The international Working Group on Digital Soil Mapping (WG-DSM) defines digital soil mapping as "the creation and the population of a geographically referenced soil databases generated at a given resolution by using field and laboratory observation methods coupled with environmental data through quantitative relationships.
Non-digital soil maps produced as result of manual delineation of soil mapping units may also be digitized or surveyors may draw boundaries using field computers, hence both traditional, knowledge-based and technology and data-driven soil mapping frameworks are in essence digital.
Products of the data-driven or statistical soil mapping are commonly assessed for the accuracy and uncertainty and can be more easily updated when new information comes available.
[8] Another example, are maps of soil properties for the entire world (250 m cell size), with quantified uncertainty, generated by ISRIC - World Soil Information [9] using state-of-the-art machine learning methods (SoilGrids)[10] that use as inputs point data from a large global soil profile database (WoSIS)[11] and over 400 global environmental covariates.
It is an adaptation of Hans Jenny's five factors not for explanation of soil formation, but for empirical descriptions of relationships between soil and other spatially referenced factors.