关键词: 粮食体系韧性, 粮食安全, 时空演变, 多维评估, 动态GMM

Abstract: In view of the limitations of existing research on the resilience of the food system, which mainly focuses on three-dimensional static evaluations, this study innovatively constructs a four-dimensional dynamic evaluation framework encompassing "resistance, adaptation, recovery, and transformation." By integrating the Dagum Gini coefficient method, Kernel density estimation, Markov chains, and a dynamic GMM system, it systematically analyzes the spatiotemporal evolution patterns and driving mechanisms of grain system resilience in China. Key findings reveal: (1) While overall national resilience has improved, significant gradient differentiation exists across regions (Major Production Areas: 0.42>Major Consumption Areas: 0.31>Production-Consumption Balanced Areas: 0.28). Inter-regional disparities, contributing 64.27% to the overall differentiation, emerge as the dominant factor. Spatial proximity effects significantly drive resilience state transitions, highlighting the spatial dependency inherent in resilience dynamics. (2) Severe imbalance characterizes the development of the four resilience dimensions: Resistance (0.290) and Recovery (0.265) are relatively strong, while Transformation capacity (0.148) constitutes a core deficiency, constraining overall resilience levels. (3) Distinct regional characteristics are evident: Major Production Areas exhibit clear comprehensive advantages but decelerating growth momentum; Major Consumption Areas demonstrate weaknesses in resistance and recovery; Balanced Areas show significant deficiencies in adaptation and transformation. (4) Transportation infrastructure and technological advancement are identified as key drivers enhancing resilience. These results provide a precise scientific foundation for formulating differentiated regional policies, optimizing agricultural resource allocation, and building a resilient grain security system.

Key words: food system resilience, food security, temporal and spatial evolution, multidimensional assessment, dynamic GMM