graphic
View largeDownload slide

Modeling and mapping hydrological responses of runoff and sediment yield to spatiotemporal land use changes are crucial concerning environmental sustainability. The research was aimed at quantifying the spatiotemporal effects of land use on runoff and sediment yields using three land use satellite images and the SWAT+ model. The increase in agriculture, settlement, and decreasing forest goes to the possibility of increasing sediment yield and runoff by 53.2 and 56.5%, respectively, affecting ecosystems. The areas vulnerable to high runoff were found at the lower and middle reaches with the annual average runoff of 10,825.1, 11,972.9, and 13,452 mm for each respective scenario. On the other hand, most of the soil erosion-prone areas designated as severe in the second and third scenarios were covered by agriculture and shrubland, with annual sediment yields of 301.5 and 267.5 tons, respectively. Deforestation for agricultural expansion has a significant role in environmental degradation, as forests play an irreplaceable role in ecological resilience. Generally, the dominant land uses that instigate soil erosion, runoff, and sediment yield are agriculture, shrubland, and deforestation. The simulation of runoff and sediment yield in response to land use change using the SWAT+ model is more scientifically reliable and acceptable.

  • The main new part of this study focused on the applicability of the SWAT+ model in examining the response of land use change to sediment yield and runoff.

  • In addition, the novel metric in this study was the assessment of the role of each LULC change response to extreme hydrology.

  • The SWAT+ model has not yet been applied in the upper Blue Nile Basin.

environmental sustainability, hotspot area, sediment yield, soil erosion, SWAT +

Over time, human population growth and technological advancements have escalated the utilization of resources like land, water, energy, minerals, and biological elements to fulfill socioeconomic needs (Kiprotich et al. 2021). The resulting expansion of agriculture and urbanization has directly led to amplified water use for irrigation and domestic purposes (Tumsa et al. 2022), thereby contributing to changes in land use and cover, water scarcity, flood risks, and soil erosion, ultimately affecting living conditions (Berihun et al. 2022). Concurrently, alterations in land use and land cover (LULC) influence vital hydrological processes such as infiltration, groundwater recharge, base flow, and surface runoff by modifying surface characteristics (Elif Sertel et al. 2019; Umukiza et al. 2021), subsequently impacting rainfall pathways and basin runoff generation through interactions among climate, LULC, erosion, and sediment loading (Gashaw et al. 2021). Furthermore, spatiotemporal changes in land use have a significant impact on environmental sustainability and runoff and sediment yield in the catchment (Yigez et al. 2021). Significant increases in surface runoff and sediment yield resulted in the loss of fertile soil, downstream sedimentation, and a reduction in farmland productivity, which brought a food shortage to the local farmers (Belay & Mengistu 2021). The central highlands of Ethiopia are characterized by a high rate of land degradation and soil erosion (Kidane et al. 2019). There was a rapid expansion of cultivated land at the expense of forest land, mostly in the highlands of the country (Gashaw et al. 2018). In Ethiopia, land use and land cover changes (LULCCs) are associated with large negative impacts on ecosystems observed at local, regional, and global scales (Moisa et al. 2021). The Guder watershed is one of the tributaries of Ethiopia's Abbay River basin, located in the country's central highlands, and it is characterized by a variety of topographic conditions ranging from flat plains to steep areas (Duguma 2022). The catchment's topographical settings exposed the watershed to significant soil erosion, surface runoff, and sediment yields (Berihun et al. 2020).

Several studies have delved into the quantification of potential effects arising from land use and land cover (LULC) dynamics on runoff and sediment yield (Getachew & Melesse 2012). In a notable study by Saddique et al. (2020) conducted in an Indian river sub-basin, it was demonstrated that alterations in LULC led to proportional increases in surface runoff, water yield, and sediment yield, ultimately influencing the functioning of ecosystems (Gurara et al. 2021). The pivotal role of surface runoff as a driving force for sediment transport became evident, serving as a prerequisite for the processes involved in transporting sediment-laden flows during discrete flood events (Worku et al. 2017). Alarming consequences arise from the escalated soil erosion, posing not only a threat to on-site agricultural sustainability through the detachment of fertile topsoil from uplands but also inducing reservoir siltation and water pollution as sediments and soil nutrients are transported off-site, thereby jeopardizing environmental sustainability (Tan et al. 2022).

The comprehension of the ramifications arising from spatiotemporal fluctuations in LULC for runoff and sediment yield assumes paramount importance. This research concern stems from its relevance to environmental sustainability indicators and the overarching context of watershed management (Li et al. 2007; Saddique et al. 2020). The intricate interplay between these land use dynamics and their influence on runoff and sediment yield necessitates a comprehensive investigation to inform effective management strategies. As such, understanding these intricate interactions contributes significantly to the field, as they hold the key to sustainable practices that can mitigate the negative impacts on ecosystems, water quality, and the long-term stability of the landscape.

Zhang et al. (2019) and Burgan (2022) performed separate investigations studying sediment yield trends and the impacts of land use on runoff and sediment yields in southwest China and Turkey, respectively. The research by Zhang et al. (2019) simulated runoff and sediment yield responses to land use change using the Soil and Water Assessment Tool (SWAT) model in Northeast China. On the other hand, Burgan (2022) concentrated on sediment discharge patterns in the Mediterranean region. Both studies give considerable insights into the intricacies of sediment output patterns and the impacts of land use change's implications for regional water resource planning.

A hydrological model was developed to assess water resources and predict the impacts of LULC changes and land management practices on soil erosion, sedimentation, and non-point source pollution in watersheds (Ayana et al. 2012; Gavit et al. 2020; Wagner et al. 2022). To address current and future challenges in runoff and sediment yield modeling, the SWAT+ model was developed recently with input files structured for visualization, and spatial representation of elements and processes within watersheds (Wu et al. 2020). The associated modular codes were designed to facilitate future applications and development (Bieger et al. 2019; Yen et al. 2019; Kiprotich et al. 2021). So far, only a few applications of SWAT + , primarily in catchments of Africa, have been reported in the literature (Ougahi 2022). In this study, the spatiotemporal change of sensitive land use on runoff and sediment yield was evaluated for environmental sustainability.

The study makes important strides in understanding the complex relationship between shifting land use and its effects on the ecosystem. The study sheds light on how these changes affect runoff and sediment yield patterns by conducting a thorough assessment of land use dynamics in the basin. As it informs policymakers and local stakeholders about the potential effects of human activity on hydrological systems and sediment movement, this knowledge is vital for creating environmental sustainability strategies. The research also aids in the establishment of targeted and basin-specific management strategies, enabling the development of effective interventions that support sustainable land use practices, protect water quality, and lessen soil erosion in the region. Notably, the study introduces an innovative metric, assessing the role of each LULC change in extreme hydrological responses. This study uniquely investigated the impacts of dynamic land use on hydrological extremes at the HRU level than traditionally well-known at the sub-basin. The study clearly demonstrates the consequences of the hydrological extremes and informs water resource management and environmental protection sectors to target early sustainable solutions to the root cause. This research extends the applicability of the SWAT+ model to quantify the influence of LULC dynamics on runoff and sediment yield, thereby enhancing environmental sustainability indicators in the upper Blue Nile Basin.

Study area

The catchment has historically engaged in traditional agricultural practices, which have led to soil erosion primarily due to the substantial generation of runoff. It is geographically situated in Oromia, and its coordinates range from 70°30′ to 90°30′N latitude and 37°00′ to 39°00′E longitude, as depicted in Figure 1. The geographical expanse of the Guder catchment is estimated to cover 6,764.7 km2 within the broader context of the Blue Nile River basin. Within this catchment, the Guder River, a perennial watercourse, meanders from the south to the north before joining the Blue Nile River. This river not only contributes to the regional watercourse but also serves as an essential source of irrigation for crop cultivation, thus playing a pivotal role in sustaining agricultural activities (Duguma 2022).
Figure 1
Map of the study area.
View largeDownload slide

Map of the study area.

Figure 1
Map of the study area.
View largeDownload slide

Map of the study area.

Close modal

Climate conditions

The sub-basin is characterized by seasonal rainfall patterns, which are a frequent characteristic of tropical wet to semi-arid and arid climatic zones (Tumsa et al. 2022). The catchment's climatic conditions are impacted by its lofty elevation, running from 928 to 3,500 m. This high elevation adds to a humid environment, resulting in the bulk of the yearly rainfall happening from May to September. The watershed sees a mean annual rainfall of 1,228 mm, while the average maximum and lowest temperatures are at 21.47 and 9.82 °C, respectively, as indicated in Figure 2. Notably, the dry season runs from November to April, with minor rainfall seen in February and April. During this season, the northeastern portions of the watershed endure chilly and dry weather with moderate wind rates.
Figure 2
Average monthly rainfall in the Guder catchment.
View largeDownload slide

Average monthly rainfall in the Guder catchment.

Figure 2
Average monthly rainfall in the Guder catchment.
View largeDownload slide

Average monthly rainfall in the Guder catchment.

Close modal

Data sources and acquisitions

The raw data inputs for the SWAT+ model comprised DEM, soil data, land use/land cover, and meteorological data. Precipitation, maximum and minimum temperatures, relative humidity, and wind speed came from six synoptic stations. All five stations inside the watershed generated full yearly data, except for temperatures, covering the years 1990 to 2019 (Kidane et al. 2019). Meteorological data were obtained from the Ethiopian Minister of Water and Energy (MoWE). To overcome data limitations such as streamflow and sediment data which have been recorded at Guder outlet and ‘N’ the Guder stations, respectively, we used continuous recorded time series to be acquired and utilized for the calibration and validation of the model (Table 1).

