Title page for ETD etd-04142004-140341


Type of Document Master's Thesis
Author Johnson, Kurt
Author's Email Address kjohns6@lsu.edu
URN etd-04142004-140341
Title Estimating Impervious Surfaces from a Small Urban Watershed in Baton Rouge, Louisiana, Using LANDSAT ThematicMapper Imagery
Degree Master of Science (M.S.)
Department Geography & Anthropology
Advisory Committee
Advisor Name Title
Michael Leitner Committee Chair
Cornelis De Hoop Committee Member
DeWitt Braud, Jr. Committee Member
Keywords
  • subpixel analysis
  • environmental impacts
  • unsupervised classification
  • supervised classification
Date of Defense 2004-03-23
Availability unrestricted
Abstract
Many urban areas are using estimations of impervious surfaces as a means for better environmental management. This is because research over the last two decades indicate a consistent, inverse relationship between the percentage of impervious surfaces in a watershed and the environmental problems urban areas are experiencing. Although various methods for estimating impervious surfaces can be identified, few result in accurate and defensible estimations by which environmental problems can be assessed.

This is especially important to rapidly expanding urban areas such as Baton Rouge, Louisiana where detailed records and planimetric data are lacking. Numerous studies have shown a potential for estimating impervious surfaces using remotely sensed satellite imagery however, none were performed in a sub-tropical geographical area such as southern Louisiana. Three different dates of Landsat TM multi-spectral imagery, corresponding to seasonal differences, were acquired for land cover type classification purposes. Seasonal dates of imagery were used to determine tree canopy effects and the optimum season for estimating impervious surfaces from satellite imagery. Unique to this study, the derived classified estimates were compared to an impervious surfaces reference estimate developed from high resolution, true color aerial photography. The impervious surfaces reference estimate was developed by digitizing over 15,000 polygons of impervious features throughout the watershed such as roads, buildings, and parking lots. Statistical evaluation of the seasonal classified images included the error matrix analysis, Kappa analysis (both overall and conditional), and the Pair-Wise Z test statistic. Results obtained in this research indicate overall accuracies of the derived classified estimates ranged between 75.33 percent and 81.33 percent while differing from the reference estimate by 10 percent or less.

Files
  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  Johnson_thesis.pdf 3.86 Mb 00:17:51 00:09:11 00:08:02 00:04:01 00:00:20

Browse All Available ETDs by ( Author | Department )

If you have more questions or technical problems, please Contact LSU-ETD Support.