Modeling hydrological responses of watershed under climate change scenarios using machine learning techniques

Abstract Climate change is the most important problem of the earth in the current century. In this study, the effects of climate change on precipitation, temperature, wind speed, relative humidity and surface runoff in Saghez watershed in Iran investigated. The main methods were using the Coupled Model Intercomparison Project phase 6 (CMIP6), the Soil and Water Assessment Tool (SWAT) and the Artificial Neural Network (ANN) model under the Shared Socio-economic Pathway scenarios (SSPs) using the Linear Scaling Bias Correction (LSBC) for the future period (2021–2050) compared to the base period (1985–2014). MAE, MSE, RMSE and R2 indices used for model calibration and validation. The results of forecasting temperature, precipitation, wind speed and relative humidity showed the average percentage of precipitation decrease in the future period will be 6.1%. In terms of the temperature, 1.4, 1.5 and 1.6 Cº increase predicted for minimum, average and maximum temperatures, respectively. The results of studying the surface runoff changes using the SWAT model also illustrated that based on all three scenarios SSP1-2.6, SSP3-7.0 and SSP5-8.5 in the future period, the amount of surface runoff will decrease, which based on three mentioned scenarios is equals to 17.5%, 23.7% and 26.3% decrease, respectively. Furthermore, the assessment using the artificial neural network (ANN) also showed that the parameters of precipitation in the previous two days, wind speed and maximum relative humidity have the greatest effect on the watershed runoff.