Document Type : Original Article
Authors
1
Assistant Professor, Department of Civil Engineering, Materials and Energy Research Center, Dez. C., Islamic Azad University, Dezful, Iran.
2
Associate Professor, Department of Civil Engineering, Islamic Azad University, Khorramabad branch, Khorramabad, Iran
3
PhD in Water Sciences and Engineering, Department of Soil Conservation and Watershed Management, Lorestan Province Agriculture and Natural Resources Research and Education Center, Areeo, Khorramabad, Iran
10.22034/hyd.2025.69085.1814
Abstract
Firstly, it compares and selects the most efficient hybrid artificial intelligence model (Wavelet Support Vector Regression, Whale Optimization Algorithm-Support Vector Regression, and Particle Swarm Optimization-Support Vector Regression) for accurate discharge estimation based on historical data from 1992-2022. In the next step, climate projections from General Circulation Models (GCMs) under various greenhouse gas emission scenarios are used to predict the future trend of river discharge from 2023-2043. Additionally, statistical indices such as the correlation coefficient, root mean square error, mean absolute error, and Nash-Sutcliffe efficiency coefficient were used to compare the performance of the hybrid models investigated. The results from evaluating the hybrid models showed that the Wavelet-Support Vector Regression model demonstrated better performance compared to other models studied, with the highest correlation coefficient of 0.980, the lowest root mean square error of 0.372, the lowest mean absolute error of 0.174, and the highest Nash-Sutcliffe efficiency coefficient of 0.985. Furthermore, the results from the evaluation of the LARS-WG model indicated that the CanESM5.0 model accurately predicts maximum and minimum temperatures but has errors in precipitation estimation, while the BCC-CSM2-MR model estimates higher precipitation during warmer seasons. Projections for the period 2023 to 2043 under different greenhouse gas emission scenarios, SSP126 and SSP585, suggest an increase in temperature, particularly in the high emission scenario SSP585, and uncertain fluctuations in precipitation amounts, with significant differences in these fluctuations among different models. Overall, the results of the flow prediction in the coming years indicated a significant decrease in the river’s discharge in the future.
Keywords
Main Subjects
)