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


Degree Name

Master of Science (MS)

Department / School


First Advisor

Matthew C. Hansen


datasets, geospatial, Landsats, mapping


Land cover datasets are important for environmental monitoring and research purposes, and can influence policy regarding natural resource allocation and exploitation. For continental scale mapping, datasets with continuous spatial coverage, a spatial resolution good enough to represent the land cover heterogeneity, and temporal resolution that allows for regular mapping for monitoring purposes are needed. Maps produced using the relatively higher spatial resolution data, for example the Landsat series data, more often than not, accurately represents the heterogeneity, but these datasets lack spatial and temporal continuity. On the other hand, maps produced using the relatively lower spatial resolution data, for example using A VHRR data, have the advantage of spatial and temporal continuity, but lack the ideal spatial resolution to accurately depict heterogeneity. This study is in support of the United Nations Food and Agriculture Organization's medium resolution database. The medium resolution was arrived at as a compromise to overcome the short comings of the high and coarse spatial resolution data in large area land cover mapping. 250m spatial resolution MODIS data have been used alongside classification tree and regression tree algorithms to produce a medium resolution land map of Africa. In doing this, 5 VCF maps were produced, and then thresholded to result in 9 broad land cover classes. The results show that the data and the methodology employed are capable of accurately depicting the 9 broad land cover classes mapped. In the visual and numeric comparisons that are presented, the dataset scores highly in terms of replication of the training data, and in reproducing fine spatial patterns.


Includes bibliographical references (pages 71-74)



Number of Pages



South Dakota State University


Copyright © 2007 Eugene Apindi Ochieng