Introduction

Many areas worldwide are exposed to the problem of flooding during the rainy seasons, specifically in areas near riverbeds, as a result of high water levels within the water channel and thus the flow of water towards areas adjacent to the river. One of the problems that causes floods is the sedimentary load within the riverbed, which is mainly due to the heavy rain falling on the river basin, which causes the erosion of the soil and fragile rocks. The eroded soil and rock is transported in the form of a sedimentary load by the running water and deposited in the riverbed. These sediments reduce the water capacity of the stream, causing excess water to flow towards areas adjacent to the river in the form of floods. To study this problem, find real solutions to it, reduce its recurrence, and overcome it, it is necessary to use modern and advanced hydrological models relating to the management of water basins.

The development of hydrological models and recent developments in the use of geographic information system (GIS) software tools have led to new ways to study water resources and their environmental assessment. Hydrological models and GIS software tools have been widely applied in many studies on water resource assessment in recent decades (Krysanova and White 2015; Saidi et al. 2017). Hydrological models simplify real-world processes (Schürz et al. 2019). These models are used in hydrological simulation, which is a simplified software representation of the hydrological process that occurs within a basin boundary and generates surface runoff at the required locations of the river drainage basin. The SWAT hydrological model has been widely used by the scientific community to simulate changing hydroclimatic conditions and the geophysical environment (Marahatta et al. 2021). Water assessment modeling, simulation, and associated processes have been widely used to address a wide range of water resource challenges globally (Akoko et al. 2021).

The conservation of the water component is one of the main indicators of the hydrological regulation capacity of terrestrial ecosystems (Zhai et al. 2023). Physical and hydrological properties of soil influence the processes of modeling, planning, and soundly managing water resources as well as soil conservation (Ribeiro Filho et al. 2023). Studies of soil erosion and sediment transport are very important and necessary because soil erosion is a major component of source pollution and is one of the core procedures for calculating sediment transport and its products (Gassman et al. 2007). The use of hydro-sedimentary models can be necessary to reveal the effectiveness of soil and water conservation practices in watersheds (Pontes et al. 2021). Soil erosion poses a serious challenge to land productivity all over the world, and this alarming situation requires urgent interventions by specialists in order to conserve water and soil resources (Liu and Jiang 2019). A set of advanced techniques have been developed to estimate the accuracy of calculations of surface runoff and the sediment yield from it in river drainage basins. The hydrological process usually depends on the local conditions, which are represented by natural, biotic, and climatic factors. Several studies have indicated that climate change causes an increase in annual rainfall intensity, which leads to increased sediment yield and soil erosion rates (Mullan et al. 2019). Climate change is also increasing the intensity and frequency of heavy rainfall (Chiang et al. 2021; Al-Hussein et al. 2022a). So, rainfall plays the main and most influential role in soil erosion due to water erosion in the watershed (Al-Hussein 2022). Simulation of the sediment yield by the SWAT model provides useful information for targeted management and helps stakeholders involved in soil and water conservation activities to choose the most appropriate watershed practice (Khanchoula et al. 2020). Several studies have shown the power of the model for predicting surface runoff and sediment yield in different watersheds.

The assessment of sediment yield and surface runoff using the SWAT model has been the subject of various scientific studies, especially for North African basins. The modeling of surface runoff and sediment yield was conducted using the SWAT model for the Wadi Hatab basin in central Tunisia. It was clear from the interpretation of the results for the surface runoff and sediment yield data for daily and monthly time periods that the basin is characterized by great climatic variation, a steep topography, and soil fragility, which causes flash floods and high water erosion rates (Salah and Abida 2016). Furthermore, a study undertaken in the Harraza basin, northwestern Algeria, showed that the SWAT model is effective and provides a better understanding of the behavior of the hydrological system, facilitating optimal management of water and soil resources in the basin (Hallouz et al. 2017). Surface runoff estimation was performed using the SWAT model for one of the dams built at the mouth of the Beni Haroun basin, northeastern Algeria. The water productivity was calculated and vulnerable spots were identified using two scenarios with two types of data relating to soil types and land uses. The first type represents a land cover map from GlobCover (created by the European Space Agency) with the Harmonized World Soil Database (HWSD; created by the FAO and IIASA to develop regional databases and global soil information) and the second type corresponds to maps extracted from satellite images (Landsat 8). The simulation results indicate that the second scenario gives a better representation of the watershed in daily time steps. The first scenario provides acceptable results that allow for a general prediction of the status of the basin in the future (Kateb et al. 2019). An assessment of the sediment yield and erosion susceptibility of the Kebir basin, northeastern Algeria, using a SWAT model showed that 50% of the sediment yield generated in the basin was produced during the 5 months of the winter and autumn seasons. The sub-basins that are the most vulnerable to water erosion and thus could have a significant impact on sediment yield were identified in reservoirs at the mouths of wadis (Khanchoula et al. 2020). On the other hand, an evaluation of the soil erosion potential, sediment productivity, and surface runoff within the N'fis basin, High Atlas, Morocco, was achieved by the use of the SWAT model. The process of evaluating and simulating the productivity of surface runoff and the sediment transported was conducted using graphical and statistical methods. The phenomenon of deforestation in the basin led to easy water flow and thus soil erosion and transport and siltation accumulation in the reservoirs. This study indicated that better characterization of the sediment yield is important for reducing the effect of sedimentation in the reservoir and extending the lifespan of the dams (Markhi et al. 2019). Also, hydrological modeling of the Ouergha basin, northwest Morocco, was conducted using a physical and semi-distributed SWAT model to evaluate hydrological simulations and specifically the flow in the basin. Results indicated that the basin's water resources can be better protected in the future through research that takes into consideration changes in climate and cultural practices (Erraioui et al. 2023). In addition to the above, some hydrological studies measure uncertainties in surface runoff and sediment yield and use a simplified and standardized sampling approach that focuses on strategic basin planning without taking into account climate and land-use changes, which may not help decision-making in the real world due to the spread of uncertainties resulting from different measurements and inference methods.

