The occurrence of seasonal drought has led to periodic water shortages in irrigation districts of south China, posing significant agricultural challenges on water and food security. Consequently, there is a pressing need for high-efficient water resources management from a sustainability perspective. To address this issue, a copula-based fuzzy-flexible stochastic multi-objective programming (CFSMP) model was proposed to facilitate irrigation planning under the combined influence of seasonal dry and wet conditions. The model was aimed to simultaneously achieve maximum economic benefit, water productivity and green water use efficiency, considering the economic, social, and environmental spheres of the irrigation district. The CFSMP model represented an innovative solution that was capable of (1) precisely describing the response of crop yield to water supply at ten-day time scale, which can be matched with practical irrigation demand; (2) characterizing joint distributions of seasonal dry and wet conditions of precipitation during different crop growth periods based on the copula function; (3) reconciling conflicts on objectives among economic, social, and environmental spheres of the irrigation district in complex and uncertain environments. The proposed model was applied to the Dongfeng Reservoir Irrigation District in south China. The downscaled ten-day Jensen water production functions of main grain and cash crops were fitted to recognize precise distributions of crop water demand and water sensitivity index. The Frank copula, with the smallest value of squared Euclidean distribution, was selected to describe the joint distribution of seasonal precipitation and gain insights into actual seasonal water availability and water scarcity. The results showed that the proposed model produced more water-saving and yield-promoting irrigation planning policies than the current quota. The average water allocation of rice, wheat, maize, rape, and citrus decreased by 352.6 mm, 10.6 mm, 67.7 mm, 7.44 mm, and 54.6 mm, respectively. Except for a yield reduction of 8.8% for rape, crop yields of rice, wheat, maize, and citrus were improved by 33.3%, 45.4%, 29.9%, and 12.3%, respectively. The coordination degree of the model reached 0.827, demonstrating excellent performance on generating coordinated and robust solutions. Therefore, the CSFMP model has the potential to alleviate seasonal drought in south China and can be applied in similar regions with comparable resources crisis.