TROPICAL FOREST AREA MEASURED FROM GLOBAL LAND-COVER CLASSIFICATIONS - INVERSE CALIBRATION MODELS BASED ON SPATIAL TEXTURES

Retrieving area estimates from broad scale land-cover maps is generall y inaccurate due to the effect of spatial aggregation on class proport ions. In a previous study, we tested a method to calibrate area estima tes of tropical forest cover by inverting a model of the influence of the forest spatial fragmentation on the spatial aggregation bias, as c haracterized by two nested regression models. This was based on a samp le of high resolution land-cover classifications, distributed across t he tropical belt. In this study, improvements of this previous model a re sought, first, by better accounting for the spatial variability of landscape characteristics using texture measures and, second, by integ rating spatial information in the mixed pixel estimator-that is, the m odeling of spectral mixtures at the scale of coarse resolution pixels as a function of the proportion of land-cover types. These improvement s were tested using NOAA's Advanced Very High Resolution Radiometer da ta at 1.1 km resolution, Landsat Thematic Mapper-based classifications , and data simulated at the 250 m resolution of the forthcoming Earth Observing System's Moderate Resolution Imaging Spectroradiometer (MODI S). The integration of spatial information into a correction model to retrieve fine resolution cover-type proportions from coarse resolution data can improve by up to 35% the reliability of the estimates. The r esults also demonstrate that the integration of spatial information in the mixed pixel estimator controls for the variability due to differe nt landscape characteristics. This study improves our capability to es timate tropical forest cover from coarse resolution remote sensing dat a at a global scale. (C) Elsevier Science Inc., 1997