Return to search

A Comparison of Fuzzy Models in Similarity Assessment of Misregistered Area Class Maps

abstract: Spatial uncertainty refers to unknown error and vagueness in geographic data. It is relevant to land change and urban growth modelers, soil and biome scientists, geological surveyors and others, who must assess thematic maps for similarity, or categorical agreement. In this paper I build upon prior map comparison research, testing the effectiveness of similarity measures on misregistered data. Though several methods compare uncertain thematic maps, few methods have been tested on misregistration. My objective is to test five map comparison methods for sensitivity to misregistration, including sub-pixel errors in both position and rotation. Methods included four fuzzy categorical models: fuzzy kappa's model, fuzzy inference, cell aggregation, and the epsilon band. The fifth method used conventional crisp classification. I applied these methods to a case study map and simulated data in two sets: a test set with misregistration error, and a control set with equivalent uniform random error. For all five methods, I used raw accuracy or the kappa statistic to measure similarity. Rough-set epsilon bands report the most similarity increase in test maps relative to control data. Conversely, the fuzzy inference model reports a decrease in test map similarity. / Dissertation/Thesis / M.A. Geography 2010

Identiferoai:union.ndltd.org:asu.edu/item:8672
Date January 2010
ContributorsBrown, Scott Benjamin (Author), Wentz, Elizabeth A. (Advisor), Myint, Soe W. (Committee member), Anderson, Sharolyn (Committee member), Arizona State University (Publisher)
Source SetsArizona State University
LanguageEnglish
Detected LanguageEnglish
TypeMasters Thesis
Format53 pages
Rightshttp://rightsstatements.org/vocab/InC/1.0/, All Rights Reserved

Page generated in 0.0018 seconds