Climate change impact assessment on water resources in the mountainous Karaj Dam Basin using an integrated weather generator-machine learning framework

This study assesses the impacts of climate change on water resources and droughts in the Karaj Dam basin, a key water source for Tehran, Iran. Sixteen climate projections from four GCMs under four SSPs were downscaled using the M-LARS-WG model. Monthly streamflow was simulated using the Random Forest (RF) model, with NSE of 0.96 (calibration) and 0.95 (validation). The models demonstrated strong performance in reproducing weather, streamflow, and drought characteristics, with droughts assessed by SPI, SPEI, and SDI indices. Under the median of future projections, annual streamflow and precipitation rise by ~ 4%, from 12.7 to 13.3 m3/s and 687 to 714 mm, while the probability of moderate or more intense droughts decreases by 3–4% (e.g., from 0.16 to 0.13 based on SPEI). Rising temperatures and increased snowmelt are expected to alter seasonal streamflow more than precipitation, leading to higher winter flows and lower spring flows. Based on the 12-month SDI, average drought duration and magnitude are projected to decrease from 23.4 months and 22.8 to 20.6 months and 20.2, respectively. These findings for the Karaj basin—due to differences in baseline climate conditions—contrast with projections for Tehran, alleviating concerns over severe future water shortages in the capital.