Accurately characterizing conterminous soil pH dynamics is essential for assessing soil health risk and land degradation trends. However, the lack of robust modeling of temporal pH change from legacy observations has hindered quantifying pH risk and estimating time-to-critical (TTC), the time remaining until soils reach crop-limiting pH thresholds, thereby delaying identification of hotspots requiring timely intervention. To address this gap, we developed and validated an ensemble modeling framework using legacy soil profile data for projecting time-adjusted historical and current soil pH conditions across the conterminous United States (CONUS). Our framework combines spatial-temporal harmonization with machine learning to predict pH trajectories and estimate the TTC. Results showed that model performance was consistent across time periods, yielding mean absolute errors (MAE) of 0.51-0.55, root mean square errors (RMSE) of 0.66-0.72, and coefficients of determination (R2) of 0.64-0.66 from 1980 to 2025. TTC analysis indicated that approximately 59.0% (62.91 M ha) of U.S. croplands were projected to remain within safe pH thresholds for more than 20 years. However, nearly one-third of croplands (33.8%, 35.99 M ha) were projected to reach critical pH conditions within 0-5 years, indicating an urgent need for region-specific soil management. Acid-sensitive crops such as soybeans and peanuts were most vulnerable in acidic soils, while barley, oats, and canola faced strong alkalinity constraints. Spatially, alkalinity risks were widespread in the western and Midwest and Southwest regions, whereas acidity risks dominated in the Midwest, Southeast, and northern U.S. The study results demonstrated that soil pH risk modeling and assessment of TTC is an effective framework for quantifying temporal risks and identifying near-term hotspots, thereby guiding targeted management, land-use planning, and policy interventions.
Thai et al. (Tue,) studied this question.