Spatial optimization of water-saving irrigation in Chinese rice paddies: Balancing yield, greenhouse gases, and cost using NSGA-II

CONTEXT Optimizing the spatial allocation of water-saving irrigation (WSI) promotion is critical for balancing rice production, greenhouse gas (GHG) mitigation, and economic costs, yet remains challenging at a national scale due to computational intractability. OBJECTIVE This study aims to develop a spatially explicit optimization framework to identify WSI promotion pathways across China's rice paddies that simultaneously maximize yield gain, maximize GHG reduction, and minimize implementation cost. METHODS We coupled K-means clustering with a multi-objective evolutionary algorithm (NSGA-II). Based on machine-learning predicted yield and GHG emissions for 157,417 flooded irrigation grids, we first clustered these grids into 500 environmentally and agronomically homogeneous groups. We then formulated a continuous optimization problem with cluster-level conversion rates as decision variables, simultaneously maximizing national yield gain, maximizing GHG reduction, and minimizing implementation cost. RESULTS AND CONCLUSIONS The Pareto front comprised 100 non-dominated solutions spanning promotion rates from 61.8% to 80.3%. The optimal solution, selected by normalized scoring, achieved 6.47 Mt. yield gain and 63.5 Mt. CO 2 e GHG reduction at a cost of 10.14 billion CNY, corresponding to a national promotion rate of 80.3%. Cluster-scale conversion rates exhibited significant spatial heterogeneity and positive correlation with cluster size ( r = 0.23), revealing economies of scale in WSI promotion. Compared with a random promotion strategy at 90% adoption, our optimized solution delivered 74% higher yield gain with 9.7 percentage points lower promotion effort while achieving comparable GHG reduction. SIGNIFICANCE Our framework provides a spatially explicit decision-support tool for precision agricultural policy, demonstrating that smart spatial allocation can substantially enhance the efficiency of limited resources in scaling climate-smart agricultural practices.