With climate variability posing significant global risks, semi-arid tropics are particularly vulnerable to the projected rise in radiative forcing, expected to shift monsoon rainfall patterns. Consequently, this heightened variability makes the development of climate adaptation increasingly challenging for rainfed rice. This study integrated bias-corrected CMIP6 projections (SSP245 and SSP585) with the FAO AquaCrop model and a Principal Component Analysis (PCA) to determine how climate adaptation strategies can close the yield gap (YG) and moderate water demand (WD) in the Upper Noyyal Basin, India (2020 2059). Six sowing weeks (SW), including two farmer-informed counterfactuals, were tested for both Kharif and Rabi growing seasons. Uncertainty analysis revealed that ensemble-averaged climate inputs reduced yield projection uncertainty. Under SSP245, water stress events become less frequent, with a 66 % reduction during the first two decades (2020-2039); SSP585 instead mitigates excessive rainfall events by 57 % in the later decades (2040-2059). Kharif rice sown in the counterfactual window outperforms shifted SWs, underscoring the value of indigenous knowledge. For Rabi rice, a two-week advance (SW38) narrows the YG by -65 % (SSP245) and -68 % (SSP585). PCA indicates that Water Stress Index (WSI) and Dry Day frequency (DD) jointly explain >90 % of the variance in YG and WD; under SSP585 these two variables dominate, underscoring a pivot toward high-frequency moisture shocks. This study pioneers a site-specific assessment through innovative counterfactual scenario, decadal-scale analysis to prevent maladaptation, and linking extreme indices to YG and WD under the effective adaptation strategy to quantify adaptation efficacy for policy guidance. These findings provide site-specific, time-bounded guidance for policymakers targeting resilient rice production under escalating climate uncertainty.