Date of Award

5-2015

Degree Type

Thesis

Degree Name

Master of Science

Major

Geography

Major Professor

Nicholas N. Nagle

Committee Members

Henri D. Grissino-Mayer, Liem T. Tran

Abstract

Mali is a country in sub–Saharan Africa where monitoring of cropped land area would greatly benefit food security initiatives and aid organizations. More importantly village–scale studies on cropped land are fundamental to making a difference in the way we look at cropped land area and food availability in this region of the world. Using Landsat surface reflectance imagery and World View–2 derived labeled data, this study focuses on accuracy of supervised classification methods while addressing various levels of scale. Several classification methods are taken into account to determine the best method possible to produce cropped area estimates using this data. The relationship between classification and scale is addressed by taking into account how distance and proximity affect accuracy. Accuracy is measured by kappa coefficients, and results among the different methods vary. Kappa coefficients generated are very low, and results suggest that estimates between labels are more accurate than estimates far from labels.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS