Improvement of soil properties maps using an iterative residual correction method

Authors: XU, CHENGCHENG - Duke University; SCUDIERO, ELIA - University Of California, Riverside; Anderson, Raymond; CHANEY, NATHANIEL - Duke University

Submitted to: Soil
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/30/2026
Publication Date: 5/19/2026
Citation: Xu, C., Scudiero, E., Anderson, R.G., Chaney, N.W. 2026. Improvement of soil properties maps using an iterative residual correction method. Soil. 12(1):665-687. https://doi.org/10.5194/soil-12-665-2026.
DOI: https://doi.org/10.5194/soil-12-665-2026

Interpretive Summary: Accurately mapping soil properties is critical for many agricultural applications, including irrigation scheduling and fertilization. Inaccurate soil maps can lead to under application of fertilizers or irrigation water, resulting in crop stress or over application, resulting in wasted resources and potential pollution. Commonly used soil maps lack uncertainty estimates in key soil physical and hydrologic properties, thus reducing the confidence for soil modeling and management applications. In this study, we used an artificial intelligence (AI) approach (pruned hierarchical Random Forest) to generate uncertainty distributions for soil properties and then refined those uncertainty distributions using observed soil data. The AI-informed refining process reduced the error in the soil property distribution by over 40%. These results are of interest to soil modelers and agronomists who need accurate digital soil maps for consulting and management purposes.

Technical Abstract: Accurate mapping of soil properties is vital for many applications, yet existing models can underestimate their spatial variability or prediction uncertainties. This study introduces a hybrid approach that combines prior soil predictions with iterative residual correction to improve soil mapping performance, with a case study in California demonstrating its application. We first generate prior probabilistic soil property maps using a pruned hierarchical Random Forest (pHRF) method. These prior estimates are then refined by integrating additional soil profile data and iteratively adjusting residuals of distribution of soil properties (differences between observation and prior predictions) pixel by pixel. The process employs non-parametric models to adaptively correct residuals without assuming predefined soil property distributions. It gradually adjusts the statistical shape of soil property distributions and incrementally corrects bias of the prior knowledge with observed soil information. The results demonstrate improved accuracy and reduced uncertainties in the posterior predictions. We currently perform soil mapping over California and at 1-km resolution to test the methodology. For residual correction, we compiled laboratory-measured soil profile data from three primary sources: the World Soil Information Service (WoSIS), the National Soil Characterization Database (SCD), and field measurements conducted by the University of California, Riverside (UCR) and the USDA-ARS United States Salinity Laboratory. From the evaluation, the posterior soil texture predictions show an RMSE of less than 10%, a 7% reduction compared to the priors. For soil organic matter (SOM) and oven-dry bulk density (BD), the RMSE also decreased, as the priors initially underestimated their spatial variation. Though posterior SOM and BD predictions were less accurate than other soil properties, this was expected since they are dynamic soil properties and their response to environment and anthropogenic activities is more difficult to simulate. The residual correction also showed reduced uncertainties, as demonstrated by narrower prediction intervals compared to the priors. This method also applied optimization with physical constraints, such as ensuring the bounds of soil property values. This study presents a two-step framework that improves accuracy and reduces uncertainty for DSM applications.