Remembering Knowledge: An Expert Knowledge Based Approach to Digital Soil Mapping

Abstract

A paradigm shift has occurred in the field of pedolody, moving away from traditional, tacit knowledge based soil survey methods to new, digital soil mapping (DSM) approaches. Although a variety of DSM techniques have been developed and are widely available, it is important not to lose sight of the wealth of existing soils information captured by historic soil survey and soil scientists. This paper presents a DSM method that combines expert knowledge with fuzzy logic and selective soil sampling to generate soil class and property maps for an approximately 16-ha (?40-acre) farm field in southeastern Indiana. Five soil classes were produced by implementing a fuzzy membership based approach utilizing four digital elevation model (DEM) derived terrain attributes: SAGA topographic wetness index (TWI), modified catchment area (MCA), slope percentage, and elevation. From the fuzzy class membership values, in combination with selective soil sampling and model calibration, calibrated and uncalibrated loess depth maps were generated. The mean absolute error (MAE) of predicted loess depth was 27 cm for uncalibrated prediction and 29 cm for calibrated prediction. Normalized yield data from 2011 found all five soil classes differed significantly in yield, giving evidence the five classes function differently in crop production. Peer Reviewed Papers