LAUSR

In this study, the impact of climate change on streamflow is investigated using the adaptive neuro-fuzzy inference system (ANFIS) model and integrating it with metaheuristic optimization algorithms, including particle swarm optimization (PSO) and genetic algorithm (GA) under four models: MPI-ESML-2HR, MIROC6, IPSL-CM6A-IL, GFDL-ESM4, and scenarios: SSP1-26, SSP3-70, SSP5-85, for time periods (2026–2100) for which the Qazvin Plain salt marsh was investigated. LARSWG8 was used for downscaling and then bias-corrected with the quantile mapping (QM) method. Mann–Kendall and Sen's slope tests were utilized to identify the trends of climatic observational parameters. The results generally showed that among the models used, ANFIS–PSO and ANFIS–GA, respectively, showed better performance compared with ANFIS, with correlation coefficient, root mean square error (m3/s), Nash–Sutcliffe, and Kling–Gupta coefficients of 0.98, 0.19, 0.91 and 0.97 in the training period and 0.97, 0.20, 0.83 and 0.95 in the testing period. The results also indicated that streamflow will decrease under all climate change scenarios, especially during the first four months of the year in future periods. This reduction in streamflow could have widespread consequences, including negative impacts on ecosystems, economic conditions, and social structures. Therefore, optimal water-resource management, adaptation to new conditions, and precise planning for the future are essential.