Integrating remote sensing and wavelet analysis for studying fine-scaled vegetation spatial variation among three different ecosystems.

摘要

This study investigated the optimum pixel size for detecting vegetation spatial patterns in a complex environment where three different ecosystems (grassland, shrubland, and forest) are present. Wavelet analysis indicated that ground Leaf Area Index (LAI) along the study transect is significant at a scale of 8 m. Following the sampling theorem, 2 m would be an optimum pixel size to detect potentially important vegetation patterns (possible sensors include QuickBird 2.4 m and Ikonos 3.2 m). A QuickBird image was then acquired to evaluate its suitability to identify vegetation patterns for the study area. The LAI map derived from QuickBird WDRVI (Wide Dynamic Range Vegetation Index) showed distinct vegetation patterns among three ecosystems. These results confirmed that the dominant spatial scale of ground vegetation biophysical properties (e.g., LAI) can aid in the selection of appropriate image resolution for monitoring multiple ecosystems.