The current study addresses the above-mentioned research gaps by relying on a modeling framework based on real data representations of the river basin, as such modeling is crucial for sustainable development in the basin. This research will add to the growing body of literature on the use of hydrological models in basins. It will also provide basin managers with an objective method through which they can estimate the rates of hydrological processes that affect surface waters and help develop strategic plans for the integrated management of the basins. The results of the spatially distributed hydrological model and their representation in the form of maps are considered necessary to support communication efforts between the competent authorities regarding rapid and early intervention aimed at restoring degraded areas in the basins. The research provides a comprehensive representation of surface runoff rates and sediment yield and can be used by water resource authorities in decision-making regarding the design of appropriate adaptation practices and the mitigation of the effects of climate change on land degradation in the basins. In addition, the methodology proposed here can be adopted in other regional studies to estimate surface runoff and sediment yield under climate change.

The main objectives of this paper are (i) to apply the SWAT model to estimate surface runoff and sediment yield in the sub-basins (analysis and simulation are performed for the period 2003–2022 at the spatial level of the hydrological response unit, HRU), (ii) to test the SWAT CUP model in terms of evaluating accuracy, uncertainty, and parameter sensitivity by comparing observed and simulated daily data for the average surface runoff and sediment yield, (iii) to explain the relationship between precipitation and sediment yield within the basin on the one hand and surface runoff on the other hand, and (iv) to present the spatial variation of surface runoff and sediment yield at the sub-basin level in the form of maps.

Materials and methods

Study area

The study area corresponds to the Khazir River basin, which is located in the far north of Iraq (Fig. 1). This river is formed from the confluence of two rivers, the Roser and Kafia, and the waters of the Komel River flow into the course of the Khazar River before the latter meets the course of the Great Zab River. The length of the Khazir River is 123.58 km, and a large network of seasonal valleys flows into its course. The area of ​​the drainage basin is 3290.08 km2. The basin considered here can be divided according to topographic characteristics into a group of geomorphological units: high mountains, hills, plains, and valleys (Al-Hussein et al. 2023). These units are characterized by significant differences in terrain, ranging from mountainous heights and steep slopes in the north to flat areas with little slope in the center and south. Elevations range from 2201 m in the mountainous areas to 221 m above sea level near the mouth of the basin, while the slope of the basin varies between 0° and 81.8°. Geographically, the basin is confined between longitudes 43° 03′ 0′′ and 43° 57′ 0′′ E and the latitudes 36° 07′ 0′′ and 37° 03′ 0′′ N. Geologically, the study area is characterized by many geological formations dating from different times, starting from the Lower Jurassic period and ending with the Quaternary period. Ancient formations appear in the northern and northeastern parts of the study area, while modern formations appear in the center and south. The rock characteristics of these formations are dolomite, limestone, shales, gypsum, halite, anhydrite, marl, siltstone, mudstone, sandstone, and conglomerates (Sissakian 2013).

Fig. 1
figure 1

Map of the Khazir River basin showing the locations of weather stations and the stream network

Full size image

Data used

Before starting the modeling, analysis, and simulation processes, the geographical coordinates of some of the data used in the study—the digital elevation model (DEM), land uses, and soil types—must be unified to avoid errors that may appear during the implementation of the simulation steps. Projections in the SWAT model and units of latitude and longitude for geographic coordinates are converted to units of length in the metric system through ArcGIS 10.8 software. Table 1 shows all the data used in modeling and simulation.

Table 1 Data used in the Soil and Water Assessment Tool model
Full size table

Digital elevation model (DEM)

The digital elevation model (DEM) used in the modeling process has a spatial resolution of 30 × 30 m and was obtained from the EarthExplorer website after defining the study area. A set of important hydrological features were extracted from the DEM model by means of the toolbox in ArcGIS. These mainly included the basin topography, flow direction, flow length, flow accumulation, and flow network of surface water. Using these, the boundaries of the main basin were also defined, as were the 27 sub-basins. In addition, the slope of the basin was extracted from the DEM model. The slope varied between severe in the mountainous regions and mild in the plains (Fig. 2a).

Fig. 2
figure 2

Maps showing a slope classes, b land use, and c soil types

Full size image

Land use

The layer of land-use categories is critical when using the SWAT model to compute the water flow rate and sediment yield, which depend on the kind of land use present within the watershed. The land-use coding process is relied upon when conducting hydrological simulations. In this study, a Sentinel-2 satellite image for the year 2022 with a spatial resolution of 10 m was taken. We relied on the ArcGIS software to process, analyze and classify satellite images, where the directed classification (supported by a field study) was used to determine the types of land uses present. Six categories were identified within the drainage basin (Table 2): water areas, shrublands, croplands, built areas, bare ground, and pasture (Fig. 2b).

