MODEL INVERSION OF SATELLITE-MEASURED REFLECTANCES FOR OBTAINING SURFACE BIOPHYSICAL AND BIDIRECTIONAL REFLECTANCE CHARACTERISTICS OF GRASSLAND

An inversion method that can be used to derive information about bioph ysical and radiative conditions of a grassland surface from remotely s ensed surface spectral reflectances is described. A previously develop ed simple surface bidirectional reflectance model was expanded for use in this inversion process. In the inversion process, a multidimension al optimization scheme is used to obtain a set of model parameters tha t minimizes the difference between modeled and observed reflectances. The effect of the slumber of data points used in the inversion on the accuracy of the model inversion was investigated with ground based hem ispheric scan measurements of surface spectral reflectances over a tal lgrass prairie in Kansas. The result shows that the total inversion er ror consists of minimum or systematic error and ambiguity error The fo rmer result from uncertainties in the accuracies of the model and meas urements, whereas the latter is caused by the ambiguity of the model i nversion when a limited number of data points are used The systematic error is independent of the number of data points, whereas the ambigui ty error decreases sharply with increasing data points when the num be r of data points is small (fewer than about seven). An inversion metho d with appropriate data grouping applied to the reflectance data from the Advanced Very High Resolution Radiometer is shown to improve estim ates of surface parameters over the grassland site. Relative differenc es between measured surface bidirectional reflectances and those calcu lated with the derived surface parameters range from 6% to 15% (12.9% for band 1 [0.65-0.67 mu m] 6.9% for band 2 [0.81-0.84 mu m] and 7.6% for band 3 [1.62-1.69 mu m]). (C)Elsevier Science Inc., 1997.