Environmental Variables in Predictive Soil Mapping: A Review

In the well-known conceptual model SCORPAN, a given soil property is considered as dependent on the following environmental factors: soil, climate, organisms, topography, time, and space. Predictive mapping of soils in digital soil mapping is based on similar ideas, but environmental factors may include not only factors of soil formation, but also, for example, remote sensing data. At present, predictive mapping has found wide application not only in soil science but also in ecology, agriculture, and geomorphology. This paper provides a review of environmental factors that are used in predictive mapping with a special attention to situations, when wide sets of environmental factors may be used and a part of them are not quantitative, such as vegetation types. The systems of quantitative variables for topography and climate description are most developed, so special attention is paid to them. Land surface description is performed using both local and non-local variables that need integration. In climate description, variables, which give estimates of dry or wet terrain features, such as moisture ratio or water deficit, are essential. They need evaluation of potential evapotranspiration that is not measured by weather stations, but can be calculated. The possibilities of accounting for these and other environmental factors, including non-quantitative ones, in quantitative statistical models of predictive mapping are described together with principles of their construction, verification, comparison, and choice of appropriate models. Examples of predictive soil mapping applications are given for various scales, and their specificity for different scales is outlined. Some aspects of remote sensing data usage are discussed.