A regional soil classification framework to improve soil health diagnosis and management
Scott Devine, Kerri L. Steenwerth, Anthony Debons
Abstract
Abstract Soil health research has developed soil tests to validate management strategies for creating healthier soils, such as cover crops, compost, and reduced tillage. Missing in this research agenda are soil health management strategies and diagnostics tailored to the systematic variation of soils. The Soil Survey Geographic (SSURGO) database offers comprehensive estimates for many properties linked to soil health indicators. This research uses unsupervised classification of soil surface (0–30 cm) properties and root restrictive horizons in an internationally important agricultural region, the Central and Coastal Valleys of California, covering 5.6 million ha of mostly cropland. K ‐means clustering of 10 soil properties derived from SSURGO revealed groups distinct from USDA‐NRCS Soil Taxonomy. Clustering metrics, data visualization, and validation tests justified a seven‐region soil health conceptual model that divided the landscape into two broad classes—those with and without performance limitations. Results revealed three pedogenic conditions limiting soil performance: (a) root restrictive horizons that impede root growth into deeper soil; (b) salinity and alkalinity harmful to crops; and (c) shrink–swell properties making soils challenging to cultivate. Soils without performance limitations, soils with restrictive horizons, and salt‐affected soils are further divided by texture and/or organic matter for a total of seven soil health regions. In reality, variation in soil properties is often continuous across the landscape and sometimes fractal in nature, but this research establishes a testable framework for evaluating soil health diagnostics according to geographically coherent differences in soil properties, demonstrating potential for how soil health management and interpretation can be tailor‐made to the landscape.