Table 2 The percentage of each land-use class and soil type that was simulated in the SWAT model
Full size table

Soil type

Soil plays an influential role in the occurrence of surface runoff and the resulting waterborne sediments. The soil quality of the basin significantly affects the increase in the volume of surface runoff (Al-Hussein et al. 2022b). The soil of the basin varies in terms of minerals, texture, and hydrology. The study area is covered in soils from two hydrological groups (Table 2). The first soil group is C, which covers most of the drainage basin and is characterized by a mixture of sand, clay, and silt, while the second soil group is D, which covers a very small percentage of the basin and contains sand and clay. These two soil groups greatly promote surface runoff. On the other hand, there are four types of soils within the watershed, namely lithosols (54.094%), calcaric fluvisols (0.078%), chromic vertisols (37.04%), and calcic xerosols (8.098%) (Fig. 2c).

Hydro-meteorological data

The SWAT model handles daily climate data. The meteorological data used in the model are the daily rainfall (mm) and the daily maximum and minimum temperatures (°C). Meteorological data were collected from six climatic stations (Table 3) in and around the study area basin for the period 2003–2022, and these data were used as inputs to the model. The study area is located within the climate of the Mediterranean region, which is characterized by great variations in climatic conditions during the seasons of the year. The average annual rainfall is 252.55 mm, while the maximum and minimum temperatures are 46.64 and 16.54 °C, respectively, and the average annual temperature reaches 21.49 °C.

Table 3 Spatial locations of the meteorological stations used
Full size table

Methodology

SWAT model

In the last few decades, many methods to estimate the amount of erosion and the sediment yield have emerged. This has led to the need to build applied mathematical models and equations to evaluate the Earth’s surface and reduce the errors that arise from the model itself. Among the most common methods used are (i) the erosion potential method (EPM), known as the Gavrilović model, which is considered an experimental and semi-quantitative method and can estimate the sediment yield, determine the severity of water erosion, and indicate areas that are likely to be threatened by erosion; (ii) the intensity of erosion and outflow method (IntErO), which is used to predict the maximum flow, soil erosion intensity, and sediment yield; (iii) the universal soil loss equation (USLE), which has the ability to predict soil erosion; and (iv) the revised universal soil loss equation (RUSLE), which is also used to estimate soil erosion in watersheds. All of the methods mentioned rely on mathematical equations applied manually within the ArcGIS software to obtain results, unlike the SWAT model, which contains mathematical equations and modeling tools that are included with the ArcGIS software and is connected to the SWAT CUP model for calibration and the validation of results. The results extracted from the SWAT model are more accurate than the results of the aforementioned models, most of which are speculative and concerned with estimating soil erosion only (i.e., without calculating surface runoff). The SWAT model is considered one of the most modern and internationally approved models for calculating sediment yield and surface runoff for large, medium, and small basins in arid and semi-arid areas.

The current study relied on the Soil and Water Assessment Tool (SWAT) that was developed in 1994. This is a continuous, semi-distributed, process-based temporal simulation model for simulating river drainage basins and operates in daily or near-daily time steps (Srinivasan et al. 1998; Van Liew et al. 2005). This tool can simulate hydrological, sedimentary, and nutrient cycle processes in a watershed at various spatial and temporal scales (Chiang et al. 2021). Developed by the United States Department of Agriculture (USDA) in collaboration with Texas A&M University, this model can simulate the hydrologies of small watersheds and river basins (Neitsch et al. 2011; Abbaspour et al. 2015). The SWAT model belongs to the category of ecohydrological river basin models based on a set of natural processes, which can be defined as dynamic, continuous-time models that are based on mathematical descriptions of physical, chemical, biotic, and hydrological processes and combine both physical and semi-empirical elements (Krysanova and White 2015). The SWAT model has been used worldwide to simulate the impacts of climate change and land-use change on stream flow, sediment, and nutrient output and is a best practice for watershed management in many countries (Yevenes and Mannaerts 2011).

The SWAT model has several versions, such as Arc SWAT v. 2012, which is used to enter the different types of model data needed to perform the simulation (Kateb et al. 2019). The SWAT model simulation of the hydrological process in the watershed is divided into the surface water cycle, which controls the water and sediment inputs of the main river channel in each sub-basin, and the confluence water cycle, which determines the movement of water and sediment transport from the river network to the basin outlet (Liu and Jiang 2019).