Table 1

Meteorological, spatial, and hydrological data inputs for the SWAT+ model

Data typesStationsLength of recordsSource of data
Meteorological data Ambo 1990–2019 Ethiopian National Meteorological Agency (NMA) 
Gedo 
Incinni 
Jeldu 
Kachise 
Shukute 
Streamflow data Guder outlet 1992–2009 Ethiopian Ministry of Water, Irrigation, and Electricity (MWIE) 
Sediment data ‘N’ Guder 2000–2009 
Soil map  2013 
DEM (12.5 by 12.5 m)   http://vertex.daac.asf.alaska.edu 
Data typesStationsLength of recordsSource of data
Meteorological data Ambo 1990–2019 Ethiopian National Meteorological Agency (NMA) 
Gedo 
Incinni 
Jeldu 
Kachise 
Shukute 
Streamflow data Guder outlet 1992–2009 Ethiopian Ministry of Water, Irrigation, and Electricity (MWIE) 
Sediment data ‘N’ Guder 2000–2009 
Soil map  2013 
DEM (12.5 by 12.5 m)   http://vertex.daac.asf.alaska.edu 
View Large

Digital elevation model (DEM) and soil data

One of the main inputs to the SWAT+ model is a digital elevation model (DEM), which defines the terrain and specifies the elevation of any point inside the catchment at a certain spatial resolution. Topographic data is very important to create topographic features like the floodplain and terrain settings of the watershed.

The soil textural and physicochemical characteristics the SWAT+ model requires as input include soil texture, available water content, hydraulic conductivity, bulk density, and organic carbon (Teshome et al. 2022). Haplic Alisols and Eutric Leptosols are projected to cover 18 and 17% of the watershed, respectively (Negese 2021). On the other hand, Eutric Fluvisols, Haplic Arenosols, and Rendzic Leptosols are soils that each covered 3% of the total area as shown in Table 2.

Table 2

Dominant soil type distributed in the catchment (WRB database)

OrderMajor soil typesWRB_GroupSoil codeArea (km2)% Coverage
1 Calcic Vertisols Vertisols VkVr 314 5 
2 Chromic Luvisols Luvisols RxLv 551 8 
3 Dystric Cambisols Cambisols RdCm 408 6 
4 Dystric Leptosols Leptosols RdLp 983 15 
5 Eutric Cambisols Cambisols VeCm 427 6 
6 Eutric Fluvisols Fluvisols ReVr 178 3 
7 Eutric Leptosols Leptosols V/SeLp 1,119 17 
8 Eutric Vertisols Vertisols VeVr 119 2 
9 Haplic Alisols Alisols VhAl 1,213 18 
10 Haplic Arenosols Arenosols RhAr 224 3 
11 Haplic Luvisols Luvisols RhLv 600 9 
12 Haplic Nitisols Nitisols ShNt 368 5 
13 Rendzic Leptosols Leptosols RkLp 194 3 
OrderMajor soil typesWRB_GroupSoil codeArea (km2)% Coverage
1 Calcic Vertisols Vertisols VkVr 314 5 
2 Chromic Luvisols Luvisols RxLv 551 8 
3 Dystric Cambisols Cambisols RdCm 408 6 
4 Dystric Leptosols Leptosols RdLp 983 15 
5 Eutric Cambisols Cambisols VeCm 427 6 
6 Eutric Fluvisols Fluvisols ReVr 178 3 
7 Eutric Leptosols Leptosols V/SeLp 1,119 17 
8 Eutric Vertisols Vertisols VeVr 119 2 
9 Haplic Alisols Alisols VhAl 1,213 18 
10 Haplic Arenosols Arenosols RhAr 224 3 
11 Haplic Luvisols Luvisols RhLv 600 9 
12 Haplic Nitisols Nitisols ShNt 368 5 
13 Rendzic Leptosols Leptosols RkLp 194 3 
View Large

Land use classification and accuracy assessment

The satellite image of the LULC data must be rectified for any distortions and anticipated cloud cover before being used as an input for hydrological modeling (Leta et al. 2021). This improves the study's credibility and better illustrates the effects of the LULC scenario and the responses of the hydrological systems. The ETM + , TM, and OLI were the sensors that were used to download the satellite images of land use from the USGS website as depicted in Table 3. ERDAS 2015 software was used for the pixel cell mosaic, layer stack, image classification, and confusion matrix procedures for each LULC map (Sulamo et al. 2021), as shown in Tables 4–6. The general accuracy assessment using the kappa coefficient for each land use has been shown in Table 7.

Table 3

Satellite imagery data for each LULC

ScenarioBandsSensor typesPath/RowAcquisition dateResolutionCloud cover (%)
2003 7 ETM + 172/055 22/05/2003 30 m <1 
2013 8 TM 158/064 31/10/2013 15 m <1 
2021 8 OLI 169/059 01/10/2022 15 m <1 
ScenarioBandsSensor typesPath/RowAcquisition dateResolutionCloud cover (%)
2003 7 ETM + 172/055 22/05/2003 30 m <1 
2013 8 TM 158/064 31/10/2013 15 m <1 
2021 8 OLI 169/059 01/10/2022 15 m <1 
View Large
Table 4

Confusion matrix for LULC of 2021

Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 365 10 5 0 0 0 380 
FRST 10 107 16 1 0 3 137 
RNGB 10 0 350 0 6 18 384 
WATL 0 0 14 108 21 0 143 
WETL 0 12 3 21 82 0 118 
SETL 0 1 18 1 0 393 413 
Total 385 131 406 131 109 414 1,635 
Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 365 10 5 0 0 0 380 
FRST 10 107 16 1 0 3 137 
RNGB 10 0 350 0 6 18 384 
WATL 0 0 14 108 21 0 143 
WETL 0 12 3 21 82 0 118 
SETL 0 1 18 1 0 393 413 
Total 385 131 406 131 109 414 1,635 
View Large
Table 5

Confusion matrix for LULC of 2013

Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 360 10 5 0 3 2 380 
FRST 10 105 16 2 1 3 137 
RNGB 12 2 354 0 6 18 392 
WATL 0 0 14 104 24 2 144 
WETL 0 12 3 21 82 0 118 
SETL 0 1 11 7 4 390 413 
Total 382 130 403 134 120 415 1,584 
Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 360 10 5 0 3 2 380 
FRST 10 105 16 2 1 3 137 
RNGB 12 2 354 0 6 18 392 
WATL 0 0 14 104 24 2 144 
WETL 0 12 3 21 82 0 118 
SETL 0 1 11 7 4 390 413 
Total 382 130 403 134 120 415 1,584 
View Large
Table 6

Confusion matrix for LULC of 2003

Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 365 10 5 0 0 0 380 
FRST 10 107 12 0 6 3 138 
RNGB 2 8 355 0 6 13 384 
WATL 0 0 14 112 14 3 143 
WETL 0 12 3 21 82 0 118 
SETL 0 1 18 1 0 393 413 
Total 395 131 406 131 109 416 1,576 
Class nameAGRLFRSTRNGBWATLWETLSETLTotal
AGRL 365 10 5 0 0 0 380 
FRST 10 107 12 0 6 3 138 
RNGB 2 8 355 0 6 13 384 
WATL 0 0 14 112 14 3 143 
WETL 0 12 3 21 82 0 118 
SETL 0 1 18 1 0 393 413 
Total 395 131 406 131 109 416 1,576 
View Large
Table 7

Summary of each LULC classification accuracy assessment

LULC types2003
2013
2021
PA%UA%OA%KC%PA%UA%OA%KC%PA%UA%OA%KC%
AGRL 0.91 0.91 0.90 0.89 0.91 0.91 0.91 0.9 0.94 0.94 0.92 0.9 
FRST 0.82 0.73   0.9 0.84   0.82 0.73   
RNGB 0.84 0.85   0.78 0.78   0.83 0.86   
WATL 0.9 0.85   0.81 0.74   0.82 0.76   
WETL 0.79 0.79   0.87 0.72   0.75 0.84   
SETL 0.87 0.88   0.9 0.91   0.92 0.89   
LULC types2003
2013
2021
PA%UA%OA%KC%PA%UA%OA%KC%PA%UA%OA%KC%
AGRL 0.91 0.91 0.90 0.89 0.91 0.91 0.91 0.9 0.94 0.94 0.92 0.9 
FRST 0.82 0.73   0.9 0.84   0.82 0.73   
RNGB 0.84 0.85   0.78 0.78   0.83 0.86   
WATL 0.9 0.85   0.81 0.74   0.82 0.76   
WETL 0.79 0.79   0.87 0.72   0.75 0.84   
SETL 0.87 0.88   0.9 0.91   0.92 0.89   
View Large

To increase the precision of each image, 1,635 ground control points were collected to compare how well the pixels represent the actual land use of the catchment as seen in Figure 3. To evaluate the precision of the final classified image, user accuracy, producer accuracy, and Kappa coefficient (k) were used (Belay & Mengistu 2021) and computed using Equations (1) and (2):
formula
(1)
formula
(2)
where N is the total number of sites in the matrix, r is the number of rows in the matrix, is the number in row i and column i, is the total for row i, and is the total for column.
Figure 3
Types of land use and land cover.
View largeDownload slide

Types of land use and land cover.

Figure 3
Types of land use and land cover.
View largeDownload slide

Types of land use and land cover.

Close modal

Characteristics of runoff and sediment data

Sediment transportation is highly correlated to surface runoff data. The intensity of rainfall in the catchment is the one that can determine the magnitude of runoff to occur. On the other hand, the volume of surface runoff within the drainage area determines the magnitude of sediment yield to be initiated and transported. This depends on the nature of LULC in the drainage area and the coefficient of surface runoff. The average annual discharge of Guder River is estimated to be a 2.9-billion-meter cub. The catchment's minimum and maximum areas in the drainage area are 3 and 455 km2 with minimum sediment yields of 116 and 34,710 tons, respectively. The statistical characteristics of monthly sediment data and runoff in the Guder catchment have been illustrated in Table 8. The correlation coefficient (r) between the two gauging stations for both sediment and flow data were 0.46 and 0.27 for the maximum gauged sediment and flow data.

