Plant hardiness zone mapping for the conterminous USA using the Empirical Bayesian Kriging Regression Prediction method

Abstract

Mapping plant hardiness zones (PHZs) is crucial for understanding and predicting the spatial distribution of plants in response to climatic conditions. The United States Department of Agriculture (USDA) developed the plant hardiness zones for 2012 and 2023 based on the average of annual extreme minimum temperatures using the Parameter elevation Regression on Independent Slope Model (PRISM). While the USDA's 2012 and 2023 PHZ maps remain standard references, they offer limited geostatistical transparency. This study applies the Empirical Bayesian Kriging Regression Prediction (EBKRP) method, using elevation as the only explanatory variable, to generate alternative PHZ maps for the conterminous United States, aiming to improve spatial prediction and facilitate methodological reproducibility. We applied EBKRP to generate PHZ maps for 2012 and 2023 and validated the outputs using five standard metrics: mean error, average standard error, root mean square error (RMSE), Continuous Rank Probability Scores (CRPS), and R2. The results demonstrate high model accuracy, with RMSEs under 2°C and R2 values above 0.96 for both years. Confusion matrix analyses show strong agreement with USDA's PHZ classifications, achieving overall accuracies of 88.2% (2012) and 87.0% (2023). Spatial change analysis between 2012 and 2023 revealed a distinct warming pattern: approximately 800,000 square miles (26% of the conterminous USA) transitioned into warmer PHZs, particularly in the central and southeastern U.S., while only about 66,000 square miles shifted into cooler zones, primarily in portions of the western US. Overall, this study demonstrates that EBKRP provides a robust, spatially detailed, and replicable approach for PHZ mapping, offering a complementary alternative to PRISM.

Department(s)

School of Earth, Environment and Sustainability

Document Type

Article

DOI

10.1016/j.apgeog.2026.103927

Keywords

Agriculture, Climate change, Empirical Bayesian Kriging, Horticulture, Plant hardiness zones, USA

Publication Date

4-1-2026

Journal Title

Applied Geography

Link to Full Text

Share

COinS