Comparison of DEM-derived determinants for modelling of long-term land cover change in a large scale: case studies from Slovak Western Carpathians

Abstract

Studies detecting land use/cover change (LUCC) in large scales are increasing in number, and so are the studies identifying spatial determinants of these changes and creating their models. Raster datasets derived from digital elevation models (DEM) belong to a limited group of determinants that are relatively available for LUCC modelling in large scales. This study compares the performance of 12 DEM-derived determinants in models of six distinct land cover changes: urbanisation, industrialisation, agricultural intensification and extensification, afforestation, and deforestation. The changes were identified in the 1949-2010 period in a reference scale of 1:10 000 on a total area of 176 km2 of 12 municipalities systematically selected to partially represent Western Carpathians in Slovakia. Nearly 45% of the area changed; afforestation, agricultural extensification and intensification were the most prevalent changes. Logistic regression and hierarchical partitioning were used to quantify the influence of the determinants on them. Among other commonly used determinants (elevation, slope, cost distance), vertical dissection and duration of solar radiation had an unexpectedly high influence, mostly on agricultural intensity and forest changes. However, further research is needed to verify these influences in other areas and to provide their sufficient causal interpretation.