The Agricultural water interventions can trigger human-water feedback, including unintended supply demand feedback—where increased water availability drives greater water use. In the Kamadhiya catchment, India, the introduction of check dams (CDs) led to a shift toward more water-intensive crops like cotton and wheat. This study formulates and tests hypotheses to understand these dynamics using an agent-based model (ABM) that integrates a spatially explicit hydrological model with a farmer behavior module. The ABM simulates 38,447 farmers using the RANAS behavioral framework, based on household surveys and observed data. Model results confirm the hypothesized feedback: increased water from CDs led to an 11.9% rise in cotton and 36.1% in wheat areas, boosting incomes and increasing adoption of drip and borewell irrigation, particularly near CDs. While drip irrigation systems improve water efficiency and post-monsoon groundwater levels, the saved water enables further wheat expansion—triggering a second supply demand feedback loop. These changes are spatially concentrated near CDs, exacerbating within-catchment disparities. Overall, about 54% of the additional recharge is used for irrigation expansion, lowering groundwater levels by 1.0 m and reducing the net benefit of recharge interventions. These findings underscore the need to critically understand human-water feedback and value of ABM as a tool to support more informed planning by offering strategies that mitigate negative externalities.