Hydrological modeling

The surface runoff and sediment yield modeling of the Khazir River basin was achieved using a SWAT model. This model is an extension that is added to the interface of the ArcGIS 10.8 software, through which the data necessary to perform the simulation process can easily be entered. The model was designed using correctly extracted and predicted Khazir River basin data, such as a digital elevation model, land uses, soil types, and weather data (Fig. 3). The first spatial estimation unit is the sub-basin. The main basin is divided into a number of sub-basins based on digital elevation model (DEM) data and an outlet is created for each sub-basin. The minimum drainage area is determined in order to form the watercourse, where we define the hydrographic network that we want to model. The process of dividing the main basin into sub-basins depends on the hydrographic network of the basin because two outlets are placed at each junction, with each outlet corresponding to the sub-catchment area (Fig. 4). Each sub-basin is subdivided into several hydrological response units (HRUs) to increase the accuracy of flow prediction and provide a better physical description of the water balance (Roti et al. 2018). This is done based on the entered data, such as slope, soil-type, and land-use data. The cyclic movement of water in nature is determined by simulating the hydrological cycle with the SWAT model and relying on the following water balance equation (Neitsch et al. 2011):

$${\text{SW}}_{t} = {\text{SW}}_{0} + \sum\limits_{i = 1}^{t} {(R_{{{\text{day}}}} } - Q_{{{\text{surf}}}} -E_{a}- W_{{{\text{sweep}}}} - Q_{gw} ),$$
(1)
Fig. 3
figure 3

Methodological chart for the application of the SWAT hydrological model

Full size image
Fig. 4
figure 4

Partition mapping of sub-basins in the Khazir River basin

Full size image

where t = time (days), SWt = final soil moisture content (mm); SW0 = initial water content of the soil (mm); Rday = amount of precipitation on day (mm); Qsurf = quantity of runoff water on day (mm); Ea = amount of evapotranspiration on day (mm); Wsweep = quantity of water entering the unsaturated zone of the soil profile on day (mm); Qgw = the amount of return flow on day (mm).

The surface runoff is allocated separately for each HRU and channeled to obtain the total catchment surface runoff, which increases the accuracy and gives a better physical description of the water balance.

Surface runoff is the main component of the hydrological cycle, and its prediction has been and still is one of the major issues in hydrology (Lu et al. 2023). Surface runoff is the flow that occurs on a surface according to the degree of slope. The SWAT model simulates runoff values and maximum runoff rates per HRU based on the daily or near-daily rainfall amounts (Neitsch et al. 2011). The volume of surface runoff from daily precipitation is predicted using the curve number method (Soil Conservation Service 1972). The SWAT model calculates the peak runoff rate per HRU in the catchment using a modified Boolean formula. The sediment yield per HRU due to precipitation and surface runoff is estimated and then combined to determine the total sub-basin sediment (Arnold et al. 1998). The eroded soil that is not deposited in the watershed and reaches the downstream region is referred to as the sediment yield (Sil and Pathan 2022). Hydrological, geomorphological, and climatic parameters play influential roles when analyzing the density and diversity of the sediment yield (Berteni et al. 2021). The watershed sediment yield is extracted using the modified universal soil loss equation (MUSLE) (Williams 1975):

$${\text{Sed}}_{i} = 11.8(Q_{{{\text{surf}}}} .q_{{{\text{peak}}}} .A_{{{\text{hru}}}} )^{0.56} \times K_{{{\text{USLE}}}} \times C_{{{\text{USLE}}}} \times P_{{{\text{USLE}}}} \times {\text{LS}}_{{{\text{USLE}}}} \times {\text{CFGR}},$$
(2)

where 11.8 = is the unit conversion factor, Sed = sediment yield in tons per day (ton/day), Qsurf = surface runoff volume (mm/ha), qpeak = peak runoff rate in m3/s, Ahru = the area of the HRU (ha), KUSLE = soil erodibility factor, CUSLE = cover and management factor, PUSLE = support practice factor, LSUSLE = topographic factor, and CFRG = course fragment factor.

The MUSLE equation provides estimates of sheet and rill erosion and considers a range of factors that influence soil loss and sedimentation for watersheds, such as soil erosion, crop management, erosion control practices, slope length, and slope (Mapes and Pricope 2020). In addition to the above, the study of reliable estimates of sediment yield is very necessary when conducting investigations into river morphology and water resource management (Keshtegar et al. 2023).

Calibration and validation of the SWAT model

The calibration is a very important step in determining the accuracy of the information obtained from the SWAT model simulation. Calibration of the results of any study is accomplished by comparing the output of the SWAT model with the data observed under the same conditions (Arnold et al. 2012). Calibration of SWAT sample data is performed by the SWAT CUP model, standalone calibration software developed for the SWAT form (Abbaspour et al. 2007a). The universal sensitivity unconfirmed fit (SUFI-2) method was used within SWAT calibration and uncertainty (SWAT CUP) to determine the most sensitive parameters. When using the (SUFI-2) method, the uncertainty in the input parameter is expressed in the form of uniform distributions, while the uncertainty in the output is represented by a 95% normal probability distribution (Abbaspour et al. 2007b). The model was calibrated and validated for the sediment yield and surface runoff data of the Khazir River basin by calibrating the observed daily data for the period from 2003 to 2008 and validating for the period from 2009 to 2012. These data were obtained from the output of the SWAT model in ArcGIS 10.8 software.

