Scaling properties of NDVI and their relationship to land-cover spatial variability

A major issue in remote sensing involves the relationship between sensor resolution and the scales of variability of scene elements. The authors present four methods for understanding the relationship between vegetation and TM-based NDVI across a range of spatial scales for a rugged mountainous environment. They analyze spatially degraded NDVI and vegetation data by a) determining the partitioning of total image variance into different landscape scales and b) quantifying the ability of classification to absorb total NDVI-image variance across scales. For the given definition of vegetation, and within the the range of scales analyzed, 90 m emerges as a fundamental scale of variability in the landscape. This scale may provide the most parsimonious resolution for linking vegetation type to NDVI in this particular landscape