Table 8

The statistical characteristics of sediment and runoff data

TypeDrainage area (km2)Mean (tons/month)Min (tons/month)Max (tons/month)Standard deviationCVMedianr
Sediment (tons) Min 3 621.12 116 13,421.4 8,561.79 1.38 214 0.04 
Max 455 3,471 15,671 34,710 6,124.5 0.82 1,078 0.36 
Runoff (m3) Min 3 12.6 0.35 44.62 14.32 1.34 1.26 0.01 
Max 455 16.4 2.61 74.08 16.92 1.36 2.19 0.06 
TypeDrainage area (km2)Mean (tons/month)Min (tons/month)Max (tons/month)Standard deviationCVMedianr
Sediment (tons) Min 3 621.12 116 13,421.4 8,561.79 1.38 214 0.04 
Max 455 3,471 15,671 34,710 6,124.5 0.82 1,078 0.36 
Runoff (m3) Min 3 12.6 0.35 44.62 14.32 1.34 1.26 0.01 
Max 455 16.4 2.61 74.08 16.92 1.36 2.19 0.06 
View Large

Methodology

Areal rainfall estimation by the Thiessen polygon

The magnitude of rainfall across the catchment significantly influences both runoff and soil erosion. Meteorological stations frequently measure a point of rainfall that may not accurately reflect the total rainfall over the catchment. To determine the overall amount of rainfall, it is crucial to consider the average value of each station in the catchment. By assuming that the rainfall is the same at the closest gauging stations, the contribution of each synoptic station in the catchment was converted to an average value by using the Theissen polygon method (Zeberie 2019). In general, this method uses Equation (3) to estimate the mean annual aerial rainfall close to the gauged station:
formula
(3)
where represents the average areal rainfall (mm), is the precipitation of stations 1, 2 … n, respectively, and is the area coverage of stations 1, 2, 3 … , n in the Theissen polygon. The approach includes computing the average weight for each station's data in relation to their distance from one another. Notably, Kachise, Ambo, and Gedo stations account for 26.29, 23.23, and 23.12% of the total area of the watershed, respectively. For a more detailed picture, the watershed is split into 33 sub-basins, deliberately scattered throughout six unique places within the larger watershed. These chosen sub-basins are largely contained by the stations that are expected to generate the most substantial rainfall contributions and span considerable areas of the catchment region. A detailed description of this distribution can be seen in Table 9, while a visual portrayal of the Theissen polygon approach is shown in Figure 4. The application of the Theissen polygon to estimate the area of rainfall from each available station enables the study to investigate the spatial distribution of rainfall over the catchments, which is time and space-dependent.
Table 9

Aerial coverage and annual rainfall contribution of the stations

PolygonStationsLatitudeLongitudeArea (km2)% CoverageSub-basins in stations
1 Ambo 8.99 37.84 1,571.41 23.23 [14, 23, 24, 26, 27, 29, 31] 
2 Gedo 9.02 37.46 1,563.75 23.12 [28, 30, 32, 33] 
3 Incinni 8.84 37.67 737.78 10.91 [12, 13, 15, 16, 18, 19, 20, 25] 
4 Jeldu 9.26 38.09 517.83 7.66 [17, 21, 22] 
5 Kachise 9.61 37.86 1,778.37 26.29 [1, 2, 3, 4, 5, 6, 7, 9, 10] 
6 Shukute 9.78 38.04 594.58 8.79 [8, 11] 
PolygonStationsLatitudeLongitudeArea (km2)% CoverageSub-basins in stations
1 Ambo 8.99 37.84 1,571.41 23.23 [14, 23, 24, 26, 27, 29, 31] 
2 Gedo 9.02 37.46 1,563.75 23.12 [28, 30, 32, 33] 
3 Incinni 8.84 37.67 737.78 10.91 [12, 13, 15, 16, 18, 19, 20, 25] 
4 Jeldu 9.26 38.09 517.83 7.66 [17, 21, 22] 
5 Kachise 9.61 37.86 1,778.37 26.29 [1, 2, 3, 4, 5, 6, 7, 9, 10] 
6 Shukute 9.78 38.04 594.58 8.79 [8, 11] 
View Large
Figure 4
(a) Thiessen polygons of the study area and (b) sub-basins.
View largeDownload slide

(a) Thiessen polygons of the study area and (b) sub-basins.

Figure 4
(a) Thiessen polygons of the study area and (b) sub-basins.
View largeDownload slide

(a) Thiessen polygons of the study area and (b) sub-basins.

Close modal

Description of the modified SWAT model

SWAT has been totally modified and named the SWAT+ model which makes use of connection files to increase its flexibility with regard to the delineation and interaction of spatial objects in a watershed (Yen et al. 2019). The model was modified to improve the runoff routing capabilities of the SWAT model, allowing flow to be routed between the separated sub-basins to landscape units (LSUs) with upland areas and floodplains (Bieger et al. 2019). To further discretize sub-basins and enable the separation of upland processes from wetlands, the model incorporates LSUs. Sub-basins are defined during model setup using stream thresholds, while LSUs are defined using channel thresholds (Aredehey et al. 2020). HRUs are defined following LSU delineation, which has its own set of HRUs. The flow produced by the HRUs accumulates at the LSU level and can be directed from the LSU to any other spatial object situated within the watershed. The primary goal of this study is to estimate the amount of sediment generated by runoff, which causes soil erosion. The hydrologic cycle in a catchment is simulated based on the water balance (Equation (4)):
formula
(4)
where is the final soil water content (mm) on day i, is the initial soil water content on day i (mm), t is the time (days), is the amount of precipitation on day i (mm), is the amount of surface runoff on day i (mm), is the amount of evapotranspiration on day i (mm), is the amount of water entering the vadose zone from the soil profile on day i (mm), and is the amount of return flow on day i (mm) (Sitotaw et al. 2021).
The model uses daily rainfall data and the soil conservation service (SCS)–curve number (CN) to estimate surface runoff and peak runoff rates for each HRU (Leta 2020), using Equation (5). The SCS method is a simple, widely used, and effective method for determining runoff from rainfall and the main limitation is that it is sensitive to CN values and adapted considering the initial abstraction with equal proportions:
formula
(5)
where is the accumulated runoff or rainfall excess (mm), is the rainfall depth for the day (mm), is the initial abstraction which includes surface storage, interception, and infiltration before runoff (mm H2O), and S is the retention parameter (mm H2O).
For each hydrological response unit (HRU), the SWAT+ model calculates sediment yield using the modified universal soil loss equation (MUSLE). MUSLE was used in this study due to its simplicity and flexibility in simulating sediment yield and can also be used as a sub-model within an established model such as SWAT+. Moreover, this method primarily considers the effects of runoff on soil detachment due to discharge and the runoff volume for heavy rainfall events (Zhang et al. 2019). MUSLE uses peak discharge and surface runoff, which improves model forecast precision. The sediment yield computation is given in Equation (6):
formula
(6)
where is the sediment yield (tons/day), is the surface runoff volume (mm/day), is the peak runoff rate (m3/s), is the area of the HRU (ha), is the universal soil loss equation (USLE) soil erodibility factor, is the USLE crop management factor, is the USLE topographic factor, and is the coarse fragment factor.

Sediment rating curve

For the SWAT+ model calibration and validation, the sediment data collected from MOWIE was insufficient. Because the recorded sediment data at the catchment outflow is not time-step recorded and should be generated from the plot of streamflow versus sediment yield using Equations (7) and sensitivity analysis was properly conducted and stated in the results.
formula
(7)
where Qs is the sediment load in tons/day, Qi is the instantaneous stream discharge in m3/s, and a and b are regression constants. As shown in Figure 5, the values of ‘a’ and ‘b’ in Equation (8) were determined to be 106.56 and 0.95, respectively.
formula
(8)
Figure 5
Sediment rating curve for the watershed.
View largeDownload slide

Sediment rating curve for the watershed.

Figure 5
Sediment rating curve for the watershed.
View largeDownload slide

Sediment rating curve for the watershed.

Close modal

Model performance evaluation

In this study, the SWAT+ model calibration and validation were evaluated using the coefficient of determination (R2), Nash–Sutcliffe efficiency (NSE), percent bias (PBIAS), mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE) using Equations (9)–(14). R2 describes the degree of collinearity between simulated and measured data. R2 ranges from 0 to 1, with higher values indicating less error variance, and a value close to 0 means a very low correlation between observed and simulated quantities. NSE shows how well the plot of observed versus simulated data fits the 1:1 line. NSE ranges between and 1 with NSE = 1 being the optimal value. PBIAS indicates the average tendency of simulated results to be greater than their observed data. It measures the difference between simulated and observed quantities and its optimal value is 0. A positive value of the model represents an underestimation while a negative value represents an overestimation (Moriasi et al. 2007). Furthermore, MSE and MAE are used to measure the difference between the mean of simulated and observed data. The optimal value of both MAE and MSE equal to zero indicates the simulated value fits the observed. The optimal values of RMSE are between 0.2 and 0.5. General performance ratings of statistical parameters such as R2, NSE, RMSE, MAE, MSE, RSR, and PBIAS are given in Table 10:
formula
(9)
formula
(10)
formula
(11)
formula
(12)
formula
(13)
formula
(14)
where and represent the observed data and model-simulated value, respectively; and represent the mean of observed and simulated values for the entire period of the evaluation, respectively; and n is the total number of observed and simulated data pairs.
Table 10

