Spatio-temporal analysis of sediment yield with a physically based model for a data-scarce headwater in Konya Closed Basin, Turkey

There are many empirical, semi-empirical and mathematical methods that have been developed to estimate sediment yield by researchers. In the last decades, the advancement in computer technologies has increased the use of mathematical models as they can solve the system more rapidly and accurately. The Soil and Water Assessment Tool (SWAT) is one of the physically based hydrological models that is preferred to compute sediment yield. In this study, spatial and temporal analysis of sediment yield in the Çarşamba Stream located at the Konya Closed Basin has been investigated using the SWAT model. Streamflow and sediment data collected during the 2003–2015 time period have been used in the analysis. Consequently, the SWAT presented satisfactory results compared with R1⁄4 0.68, Nash–Sutcliffe Efficiency (NSE)1⁄4 0.68 in calibration and R1⁄4 0.76, NSE1⁄4 0.66 in validation. According to the model results, spatial asymmetry in terms of sediment yield was determined in the sub-basins of the study area.


INTRODUCTION
Only 3% of the water in the world can be used by humans.
Water resources are rapidly decreasing due to industrialization, population growth, water pollution, unknown water consumption and climate change. This makes it important to plan, design and manage water structures correctly to use the available water resources most efficiently. Sediment transported in rivers and accumulated at the upstream of dam reservoirs reduces the economic life of the water structures and this accumulation is a serious problem and has severe consequences for water management, flood control, producing energy, irrigation and drainage systems, domestic water supplies, river transportation systems and recreation works. Additionally, the morphology and ecology of the river are affected negatively by sediment. Due to the strong relationship between sustainability of water resources management and development of water resources, the estimation of sediment transport rate along the rivers and erosion control in the streams and tributaries is very important (Öztürk et al. ; Sivakumar ; Dogan ).
Sedimentation accumulation in dam reservoirs as a result of erosion from catchments during relatively short periods of flood discharges and also the gradual accumulation of the incoming sediment load from a river basin emerge as an important problem as this decreases the economic life of dam reservoirs. The worldwide loss in reservoir storage capacity is reported to be approximately 1.0% per annum (Mahmood ). If the sediment transport rate is high, sediment will eventually fill a reservoir earlier than its economic life. It is estimated that on a worldwide basis the cost of the annual capacity loss because of the siltation is around US $6 billion (Mahmood ). In Turkey, the lack of measurements taken for land use and erosion control results in the deposition of sediments and loss of capacity of dam reservoirs (Medik Dam, Akkaya Dam, Cubuk-I Dam, etc.).
It is very difficult to calculate the sediment yield analytically because of the complexity of phenomena affected by many parameters and resulting from geological, topographical and climatological factors. Although the measurements taken by sediment observation stations are the most reliable data, these may have disadvantages in cost and time. Generally, suspended sediment measurements are observed along the river on any day of each month, this causes a limited number of sediment data. In addition to direct measurement along the river, sediment load can also be estimated by rating curve method, regression analysis, artificial neural network (ANN) and empirical methods. The Soil and Water Assessment Tool (SWAT) is one of the physical-based tools used in estimating sediment transport rate. Although SWAT has been widely used for streamflow, there is a scarcity of literature applicable to the accuracy of sediment simulations (Borah &   Basin, there is no study in which spatial and temporal analysis of sediment movement is performed. Therefore, it was thought that selection of the Çarsamba Stream Basin, which is an upstream basin as a study area, would constitute a successful example to cover these gaps. Also, when it comes to the regional importance of the study, the Konya Closed Basin is an endorheic basin where it is quite insufficient in terms of water resources, and has a cold-semi-arid climate and intensive agricultural activities. The headwater of the Çarsamba Stream, which is used as a study area, feeds the important water resources at downstream. The inefficiency of the Çarsamba River will cause important water resources and dam reservoirs to lose their effectiveness. Considering its global importance as well as its regional importance, the originality of the study comes to the fore. In this study, the aim was to make a spatial and temporal analysis of the sediment movement for the Çarsamba River upstream basin, which has an important role for the Konya Closed Basin. For this main purpose, objectives were as follows: (i) to establish, calibrate and validate a physically based model for the study area; (ii) to analyze sediment movement as spatially and temporally; and (iii) to suggest strategies for controlling sediment movement to authorities.

Study area and data
The study area covers the drainage area of D16A115 sediment and flow measurement station, which is located in In order to set the SWAT model, meteorological data like precipitation, temperature (maximum and minimum), relative humidity, wind speed and solar radiation have been used. As there was no suitable station that had enough data for modeling inside the study area, the data were taken from Hadim (Station number: 17928) and Seydisehir (Station number: 17898) gauging stations, which are very close to the study basin and operated by the State Meteorological Service.
The meteorological data belonging to the study basin were determined by the Thiessen method. Mean maximum temperature, mean minimum temperature and mean annual precipitation were 17.5 C, 6.1 C and 785 mm, respectively.
The data, which were used for calibrating streamflow and sediment transport rate, were observed from D16A115 streamflow gauging station. The information belonging to the stations used in this study is given in Table 1 and the sediment rating curve of D16A115 streamflow gauging station is given in Figure 2.

