Premium
Mapping with words: A new approach to automated digital soil survey
Author(s) -
Liu Jian,
Zhu AXing
Publication year - 2009
Publication title -
international journal of intelligent systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.291
H-Index - 87
eISSN - 1098-111X
pISSN - 0884-8173
DOI - 10.1002/int.20337
Subject(s) - soil map , fuzzy logic , inference , computer science , fuzzy inference , digital soil mapping , data mining , artificial intelligence , natural language processing , adaptive neuro fuzzy inference system , soil science , environmental science , fuzzy control system , soil water
Soil survey (soil mapping) is based on soil‐landscape knowledge of soil scientists. Current automated approaches to soil survey cannot take such knowledge as direct input, because the knowledge is descriptive in nature. This paper presents a “mapping with words” solution by using fuzzy logic. Environmental variables used to describe landscape conditions are treated as linguistic variables. Each descriptive term used to characterize an environmental variable is treated as a fuzzy granule and is represented with a fuzzy membership function. Fuzzy membership functions are defined through gathering samples of expert perception on the landscape. Using the granule—fuzzy membership functions pairs as a dictionary, an inference can decode input descriptive knowledge accordingly and conduct soil inference. The proposed approach has been tested in a case study in Dane County, Wisconsin, USA via a soil inference approach (soil–land inference model, SoLIM). The mapping result shows that the mapping with word version of SoLIM has an 85% accuracy based on collected field points, better than a comparable earlier version (about 78%). Traditional soil survey maps usually have a mapping accuracy about 60%. The proposed methodology can be adapted to other knowledge‐based natural resource mapping with slight modifications. © 2009 Wiley Periodicals, Inc.