General performance ratings for calibration and validation of the model

PBIAS
RSRRMSENSER2Performance rating
SedimentStreamflow
PBIAS < ±15 PBIAS < ±10 0 ≤ RSR ≤ 0.5 RMSE ≤ 0.75 0.75 < NSE ≤ 1 0.75 < R2 ≤ 1 Very good 
±15 ≤ PBIAS < ±30 ±10 ≤ PBIAS < ±15 0.5 ≤ RSR ≤ 0.6 0.75 ≤ RMSE ≤ 1 0.65 < NSE ≤ 0.75 0.65 < R2 ≤ 0.75 Good 
±30 ≤ PBIAS < ±55 ±15 ≤ PBIAS < ±25 0.6 ≤ RSR ≤ 0.7 1 ≤ RMSE ≤ 2 0.50 < NSE ≤ 0.65 0.50 < R2 ≤ 0.65 Satisfactory 
PBIAS ≥ ±55 PBIAS ≥ ±25 RSR > 0.7 RMSE ≥ 2.0 NSE ≤ 0.50 R2 ≤ 0.50 Unsatisfactory 
PBIAS
RSRRMSENSER2Performance rating
SedimentStreamflow
PBIAS < ±15 PBIAS < ±10 0 ≤ RSR ≤ 0.5 RMSE ≤ 0.75 0.75 < NSE ≤ 1 0.75 < R2 ≤ 1 Very good 
±15 ≤ PBIAS < ±30 ±10 ≤ PBIAS < ±15 0.5 ≤ RSR ≤ 0.6 0.75 ≤ RMSE ≤ 1 0.65 < NSE ≤ 0.75 0.65 < R2 ≤ 0.75 Good 
±30 ≤ PBIAS < ±55 ±15 ≤ PBIAS < ±25 0.6 ≤ RSR ≤ 0.7 1 ≤ RMSE ≤ 2 0.50 < NSE ≤ 0.65 0.50 < R2 ≤ 0.65 Satisfactory 
PBIAS ≥ ±55 PBIAS ≥ ±25 RSR > 0.7 RMSE ≥ 2.0 NSE ≤ 0.50 R2 ≤ 0.50 Unsatisfactory 
View Large

Sensitivity analysis, calibration, and validation of the SWAT+ model

There are numerous sources of uncertainty that relate to model assumptions and input data. The streamflow and other components are being significantly impacted by these critical scenarios (Hosseini & Khaleghi 2020). Additionally, it has an impact on the catchment-specific hydrological process and sediment distribution. Therefore, it is important to identify these sensitive parameters so that they may be improved and validated for the entire watershed's surface runoff and sediment yield scenarios. Sobol, Fourier Amplitude, Random Balance Design Fourier Amplitude, and Delta Moment Independent Measurements are popular sensitivity analysis techniques in SWAT+ (Wu et al. 2020).

The Sobol approach was chosen because of its wide use and ability to estimate significant sensitivity parameters using the P-factor and t-test (Kakarndee & Kositsakulchai 2020). The SWAT+ model's capability to calibrate streamflow and sediment output was also examined. For these reasons, three models were developed and evaluated to select the best calibration parameter values for each land use map corresponding to that climate period (Table 11). The flowchart with the necessary steps that shows the overall study sequence is given in Figure 6.
Table 11

Developed scenarios in response to each land use along with weather data

ScenariosLULC mapsWeather data
 2003 1990–2019 
 2013 1990–2019 
 2021 1990–2019 
ScenariosLULC mapsWeather data
 2003 1990–2019 
 2013 1990–2019 
 2021 1990–2019 
View Large
Figure 6
Methodological flowchart.
View largeDownload slide

Methodological flowchart.

Figure 6
Methodological flowchart.
View largeDownload slide

Methodological flowchart.

Close modal

Spatial map analysis

The evaluation of hydrological processes at the catchment or watershed level requires spatial and statistical analysis. These methods are mostly used to evaluate the correlation of hydrological processes that are interdependent. In this study, spatial maps were developed to show the distribution of runoff and sediment yield in the catchment for best management practices within the specified location. This spatial map was developed using inverse distance weighting (IDW), which is a popular interpolation method relative to others like kriging (Zhang et al. 2018). The IDW method is mathematically defined as shown in Equations (15) and (16):
formula
(15)
where Vf is the interpolated value at the considered station, Vi is the data at grid point i, di is the distance from grid point i to the station, n is the total number of grid points surrounding the stations.
formula
(16)
where d is the distance between two points on the earth's surface and r is the earth's radius.

Indicators of LULCCs

Since 1996, the watershed has experienced changes in LULC as agriculture and settlement increased. From 2003 to 2013, the rate of change in agricultural land was 16.8% and from 2013 to 2021, it was 6.94%. As the food security of the people depends mainly on agriculture, this proves that the river basin has been under considerable pressure from socioeconomic mobilization. Watershed ecosystems are severely affected by the widespread conversion of forests and shrublands to agricultural land. Over the past 10 years, the area of forest land has decreased, from 2003 to 2013 by 4.74% and from 2013 to 2021 by 7.12%. This indicates a large influx of people into the catchment area and increased deforestation activities. Due to the significant changes in land use, there has been severe soil erosion and high surface runoff in the river basin, which has caused the sediment load to enter the river basin. After nearly 20 years, 81.2% of agricultural land, 9.15% of forest land, and 5.94% of the area have been recently occupied by settlements. For the three different maps, an analysis of changes in land use was performed using scenarios based on replacing one land use with another, considering the main effect and roles of each land use in causing surface runoff and soil erosion.

Scenario 1: Deforestation of agricultural land between 2003 and 2021

Since 1996, due to socioeconomic movements for food security, massive agricultural activities have taken place in the Guder catchment. This scenario suggests that most of the forest land, especially in the upper part of the river basin, has been converted to agricultural land as shown in Figure 7. Agricultural land expansion increased by 17.19% and deforestation increased by 11.85%. This suggests that the expansion of agricultural activities in the basin leads to partial deforestation. The increase in the agricultural land area caused the increased surface runoff by 260.56 mm, water production by 287.16 mm, and soil erosion by 138.92 tons/ha, resulting in sediment loads downstream of the river section.
Figure 7
Spatiotemporal of major land use and land cover variability 2003 to 2021.
View largeDownload slide

Spatiotemporal of major land use and land cover variability 2003 to 2021.

Figure 7
Spatiotemporal of major land use and land cover variability 2003 to 2021.
View largeDownload slide

Spatiotemporal of major land use and land cover variability 2003 to 2021.

Close modal

Scenario 2: The replacement of shrubland by settlements and agricultural land between 2003 and 2021

Some land covers and land use that are important in maintaining ecosystem stability, such as shrubs, forests, or wetlands, have declined over the past two decades. The interest in expanding agricultural land has led to environmental degradation due to massive population mobility in the river basin. As a result, shrubland has been reduced by 9.59% and is eligible for conversion to residential and agricultural land. On the other hand, the settlement rate increased by 5.22% along with urbanization and rural settlement. In this case, surface runoff intensity increased due to urban expansion and soil erosion started due to agricultural expansion as shown in Table 12 and Figure 8.
Table 12

Annual land use and land cover changes

LULC classes200320132021Annual change (%)
Rate of change (%)
Area (km2)Area (km2)Area (km2)2003–20132013–20212003–20132013–2021
Agriculture 4,333 5,127.3 5,496 11.74 5.45 16.80 6.94 
Forest 1,421 1,100.5 619 −4.74 −7.12 −2.56 −5.75 
Shrubland 852 410.0 203 −6.53 −3.06 −7.30 −3.02 
Waterbody 25 18.0 14 −0.10 −0.06 −3.29 −2.51 
Wetland 87 43.2 33 −0.65 −0.15 −7.00 −2.69 
Settlements 49 68.0 402 0.28 4.94 3.28 9.54 
LULC classes200320132021Annual change (%)
Rate of change (%)
Area (km2)Area (km2)Area (km2)2003–20132013–20212003–20132013–2021
Agriculture 4,333 5,127.3 5,496 11.74 5.45 16.80 6.94 
Forest 1,421 1,100.5 619 −4.74 −7.12 −2.56 −5.75 
Shrubland 852 410.0 203 −6.53 −3.06 −7.30 −3.02 
Waterbody 25 18.0 14 −0.10 −0.06 −3.29 −2.51 
Wetland 87 43.2 33 −0.65 −0.15 −7.00 −2.69 
Settlements 49 68.0 402 0.28 4.94 3.28 9.54 
View Large
Figure 8
Annual rate of change in LULC.
View largeDownload slide

Annual rate of change in LULC.

Figure 8
Annual rate of change in LULC.
View largeDownload slide

Annual rate of change in LULC.

Close modal

Scenario 3: Conversion of wetland and waterbodies to shrubland between 2003 and 2021

Some of the watersheds and wetland-covered basins have been transformed into shrublands. The area produces high runoff and water yields as shrubland is less susceptible to sediment runoff and soil erosion than other land uses. However, spatiotemporal changes in water bodies, wetlands, and shrublands have resulted in a decreasing trend as seen from historical LULC images.

