Landsat photo-optic data analysis of land cover changes in the Middle Belt region of Nigeria
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The need for a standardized land-use classification scheme and an appropriate framework for acquiring land-use data to aid national development planning is imperative if Nigeria is to develop in a positive direction. Therefore, the usability of the Level-one U.S.G.S. land-cover classification categories with visual interpretation of LANDSAT imagery for acquiring land-use data was tested.
The generalized Level-one U.S.G.S. classification model was used as a basis for inventorying land covers, and both inductive and deductive systems of land classification, based on the conventional polythetic photo interpretation of spectral discrimination, were used to analyze the photographic elements of the acquired images. Landcover patterns were inventoried, the types of land-cover changes were monitored, and the rates and magnitudes of land-use changes were assessed for the Experimental and Control Locations.
Based on the results of in-situ verification of 108 randomly selected polygon samples from the generated landcover maps, interpretation accuracy was determined using the minimum acceptable level of 85% established by Professor J.R. Anderson in 1971. The planimeter measurement accuracy was determined at a stipulated j^5% error margin of the calculated areal sizes of each study location. The land-cover change types were enumerated for each study location; then, using the planimeter measurements acquired from the generated maps for both study years, the magnitudes and rates of land-cover changes were evaluated using the mathematical model employed by Dr. Peter Adeniyi. The statistical significance of the changes that had occurred in each study location between the two study years was determined at the 95% confidence level using the t-test.
The results indicate that the U.S.G.S. land classification model is useable in a developing country. The relatively high interpretation and measurement accuracy percentages obtained imply that the interpretation technique and the measuring device are ideal for acquiring land-use data. Only the urban and farmland cover changes were of statistical significance.