Vegetation index differencing and linear regression for change detection in a Swedish mountain range using Landsat TM® and ETM+® imagery

This paper describes the extent to which the normalized difference vegetation index (NDVI) in combination with image regression used on satellite data can indicate vegetation cover decreases caused by increased exploitation of Swedish mountains. The methods outlined in this study give a basis for detection of less sustainable mountain ecosystems by using as an indicator bare patches of humus or soil where none existed previously. Landsat 5 TM® data from 1984 and 1994 and Landsat 7 ETM+® data from the year 2000 were used in the study. The results show that the NDVI significantly separates areas with vegetation cover decrease from areas with no vegetation cover decrease in sensitive, high‐latitude mountain ecosystems, such as mountainous dry heath communities.