Calibration and validation of the SWAT+ model

The assessment of the SWAT+ model's performance encompassed the utilization of statistical parameters to fine-tune river flow calibration and identify influential factors governing river flow, surface runoff, and sediment yields. With a focus on enhancing hydrological simulations, the model emphasized the 15 most responsive factors, calibrated against recorded river flow and sediment yield data from the Guder catchment outlet. Concurrently, non-sensitive parameters were held constant. The calibrated SWAT+ model effectively adjusted river flows and sediment yield, demonstrating commendable performance that aligned well with observed flow patterns over time. However, the calibration and validation phases did not yield consistent trends, occasionally resulting in overestimations and underestimations of specific events, potentially attributed to inherent uncertainties within the hydrological model. In the calibration process, the adjustment of SWAT+ parameter values was achieved through trial and error to ensure their alignment within acceptable ranges, thereby optimizing predictive hydrological processes like surface runoff and water yield. Primarily, streamflow calibrations were performed to adjust the SWAT+ parameter values within acceptable ranges to maximize the hydrological processes predicted by the model, such as surface runoff and water yield. The model's accuracy in calibrating streamflow is evident, as reflected in statistical measures such as R2 = 0.74, NSE = 0.77, and PBIAS = −3.20 for calibration and R2 = 0.82, NSE = 0.86, and PBIAS = 1.46 for validation as shown in Table 13 and Figure 9. This proficient streamflow calibration underscored the model's capacity to replicate observed flow patterns, albeit with certain discrepancies, reaffirming the intricate challenges of addressing uncertainties inherent to hydrological modeling. Whereas the performance of the SWAT+ model to calibrate and validate sediment yields was measured using average statistical tool values that represent the simulation period with R2 = 0.77, NSE = 0.85, RMSE = 3.79, and PBIAS = 2.07 for calibration and R2 = 0.9, NSE = 0.92, RMSE = 2.85, and PBIAS = 1.1 for validation, which shows strong agreement as shown in Figures 10 and 11.
Table 13

Performance of SWAT+ in calibrated and validated sediment yields and runoff

SeriesProcessR2NSERMSEPBIASMAEMSERSR
Streamflow (m3/s) Calibration 0.74 0.77 4.76 −3.2 2.82 5.63 0.52 
Validation 0.82 0.86 3.75 1.46 1.08 3.74 0.44 
Sediment (tons) Calibration 0.77 0.85 3.79 2.07 2.13 4.19 0.68 
Validation 0.9 0.92 2.85 1.1 1.02 2.43 0.32 
SeriesProcessR2NSERMSEPBIASMAEMSERSR
Streamflow (m3/s) Calibration 0.74 0.77 4.76 −3.2 2.82 5.63 0.52 
Validation 0.82 0.86 3.75 1.46 1.08 3.74 0.44 
Sediment (tons) Calibration 0.77 0.85 3.79 2.07 2.13 4.19 0.68 
Validation 0.9 0.92 2.85 1.1 1.02 2.43 0.32 
View Large
Figure 9
Streamflow calibration for each scenario (2000–2019).
View largeDownload slide

Streamflow calibration for each scenario (2000–2019).

Figure 9
Streamflow calibration for each scenario (2000–2019).
View largeDownload slide

Streamflow calibration for each scenario (2000–2019).

Close modal
Figure 10
Calibration of sediment yield with the model developed for each scenario (2000–2015).
View largeDownload slide

Calibration of sediment yield with the model developed for each scenario (2000–2015).

Figure 10
Calibration of sediment yield with the model developed for each scenario (2000–2015).
View largeDownload slide

Calibration of sediment yield with the model developed for each scenario (2000–2015).

Close modal
Figure 11
Validation graph for sediment yield.
View largeDownload slide

Validation graph for sediment yield.

Figure 11
Validation graph for sediment yield.
View largeDownload slide

Validation graph for sediment yield.

Close modal

Sensitivity parameters

The most sensitive parameters that affected the calibration and validation process for streamflow were SCS-CN for moisture condition II (cn2), soil water factor for curve number III (cn3_swf), the scope of base flow alpha factor (alpha_bf), groundwater contribution to streamflow (mm H2O) (gwflow_lte), the depth of water in the shallow aquifer required for return flow to occur (gwflow_lte), minimum shallow aquifer water depth required to return flow (mm H2O) (flo_min), universal soil loss equation p-factor (usle_p), shallow aquifer water depth required for percolation to the deep aquifer (mm H2O) (revap_min), groundwater ‘revap’ coefficient (revap_co), plant uptake compensation factor (epco), and soil evaporation compensation factor (esco) are the most sensitive parameters that affect water balance based on their maximum and minimum value output from the SWAT+ model (Table 14).

Table 14

Sensitive parameters and the calibrated value

RankParametersObjectMin valueCalibrated valueMax valueUnits
1 cn2 hru 35 63 95  
2 cn3_swf hru 0 0.467 1  
3 alpha aqu 0 0.82 1 days 
4 bf_max aqu 0.1 1.45 2 mm 
5 gwflow_lte hlt 0 6.04 10 mm H2O 
6 flo_min aqu 0 0.13 0.5 m 
7 usle_p hru 0 0.92 1  
8 revap_min aqu 0 36.1 50 mm H2O 
9 revap_co aqu 0.02 0.27 0.2  
10 epco hru 0 0.75 1  
11 esco hru 0 0.65 1  
12 perco hru 0 0.33 1 fractions 
13 gw_lte hlt 0 546 1,000 mm 
14 slope hru 0 0.52 0.9 m/m 
15 ovn hru 0.01 19.2 30 m 
16 uslek_lte hlt 0 0.42 0.65  
RankParametersObjectMin valueCalibrated valueMax valueUnits
1 cn2 hru 35 63 95  
2 cn3_swf hru 0 0.467 1  
3 alpha aqu 0 0.82 1 days 
4 bf_max aqu 0.1 1.45 2 mm 
5 gwflow_lte hlt 0 6.04 10 mm H2O 
6 flo_min aqu 0 0.13 0.5 m 
7 usle_p hru 0 0.92 1  
8 revap_min aqu 0 36.1 50 mm H2O 
9 revap_co aqu 0.02 0.27 0.2  
10 epco hru 0 0.75 1  
11 esco hru 0 0.65 1  
12 perco hru 0 0.33 1 fractions 
13 gw_lte hlt 0 546 1,000 mm 
14 slope hru 0 0.52 0.9 m/m 
15 ovn hru 0.01 19.2 30 m 
16 uslek_lte hlt 0 0.42 0.65  

Note: hru, hydrological response unit; aqu, aquifer; hlt, hru_lte; bsn, basin; rte, routing.

View Large

Prioritization of sensitive parameters that affect sediment yields

Simulated sediment yields were calibrated against the observed data in favor of 11 sensitive parameters that influence sediment yield in submarine catchments. The most sensitive parameters are the USLE support p-active factor (usle_p), USLE land cover management factor (usle_c), channel sediment routing linear factor (sp.con), USLE soil erosion factor (uslek_lte), and LAT and GW (lat_sed), newly entered channel deposition amount (spex.bsn), channel deposition parameter (sp_con.), and channel erosion coefficient (ch_cov2). The calibrated values for each sensitivity parameter shown in Table 15 indicate that the model was able to simulate catchment sediment loading relative to the record. The rest of the remaining parameters are sensitive, but not as sensitive as the first five selected parameters.

Table 15

The best calibrated parameters for sediment yield and their fitted values

RankParametersObjectDescriptionMinFittedMax
1 usle_p hru USLE support practice factor 0 0.72 1 
2 usle_c hru USLE land cover management factor 0.03 0.26 0.6 
3 sp.con hru Linear factor for sediment routing 0.001 0.014 0.01 
4 uslek_lte hlt USLE soil erodability factors 0 0.482 0.65 
5 Lat_sed hru Sediment intensity in LAT and GW 0 64 120 
6 spex.bsn bsn Re-entrained channel sediment routing 1 1.134 2 
7 sp.con bsn Parameter for channel sediment routing 0 0.002 0.01 
8 ch_cov2 rte Channel erodability factors 0.6 0.88 1 
9 ch_eqn rte Sediment channel routing method 0 0.0065 0.001 
10 slope hru Slope intensity at channel HRUs 0 0.76 0.9 
11 surlags bsn Time sediment concentration lags 0.05 13.4 24 
RankParametersObjectDescriptionMinFittedMax
1 usle_p hru USLE support practice factor 0 0.72 1 
2 usle_c hru USLE land cover management factor 0.03 0.26 0.6 
3 sp.con hru Linear factor for sediment routing 0.001 0.014 0.01 
4 uslek_lte hlt USLE soil erodability factors 0 0.482 0.65 
5 Lat_sed hru Sediment intensity in LAT and GW 0 64 120 
6 spex.bsn bsn Re-entrained channel sediment routing 1 1.134 2 
7 sp.con bsn Parameter for channel sediment routing 0 0.002 0.01 
8 ch_cov2 rte Channel erodability factors 0.6 0.88 1 
9 ch_eqn rte Sediment channel routing method 0 0.0065 0.001 
10 slope hru Slope intensity at channel HRUs 0 0.76 0.9 
11 surlags bsn Time sediment concentration lags 0.05 13.4 24 
View Large

The effects of LULC changes on surface runoff and sediment yield

The basins exhibit an average elevation ranging from 895 to 3,330 m. Each basin contributed to surface runoff, soil erosion, and water development processes. Within the watershed, a configuration of 1,761 HRUs and 184 LSUs is observed. The distribution of HRUs is primarily influenced by variations in land cover, slope, and soil properties. Notably, the hydrological attributes of each sub-basin encompass diverse combinations of land uses, land covers, soil groups, and slope characteristics.

The calculation of annual surface runoff was executed through the employment of the SCS-CN method, while sediment yield was determined utilizing the MUSLE. Remarkably, surface runoff and water production exhibited an increasing trend, particularly in response to an average annual rainfall of 1,338.05 mm. This observed correlation between precipitation and surface runoff aligns with the anticipated hydrological response, showcasing the watershed's sensitivity to varying levels of rainfall.

