Development of an Improved GSGMDH Model With the RSA Algorithm for Probabilistic River Discharge Prediction Under the Uncertainty of Climate Change Models Compared to the SWAT Model

ABSTRACT The aim is to predict river probable discharges under the uncertainty conditions of climate change models via the Soil and Water Assessment Tool (SWAT) model, as well as the generalized structure of the group method of data handling improved with the RSA algorithm (RSA‐GSGMDH). To study the impact of climate change on regional groundwater fluctuations in the next period, the CMIP5 model is utilized under three scenarios, that is, RCP2.6, RCP4.5 and RCP8.5. The results showed that the RSA‐GSGMDH model is able to predict the outflow from the basin on the basis of precipitation and temperature data under possible climate change scenarios without the need for various data and maps, which are not available most of the time and are based only on simple input layers. The differences in the annual runoff volume predicted by the two SWAT and RSA‐GSGMDH models in the 90% and 50% probability scenarios are 4.1% and 8%, respectively. Therefore, in areas that are faced with data deficiencies, instead of the SWAT model, the RSA‐GSGMDH model can be used to predict the effects of climate change on runoff. These results are very important for water resource planners because the results of both models show that they will face serious problems in the coming years for the supply of drinking water and industry in the region.