Assessing and Modeling Landscape Change in a Sensitive High-Elevation Region of the Bolivian Andes

Abstract

This study used remotely sensed land cover and topographic data, maximum likelihood classification, and spectral mixing analysis to characterize current landscape patterns and quantify land cover change from 1985 to 2003 in the Southeastern Bolivian Andes. Current land cover was mapped into 9 classes with an overall accuracy of 89%. The change analysis demonstrated significant gains in bare and cultivated land (4.4% and 4.1%, respectively) at the expense of forest and pasture (losses of 4.8% and 3.9%, respectively). Spectral mixture analysis indicated that communal rangeland degradation (as measured by changes in proportions of green vegetation, non-photosynthetic vegetation and bare soil on the landscape) may have occurred, especially where conversion of land to more productive uses is restricted by soil fertility, topography, and climate. The study demonstrated that remotely sensed data and traditional image analysis techniques can be used to characterize land cover and land cover change in remote, mountainous areas.