Journal of Watershed Management Research

Extended Abstract
Background: The application of hydrological models in watersheds has always been considered by water resources researchers. This subject is of special importance in arid and semi-arid regions due to the greater complexity of hydrological processes in these regions. Optimal utilization and consumption of water resources and their optimal management require a better understanding of the hydrological model. Precipitation and subsequently surface runoff are important phases of the hydrological cycle. Given the present situation in most of Iran's watersheds in terms of missing statistics and high complexity and impossibility of fully understanding hydrological ecosystems, it is essential to use some approaches and methods to estimate runoff in watersheds with missing data or incomplete statistics. Rainfall-runoff models can simulate processes within a watershed and serve as a tool to estimate runoff and study hydrological processes. In the present research, the WetSpa distributed hydrological model was applied to simulate river flow in the Sarbaz watershed of Sistan and Baluchistan Province.
Methods: With an area of 6324.29 km², the Sarbaz watershed is located in Sistan and Baluchistan Province, the southeast of Iran.  The discharge of the Sarbaz River is more affected than the northern areas of the basin. The base discharge of the Sarbaz River is low, but its flood discharge is high even for the periods of low return. The Wetspa distributed hydrological model was used to simulate the runoff of this area. WetSpa is a distributed, continuous, and physical model with daily or hourly time step that explains precipitation, runoff, and evapotranspiration processes for both simple and complex contexts. The WetSpa model first calculates water balance in the root zone because this is the most important area in water retention and, at the same time, it controls surface and subsurface runoff, evapotranspiration, and underground water flow. The model uses the modified rational method to calculate runoff and to estimate snowmelt runoff growth; it applies the day-degrees coefficient method. After determining the initial values of the parameters, the model was calibrated manually in this study. Then, PEST software was used to obtain more accurate results in the calibration. PEST software adjusts the parameters step by step according to its algorithm and the program execution is repeated using the local search method until the best parameter values are obtained. To run the model, daily data on flow, precipitation, temperature, and evaporation for years (2018-2022) were considered for calibration and validation. From this statistical period, the years 2018-2020 and 2020-2022 were considered for model recalibration and model evaluation, respectively. The efficiency of the model in the calibration and validation stage was evaluated using Nash-Sutcliffe evaluation indices, the mean squared error, and the bias model.
Results: After preparing the WetSpa model based on the daily data of discharge, precipitation, temperature, evaporation, and maps of land use, soil texture, and digital elevation map, the model was implemented for two periods of calibration and validation. Based on the Nash-Sutcliffe general criterion, the model showed daily hydrographs with an accuracy of more than 69.02 percent for the calibration period, and with an accuracy of 88.18 percent based on the adapted Nash-Sutcliffe criterion for maximum discharges. In the validation period, the model based on the general Nash-Sutcliffe criterion simulated daily hydrographs with an accuracy of more than 57.4 percent, and with an accuracy of 69.39 percent based on the adapted Nash-Sutcliffe criterion for maximum flow rates. Moreover, the model is more effective in estimating high flows than low flows, but overall, the model has simulated the total flow with acceptable accuracy. The model did not simulate the discharge well at some points during the validation stage, which could be due to the structure of the model, the field conditions, or the data. The response of the model to precipitation in the area is reasonable; thus, the lack of simulation in the specified areas can be caused by the conditions of the area or errors in the recorded data. The investigations carried out at the field level revealed that a diversion dam was built to extract water before the outlet hydrometric station, which caused the simulated flow rate to be higher than the observed amount, and this lack of accurate simulation is due to the construction of this dam. Furthermore, this shows that when the simulation results are not suitable, it is not related to the inefficiency of the model.
Conclusion: The results of the evaluation indicators in the calibration and validation stage demonstrate that the model has the necessary efficiency to simulate the flow in the Sarbaz watershed. According to the investigations, most of the studies on the WetSpa model in humid and semi-humid conditions of most European countries have confirmed the appropriate use and efficiency of the model in these conditions. However, considering that a large part of our country has a dry and semi-arid climate, it is necessary to evaluate the effectiveness of this model in dry and semi-arid regions of the country. Because the climatic conditions of the Sarzab watershed are dry to semi-arid, the research results indicate that the WetSpa model has the necessary efficiency for hydrograph simulation in these climatic conditions.