Soil and Water Assessment Tool (SWAT)
The SWAT, which is widely and effectively used in assessing sediment yield, was used in this study. SWAT is a continuous physically based semi-distributed model and an effective tool for assessing changes in hydrological processes devel-  in the hydrological cycle is given as follows: where:

SWAT model implementation
The SWAT model is used effectively in estimating sediment transport rate, land use management and estimating avail-  (Figure 1(b)).
The study area was divided into eight sub-basins. SWAT codes and areas for LULC are seen in Table 2.
The dominant LULC is seen as SWRN having 41% in the overall area. It is followed by RNGB and RNGE with 34% and 14%, respectively. The rest of the LULC is not classified as their ranges are small and given totally as Generally, there are three soil types in the study area.
Soil types belonging to sub-basins and overall study area are given in Figure 4. Soil types in the basin consisted of   If the sub-basins are analyzed, the steepest one is the 4th sub-basin, which is steeper than 25% in 71% of its area. It is followed by the 2nd, 5th and 7th sub-basins, in which 66% is steeper than 25%.
The majority of the study area is covered by mountainous and has steep slopes. So it is expected that sediment transport rate significantly increases because of the erosion of the land surface.

Sediment model
The where sed is the sediment yield, CFRG is the coarse fragment factor, area hru is an area of the HRU, Q surf is surface runoff volume, q peak is the peak runoff rate, K USLE is Universal Sediment Loss Equation (USLE) soil erodibility factor, P USLE is USLE support practice factor, C USLE is USLE cover and management factor, and LS USLE is USLE topographic factor (Neitsch et al. ).

Calibration and validation
In this study, automatic calibration and validation steps were performed using Sequential Uncertainty Fitting Version 2 (SUFI-2) included in the SWAT-CUP program. The SUFI-2 algorithm is frequently preferred due to its advantages in calibration and uncertainty analysis processes (Khoi & Thom ). In SUFI-2, the uncertainty of the parameters is confidence interval is intended to include the measured data.
With the SUFI-2 algorithm, two different statistics are taken into account during the solution phase. The first is the Pfactor and the second is the R-factor. The P-factor is the percentage of actual data covered by the 95PPU. The R-factor is the thickness of the 95PPU range. The algorithm works on the principle of obtaining the most P-factor with the least R-factor. Theoretically, the P-factor ranges between 0 and 100%, and the R-factor ranges between 0 and infinity. In cases where the P-factor is 100% and the R-factor is 0, the simulation data overlap with the measured data.

Performance criteria
In this study, Root Mean Square Error ( (Table 3).

RESULTS AND DISCUSSION
In this study, the SWAT model was simulated  Table 4.
The performance metrics obtained after the calibration phase of the sediment model are given in Table 5 for both streamflow and sediment. According to   both parameters for both the calibration and validation period can be seen in Figure 6. When the sediment data in Figure 5 are analyzed, it is seen that the highest sediment yield was simulated as 214.3 ton/day in 2012. According to Figure 5, the shape of the streamflow and sediment yield is almost the same.
The annual averages of sediment inflow and sediment outflow amounts in the sub-basins are given in Figures 7     • It was determined that sediment production was not significant in 1-6 sub-basins. It is thought that the bushes and forests in the land cover have a positive effect on this situation.
• There is a significant amount of sediment inflow from the 7th and 8th sub-basins to the 6th sub-basin. However, the 6th sub-basin holds a large part of the sediment it receives. This is an important factor in explaining the fact that there are more slightly sloping areas in the 6th sub-basin and the increase of shrubs, agricultural land and forests in the land cover north of the sub-basin.
• Looking at the whole basin, the fact that the dominant soil type is one of the easily degradable soil classes, lithosols, shows that the basin has a soil structure prone to sediment production.
• In 2009 and 2011, quite a lot of sediment production was observed in the 7th and the 8th sub-basins compared with other years.
By the obtained results, it was determined that the 7th and 8th sub-basins in the southern basin produced a significant amount of sediment. It was evaluated that this situation is caused by the slope and land cover characteristics of these basins. The sediment transmitted from the Çarsamba Stream upstream affects the lifetime of the Apa Dam, which is an important water source for the Konya Closed Basin. In the short term, the fact that the 6th lower basin holds a large part of the sediment coming from the other sub-basins in the upper parts may create the impression that there is no problem. However, in the long term, the river structure of the 6th sub-basin is foreseen to show significant changes with sediment accumulating. Changing river structure may affect flow efficiency negatively or cause floods to occur.
The solutions and research suggestions that can be developed for the sediment accumulation problem in the sixth basin are given below.
• Terracing can be done in places where the slope is high in the south of the basin.