Sensitivity analysis of the SWAT model

A sensitivity analysis is performed for the calibration phase to reveal the most influential parameters in the chosen model (Martínez-Salvador et al. 2021; Shoemaker and Benaman 2003). In this study, calibration, testing, sensitivity, and uncertainty analysis were conducted using the SUFI-2 algorithm from the SWAT-CUP model. Calibration is the process of adjusting a model by changing its sensitive parameters to obtain the best relationship between a given time series of observed data and the simulated data obtained during the modeling. The SWAT-CUP model contains a set of ready-made parameters that aim to perform hydrological calibration by comparing the estimated surface runoff and sediment yield data with those calculated by the model. Once the model is calibrated, validation is performed over the required period. The most sensitive calibration parameters are noted and can be modified by changing the non-sensitive parameter to a sensitive parameter during calibration and verification in order to obtain the best results. Statistical measures and graphical representations are used when evaluating model performance during calibration and validation. The calibration results were evaluated using five statistical methods, as recommended by Moriasi et al. (2007), through the application of the following equations:

$${\text{NSE}} = 1 - \left[ {\frac{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} - Y_{i}^{{{\text{sim}}}} )^{2} } }}{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} - Y_{{{\text{obs}}}}^{{{\text{mean}}}} )^{2} } }}} \right],$$
(3)
$${\text{PBIAS}}(\% ) = \left[ {\frac{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} - Y_{i}^{{{\text{sim}}}} )\cdot(100)} }}{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} )} }}} \right],$$
(4)
$$R^{2} = \frac{{\left( {\sum\nolimits_{i = 1}^{n} {\left( {Y_{i}^{{{\text{obs}}}} - Y_{{{\text{obs}}}}^{{{\text{mean}}}} } \right)\left( {Y_{i}^{{{\text{sim}}}} - Y_{{{\text{sim}}}}^{{{\text{mean}}}} } \right)} } \right)^{2} }}{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} - Y_{{{\text{obs}}}}^{{{\text{mean}}}} )^{2} \sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{sim}}}} - Y_{{{\text{sim}}}}^{{{\text{mean}}}} )^{2} } } }},$$
(5)
$${\text{RMSE}} = \sqrt {\frac{{\sum\nolimits_{i = 1}^{n} {(Y_{i}^{{{\text{obs}}}} - Y_{i}^{{{\text{sim}}}} )^{2} } }}{n}} ,$$
(6)
$${\text{RSR}} = \frac{{{\text{RMSE}}}}{{{\text{STDEV}}_{{{\text{obs}}}} }},$$
(7)

where NSE = the Nash–Sutcliffe efficiency; R2 = the coefficient of determination; PBIAS = the percentage bias; RSR = the standard deviation of the measured data; RMSE = root mean square error; \(Y_{i}^{{{\text{obs}}}}\) and \(Y_{i}^{{{\text{sim}}}}\) = the evaluated observation and simulation, respectively; \(Y_{{{\text{obs}}}}^{{{\text{mean}}}}\) \(Y_{{{\text{sim}}}}^{{{\text{mean}}}}\) = the mean of the evaluated observed and simulated data; STDEV obs = the standard deviation of the observed data; and \(n\) = the number of observations.

The conceptual methodology applied to three African basins (Tunisia, Algeria, and Morocco) is the same as that applied to the basin in the study area (Iraq). The SWAT hydrological model added to the ArcGIS 10.8 software interface was used. The SUFI-2 method was used to calibrate the model and verify the results of the sensitivity analysis within the SWAT CUP model. The differences were in the data on the climate elements, land uses, and soil type used in the SWAT model, in addition to the differences in the parameters entered into the SWAT CUP model. All of these differences led to contrasts between the results of the study of the African basins and the results for the basin in the study area.

Results

The study's findings led to the preparation of a SWAT model in the ArcGIS software. This model divided the Khaziar River basin into 27 sub-basins and 27 HRUs. The SWAT model estimated the precipitation rate to be about 371.320 mm/year, the surface runoff rate to be about 33.201 mm/year, and the sediment yield rate to be 0.070 tons/ha/year. The rates of sediment yield and surface runoff vary greatly amongst the Khazir River's sub-basins due to differences in soil type, land use, slope, and the quantity of rainfall falling on these sub-basins, with the volume of surface runoff significantly increasing as it moves towards the main basin outlet. The cause of this rise is the linking of the little valleys' waters with the waters from the large valleys as well as the gathering together of these valleys' waters along the course of the Khazar River and then the emptying of this water into the river's mouth. As the increase in the volume of surface runoff was accompanied by an increase in the volume of sediment yield, there is a direct relationship between them, and the value of the corresponding correlation coefficient reached 0.813.

The spatial distribution of sediment yield and surface runoff was not the same for all sub-basins. Rather, it differed from one basin to another due to differences in the distribution and intensity of precipitation. The erosion and surface runoff was affected by not only the amount and distribution of precipitation but also its intensity (Nearinga et al. 2005). Land use played a positive role in reducing sediment yield and surface runoff. It is noted from Fig. 5a that most of the sediment production occurred within sub-basins 1, 2, 3, 4, 7, 8, 9, 25, 26, and 27 because the type of land cover for these basins was pasture (Fig. 2b), which is characterized by a lack of vegetation, facilitating water erosion. These basins were characterized by the maximum precipitation (Fig. 5b) and thus an increase in the surface runoff volume (Fig. 5c).

Fig. 5
figure 5

Annual average estimates of a sediment yield (SDYLD), b surface runoff (SURQ), and c precipitation (PRECIP) for the period 2003–2022

Full size image

The increase in the volume of surface runoff and the sediment yield towards the mouth of the main basin led to many floods over time in the study area, resulting in material losses represented by personal property, residential buildings, and agricultural crops. These floods occurred for many reasons, but the main reasons were (i) the flatness of the land surface, (ii) the high discharge generated by surface runoff, and (iii) the low efficiency of the river course in carrying water due to it becoming filled with sediments, which leads to the flow of water towards the two banks of the river, thus exposing the areas close to the river to the risk of flooding.

