Agriculture consumes huge amounts of water in China and is profoundly affected by climate change. This study projects the agricultural water use towards 2030 under the climate change mitigation target at the provincial level in China by linking a computable general equilibrium (CGE) model and a regression model. By solving the endogeneities amongst agricultural water use, output and climate factors, we explore how these variables affect water use and further predict future trends through soft-link with the IMED|CGE model. It is found that sunshine duration has a slightly positive impact on water use. Furthermore, agricultural output will significantly drive agricultural water use based on historical data of the past 16 years. Results also show that carbon reduction would have a trade-off or co-benefit effect on water use due to regional disparity. Provinces with increasing agricultural exports, such as Xinjiang and Ningxia, would anticipate considerable growth in agricultural water use induced by carbon reduction. The soft-link method proposed by this study could be applied for future studies that aim to incorporate natural and geographical factors into human activities, and vice versa, for assessing sustainable development policies in an integrated way.