Answers are grounded in this paper and cite their sources.
High-resolution 3D soil maps are essential for agricultural and environmental management and help understand interactions between the pedosphere and other Earth system components. Soil depth functions are a primary tool for characterizing the vertical distribution of soil properties. However, their applications are often constrained by limited functional forms, insufficient physical interpretability of parameters, and weak vertical predictive capacity. These limitations restrict our ability to understand the depth-dependent patterns and mechanisms of soil information under complex topography and diverse pedogenic environments. To address these challenges, this study selected the geomorphologically complex black soil region of Northeast China as the study area and collected data from 380 typical soil profiles covering 1,534 soil genetic horizons. Based on a candidate library of eight fundamental depth functions, we constructed an interpretable soil-adaptive depth function (SADF) selection framework, where the optimal function was chosen for each profile, and the key parameter K was standardized as an auxiliary descriptor of vertical soil-property differentiation. By integrating multi-source environmental data with four machine learning models on the Google Earth Engine platform, we produced high-resolution (30 m) 3D maps for nine soil attributes: pH, soil organic carbon, total nitrogen, total phosphorus, total potassium, cation exchange capacity, clay, sand, and silt. Results indicated the following: (1) The SADF can establish optimal soil depth functions tailored to various soil properties across distinct pedogenic environments. (2) The standardized K parameter reveals distinct differences in the influence of pedogenic factors on various soil properties under different land-use types. (3) In the SOC-based horizon-level comparison with EA-Spline, SADF performed better in the 10–20, 20–30, and 30–50 cm groups, whereas EA-Spline performed better at 50–100 cm. (4) Under random ten-fold cross-validation, prediction performance varied among the nine soil properties and across depths, with R 2 values ranging from 0.11 to 0.58 and relatively low accuracy in some deeper soil layers. Independent external validation of SOC yielded R 2 values of 0.53–0.78, while the importance of environmental factors varied with soil property and depth. The SADF framework provides a flexible, interpretable approach for characterizing soil vertical differentiation and supporting high-resolution 3D soil mapping in the black soil region of Northeast China.
No takes yet. Share an insight, caveat, or question.
Gao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: