Theses / Dissertations >
College of Engineering >
Please use this identifier to cite or link to this item:
|Title: ||Comparison of Change Detection Techniques in the Ejisu-Juaben District|
|Authors: ||Adjapong, Adwoa Sarpong|
|Issue Date: ||11-Aug-2009|
|Abstract: ||Land cover/use change information has become important in the life of every nation due to its negative and positive effects. To be able to account for change a technique, which is convenient for the landscape in question has to be used to do the change analysis.
The main objective of this study was to compare various change detection techniques for two dates and find out the optimum one suitable for the chosen study area, which is the Ejisu - Juaben district of the Ashanti region. The methods compared were image differencing, tasseled cap, principal component analysis (PCA), change vector analysis, normalized difference vegetative index (NDVI) and post classification. 1986 TM and 2007 ETM+ images were used for the analysis.
The images underwent geometric and radiometric corrections before the data were processed to reveal change using the various techniques. The various techniques each showed its ability to depict change over the chosen duration. After assessing their accuracies using the error matrix NDVI came out to be the best with a Kappa of 0.5139. A land cover/use projection for 2020 was also attempted using Markov chain and cellular automata. This produced a land cover/use map showing the changes that might occur in the year 2020 if things continue the way they are without any monitoring and management of the various land covers.
Keywords: Change detection, Land covers/use, post classification, Markov chain.|
|Description: ||A Thesis submitted to the department of Geomatic Engineering, Kwame Nkrumah Uarversity of Science and Technology in partial, fulfilment of the requirements for tfce degree of Master of Science in Geomatic Engineering, 2009|
|Appears in Collections:||College of Engineering|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.