Sediment yield simulations

Sensitivity analysis

Sensitivity analysis was applied in the SWAT CUP model to determine the most important parameters during calibration and to validate the results. It was found to rely on six default parameters for the analysis of basin sediment yield and showed high sensitivity in the analysis (Table 4). The soil erodibility factor equation (USLE_K), channel cover factor (CH_Cov2), channel erodibility factor (CH_Cov1), and linear parameters for calculating the channel sediment rooting (SPCON) all showed that the calibration of the carried sediments was particularly important, and the most sensitive parameters were those for the drainage basin sediment yield.

Table 4 Calibration parameters for the sediment yield
Full size table

Calibration and validation

The first step in the CUP SWAT calibration and validation process is to identify the most sensitive watershed parameters which significantly impact specific model outputs such as the runoff and sediment yield (Ma et al. 2000). Calibration was performed on daily data covering 6 years (2003–2008), whereas validation was undertaken on data from 2009 to 2012. Sediment yield histogram plots were developed to compare observed and simulated sediment yield values for the calibration and validation periods in daily time steps. During the calibration procedure, a relative mismatch appeared between the simulated and observed sediment yield parameter data. However, after adjusting the sensitive sediment parameters, there was good agreement between the model output data and the observed sediment output data (Figs. 6, 7). In addition, there was great agreement between the sediment yield data and the amount of precipitation falling on the watershed.

Fig. 6
figure 6

Daily sediment yield calibration for the Khazir River basin

Full size image
Fig. 7
figure 7

Daily sediment yield validation for the Khazir River basin

Full size image

Table 5 shows the performance indicators calculated for sediment yield. There was good agreement between the observed data and the simulated data for daily sediment transport. With NSE values between 0.71 and 0.75 and correlation coefficients between 0.75 and 0.80, the PBIAS values were between 27.6 and 37.4 for the two calibration and validation periods. This indicates satisfactory analysis results relative to the observed data.

Table 5 Statistical evaluation of sediment yield performance
Full size table

Surface runoff simulations

Sensitivity analysis

The initial step of calibration included a sensitivity study of 12 surface runoff parameters. The sensitivity of each parameter was evaluated according to Van Griensven et al. (2006). In this study, the global classification method was used, where ranks 1–2 were classified as important, ranks 3–7 were considered somewhat important, and ranks 8–12 were not important. This means that, according to the sensitivity analysis, parameters with global ranks between 1 and 7 are sensitive, while parameters with ranks between 8 and 12 are considered insensitive (Table 6). The soil evaporation compensation coefficient (ESCO), the surface runoff lag coefficient (SURLAG), the average slope length (SLSUBBSN), the plant uptake compensation coefficient (EPCO), and Manning's n value for overland flow (OV_N) showed high sensitivity when calibrating the surface runoff within a watershed.

Table 6 Calibration parameters for the surface runoff
Full size table

Calibration and validation

Daily surface runoff data obtained from the SWAT model from the period 2003–2008 were used for calibration and those from 2009–2012 were used for validation. The daily surface runoff that was simulated and observed during the calibration and validation periods is presented in Figs. 8 and 9. Simulated results were consistent with the observed data; the simulated runoff was consistent with low runoff and most peaks. The model slightly overestimated the simulated runoff results on some days during the calibration period and underestimated the simulated results on other days. This was the result of the heavy rains during the wet season days, which led to a record high period of surface runoff, in contrast to the dry season days, which did not witness any surface runoff. The trend for the time series of measured steps was a good estimator of the daily time steps. A reasonably good agreement emerged between the observed and simulated daily time-step surface runoff data during the validation period.

Fig. 8
figure 8

Daily surface runoff calibration for the Khazir River basin

Full size image
Fig. 9
figure 9

Daily surface runoff validation for the Khazir River basin

Full size image

Table 7 shows the performance indicators calculated for the daily surface runoff, where good agreement between the observed data and the simulated data is observed. With NSE values of between 0.77 and 0.82 and correlation coefficients between 0.85 and 0.88, the PBIAS values were between 9.7 and 10.5 for the calibration and validation periods. These values represent satisfactory analysis results for the observed data.

Table 7 Statistical evaluation of surface runoff performance
Full size table

Evaluation of the SWAT model

A statistical evaluation of the accuracy of the SWAT model was conducted by analyzing observed and simulated data during the calibration and validation periods for both sediment yield and surface runoff. Quantitative statistical tests included a set of metrics, such as the mean absolute error (MAE), standard error (SE), standard deviation (SD), coefficient of determination (R2), root mean square error (RMSE), standard deviation of the measured data (RSR), and Nash–Sutcliffe efficiency (NSE). The results of the statistical analysis of the SWAT model (Table 8) showed that the calibration and validation data for sediment yield were consistent with the data from the SWAT CUP model, as the average coefficient of determination was 0.76 and the mean absolute error was 2.85%. Also, the calibration and validation data for surface runoff were consistent with the SWAT CUP model data, as the average coefficient of determination was 0.84 and the mean absolute error was 0.62%. The mean absolute error values for the sediment yield and surface runoff results were very small. In general, the results of the statistical analysis show that there is a strong relationship between the calibration and validation data, with similarities between the results of the SWAT model and the results of the SWAT CUP model. Using the performance rating, the MAE, SE, SD, R2, RMSE, RSR, and NSE can be considered to be very good for the calibration and validation periods.

