Abstract

Deforestation has been practiced for millennia in Ethiopia. The power of GIS and remote sensing was combined with SPSS to investigate the effect of physical factors on the deforestation of key species. Landsat images of (1972 and 2013) were obtained and Iimagine Subpixel classifier was used to detect species and estimate their respective area in the respective years. Logistic regression in SPSS was used to identify the deforestation deriving physical factors. Five percent of the total pixels (100,000 random points) were sampled and taken as permanent plots for change detection. Cover status were determined using the extract values to points function in Arcmap10 by overlaying these random points on the classified images. Elevation, aspect, altitude, and slope were derived from Aster DEM, and distance from settlement, roads were digitized from Topo map at scale of 50,000. For each random point response value was determined using the 1972 and 2013 classified images. The accuracy of the results of Subpixel classifications were assessed using 200 points, 100 for each species This yielded 86% accuracy level with 5.5% and 8.5% commission and omissions respectively. All physical factors except aspect entered in SPSS for deforestation prediction were significant and have negative correlation coefficient indicating higher deforestation probability as the value of the physical factors decreases. The influence of slope seems to be constant with unit increment of its value. However, using the value of the coefficients; slope, distance from road, elevation, distance from deforested/forested edge, and distance from settlement area were found to influence the spatial distribution of deforestation in decreasing order. While creating ‘social fencing’ mentality is demanded, the current management system of this forest focuses on guarding and command posts along the main roads. Therefore, considering the influence of the physical characteristics of the area is very important for the effectiveness this management system.

Authors and Affiliations