Rapid expansions in agricultural land, urban settlements, and deforestation significantly amplify the potential for escalated runoff. In specific instances, the sediment yields, and runoff witnessed a noteworthy surge, soaring by 53.2 and 56.5% for Scenario 1 and by corresponding 53.2 and 56.5% for Scenario 2. However, the transition in LULC from Scenario 2 to Scenario 3 triggered a gradual yet discernible elevation in sediment concentration and surface runoff, reaching 4.33 and 2.31%, respectively, under comparable precipitation conditions (Table 16). These findings are graphically illustrated in Figure 12.
Table 16

Changes in runoff and sediment yield

ScenariosRainfall (mm)Runoff (mm)WYLD (mm)Sediment yield (tons/ha)CN
 1,338.6 94.05 106.45 54.15 75.23 
 1,338.6 338.61 390.51 177.05 79.31 
 1,338.6 354.61 393.61 193.07 79.45 
Changes – (+) 260.56 (+) 287.16 (+) 138.92 (+) 4.08 
ScenariosRainfall (mm)Runoff (mm)WYLD (mm)Sediment yield (tons/ha)CN
 1,338.6 94.05 106.45 54.15 75.23 
 1,338.6 338.61 390.51 177.05 79.31 
 1,338.6 354.61 393.61 193.07 79.45 
Changes – (+) 260.56 (+) 287.16 (+) 138.92 (+) 4.08 
View Large
Figure 12
Long-term changes of runoff, water yield, and sediment yields.
View largeDownload slide

Long-term changes of runoff, water yield, and sediment yields.

Figure 12
Long-term changes of runoff, water yield, and sediment yields.
View largeDownload slide

Long-term changes of runoff, water yield, and sediment yields.

Close modal

The spatiotemporal alterations in land use, when progressing from Scenario 1 to Scenario 2, resulted in substantial sediment loads and elevated runoff within the watershed. This dynamic response underscores the rapid adaptability of hydrological processes to shifts in LULC. Additional markers of heightened surface runoff are reflected in augmented mean curve numbers and elevated water yield values for the soil moisture precursor II (cn2), a correlation also identified by Leta et al. (2021). Over the span of a year, water yield exhibited an annual increment of 284.06 mm between Scenario 1 and Scenario 2, with a corresponding elevation of 3.1 mm between Scenario 2 and Scenario 3. These findings collectively highlight the sensitivity of hydrological systems to changes in land use and their implications for surface runoff dynamics.

Potential runoff-contributing areas in the catchment under LULC changes

The utility of SWAT+ models in accurately simulating and forecasting hydrological processes at both HRU and LSU levels within watersheds is well-established. In the context of this study, the central objective of the model was to comprehensively simulate and estimate surface runoff in response to spatiotemporal land use alterations, encompassing three distinct, independently developed scenarios. This analysis consistently indicated an upward trend in runoff, with substantial runoff occurrences across each LSU within the basin, precipitating consequential soil erosion. In the initial scenario, the annual runoff encompassed an area of 584.53 km2 (8.64%) relative to the total catchment area. The quantitative breakdown of the LSUs exhibiting the maximum potential runoff contribution is provided in Table 17, with Scenario 2 and Scenario 3 displaying progressively larger coverage areas of 620.13 (9.17%) and 696.5 km2 (10.3%), respectively, thereby reinforcing the trend of heightened runoff.

Table 17

Potential high runoff-generating area in the catchment

LSU codeArea (km2)Runoff (mm)Landscape unitsArea (km2)Runoff (mm)Landscape unitsArea (km2)Runoff (mm)
310 16.06 709.8 310 16.06 727.8 280 20.12 721.7 
320 41.4 718.7 320 41.4 710.9 310 16.06 727.8 
380 28.99 702.2 380 28.99 725.7 320 41.4 710.9 
490 135.6 705.7 490 135.6 775.8 380 28.99 725.7 
570 24.23 676.3 920 41.69 724.6 490 135.6 775.8 
940 76.34 751.6 940 76.34 793.3 570 24.23 767.4 
1,010 10.17 703.2 1,010 10.17 774.7 770 96.34 688.7 
1,120 13.13 829.1 1,120 13.13 862.2 940 76.34 793.3 
1,210 30.7 757.1 1,210 30.7 685.3 1,010 10.17 774.7 
1,630 47.71 763.8 1,250 18.32 699.1 1,120 13.13 862.2 
2,330 58.5 664.8 1,630 47.71 768.9 1,210 30.7 685.3 
2,460 13.93 668.7 2,330 58.5 720.4 1,250 18.32 699.1 
2,500 48.64 778.1 2,460 13.93 697.3 1,630 47.71 768.9 
2,510 15.46 665.4 2,500 48.64 821.2 1,720 23.67 725.5 
2,540 23.67 730.6 2,510 15.46 723.1 2,460 13.93 697.3 
   2,540 23.67 762.6 2,500 48.64 821.2 
      2,540 23.67 762.6 
      2,620 27.48 743.9 
Total 584.53   620.31   696.5  
LSU codeArea (km2)Runoff (mm)Landscape unitsArea (km2)Runoff (mm)Landscape unitsArea (km2)Runoff (mm)
310 16.06 709.8 310 16.06 727.8 280 20.12 721.7 
320 41.4 718.7 320 41.4 710.9 310 16.06 727.8 
380 28.99 702.2 380 28.99 725.7 320 41.4 710.9 
490 135.6 705.7 490 135.6 775.8 380 28.99 725.7 
570 24.23 676.3 920 41.69 724.6 490 135.6 775.8 
940 76.34 751.6 940 76.34 793.3 570 24.23 767.4 
1,010 10.17 703.2 1,010 10.17 774.7 770 96.34 688.7 
1,120 13.13 829.1 1,120 13.13 862.2 940 76.34 793.3 
1,210 30.7 757.1 1,210 30.7 685.3 1,010 10.17 774.7 
1,630 47.71 763.8 1,250 18.32 699.1 1,120 13.13 862.2 
2,330 58.5 664.8 1,630 47.71 768.9 1,210 30.7 685.3 
2,460 13.93 668.7 2,330 58.5 720.4 1,250 18.32 699.1 
2,500 48.64 778.1 2,460 13.93 697.3 1,630 47.71 768.9 
2,510 15.46 665.4 2,500 48.64 821.2 1,720 23.67 725.5 
2,540 23.67 730.6 2,510 15.46 723.1 2,460 13.93 697.3 
   2,540 23.67 762.6 2,500 48.64 821.2 
      2,540 23.67 762.6 
      2,620 27.48 743.9 
Total 584.53   620.31   696.5  
View Large

An additional indicator of augmented water flow and runoff materialized through the escalation in LSUs. As visually represented in Figure 13, spatial maps for each scenario spotlight regions of elevated exposure, potential water production, and runoff are denoted by the color red. Among the 184 LSUs delineated within the basin, 15, 16, and 18 LSUs were identified as being susceptible to high-potential runoff, characterized by annual averages of 10,825, 11,972.9, and 13,452 mm for Scenarios 1, 2, and 3, respectively. These findings collectively emphasize the cumulative annual potential runoff from these high-risk regions during Scenarios 1 through 3, effectively underscoring the basin's vulnerability to flooding events. In a broader context, the modifications in LULC within the Guder catchment have directly contributed to heightened water yields, attributed to the expansion of agricultural land, ensuing soil erosion, and intensified surface runoff.
Figure 13
Spatial variability maps of water yield and potential runoff-generating areas.
View largeDownload slide

Spatial variability maps of water yield and potential runoff-generating areas.

Figure 13
Spatial variability maps of water yield and potential runoff-generating areas.
View largeDownload slide

Spatial variability maps of water yield and potential runoff-generating areas.

Close modal

Sediment yields prone areas and severity index

Spatial variability of sediment loading in watersheds has been of great importance for identifying hotspot regions of soil erosion that change in LULC. The yield of this sediment was driven by the combined effects of LULC change, soil type, climate change, and increased runoff. In this study, annual averages of sediment generation were used to create spatial variation maps of sediment concentrations generated from each LSU, indicating the most severe areas of the watershed for recommending remedial measures which came to the same conclusion (Kefay et al. 2022). This spatial map was created to identify different sediment yield areas based on the number of LSUs depicted in Figure 14.
Figure 14
Spatial variability maps of the susceptible watershed to sediment yields.
View largeDownload slide

Spatial variability maps of the susceptible watershed to sediment yields.

Figure 14
Spatial variability maps of the susceptible watershed to sediment yields.
View largeDownload slide

Spatial variability maps of the susceptible watershed to sediment yields.

Close modal
According to this severity class, 96 LSUs were designated as areas of low risk of soil erosion, 7 LSUs were highly exposed to sedimentation, and most of the LSUs highly susceptible to runoff were covered with agriculture, scrubland, and some woodlands. It is in the lower part of the catchment, with a total endangered area of 348.89 km2 (5.14%) and a total annual sediment yield of 253.7 tons. On the other hand, most of the soil erosion areas classified as severe in Scenarios 2 and 3 are covered by agriculture and shrubland. The area under sediment yield risk was 353.26 km2, with an annual total sediment yield of 301.5 tons for Scenario 2, and 387.6 km2 total area with an annual sediment yield of 267.5 tons for Scenario 3, as shown in Table 18 and Figure 15. It is justified that steep ruby slopes, Leptosols, Cambisols, and consolidated soil types in the lower reaches of the watershed are the main contributors to soil erosion. It also shows how changes in LULC accelerated the increase in sediment yields due to increased runoff at the basin level.
Table 18

Vulnerable LSU to sediment yields under each scenario

LSUs codeArea (km2)Sediment yield (tons/ha)LSUsArea (km2)Sediment yield (tons/ha)LSUsArea (km2)Sediment yield (tons/ha)
10 17.66 37.3 10 17.66 37.3 490 135.63 41.1 
20 32.57 33.7 490 135.63 41.1 920 41.69 57.3 
490 135.63 31.1 600 37.94 38.1 1,630 41.71 36.3 
920 41.69 44.1 920 41.69 57.3 1,970 90.02 41.4 
1,630 41.71 36 1,630 41.71 36.3 2,280 20.13 46.8 
2,280 20.13 35.9 2,280 20.13 46.8 2,330 58.5 44.6 
2,330 58.5 35.6 2,330 58.5 44.6 – – – 
Total 347.89 – – 353.26  – 387.68 – 
LSUs codeArea (km2)Sediment yield (tons/ha)LSUsArea (km2)Sediment yield (tons/ha)LSUsArea (km2)Sediment yield (tons/ha)
10 17.66 37.3 10 17.66 37.3 490 135.63 41.1 
20 32.57 33.7 490 135.63 41.1 920 41.69 57.3 
490 135.63 31.1 600 37.94 38.1 1,630 41.71 36.3 
920 41.69 44.1 920 41.69 57.3 1,970 90.02 41.4 
1,630 41.71 36 1,630 41.71 36.3 2,280 20.13 46.8 
2,280 20.13 35.9 2,280 20.13 46.8 2,330 58.5 44.6 
2,330 58.5 35.6 2,330 58.5 44.6 – – – 
Total 347.89 – – 353.26  – 387.68 – 
View Large
Figure 15
The sediment yields severity index at the LSU level.
View largeDownload slide

The sediment yields severity index at the LSU level.

