An expert system for allocating soils to landforms through the application of soil survey tacit knowledge

Soil survey is a paradigm-based science that relies heavily on the application of conceptual soil-landscape models, which in turn are based upon tacit pedological knowledge. This tacit knowledge is generally acquired by systematic field observation and recording the relationships between the occurrence of soils and associated landform positions. Soil survey databases identify the types of soils within a delineated area but they do not generally describe the relationship of specific soils with specific landscape positions. A case in point is the recently completed 1:100 000 scale soil landscape database prepared for the agricultural region of Alberta, Canada. In order to utilize this database with various interpretative algorithms a procedure for allocating soils to specific landform positions needed to be developed. The development of this procedure initially involved capturing the local tacit pedological knowledge in a series of tables and programs. The procedure was then applied to the Alberta soil survey database to automatically assign soils to landform positions and then to assign specific slope characteristics to the individual soils. The resulting soil-landform product was more useable than the original data for input to land based process models. Key words: Soil survey, tacit knowledge, soil-landscape modeling, heuristic rule base, predictive mapping