Climate and urban land-use controls on potential groundwater recharge under CMIP6 futures: multi-model uncertainty in a monsoon-dominated urban region

Urban groundwater sustainability depends on how climate change and land-use transformation alter recharge, yet these controls are often assessed separately and with limited representation of climate-model uncertainty. This study developed a 30-m Soil-Water-Balance (SWB2) framework for Sylhet City Corporation, Bangladesh, integrating annual historical land use, FLUS-derived future land use, bias-corrected daily climate forcing from 14 CMIP6 global climate models, parameter-sensitivity experiments, and independent groundwater observations. Potential groundwater recharge was represented by SWB2 net infiltration and evaluated against a 2000–2020 baseline. Historical mean recharge was 517.75 mm yr−1. Under SSP2-4.5, ensemble-mean recharge declined by 16.3–30.9% across four future periods through 2100, while SSP5-8.5 produced reductions of 14.7–22.9%. Recharge remained strongly concentrated in the monsoon season, although all seasons showed future declines. Spatial model-sign analysis identified extensive areas of robust recharge decrease, while substantial inter-model spread persisted in projected magnitudes. Sensitivity experiments showed that recharge was most responsive to curve number and maximum net infiltration, highlighting the importance of land-surface controls on runoff–infiltration partitioning. At the near-boundary groundwater station, recharge anomalies were positively associated with groundwater response at a one-month lag (Spearman ρ = 0.466), with stable leave-one-year-out behaviour. These results demonstrate that future urban recharge emerges from interacting climatic, land-surface, seasonal, and model-structural controls, and that multi-model uncertainty should be explicitly considered when assessing recharge vulnerability in rapidly urbanizing monsoon environments.