these models is critical for ensuring their accuracy and reliability in predicting hydrological processes within catchments.

The Bilate Catchment, located in the Rift Valley Lakes Basin of Ethiopia, is an agriculturally important area and one of the most heavily populated regions in the Rift Valley Basin (Adugna, 2014; CSA, 2013). The catchment faces numerous environmental challenges, including deforestation, overgrazing, soil erosion, and the expansion of agricultural areas (Elias et al., 2019; Garedew et al., 2009; Godebo et al., 2018). In addition, subsistence farming in the catchment is vulnerable to variable rainfall patterns, as evidenced by the crop failure experienced during the 2021 short rainy season (February to May) (Kuma et al., 2021). While previous studies in the area have focused on identifying drivers of land use and land cover (LULC) changes, the present study aims to assess the impact of LULC alterations on hydrological processes through the use of the Soil and Water Assessment Tool (SWAT) model in the Bilate Catchment.

2. Materials and methods

2.1. Study area

The study was conducted in the Bilate River Basin, situated in the Southern Nations, Nationalities, and Peoples’ Region of southern Ethiopia. The catchment shares borders with Lake Abaya to the south, Omo Ghibe Basin to the west, Awash Basin to the north, and Rift Valley Lakes, including Awassa, Shala, and Abiyata to the east. The catchment spans an area of 562,560 ha (Figure 1) and experiences a mean annual rainfall ranging from 854 to 1039 mm. The mean annual maximum and minimum temperatures recorded in the catchment range from 22.8 to 30.4°C and 10.7 to 16.8°C, respectively. The altitude of the catchment varies from 1,174 meters at Lake Abaya to 3,330 meters above sea level. The diverse landscape and topography of the catchment provide habitats for different types of vegetation, including indigenous and non-indigenous species. The lowlands are dominated by semi-arid biomass, such as acacia, scrub, and grass. In contrast, the middle and highlands feature non-indigenous vegetation like eucalyptus. To address the shortage of construction material and firewood and generate additional income, farmyard eucalyptus plantations and agroforestry are commonly practiced. Natural woodlands are also present along the ridges, the Bilate River, and its tributaries. Land use in the catchment is mainly subsistence and rain-fed agriculture in the western and northern parts, while the eastern and lower lowlands are utilized for crop production, pasture, and small-scale irrigated commercial farms.

Figure 1. Location map of the study area.

2.2. SWAT model database

To conduct this study, a 30 m by 30 m resolution digital elevation model (DEM) of the Bilate catchment was obtained from the United States Geological Survey (USGS) at https://earthexplorer.usgs.gov. The DEM was utilized to delineate the watershed using the Soil and Water Assessment Tool (SWAT) and to analyze the drainage forms of the land surface terrains. To supplement this data, a soil map of the Bilate catchment with comprehensive soil physical and chemical properties was extracted from the Food and Agricultural Organization (FAO, 2002) database. The dominant soils in the Bilate catchment were identified and coded according to the SWAT database, which includes Vitric Andosols, Pellic Vertisols, Chromic Vertisols, Orthic Solonchaks, Chromic Luvisols, Eutric Fluvisols, Eutric Nitisols, and Dystric Nitisols. Additionally, the less dominant soils identified in the catchment include Eutric Regosols, Calcic Fluvisols, Calcaric Fluvisols, Luvic Phaeozems, and Mollic Andosols.

To evaluate land use and land cover (LULC) changes in the Bilate catchment, we developed land-use maps for 1986, 2002, and 2018 using Landsat imagery from TM, ETM+, and OLI, respectively. To preprocess the imagery, we performed layer stacking, sub-setting, mosaicking, and radiometric adjustments. Then, we employed the maximum likelihood classification algorithm for supervised classification of the images into six LULC classes, namely water bodies (WATR), urban and built-up areas (URBN), grazing lands (PAST), forest lands/scattered (FRST), barren lands (BARR), and cultivation lands (AGRC). To ensure accuracy of the classifications, we collected reference data using Global Positioning System and employed Google Earth reference data to moderate errors. The LULC classes were coded according to the SWAT database. Additionally, we utilized ERDAS Imagine 2015 and ArcGIS 10.3 software to develop the land-use maps. The spatial data, including digital elevation model (DEM), soil, and LULC maps, were projected to UTM zone 37 North on the spheroid of D-WGS 1984.