Table 8 Statistical evaluation of the results of the SWAT model
Full size table

Discussion

Summary of the application of the SWAT and SWAT CUP models

The current study focused on applying the SWAT model to study the Khazir River basin based on climate elements (rainfall and temperatures), soil types, the topography, and land use in order to estimate the sediment yield and surface runoff and to evaluate them using a set of mathematical and statistical equations for modeling and simulation. The model was applied to understand the hydrological processes and responses in the basin. The results of the SWAT model analysis indicated that the primary flow resulting from rainfall is an important component of the total drainage within the basin, as it contributes more than the surface runoff and thus helps with the sediment yield and transport. Rainfall data are also essential in distributed hydrological studies and were used as input data for SWAT when it was applied to analyze and calculate the sediment yield and surface runoff. In addition, the basin's soil and land use played influential and effective roles in varying the distribution of surface runoff volume and the amount of sediment within the sub-basins.

The calibration testing and validation were done using the SWAT CUP model. After successful calibration and verification, the average daily surface runoff of the drainage basin at the outlet was 1.254 m3/s over the selected period, while the average daily sediment yield of the outlet drainage basin was 2.691 tons/day. The wet seasons, especially the rainy months, witnessed a high sediment yield, in contrast to the dry months, in which the values were zero. Sediment yield is affected by surface runoff rates and various soil properties, particularly the hydrological group and soil texture.

The SWAT model has proven its ability to model and simulate sediment yield and surface runoff. It produces satisfactory results and can be used in future flood planning and management and in the forecasting of climate variability by selecting a range of scenarios. The results of this paper can be relied upon when studying neighboring basins that need integrated management of water resources. The outputs generated from this model can also help to determine the relative rates of hydrological processes and can help basin managers to identify areas at risk of land degradation. Modeled rates of sediment yield and surface runoff may serve as a guide for an additional hydrological analysis or field investigation of observed values of hydrological processes, which could be used to definitively evaluate the accuracy of our model predictions.

Challenges of the application of the SWAT and SWAT CUP models

The process of preparing the necessary data is a major challenge in studies that cannot rely on traditional data. Many modeling and simulation studies have reported that the low quality of temporal and spatial data represents a major challenge to obtaining accurate results, especially in studies related to land-use assessments. There is a great need to support the improvement of all data and their analysis for the sustainable development of water resource management in the basins. In light of the availability of many types of data, especially climate, soil, land-use, and topographic data, it was necessary to choose data with high accuracy and from reliable sources in order to ensure their validity and to learn how these data were measured. This stage took the researchers a long time and required considerable effort due to the need to process the data and prepare them in a format that the SWAT model accepts by using a set of computer programs, thus making the data ready for modeling and simulation. The model requires pre-processing of the input data, which is computationally laborious for large basins. Then the data were entered into the SWAT model within the ArcGIS 10.8 software. In addition to this issue with the data, the absence of sufficient hydrological studies for the study area represented another challenge in the research described in this paper.

Choosing sensitive parameters from the SWAT CUP model was one of the difficulties that the researchers faced. There are a large group of parameters, some of which have nothing to do with surface runoff and sediment yield and some of which are directly related to them. The process of choosing the most appropriate and sensitive parameter took a long time and a lot of effort during the calibration and validation of the results. To further improve the results, the calibration process was repeated more than once. Therefore, the researchers suggest that SWAT CUP model designers should identify sensitive parameters for sediment yield and surface runoff under particular climatic conditions and swap them with other parameters in the event of future climate change in order to allow for more comprehensive reviews and careful future analysis using other hydrological models. There is an urgent need to make some additional improvements to the SWAT CUP model to better predict sediment yield and surface runoff when calibrating and validating the results for river basins. Given the technology available in terms of model-building, calibration, and validation tools and the availability of freely available data, it is possible to build a hydrological model with high spatial and temporal resolution, and better data availability will help make model predictions more accurate and reduce uncertainties.

Comparison of the research results with previous studies of North African basins

The sediment yield and surface runoff of the Khazir River basin were studied and evaluated, and the modeling and simulation results were compared with the results of previous studies dealing with the application of the SWAT model, especially scientific studies of North African basins. The obtained results showed a great difference between the climatic conditions that exist in the Middle East, specifically in Iraq, on the one hand, and those in North African countries, especially Algeria, Tunisia, and Morocco, on the other hand. This variation includes all climatic elements, such as rainfall, temperature, humidity, evaporation, and wind. In the present study, the amount of rainfall was considered to play an important and effective role in modeling and simulation processes. Climate change increases the intensity and frequency of heavy rainfall, especially in mountainous areas with steep slopes and fast currents. In fact, heavy rainfall causes serious erosion and landslides on slopes, which increases the concentration of sediment in the depressions and valleys of river drainage basins. In addition, differences in land topography, land use, and soil type lead to a difference in sediment yield and surface runoff, and this difference was clearly seen when comparing the results of the current study with the results of studies in North African regions. This difference was revealed to be due to differences between the natural conditions of the four basins (Table 9).

