Mapping the heterogeneity of natural and semi-natural landscapes

Natural and semi-natural landscape cover is heterogeneous. Ideally, mapping land cover requires anapproach that represents both gradients and land covers spatiotemporal variability. These aspects canbe visualized and depicted by applying a new spatio-temporal analysis based Landscape HeterogeneityMapping (LaHMa) method to natural and semi-natural landscapes. Using MODIS NDVI 16-day imagery(February 2000-July 2009) for Crete, a 65-cluster image was selected from ISODATA classification resultsusing the separability values of the divergence statistics. The 65 clusters appropriately generalize thespatial and temporal variability in land cover. Using classified outputs from 10 to 65 clusters, the frequency of pixels identified as boundaries of homogeneous land cover classes was translated into theform of a landscape heterogeneity map, which was then validated using field data. The results show thatthe heterogeneity map had moderate correlation (R2= 0.60 and 0.63 in two transects) with the sum ofdifferences between neighbouring transect pixels in all land cover components. In general, the studyfound this new approach (LaHMa) to be suitable for mapping landscape heterogeneity in the natural andsemi-natural landscape of Crete, Greece. The new method appears to be of potential use for informinggradient analyses in landscape ecological studies.