The daily meteorological data including rainfall, temperature, humidity, wind and sunshine hours for the period of 1978 to 2017 was collected from the National Meteorological Agency of Ethiopia for input into the Soil and Water Assessment Tool (SWAT) model. To ensure completeness and consistency, the data from four representative weather stations (Hosana, Halaba, Boditi and Bilate) in Bilate catchment were selected and arranged. Among these, Hosana and Bilate stations were used as weather generators for the other two stations. The availability of all necessary weather data was confirmed for Hosana and Bilate stations while Boditi and Halaba stations only had rainfall and temperature data. Daily flow data of Bilate River from 1978 to 2017 at Halaba Kulito gauge station was used for model calibration and validation, as it was identified as the major contributor to the river flow of the catchment. The data was obtained from the Ministry of Water, Irrigation and Energy of Ethiopia.

2.3. Sensitivity analysis, calibration, validation and uncertainty analysis

The Global Sensitivity Analysis Tool in SWAT-CUP was used to select twelve parameters associated with stream flow. During model calibration, sensitivity analysis was performed to identify the most sensitive parameters using the Latin Hypercube and One-factor-At-a-Time (LH-OAT) method. This approach is an automatic sensitivity analysis tool in SWAT (Griensven et al., 2006; Lenhart et al., 2002; Ma et al., 2000).

Model calibration involves adjusting model parameters within the recommended ranges to optimize the simulated output to match the observed data. For this study, the monthly river flow data was divided into three parts for calibration and validation. Part one covering the period from 1978 to 1988, part two covering 1990 to 2001, and part three covering 2006 to 2017 were used for 1986, 2002, and 2018 land-use maps, respectively. Each LULC map was used for six years of calibration and three years of validation using SWAT-CUP. The Sequential Uncertainty Fitting version two (SUFI-2) algorithm was used for calibration and validation (Abbaspour et al., 2007).

The presence of parameter uncertainty can be attributed to both input uncertainty arising from errors in input data, including rainfall, land use, soil type, and observed data, and parameter non-uniqueness. In order to assess the quality of calibration and prediction uncertainty, two performance measures are typically used: the p-factor and r-factor. The p-factor refers to the percentage of observations that fall within the prediction uncertainty band, with a value of 100% indicating a perfect fit. Meanwhile, the r-factor represents the ratio of the width of the prediction uncertainty band to the width of the observed data, with a value of 1 indicating a small uncertainty band. These metrics have been extensively discussed in the literature, including studies by Hornberger and Spear (1981), Talebizadeh et al. (2009), and Luo et al. (2014).

2.4. Model performance evaluation

In order to evaluate the performance of the SWAT model, various statistical measures and graphical comparisons were utilized. Specifically, these evaluations were conducted by comparing the model’s simulated output with observed stream flow data. The performance measures utilized in this analysis included the coefficient of determination (R²) and Nash-Sutcliffe simulation efficiency (NSE), both of which have been widely used in hydrological modeling research (Nash & Sutcliffe, 1970; Santhi et al., 2001). These measures produce values ranging from zero to one, with values closer to one indicating a higher level of agreement between the simulated and measured flows. The results of this analysis provide valuable insights into the reliability of the SWAT model and can be used to inform future modeling efforts.

2.5. Evaluating the impacts of LULC changes on hydrology

Following the calibration of the SWAT model with sensitive parameters, the hydrologic impacts of LULC maps from 1986, 2002, and 2018 were investigated. Specifically, the model was run for each simulation period while holding constant other inputs to the model, such as the digital elevation model (DEM) and soil data. By varying only the LULC inputs, the effects of changes in land use and land cover on the hydrological response of the study area were able to be assessed. This approach provides insights into the role of LULC in shaping the hydrology of the region, which can be valuable information for land use planning and management.

