Spatiotemporal Dynamics of County-Level Grain Production Security Across Agro-Ecological Zones in Gansu Province, China

Maintaining grain production security in drylands requires improvements in land productivity and stability under climatic and resource constraints. Existing studies have often examined production dynamics, yield-related conditions, and per-capita grain availability separately, leaving their long-term relationships across contrasting agro-ecological systems insufficiently understood. Using county-level data for Gansu Province, China, from 2000 to 2024, this study integrated trend and stability analysis, Getis–Ord Gi* statistics, random forests, and principal component analysis to examine production dynamics, county yield clusters, predictor importance, and the relationship between a production input-scale index (F) and per-capita grain availability (P). Grain production increased by 81.7% and yield by 87.2%, while grain-sown area decreased by 2.9%, indicating yield-led growth. Detrended departures generally narrowed after 2016. High-yield clusters persisted in the Hexi Corridor, whereas a limited cold-spot cluster emerged in the Southern Alpine and Mountainous Region after 2018. Random forest predictive profiles differed among zones, with greater grouped climatic importance in the Hexi Corridor, greater socioeconomic importance on the Loess Plateau, and a more uncertain mixed profile in the Southern Alpine and Mountainous Region. F increased overall but retained its regional hierarchy, and its association with P remained weak to moderate (r = 0.19–0.38). Shortage counties declined from 53 to 19, while surplus counties increased from 17 to 52—most of this transition occurred on the Loess Plateau and involved both lower- and higher-F surplus types. Overall, grain production security in Gansu improved through higher productivity and smaller recent departures from long-term trends. County-level yield remained spatially clustered, and improvements in per-capita grain availability occurred under different input-scale conditions. By integrating productivity, temporal stability, spatial heterogeneity, and the relationship between input scale and per-capita availability, this study provides a more comprehensive understanding of grain production security and an empirical basis for differentiated management in resource-constrained drylands.