A Normalized Spectral Angle Index for Estimating the Probability of Viewing Sunlit Leaves From Satellite Data

The probability of viewing sunlit leaves (PT) is a crucial variable influencing observed canopy spectra. Proper determination of PT is necessary for the quantitative retrieval of vegetation parameters using remote sensing. This article describes a spectral index for estimating PT from satellite-observed canopy spectra. For this purpose, we propose a normalized spectral angle index (NSAI) at near-infrared (NIR) wavelengths, based on the spectral shapes of leaf and soil background. The performance of NSAI in estimating PT was evaluated using one ground-based high-resolution imaging dataset, one synthetic satellite dataset, and one satellite-ground synchronous observation dataset. The results demonstrate that NSAI is more suitable for estimating PT from satellite data than five commonly used spectral indices, including enhanced vegetation index (EVI), normalized difference spectral index (NDSI), normalized difference vegetation index (NDVI), simple ratio (SR) index, and photochemical reflectance index (PRI). NSAI exhibits a significant linear correlation with PT. The empirical model for estimating PT based on NSAI has the best transferability from simulated to in situ satellite data. For the fine spectral–spatial resolution (Hyperion) data, the normalized root-mean-square error (nRMSE) and adjusted $R^{2}$ of estimated PT were 14.9% and 0.744, respectively. For MODIS images, PT was estimated with satisfactory accuracy, with an nRMSE of 18.71% and an adjusted $R^{2}$ of 0.670. NSAI is potentially applicable to satellite images for direct estimation of PT to improve the inversion accuracy of vegetation parameters.