2.5.1. SWAT model setup

The SWAT (Soil and Water Assessment Tool) model is a physically-based, semi-distributed model that is designed to evaluate the impacts of climate and land use management practices on hydrologic processes occurring within basins (Arnold et al., 1998). This model employs a water balance approach to simulate the partitioning of hydrologic processes within a watershed (Neitsch et al., 2011). The hydrologic routines implemented by the SWAT model are based on an equation (Equation.1(1) $\mathrm{SWt}=\mathrm{SWo}+\sum _{\mathrm{i}=1}^{\mathrm{t}}(\mathrm{Pday}-\mathrm{Qsur}-\mathrm{Ea}-\mathrm{Wseep}-\mathrm{Qgw})$(1) ) that accounts for the major components of the hydrologic cycle, including precipitation, evapotranspiration, surface runoff, and groundwater flow. By simulating the interactions between these components, the SWAT model is able to capture the complex hydrological processes that occur within a watershed and to predict the impacts of various management practices on water resources.

(1) $\mathrm{SWt}=\mathrm{SWo}+\sum _{\mathrm{i}=1}^{\mathrm{t}}(\mathrm{Pday}-\mathrm{Qsur}-\mathrm{Ea}-\mathrm{Wseep}-\mathrm{Qgw})$(1)

Where SWt is the last soil water amount (mm), SWo is the first soil water amount on the day i (mm), t is the time (days), Pday is the quantity of precipitation on the day i (mm), Qsurfis the quantity of surface runoff on the day i (mm), Ea is the quantity of evapotranspiration on the day i (mm), Wseep is the quantity of water entering the vadose zone from the soil profile on the day i (mm), and Qgw is the quantity of return flow or base flow on the day i (mm).

To improve the resolution of the drainage networks and reduce the number of Hydrologic Response Units (HRUs), a watershed delineation process was performed. This process involved defining an outlet and partitioning the study area into smaller sub-basins, with an area of 5000 hectares or less. For the Bilate catchment, which covers an area of 562,560 hectares, there were a total of 60 sub-basins, with individual areas ranging from 103 to 29,355 hectares. To simulate the flow of the Bilate River, a multiple threshold setting was used that accounted for a combination of land use (15%), soil (15%), and slope (15%). The watershed was further subdivided into 451, 424, and 416 HRUs for the 1986, 2002, and 2018 LULC maps, respectively. This approach allowed for a more detailed analysis of the impacts of land use changes on hydrological processes within the Bilate catchment.

3. Results and discussions

3.1. LULC change

The classification of Landsat images yielded six dominant LULC classes, which were identified as cultivated land, forest land, grazing lands, water bodies, built-up areas, and barren areas. The LULC classes for each map were illustrated and labeled in Figure 2. Overall, the image classification accuracies were high, with values of 0.88, 0.89, and 0.91 for the 1986, 2002, and 2018 maps, respectively. This indicates a high level of confidence in the accuracy of the LULC data used in the subsequent analysis.

Figure 2. LULC maps of 1986, 2002 and 2018 (Kuma et al. 202; Kuma et al. 2022).

The results of the land-use evaluation indicated that between 1986 and 2018, there was an increase in cultivation land, barren land, built-up areas, and water bodies by 9.46, 0.69, and 0.19% of the catchment area, respectively. However, during the same period, there was a decline in forest and grazing lands by 8.52 and 3.46%, respectively (Table 1). These changes in land use are important for understanding the hydrologic processes in the catchment, as they can significantly impact the water balance and affect the quantity and quality of water resources in the area.

Table 1. LULC changes in 1986, 2002 and 2018

	Area of LULC (ha) and percentage share
	1986		2002		2018
Land class	ha	%	ha	%	ha	%
BL	16,370.3	2.91	21,455.2	3.813	24,044.30	4.274
BA	8,589.78	1.53	10,245.3	1.821	14,069.40	2.501
CL	338,565	60.18	407, 270	72.4	390,714	69.453
FL	177,335	31.52	115,349	20.504	129,181	22.963
GL	20,796.3	3.7	7,319.3	1.30	2,886.39	0.513
WB	903.78	0.16	922.32	0.163	1,665.36	0.296
Total	562,560		562,560		562,560

Note: BL = barren land; BA = built-up areas; CL = cultivated land; FL = forest land; GL = grazing land; and WB = water body

