|
چکیده
|
In the context of hydrological science, runoff prediction and simulation have long been acknowledged as essential research projects. However, the distribution of river runoff is very spatiotemporally variable due to the intricate interactions between basin geomorphology, climate change, and human activity. This makes hydrological process prediction extremely difficult. Consequently, predicting runoff in the Iranian Dez River Basin under climate change is the primary objective of this study. Since Tireh, Marberah, Sezar, and Bakhtiari are the four subbasins included in this case study, four synoptic and hydrometric stations were chosen (for each subbasin). For the years 1980–2012, time series data per month on precipitation, temperature, and runoff were required for the selected stations. Furthermore, for the 2020–2052 timeframe, upcoming climate data were extracted from three CMIP6 models including CANESM5, BCC-CSM2-MR and IPSL-CM6A-LR using the most recent SSPs–RCPs emission scenarios (SSP1-2.6 and SSP5-8.5) for selected station. Using observational data, the historical period's extracted monthly temperature and precipitation variations by CMIP6 models were examined. These models were evaluated using a variety of metric indices, such as the correlation coefficient (R), mean absolute error (MAE), mean bias error (MBE), and root mean square error (RMSE), Willmott index calculation (WI), and modified index of agreement (md). Among them, the CANESM5 model was selected to extract future data at four subbasin. Conversely, three Decision trees (DT), ensemble, and Gaussian Process Regression (GPR) are examples of single artificial intelligence (SAI) models—were developed for each Dez basin subbasin in order to predict runoff. Training (calibration) and testing (validation) phases were applied to the input (temperature and precipitation) and output (runoff) datasets in a ratio of 80%:20% for each subbasin. The SAI models were investigated, and Ensemble model was selected to predict runoff. Then, using training Ensemble model, runoff of each subbasin under two scenarios was predicated. The results showed that while precipitation isn't trending much, temperature is trending upward, especially at SSP5-8.5 scenario. Additionally, runoff is trending upward in the Tireh Basin whereas it is trending downward in the other sub-basins. Consequently, water harvesting is only possible at Tireh Basin
|