Table 9 Comparison of some of the characteristics of the basin in the study area with those of the basins investigated in previous studies
Full size table

There are differences in the amount of sediment yield and the volume of surface runoff between the study area and the studied basin in Tunisia. There is an increase in the volume of surface runoff resulting from an increase in the amount of rain falling on the basin in Tunisia during the months of spring, summer, and autumn, whereas rain falls in the study area during the months of winter and spring, and the quantity of rain in the study area is less than half the amount that falls on the basin in Tunisia. The increase in the volume of surface runoff resulted in the erosion of soil and fragile rocks, which caused an increase in the sediment yield and its transfer to the mouth of the basin. The topography of the basin played an influential role in the amount of sediment yield and the volume of surface runoff, as the study area was characterized by a lack of steep slopes compared to the slopes of the studied basin in Tunisia.

The amount of rain that falls over the study area is less than half of the amount of rain that falls over the studied basin in Algeria, where the intensity and type of rain and the time of its occurrence plays an effective and influential role in the volume of surface runoff and the resulting sediment yield, and the intensity of the rainfall may lead to surface runoff and thus the occurrence of floods within the basin. The rainfall, geological, biological, and physical characteristics mainly control the degradation of the soils from the mountainous regions of the studied basin in Algeria. It should be noted that, despite the small area of the basin in Algeria compared to the study area in Iraq, the volume of surface runoff and the amount of sediment yield is much greater than what is found in the Khazir River basin. This is due to several reasons, including the closeness of rainy days and the amount of rain that falls over the basin on those days, the rainy season, the type of land use, the hydrological type of the soil, the intensity of drainage, and the degrees of the slope. These characteristics differ significantly from what is found in the study area.

Furthermore, the studied basin in Morocco included mountainous areas with steep slopes, resulting in high surface runoff and very large quantities of sediment yield. Deforestation and heavy rainfall in the basin contributed to these results. This contrasts with the study area in Iraq, which is characterized by spatial variation in its geomorphology, a basin that largely consists of plains with little slope and little rainfall and confinement during the winter and spring seasons, and rainy days that are spaced such that they may not lead to the occurrence of surface runoff and sedimentary output. Some of the reasons for the lack of surface runoff and sediment yield in the study area compared to the studied basin in Morocco are the presence of agricultural lands and some shrubs that help protect the soil from erosion and the lack of water flowing too quickly in the basin. The mountainous areas of the studied basin in Morocco were characterized by high surface runoff and a large sedimentary yield, in contrast to the study area, in which the sediment yield was concentrated in the plain areas close to the mouth of the basin. The reason for this characteristic of the study area is the land-use class for these areas: pastures devoid of vegetation cover.

Conclusion and recommendations

The current study focused on evaluating the sediment yield and surface runoff of the Khazir River basin. This was done after dividing the main basin into 27 sub-basins which were spatially linked to water stream networks that contribute to the sediment yield and surface runoff. The SWAT model estimates of surface runoff and sediment yield were all within reasonable ranges for the study area. The model showed good simulation performance. However, the spatial variation of surface runoff and sediment yield was not the same for all sub-basins. This is due to the spatial variation in soil type and land use. In addition, neither surface runoff nor sediment yield showed any increase with increasing slope. Rather, they both showed a significant increase towards the mouth of the basin. There is a strong direct relationship between precipitation and surface runoff on the one hand and sediment yield on the other hand. Precipitation played the main and most influential role in surface runoff and sediment yield, as rainy days witnessed the flow of water over the surface of the earth, which led to the exposure of the fragile soil and rocks to water erosion processes and resulted in a sedimentary load with water running from areas with high slopes to areas with low slopes. This sediment-laden water flows into the river, which reduces the water capacity of the stream, and thus there will be a flow of water towards neighboring lands, which may turn into floods with a rainstorm.

The average daily sediment yield of the drainage basin at the outlet was estimated by the SWAT CUP model to be 2.691 tons/day, while the average daily surface runoff was 1.254 m3/s. The upstream sub-basins are a source of sediment due to the high surface runoff generated by the high monsoon rainfall from November to May. The performance of the SWAT CUP model with the selected parameters indicated that it has strong predictive power for modeling the hydrological process of calibration watersheds and validating sediment yield and surface runoff. There was a good agreement between the simulated data and the observed data during the calibration and validation periods. The R2 value reached approximately 0.77 during the calibration phase (2003–2008) and 0.80 during the validation phase (2009–2012) for sediment yield, while the R2 value for surface runoff reached 0.88 during the calibration phase and 0.85 during the validation phase. Predictions of sediment yield and surface runoff showed good results for the calibration (NSE = 0.74) and validation (NSE = 0.78) periods. Moreover, the values of both R2 and NSE were higher than 0.70, indicating that the SWAT model can be effectively used to simulate sediment yield and surface runoff in the study area. We can conclude that the model built in this research will be suitable for predicting hydrological responses to climate and land-use changes in the future.

The study recommends caution when using the SWAT model to estimate the surface runoff and sediment yield in watersheds; the changes that occur to climate elements and land uses over time should be taken into account, as should the use of a digital elevation model with high spatial accuracy to obtain accurate results that reflect the natural conditions of the watersheds. We also recommend that authorities concerned with water resources management should implement appropriate and accurate strategies such as wastelands, contour plowing, and soil dams with tree planting in sub-basins exposed to water erosion to reduce the sediment yield within the basins and the river course.