The changes in LULC in the Bilate catchment in Ethiopia was assessed from 1986 to 2018. The results indicated that the growth of cultivated land led to a decline in forest and grazing land, which is consistent with previous studies in Ethiopia (Berihun et al., 2019; Gashaw et al., 2017). Furthermore, a significant decline in grazing land was observed in some parts of the catchment, which aligns with other studies in Ethiopia (Mussa et al., 2016). The increase of forest and a decrease in cultivated land from 2002 to 2018 was attributed to the establishment of Eucalyptus plantations and home garden agroforestry in the catchment. However, continuous expansion of built-up areas at the expense of cultivation lands resulted from infrastructural development in the catchment. Additionally, an increase in barren land was attributed to the expansion of built-up areas and continuous use of fertilizers in agricultural lands, leading to soil salinity induced land degradation in Ethiopia (Mesene, 2017; Negasa, 2020). In conclusion, the results of this study highlight the need for sustainable land management practices in the Bilate catchment to ensure the conservation of natural resources and to mitigate the negative impacts of LULC changes on the environment.

3.2. SWAT model calibration and validation

The performance of the SWAT model in the Bilate catchment was assessed using monthly flow data from the Bilate River at Halaba gauging station, covering a period from 1978 to 2017. The model was calibrated and validated for three different periods: 1980–1988, 1993–2001, and 2009–2017, while the remaining periods were used to warm up the model. Calibration was done by using the first six years of each period, while the last three years were used for validation. Sensitivity analysis was performed to identify the most sensitive parameters affecting monthly river flow at Halaba station. The analysis was carried out using the Latin Hypercube and One-factor-At-a-Time (LH-OAT) techniques, which identified sixteen parameters as being sensitive, as shown in Table 2. The results demonstrate the suitability of the SWAT model for simulating hydrological processes in the Bilate catchment, and provide useful insights into the parameters that influence river flow in the area.

Table 2. Flow sensitive parameters and their fitted values

Parameter	Description	Range	Fitted value Periods
			1986	2002	2018
CN2	Runoff curve number to moisture condition II	(35,98)	37.61	37.61	37.691
SURLAG	Surface runoff lag coefficient	(0,12)	4.472	4.474	6.033
ESCO	Soil evaporation compensation factor	(0,1)	0.473	0.026	0.166
ALPHA_BF	Base flow recession constant	(0,1)	0.004	0.0087
EPCO	Plant uptake compensation factor	(0,1)	0.025		0.55
GW_DELAY	Groundwater delay	(0,500)	363.26	363.29	121.29
SOL_Z	Depth from soil surface to bottom of layer	(0,3500)	659.77		2371.25
GW_REVAP	Groundwater “revap” coefficient	(0.02,0.2)	0.1198	0.152	0.144
RCHRG_DP	Deep aquifer percolation fraction	(0,1)	0.0149	0.0148
CANMX	Maximum canopy storage	(0,100)	87.15	55.04	88.25
CH_K2	Effective hydraulic conductivity in main channel alluvium.	(0,500)	275.2	275.2	381.25
REVAPMN	Threshold depth of water in the shallow aquifer for “revap” to occur	(0,500)	367.7	435.7
CH-N2	Manning’s “n” value for the main channel.	(0,1)		0.183	0.627
GWQMN	Threshold depth of water in the shallow aquifer required for return flow to occur	(0,1000)		473.55
SOL_AWC	Available water capacity of the soil layer	(0,1)			0.252
SOL_K	Saturated hydraulic conductivity	(0,2000)			1045.0

Note: Numerical values 1, 2, 3 … in the superscript indicate the order of sensitive LULC parameters in each period.

Flow hydrographs were generated after identifying the significant parameters, to assess the observed and simulated flow during calibration and validation for 1986, 2002, and 2018 land use/land cover (LULC) maps (Figure 3). The results of the association between observed and simulated flow values during calibration and validation, for the three distinct periods, demonstrated the SWAT model’s ability to capture river flow at the Bilate catchment. The calibration results showed an R², ENS, p-factor, and r-factor of 0.73 to 0.79, 0.67 to 0.75, 0.81 to 0.82, and 0.73 to 0.75 in 1986, 2002, and 2018, respectively. Similarly, the validation results demonstrated an R2, ENS, p-factor, and r-factor of 0.73 to 0.82, 0.68 to 0.75, 0.81 to 0.83, and 0.74 to 0.75 in 1986, 2002, and 2018, respectively. The calibration and validation results indicate that the SWAT model achieved a relatively good fit between observations and simulations.

