Date of Award

Spring 1-1-2010

Document Type

Thesis

Degree Name

Master of Arts (MA)

Department

Geography

First Advisor

Stefan Leyk

Second Advisor

Barbara Buttenfield

Third Advisor

John Pitlick

Abstract

A multi-scale geomorphometric landform model was created through the use of fuzzy semantic import models and fuzzy overlay to measure distribution of landforms within parcels of the Conservation Reserve Program in a portion of the Delaware River Sub-basin in Northeast Kansas. Different fuzzy logic operators (intersect, algebraic mean, and fuzzy gamma) were used to test the impact of different model mechanisms on the resulting distributions of crisp and fuzzy membership values, and classification uncertainty measured by entropy values. Across scales (900 m2 to 16,900 m2 window sizes), only one crisp class (drainages) showed an optimal scale for detection area. The statistical distribution of fuzzy membership values were significantly different between the same classes derived from different overlay operators, but this had limited impact on the agreement between crisp landforms derived from the three operators at a single scale. Within each pairing of overlay operator and scale, the landform classes backslopes and flats had the highest proportional representation of classes at most entropy levels (0.95, 0.90, 0.85, and 0.75) except for locations within the single-highest entropy level (0.99), where the class proportions were more variable. This is significant, as both of these classes were dominant at different scales (backslopes at finer scales, and flats at coarser scales) within CRP Parcels. The fuzzy and multi-scale approach provides flexibility in assessing class-level uncertainty that had been previously unaddressed in the use of geomorphometric systems for mapping, modeling, and applied management applications.

Share

COinS