In this first stage of SELVANS we worked on two case studies, for two objectives:

Pedogenon, genosoil, and phenosoil mapping are soil classes that allow us to assess soil change due to human activities and land use change over time. This DSM framework was first developed at the University of Sydney in collaboration with Prof. Alex McBratney and Prof. Budiman Minasny for mapping genosoils and phenosoils at regional (New South Wales), local (Narrabri), and continental (Australia) scales. Check some references below if you want to see these examples.
 
Pedogenon classes are modelled with unsupervised classification (e.g., k-means clustering) from a set of environmental covariates that represent the soil-forming factors for a given reference time, and can include information on inherent soil properties (e.g., soil texture, mineralogy). The assumption is that pedogenon classes result from similar multimillennial natural pedogenesis and historic anthropedogenesis (i.e., legacy of human activities on soil-forming processes over centuries) up to a reference time. Subsequently, each pedogenon is divided into subclasses by incorporating information on human forcings (drivers of contemporary soil change), ranging from the least disturbed pedogenon (i.e., reference state or genosoil) to several phenosoils (i.e., variants resulting from contemporary management) depending on the type, intensity, and duration of the human activity. This novel framework produces soil spatial information at a relevant level of detail for both local and regional management.

In the context of France (and many other places), a question arises with respect to the choice of a reference time and characterization of human forcings: How can we identify reference soils when these ecosystems are the result of long-term intensive human influence? In addition, human-shaped ecosystems (e.g., high-elevation managed pastures, dehesas) are assumed to define the soil system in quasi-steady state for that reference time.