Soil inference systems

Abstract

This chapter discusses the concept of soil-inference systems and illustrates it for the prediction of various soil physical and chemical properties. By using pedotransfer functions (PTFs) as the knowledge rules for an inference engine, various important physical and chemical properties can be predicted. An ideal inference system would have a user interface, would be populated with initial values, and would return the minimum variance prediction of desired quantities. The chapter presents an example of the use of a rudimentary soil inference system, SINFERS, to predict some important soil physical and chemical properties. In this example, the chapter predicts soil physical and chemical properties of kaolinitic light clay. The data available from the laboratory analysis of this soil include particle-size distribution (PSD), bulk density, organic carbon content, and pH in water.