Figure 15
The sediment yields severity index at the LSU level.
View largeDownload slide

The sediment yields severity index at the LSU level.

Close modal

Prioritization of land use that triggers runoff and sediment yields

This watershed has been dominated by large-scale socioeconomic mobilizations that have resulted in LULC changes over the past two decades. Each LULC has a unique effect on facilitating catchment runoff and sediment concentration from large-scale soil erosion. In all three scenarios, the most vulnerable land uses in this basin are forests, settlements, agriculture, and shrublands, which are projected to cause soil erosion. The greatest amount of runoff occurred on sub-basin areas covered by forests and settlements. In addition, catchment areas heavily covered by unprotected agricultural land are susceptible to soil erosion, which contributes a significant portion of the canal's surface runoff and sediment generation.

However, shrublands and agricultural lands caused significantly higher sediment yields and soil erosion than other land uses. Simulated flood wetness records for major waterways show that surface runoff is caused by the expansion of settlements, forests, shrubs, and agricultural lands, which benefit more from increased runoff than other land uses (Figure 16). The highest potential runoff is 31.07, 23.7, and 21.37% from forest, scrubland, and settlements, respectively, and decreases for other scenarios except shrubland. On the other hand, most of the soil erosion that led to sedimentation was attributed to agriculture and shrubland in all scenarios, accounting for 13.64 and 43.8%, respectively, with these percentages showing an increasing trend. In general, forests, shrublands, and settlements were highly vulnerable to runoff events, and agriculture and shrublands were highly vulnerable to soil erosion, resulting in sedimentation in all scenarios as shown in Table 19.
Table 19

Contribution of each LULC changes in causing runoff and sediment yields

LULC types % of runoff generated
% of sediment yields
200320132021200320132021
Agriculture 13.51 14.1 14.2 13.64 12.48 11.83 
Forest 31.07 24.9 24.1 10.37 6.05 5.64 
Shrubland 23.7 22.8 22.8 43.47 61.87 64.18 
Wetland 10.36 5.3 5.2 10.28 5.92 5.52 
Settlements 21.37 32.9 33.7 11.86 7.41 6.91 
LULC types % of runoff generated
% of sediment yields
200320132021200320132021
Agriculture 13.51 14.1 14.2 13.64 12.48 11.83 
Forest 31.07 24.9 24.1 10.37 6.05 5.64 
Shrubland 23.7 22.8 22.8 43.47 61.87 64.18 
Wetland 10.36 5.3 5.2 10.28 5.92 5.52 
Settlements 21.37 32.9 33.7 11.86 7.41 6.91 
View Large
Figure 16
The contribution of each land use to generate runoff and sediment yields.
View largeDownload slide

The contribution of each land use to generate runoff and sediment yields.

Figure 16
The contribution of each land use to generate runoff and sediment yields.
View largeDownload slide

The contribution of each land use to generate runoff and sediment yields.

Close modal

So far, we have characterized the distinct impact of each land use on runoff and sediment yield, presenting our findings through regional variability maps between 2003 and 2021. Notably, significant changes in land use within the catchment have led to reductions in streamflow, evaporation, seepage loss, and overall ecosystem functioning. The escalation of runoff and sediment output due to mass soil erosion can be attributed to the influence of each specific land use and cover type. This study uniquely investigated the impacts of dynamic land use on hydrological extremes at the HRU level than traditionally well-known at the sub-basin.

The watershed has been partitioned into 184 LSUs, demonstrating that 15, 16, and 18 LSUs in Scenarios S1, S2, and S3, respectively, have seen enhanced surface runoff depths (averaging 721.67, 748.31, and 747.3 mm). The obvious variance in sediment yields has efficiently defined locations prone to soil erosion after land use alterations. Six LSUs in the first scenario and seven in the second and third scenarios have suffered considerably owing to sediment loading, according to severity categorization. Deforestation, urban, agriculture, and shrubland have been identified as the most sensitive land uses, constantly pushing runoff and soil erosion across all scenarios. The sub-catchment comprising forests and settlements has provided the largest magnitude of runoff. When grouped by vulnerability, 96 LSUs demonstrate minimal susceptibility to soil erosion, whereas seven LSUs are very sensitive to sediment yield owing to erosion in the first scenario.

Agriculture has continuously shown the largest sensitivity to sediment yield, spanning 348.89 km2 (5.14%) of the entire sensitive area. In the second and third scenarios, farmland and shrubland largely comprise the badly degraded regions, constituting 353.26 and 387.386 km2, respectively. The use of an HRU-level technique has proved its superiority in forecasting hydrological processes compared to sub-basin analysis, guaranteeing greater scientific dependability and acceptability. In conclusion, dynamic spatiotemporal land use changes have catalyzed massive soil erosion and surface runoff, presenting severe challenges to environmental sustainability. The key rests in developing well-informed land management methods to reduce these consequences and pave the path for a more resilient environment. Finally, the study elaborates on the consequences of the hydrological extremes and informs water resource management and environmental protection sectors to target early sustainable solutions to the root cause.

The authors express gratitude to editors and reviewers for taking time to improve the quality of the paper. Furthermore, special thanks are extended to the Ethiopian Ministry of Water and Energy, as well as the National Meteorological Agency, for providing all the necessary data to facilitate the completion of this study.

This research did not receive any specific funding.

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

The authors declare there is no conflict.

Aredehey
G.
,
Mezgebu
A.
 &
Girma
A.
2020
The effects of land use land cover change on hydrological flow in Giba catchment, Tigray, Ethiopia
.
Cogent Environmental Science
6
(
1
).
doi:10.1080/23311843.2020.1785780
.
Google Scholar
 
Ayana
A. B.
,
Edossa
D. C.
 &
Kositsakulchai
E.
2012
Simulation of sediment yield using SWAT model in Fincha watershed, Ethiopia
.
Kasetsart Journal – Natural Science
46
(
2
),
283
–
297
.
Google Scholar
 
Berihun
M. L.
,
Tsunekawa
A.
,
Haregeweyn
N.
,
Tsubo
M.
,
Fenta
A. A.
,
Ebabu
K.
,
Sultan
D.
 &
Yihun Dile
T.
2020
Evaluating runoff and sediment responses to soil and water conservation practices by employing alternative modeling approaches
.
Science of the Total Environment
747
,
141118
.
doi:10.1016/j.scitotenv.2020.141118
.
Google Scholar
Crossref
Search ADS
PubMed
 
Berihun
M. L.
,
Tsunekawa
A.
,
Haregeweyn
N.
,
Tsubo
M.
,
Fenta
A. A.
,
Ebabu
K.
,
Sultan
D.
 &
Yihun Dile
T.
2022
Reduced runoff and sediment loss under alternative land capability-based land use and management options in a sub-humid watershed of Ethiopia
.
Journal of Hydrology: Regional Studies
40
,
100998
.
doi:10.1016/j.ejrh.2022.100998
.
Google Scholar
 
Bieger
K.
,
Arnold
J. G.
,
Rathjens
H.
,
White
M. J.
,
Bosch
D. D.
 &
Allen
P. M.
2019
Representing the connectivity of upland areas to floodplains and streams in SWAT +
.
Journal of the American Water Resources Association
55
(
3
),
578
–
590
.
doi:10.1111/1752-1688.12728
.
Google Scholar
Crossref
Search ADS
 
Burgan
H. I.
2022
The short-term and seasonal trend detection of sediment discharges in Turkish rivers
.
Rocznik Ochrona Środowiska
24
,
214
–
230
.
Google Scholar
Crossref
Search ADS
 
Elif Sertel
M. Z. I.
,
Cuceloglu
G.
 &
Erturk
A.
2019
Impacts of land cover/use changes on hydrological processes in a rapidly urbanizing mid-latitude water
.
Water (Switzerland)
11
,
1075
.
Google Scholar
 
Gashaw
T.
,
Tulu
T.
,
Argaw
M.
 &
Worqlul
A. W.
2018
Modeling the hydrological impacts of land use/land cover changes in the Andassa watershed, Blue Nile Basin, Ethiopia
.
Science of the Total Environment
619–620
,
1394
–
1408
.
doi:10.1016/j.scitotenv.2017.11.191
.
Google Scholar
PubMed
 
Gashaw
T.
,
Dile
Y. T.
,
Worqlul
A. W.
,
Bantider
A.
,
Zeleke
G.
,
Bewket
W.
 &
Alamirew
T.
2021
Evaluating the effectiveness of best management practices on soil erosion reduction using the SWAT model: For the case of Gumara Watershed, Abbay (Upper Blue Nile) Basin
.
Environmental Management
68
(
2
),
240
–
261
.
doi:10.1007/s00267-021-01492-9
.
Google Scholar
Crossref
Search ADS
PubMed
 