Figure 3. Observed and simulated Bilate River flow for calibration and validation (a) 1986 LULC, (b) 2002 LULC, and (c) 2018 LULC.

The sensitivity analysis conducted on the SWAT model parameters revealed that it can effectively capture the river flow at the Bilate catchment, and the calibration and validation results support this conclusion. The study shows that the SWAT model can be used as an efficient tool to simulate hydrological processes in the Bilate catchment. The sources cited for this work are not provided.

3.3. Effects of LULC changes on hydrology

After the calibration of the SWAT model, three simulations were performed to determine the impact of land use/land cover (LULC) changes on hydrological processes. The SWAT modelling approach was implemented independently for the periods 1986, 2002, and 2018 to evaluate the impact of LULC changes on hydrological responses. Figure 2 and Table 1 illustrate the significant LULC changes that occurred between 1986 and 2002 and between 2002 and 2018 in the Bilate catchment. The LULC changes observed over the years have affected the hydrological processes from 1986 to 2018. The impact of LULC change on the average annual hydrological processes of the Bilate catchment is presented in Table 3. The study shows that the LULC changes that occurred between 1986 and 2018 have significantly affected the hydrological processes in the Bilate catchment. The SWAT model can be used as an effective tool to assess the impact of LULC changes on hydrological processes.

Table 3. Simulated hydrological processes of three periods

Hydrological processes	1986 LULC	2002 LULC	2018 LULC
Potential evapotranspiration (PET) mm/year	3158.05	3512.97	3751.90
Actual evapotranspiration (ET) mm/year	1336.27	1415.81	1537.98
Soil water content (SW) mm/year	6590.74	5755.49	5167.49
Water percolating root zone (PERC) mm/year	915.91	820.86	1877.93
Surface runoff (SUR_Q) mm/year	640.48	659.86	1155.34
Groundwater flow (GW_Q) mm/year	738.62	648.07	1716.16
Water yield (WYLD) mm/year	1494.2	1414.62	3069.44
Bilate River flow (average m^3/s)	32.93	31.23	29.50
Sediment yield (SYLD) (metric tons/ha)	12.20	27.78	27.77
Phosphorus transported with sediment (kg/ha)	3.60	4.96	5.50

The study findings demonstrate that there were significant changes in the average yearly surface runoff, groundwater flow, water yield, water leaving root zone, potential and actual evapotranspiration between 1986 and 2018, as a result of the LULC changes in the Bilate catchment. The average yearly surface runoff, groundwater flow, water yield, potential and actual evapotranspiration increased by 514.86, 977.54, 1575.24, 593.25, and 201.71 mm/year, respectively. However, the soil water (1423.25 mm/year) and river flow (3.43 m³/s) decreased during the same period.

The increase in surface runoff, groundwater flow, water yield, potential and actual evapotranspiration, and decline in soil water were attributed to the LULC changes that occurred in the catchment. These findings are consistent with previous studies that have demonstrated the impact of LULC changes on hydrological processes. This study findings highlight the significant impact of LULC changes on the hydrological processes in the Bilate catchment. This underscores the importance of proper land management practices that take into account the potential impact on the hydrological cycle.

After analyzing the alterations in LULC in this study between 1986 and 2018, it was found that the changes led to a rise and decline in mean annual hydrological processes. These findings are consistent with other studies conducted in different catchments. For instance, Gashaw et al. (2017) reported an increase in surface runoff in Andessa catchment due to changes in LULC. Similarly,found that changes in LULC in the upper Awash basin also increased surface runoff. In contrast, Emam et al. (2017) observed an increase in water yield and actual evapotranspiration in A-Luoi, Central Vietnam, following LULC changes.

