Monitoring tropical vegetation succession with LANDSAT data

The shadowing problem, which is endemic to the use of LANDSAT in tropical areas, and the ability to model changes over space and through time are problems to be addressed when monitoring tropical vegetation succession. Application of a trend surface analysis model to major land cover classes in a mountainous region of the Phillipines shows that the spatial modeling of radiance values can provide a useful approach to tropical rain forest succession monitoring. Results indicate shadowing effects may be due primarily to local variations in the spectral responses. These variations can be compensated for through the decomposition of the spatial variation in both elevation and MSS data. Using the model to estimate both elevation and spectral terrain surface as a posteriori inputs in the classification process leads to improved classification accuracy for vegetation of cover of this type. Spatial patterns depicted by the MSS data reflect the measurement of responses to spatial processes acting at several scales.