Digital methods for field scale soil mapping

Jenette Michelle Ashtekar, Purdue University


Fuzzy set theory is an established, well tested method for generating reliable predictive soil models and digital soil maps. Often times expert based soil knowledge is implemented as the guiding force behind fuzzy logic prediction, utilizing the soil scientist's personal knowledge of soil properties, and their relationship with covariate landscape attributes, to generate decision based fuzzy membership values. Although these methods have the capability to adequately predict soil properties, the knowledge driven definition of rules and distributions is often left murky and under defined, making it difficult for other scientists to replicate the results. In order to make accessible, cost effective field level soil maps at the detail necessary for informing management decisions, it is vital to develop a method for fuzzy classification which limits the application of individual knowledge, maximizes the use of shared knowledge, allowing for automated rule threshold setting, reproducibility, and timeliness. For this study, we developed and evaluated an approach to fuzzy logic mapping which involves predefined hard classification based on automated landform classification, followed by the generation of fuzzy membership values, through automated rule setting, driven by the distribution of landscape attributes with a given class. A method of selective sampling was also developed to determine the most ideal locations on the field to sample for model calibration, with the intention on providing the most accurate prediction using a minimal number of sampled data points. The goal of this study was to generate a flexible model that can be implemented on most landscapes typically found to occur in the glaciated region of the Midwestern United States. Mapping of a farm field level site in Southeastern Indiana for functional soil classes and properties took place, and a comprehensive validation was performed, comparing the models' ability to predict soil properties.




Owens, Purdue University.

Subject Area

Soil sciences

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server