Other studies have reported a decline in water yield due to LULC changes. For instance, Saddique et al. (2020) found that LULC changes led to a decrease in water yield by 38 mm/year and an increase in evapotranspiration by 36 mm/year in the Jhelum River Basin, Pakistan. Tena et al. (2019) also reported a decrease in actual evapotranspiration by 54.3 mm/year and an increase in stream flow at a rate of 0.13 Mm3/year in the Chongwe River catchment, Zambia. In addition, Nugroho et al. (2013) reported that a decline in vegetation cover in Goseng catchment, in Central Java, led to an increase in surface runoff. Li et al. (2019) found that the decline of woodland was the primary driving force for the decrease in soil water.

The study findings indicate that alterations in land cover conditions significantly affect the soil’s capacity to hold moisture and infiltration capacity. These changes have a significant impact on the generation of infiltration and runoff. The infiltration and water holding capacity are the most affected soil properties during land-use changes (Akale et al., 2017). Consequently, percolation decreases, leading to an increase in runoff generation, which causes topsoil movement and affects sediment yield. The contribution of annual sediment yield increased by 15.58 metric tons/ha from 1986 to 2002 and then decreased by −0.01 metric tons/ha from 2002 to 2018, which is attributed to increased cultivation land from 1986 to 2002 and a minor decrease from 2002 to 2018. Similar findings have been reported in other studies conducted in different parts of Ethiopia. For instance, Kidane et al. (2019) reported sediment yield rates of 25.8, 28.7, and 30.3 tons/ha/year in the Guder sub-watershed, Central Ethiopia, for 1973, 1995, and 2015, respectively.

During the period of 1986 to 2018, both organic and mineral phosphorus attached to the sediment were transported from the catchment by surface runoff. The results showed that the phosphorus load transported by surface runoff into the reach increased by 1.36 and 0.54 kg/ha from 1986 to 2002 and 2002 to 2018, respectively. The increase in phosphorus load is attributed to the expansion of cultivation land and urban areas, increased use of fertilizer, and increased surface runoff (Brito et al., 2019; Du et al., 2014; Fetahi, 2019).

Overall, these findings demonstrate the significant impact of LULC changes on hydrological processes in catchments across different regions. Implying the importance of considering the potential consequences of LULC changes when making land-use decisions to ensure sustainable water resource management. Besides, the findings of the study suggest that land-use changes have significant implications for soil properties and sediment yield, which should be taken into account in land-use planning and management strategies.

3.4. Sub-basins contribution to hydrological responses

The study shows that surface runoff has increased over time, with annual mean surface runoff accounting for about 640.48, 659.86 and 1155.34 mm/year in 1986, 2002 and 2018, respectively. The contribution of each sub-basin has been evaluated, and it was found that the areas upstream had the highest surface runoff. The SWAT model has identified sub-basins fourteen, seventeen, sixteen, and eighteen as major contributors to the total runoff in the catchment in the three periods. Specifically, sub-basin fourteen has shown an increasing trend in mean annual surface runoff from 41.47, 46.79, and 41.38 mm/year in 1986, 2002 and 2018, respectively. On the other hand, sub-basin five, seven, and ten have been found to be the least contributors to the total runoff in the catchment. The allocation of runoff per sub-basin is shown in Figure 4.

Figure 4. Sub-basin level surface runoff in 1986, 2002 and 2018.

The high surface runoff in the upstream areas of the basin is attributed to the high slope of the area and changes in land use and land cover. These findings are consistent with the results of previous studies (Yasarer et al. 2017; Gebrehiwot et al. 2019). The downstream areas in the catchment were found to be the least contributors to the total runoff. This observation may be due to the flat topography of the downstream areas, which may cause water to infiltrate into the ground rather than generating surface runoff.

A reduction in groundwater and water yield was observed in the downstream areas of the Bilate catchment, which can be attributed to the aridity of the climate and the uneven distribution of rainfall. As one moves downstream, the water potential decreases due to rising temperatures and decreasing rainfall, resulting in a reduction in soil moisture, which is essential for plant growth. Additionally, semi-arid conditions exacerbate the problem in the downstream areas of the catchment. The reduced availability of water for crop production is a significant concern, particularly for subsistence farmers in the catchment who rely on rain-fed agriculture (Li et al., 20212021).

