Delineating hierarchical agro-ecological zones for crop production environments in Kansas

Agro-ecological zones (AZs), defined as areas with relatively homogeneous climate and soil conditions, are important for agricultural management and planning, yet previous studies have often relied on single-step clustering, excluded edaphic attributes, or lacked validation with independent datasets. The objectives of this study were to: 1) delineate hierarchical AZs for Kansas using climate and edaphic variables, and 2) validate AZs using external datasets of environmental causes of crop yield loss and land cover. AZs were delineated using the k -means clustering with macro-AZs derived from long-term climatic features and nested micro-AZs based on soil physical attributes. Climate features included annual precipitation, annual reference evapotranspiration, and mean annual temperature. Soil physical attributes included plant available water capacity, soil organic matter, and effective soil depth. The validation consisted of a multinomial regression model with datasets of environmental causes of crop yield loss and land cover, which reflect the influence of climate and soils on crop performance. The analysis resulted in three macro-AZs partitioning the state into northwest, southwest, and east regions, each with two nested micro-AZs based on soil attributes. The multinomial model resulted in a validation accuracy of 88% for the macro-AZs and 67% for nested micro-AZs. Our study provides a scalable framework for hierarchical AZs that capture spatial variability in climate and soil conditions and can support regional agricultural planning, guide the design of area crop performance trials, facilitate scaling of point-level crop model simulations to broader regions, and inform the placement of future environmental monitoring stations.