ATMOSPHERIC EFFECTS ON SPECTRAL VEGETATION INDICES AND ITS APPLICATION OF FOREST BIOMASS ASSESSMENT FROM REMOTE SENSED DATA
2003
Zoran M. , Zoran L.F. , Dida A.
Thresholding based on biophysical variables derived from time trajectories of satellite data is a new approach to classifying forest land cover via remote sensing at coarse resolutions. This approach is attractive because it is much simpler than conventionalalternatives. Further, it operates on biophysical variables and thus should be morerobust than more data dependent techniques. The input data are composite values of the Normalized Difference Vegetation Index(NDVI). Associated with these values are radiances in three thermal bands that are used to estimate surface temperature. The classification algorithm, accepts mean growing-season NDVI, mean growing-season near-infrared radiance,NDVI amplitude and surface temperature as input parameters for the composite NDVI and surface temperature data. The units recognized are broad life-form vegetation classes, such as evergreen needleleaf forest, evergreen broadleaf forest,shrubs, etc. They are compared to a ground truth map of an area using Thematic Mapper (TM) and SAR ERS-1 imagery. Classification accuracies are variable, depending on the class and the comparison method as well as function of season of the year. Our analysis indicates a potentially application of thresholding techniques to land-cover classification as well as for forest biomass assessment. Keywords:forest biomass assessment, spectral vegetation indices, atmospheric effects, remote sensed data.
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил Forest Research and Management Institute