Off-campus South Dakota State University users: To download campus access theses, please use the following link to log into our proxy server with your South Dakota State University ID and password.

Non-South Dakota State University users: Please talk to your librarian about requesting this thesis through interlibrary loan.

Document Type

Thesis - University Access Only

Award Date

2012

Degree Name

Master of Science (MS)

Department / School

Geography

First Advisor

Darrell Napton

Abstract

The United States National Land Cover Database (NLCD), developed from the early 1990s to 2006, is an important resource for monitoring the nation’s land cover change. In this research, I developed a method that identifies and analyzes the commission and omission errors for two agricultural classes mapped in NLCD 2006, quantifies the magnitude and spatial distribution of these errors over different geographic areas, and enumerates the causes of the errors regarding the land cover characterization and mapping processes. The method focuses on integrating multiple geospatial datasets and employs a Comprehensive Change Detection Method (CCDM) that captures both spectral and land cover changes to identify the commission and omission errors for cultivated crops in NLCD 2006. We tested this method over eight Landsat path/row footprints with diverse landscapes across the conterminous United States. An accuracy assessment protocol was developed and implemented to assess the performance of the methods. The results from this research showed that the majority of the omission and commission errors identified occurred in geographic areas with nontraditional agriculture that have diverse landscapes and land cover types. Path/rows that have agriculture as the dominant land cover, however, had very little error. The agricultural class was not mapped well in areas with nontraditional agriculture mainly because the NASS Cropland Data Layer (CDL) dataset is not as accurate in these areas as in the traditional agriculture areas. The average agreement between model-identified and reference data on commission error was high (79%). On the other hand, the average agreement between model-identified and reference data on omission was somewhat low (40%). As for the source of errors, the majority of commission errors in the NLCD 2006 cultivated crops (Class 82) were from the urban classes. The majority of omission errors were from herbaceous and shrub classes. The majority of the commission and omission errors are a legacy error from NLCD 2001. In addition, some errors were caused by a change in NLCD 2006, by a misclassification in NLCD 2010 because of training data influence, or by the post agriculture classification process. Some areas were mapped correctly in NLCD 2001 and NLCD 2006 but considered a new change in NLCD 2010. The method developed in this study has a few limitations, which have implications in terms of identifying the potential omission and commission error in NLCD 2006. The method of this study can be used to analyze errors of other land cover classes in NLCD 2006.

Library of Congress Subject Headings

Land use -- United States -- Classification -- Evaluation
Land cover -- United States -- Maps -- Evaluation

Description

Includes bibliographical references (pages 117-124)

Format

application/pdf

Number of Pages

144

Publisher

South Dakota State University

Rights

In Copyright - Educational Use Permitted
http://rightsstatements.org/vocab/InC-EDU/1.0/

Share

COinS