Gavit
B. K.
,
Purohit
R. C.
,
Bhange
H. N.
 &
Ingle
P. M.
2020
Hydrological modelling using SWAT
.
Research Journal of Recent Sciences
6
(
11
),
10
–
15
.
Google Scholar
 
Getachew
H. E.
 &
Melesse
A. M.
2012
The impact of land use change on the hydrology of the Angereb Watershed, Ethiopia
.
International Journal of Water Sciences
1
(
4
),
2012
.
doi:10.5772/56266
.
Google Scholar
 
Gurara
M. A.
,
Jilo
N. B.
 &
Tolche
A. D.
2021
Modelling climate change impact on the streamflow in the Upper Wabe Bridge watershed in Wabe Shebele River Basin, Ethiopia
.
International Journal of River Basin Management
,
1
–
13
.
doi:10.1080/15715124.2021.1935978
.
Google Scholar
 
Kakarndee
I.
 &
Kositsakulchai
E.
2020
Comparison between SWAT and SWAT+ for simulating streamflow in a paddy-field-dominated basin, northeast Thailand
.
E3S Web of Conferences
187
,
06002
.
Crossref
Search ADS
 
Kefay
T.
,
Abdisa
T.
 &
Tumsa
B. C.
2022
Prioritization of susceptible watershed to sediment yield and evaluation of best management practice: A case study of Awata River, Southern Ethiopia
.
Applied and Environmental Soil Science
2022
.
doi:10.1155/2022/1460945
.
Google Scholar
 
Kidane
M.
,
Alemu Bezie
M.
,
Kesete
N.
 &
Tolessa
T.
2019
The impact of land use and land cover (LULC) dynamics on soil erosion and sediment yield in Ethiopia
.
Heliyon
5
(
12
),
e02981
.
doi:10.1016/j.heliyon.2019.e02981
.
Google Scholar
Crossref
Search ADS
PubMed
 
Kiprotich
P.
,
Wei
X.
,
Zhang
Z.
,
Ngigi
T.
,
Qiu
F.
 &
Wang
L.
2021
Assessing the impact of land use and climate change on surface runoff response using gridded observations and SWAT +
.
Hydrology
8
(
1
),
1
–
48
.
Google Scholar
Crossref
Search ADS
 
Leta
M. K.
2020
Impacts of land use land cover change on sediment yield and stream flow: A case of Finchaa Hydropower Reservoir, Ethiopia
.
International Journal of Science and Technology (IJST)
6
(
4
),
1
–
20
.
Google Scholar
 
Leta
M. K.
,
Demissie
T. A.
 &
Tränckner
J.
2021
Modeling and prediction of land use land cover change dynamics based on land change modeler (LCM) in Nashe watershed, upper Blue Nile basin, Ethiopia
.
Sustainability (Switzerland)
13
(
7
),
3470
.
doi:10.3390/su13073740
.
Google Scholar
 
Li
R.
,
Dong
M.
 &
Cui
J.
2007
Quantification of the impact of land-use changes on ecosystem services: A case study in Pingbian County, China
.
Environmental Monitoring and Assessment
128
,
503
–
510
.
doi:10.1007/s10661-006-9344-0
.
Google Scholar
Crossref
Search ADS
PubMed
 
Moisa
M. B.
,
Negash
D. A.
,
Merga
B. B.
 &
Gemeda
D. O.
2021
Impact of land-use and land-cover change on soil erosion using the RUSLE model and the geographic information system: A case of Temeji watershed, Western Ethiopia
.
Journal of Water and Climate Change
12
(
7
),
3404
–
3420
.
doi:10.2166/wcc.2021.131
.
Google Scholar
Crossref
Search ADS
 
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
.
American Society of Agricultural and Biological Engineers
50
(
3
),
885
–
900
.
Google Scholar
 
Negese
A.
2021
Impacts of land use and land cover change on soil erosion and hydrological responses in Ethiopia
.
Applied and Environmental Soil Science
2021
,
15
–
17
.
doi:10.1155/2021/6669438
.
Google Scholar
Crossref
Search ADS
 
Saddique
N.
,
Mahmood
T.
 &
Bernhofer
C.
2020
Quantifying the impacts of land use/land cover change on the water balance in the afforested River Basin, Pakistan
.
Environmental Earth Sciences
79
(
19
),
1
–
13
.
doi:10.1007/s12665-020-09206-w
.
Google Scholar
Crossref
Search ADS
 
Sitotaw
G.
,
Gebrie
G. S.
,
Gebremariam
A. G.
 &
Engida
A. N.
2021
Future climate change and impacts on water resources in the Upper Blue Nile basin
.
Hydrological Processes
36
(
5
),
e14589
.
doi:10.2166/wcc.2021.235
.
Google Scholar
 
Sulamo
M. A.
,
Kassa
A. K.
 &
Roba
N. T.
2021
Evaluation of the impacts of land use/cover changes on water balance of Bilate watershed, Rift valley basin, Ethiopia
.
Water Practice and Technology
16
(
4
),
1108
–
1127
.
doi:10.2166/wpt.2021.063
.
Google Scholar
Crossref
Search ADS
 
Tan
X.
,
Liu
S.
,
Tian
Y.
,
Zhou
Z.
,
Wang
Y.
,
Jiang
J.
 &
Shi
H.
2022
Impacts of climate change and land use/cover change on regional hydrological processes: Case of the Guangdong-Hong Kong-Macao Greater Bay Area
.
Frontiers in Environmental Science
9
,
1
–
16
.
doi:10.3389/fenvs.2021.783324
.
Google Scholar
Crossref
Search ADS
 
Teshome
D. S.
,
Moisa
M. B.
,
Gemeda
D. O.
 &
You
S.
2022
Effect of land use-land cover change on soil erosion and sediment yield in Muger Sub-Basin, Upper Blue Nile Basin, Ethiopia
.
Land
11
(
12
),
2173
.
Google Scholar
Crossref
Search ADS
 
Umukiza
E.
,
Raude
J. M.
,
Wandera
S. M.
,
Petroselli
A.
 &
Gathenya
J. M.
2021
Impacts of land use and land cover changes on peak discharge and flow volume in Kakia and Esamburmbur sub-catchments of Narok Town, Kenya
.
Hydrology
8
(
2
),
82
.
Google Scholar
Crossref
Search ADS
 
Wagner
P. D.
,
Bieger
K.
,
Arnold
J. G.
 &
Fohrer
N.
2022
Representation of hydrological processes in a rural lowland catchment in Northern Germany using SWAT and SWAT +
.
Hydrological Processes
36
(
5
).
doi:10.1002/hyp.14589
.
Google Scholar
 
Worku
T.
,
Khare
D.
 &
Tripathi
S. K.
2017
Modeling runoff–sediment response to land use/land cover changes using integrated GIS and SWAT model in the Beressa watershed
.
Environmental Earth Sciences
76
(
16
),
1
–
14
.
doi:10.1007/s12665-017-6883-3
.
Google Scholar
Crossref
Search ADS
 
Wu
J.
,
Yen
H.
,
Arnold
J. G.
,
Yang
E.
,
Cai
X.
,
White
M. J.
,
Santhi
C.
,
Miao
C.
 &
Srinivasan
R.
2020
Development of reservoir operation functions in SWAT+ for national environmental assessments
.
Journal of Hydrology
583
,
124556
.
doi:10.1016/j.jhydrol.2020.124556
.
Google Scholar
Crossref
Search ADS
 
Yen
H.
,
Park
S.
,
Arnold
J. G.
,
Srinivasan
R.
,
Chawanda
C. J.
,
Wang
R.
,
Feng
Q.
,
Wu
J.
,
Miao
C.
,
Bieger
K.
,
Daggupati
P.
,
van Griensven
A.
,
Kalin
L.
,
Lee
S.
,
Sheshukov
A. Y.
,
White
M. J.
,
Yuan
Y.
,
Yeo
I.
,
Zhang
M.
 &
Zhang
X.
2019
IPEAT + : A built-in optimization and automatic calibration tool of SWAT +
.
Water (Switzerland)
11
(
8
),
1
–
17
.
doi:10.3390/w11081681
.
Google Scholar
 
Yigez
B.
,
Xiong
D.
,
Zhang
B.
,
Yuan
Y.
,
Baig
M. A.
,
Dahal
N. M.
,
Guadie
A.
,
Zhao
W.
 &
Wu
Y.
2021
Spatial distribution of soil erosion and sediment yield in the Koshi River Basin, Nepal: A case study of Triyuga watershed
.
Journal of Soils and Sediments
21
(
12
),
3888
–
3905
.
doi:10.1007/s11368-021-03023-9
.
Google Scholar
Crossref
Search ADS
 
Zeberie
W.
2019
Modelling of rainfall-runoff relationship in Big-Akaki Watershed, Upper Awash Basin
.
Engineering, Environmental Science
27
,
108
–
120
.
Google Scholar
 
Zhang
M.
,
Leon
C.
 &
de, Migliaccio
K.
2018
Evaluation and comparison of interpolated gauge rainfall data and gridded rainfall data in Florida, USA
.
Hydrological Sciences Journal
63
(
4
),
561
–
582
.
doi:10.1080/02626667.2018.1444767
.
Google Scholar
Crossref
Search ADS
 
Zhang
L.
,
Meng
X.
,
Wang
H.
 &
Yang
M.
2019
Simulated runoff and sediment yield responses to land-use change using the SWAT model in Northeast China
.
Water
11
(
5
),
915
.
Google Scholar
Crossref
Search ADS
 
© 2023 The Authors
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).