Overall, the study highlights the importance of understanding the contributions of each sub-basin to the total runoff in the catchment, which can help in developing effective water management strategies.

3.5. Limitations of the study

Two of the primary limitations are the uncertainty of sediment and phosphorus load simulations. The study relied solely on simulated data using stream flow, as no measured data of sediment and phosphorus in the catchment were available. Additionally, changes in water consumption in different sectors are not well documented, which makes it difficult to model and plan for future water requirements. These limitations could be addressed by mounting gauge stations and installing the required instruments to gather measured data of sediment and phosphorus loads, as well as by improving documentation of water consumption changes in various sectors.

4. Conclusions

The study aimed to evaluate the impacts of spatio-temporal land use and land cover (LULC) changes on the hydrology of the Bilate catchment in southern Ethiopia from 1986 to 2018, using the SWAT model to simulate hydrological processes. The study found that the decline of soil water content (1423.35 mm/year) and the increase in potential (593.25 mm/year) and actual (201.71 mm/year) evapotranspiration could lead to reduced water availability for crop production, which is a continuing issue for subsistence and rain-fed agriculture in the Bilate catchment. Furthermore, the increasing possibility of water yield (1575.24 mm/year), groundwater flow (977.86 mm/year), and surface runoff (514.86 mm/year) could further exacerbate water stress in the area, affecting both agricultural fields and groundwater aquifers. In addition, sediment loads carrying phosphorus could cause water quality problems in the catchment.

The study recommends the implementation of comprehensive and sustainable catchment protection programs, such as rehabilitating degraded lands to increase groundwater recharge, controlling erosion and sediment loss, managing runoff water, and creating awareness among communities on environmental values. It also suggests the need for further studies on water resources planning and agricultural production. To ensure sustainable water resources and land use, all responsible bodies should commit to and work closely with communities using participatory approaches. These approaches can help restore the present LULC changing trends and develop water resources in the Bilate catchment.

4.1. Authorship statement

The authors of this study have agreed to submit a manuscript entitled “Assessing the Impact of Land Use/Land Cover Changes on Hydrological Processes in Southern Ethiopia: The SWAT Model Approach” for publication in the Cogent Engineering Journal. We, the authors, confirm that we have been actively involved in the research process, including the development of the study’s concept, design, analysis, and the drafting or revising of the manuscript, and take full responsibility for its contents. Prior to submission, the final version of the manuscript was reviewed and approved by all authors.

Acknowledgments

The authors thank Jimma and Wolaita Soddo Universities for providing materials and resource supports for the study. The authors are pleased to thank the USGS, National Meteorological Service Agency, Ministry of Water, Irrigation and Energy of Ethiopia for the provision of Landsat, weather and Bilate River flow data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The authors confirm that the data supporting the findings of this study are available within this article. The raw data are available upon the request.

Additional information

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Notes on contributors

Hailu Gisha Kuma

Dr. Hailu Gisha is an accomplished hydrologist and esteemed assistant professor at Woliata Soda University in Ethiopia. He has gained international recognition for his exceptional contributions to the field of hydrology, as evidenced by his publication record in peer-reviewed scientific journals.

Fekadu Fufa Feyessa

Dr. Eng. Fekadu Fufa Feyessa is a distinguished scholar and associate professor of Water and Environmental Engineering at Jimma Institute of Technology, Jimma University in Oromia, Ethiopia. With his extensive expertise in the field, Dr. Eng. Fekadu’s has made a significant impact through his research and contributions to a range of areas including water resources, environmental engineering, and climate change. His exceptional work has been recognized by the international community, with numerous publications featured in prestigious peer-reviewed journals.

Tamene Adugna Demissie

Prof. Dr. Eng. Tamene Adugan is a highly regarded scholar and esteemed professor of Hydraulic and Water Resources Engineering at Jimma Institute of Technology, Jimma University in Oromia, Ethiopia. Prof. Tamene’s extensive knowledge and expertise in the fields of environment, water resources, and climate change are reflected in his numerous publications in internationally renowned peer-reviewed journals. His exceptional contributions and research have earned him global recognition as a leading authority in these areas.