Assessment of badland dynamics using multi-temporal Landsat imagery: An example from the Spanish Pre-Pyrenees

This study investigated the use of remote sensing data and GIS techniques in identifying badland areas and erosion risk areas, assessing the magnitude and rate of erosion processes, and estimating the geographical factors that control erosion. The remote sensing information was generated from Landsat TM and Landsat ETM+ images for the period 1984-2006, which were merged with aerial photographs. The study was performed on a 73 km(2) area in a corridor of Eocene marls in the Pre-Pyrenees, an area highly prone to erosion processes. The results show that badland areas occupy approximately 2 km(2) and are widespread in the study area. Various spectral indices were assessed for their performance in revealing badland dynamics: all were very useful for this purpose, and few differences were found among them. From the results we recommend the use of the simplest and most widely used spectral index: the normalized difference vegetation index (NDVI). Further analyses showed that there was a clear difference between the vegetated and degraded areas. The degraded areas showed a dominant negative trend of NDVI values, which is indicative of possible degradation processes and an increase in the surface area affected by erosion. We concluded that the degraded areas, which had poor vegetation cover and extreme environmental conditions, show a spatial pattern related to topographical location, and particularly the presence of north-facing slopes. A map of possible erosion risk was generated as a function of the initial environmental conditions and the evolution of erosion from 1984 to 2006. (C) 2012 Elsevier B.V. All rights reserved.