Incorporating process-based modeling into digital soil mapping: A case study in the virgin steppe of the Central Russian Upland

Abstract

Abstract In theoretical pedology, the advantage of factor-process-properties models over factor-properties models has been confirmed for the understanding of soil genesis and its changes over time. Currently, the most frequently used model in digital soil mapping (DSM) is the SCORPAN model, which is based on the Dokuchaev-Jenny formula: soil properties are the result of the interactions among soil-forming factors over time. This work aims to incorporate processed-based modeling into DSM and investigate the use of a hydrological model to predict the soil spatial heterogeneity and simulate its spatiotemporal evolution over time in the Central Chernozem Reserve (East European Plain). Our approach, called the Nested Landscape Soil Triad: Factor-Process-Properties, is based on the subordination of soil processes to landscape processes and the nesting of the soil system into the landscape. Landscape processes result in specific landscape properties, which control the direction and intensity of soil-forming processes, and thus control soil properties. We hypothesize that, in the virgin steppe, the result of moisture redistribution via runoff along topography (landscape process) can be expressed quantitatively in the redistributed runoff value (landscape property) that controls the movement of salts within the soil profile (soil-forming processes) and soil taxa (soil properties). We directly linked a result of the landscape process and soil properties due to difficulties in soil-forming processes modeling. We conducted a prediction across the 35\xa0ha study area using a 2.5\xa0m digital elevation model (DEM) and 157 soil profile descriptions as input. The moisture redistribution process was simulated using SIMulated Water Erosion (SIMWE), implemented in open-access software (GRASS GIS). To define the optimal parameter combination for the SIMWE model, we performed multiparameter sensitivity test and optimization. We used Latin Hypercube sampling to generate the 3000\xa0×\xa06 (the size of sample per number of parameters) parameter set for Monte Carlo ensemble runs within SIMWE. The maximum correspondence between the soil cover pattern and simulated flow depth was achieved with infiltration values of 0–10\xa0mm\xa0h−1, Manning’s n of −0.3 to 1.0, water diffusion constant of −0.3 to 0.5, threshold water depth of −0.1 to 0.15\xa0m, diffusion increase constant of −3 to 6, and precipitation excess rate of 60\xa0mm\xa0h−1. The runoff redistribution values alone determine the carbonate depth in the soils (64% accuracy) and soil taxa (76% accuracy). Overall, the Nested Landscape Soil Triad: Factor-Process-Properties can explore how soil properties have changed and will change through time and identify areas with risks of soil taxa changes. The contradiction between the selected optimal precipitation excess and current climate prove the polygenetic formation of forest-steppe soils. This approach can be used to inform policymakers and large-scale management to ensure soil security.