ABSTRACT
Most methods for analyzing remote sensing to gain ecological insight about forested landscapes were developed in temperate, not tropical, forests. Until recently the image processing methods used to quantify ecologically meaningful, spatial, and temporal spectral variation across forested landscapes were not appropriate for mature, undisturbed tropical forests. For example, geometric-optical models that quantify 178variation in shadows and sunlit vegetation are ideally suited for conifer forests [1–3]. In general, tropical forests have much flatter crowns than conifers, reducing the dramatic difference in sunlit and shaded crowns that is apparent in conifer forests. Spectral mixture analysis and vegetation indices are used to detect differences in soil and green vegetation found in open forests, woodlands, and savanna [4–6]. Most humid and wet mature tropical forests have continuous vegetation coverage such that the soil is not exposed except for a brief time after gap-forming events. Leaf area indices (LAIs) in mature tropical forests are generally above the threshold where vegetation indices, such as the normalized difference vegetation index (NDVI), are responsive to leaf density changes [7]. The aforementioned techniques have been used extensively to classify the early regeneration stages of tropical forests, but, infrequently to analyze mature tropical forests.
