Propagation of shadow effects on typical remote sensing applications in forestry

Airborne LiDAR and hyperspectral data were acquired over a broadleaved forest area in Belgium. Shadow fractions were calculated, based on Sun angles and a digital surface model derived from the LiDAR data. Pixels in the hyper-spectral image were classified based on the shadow fractions to study the effect of shadow on canopy reflectance and how the effect propagated to typical remote sensing applications in forestry. As a first application, the photosynthetical reflectance index (PRI) was studied, which expresses the relative down-regulation of photosynthesis. A strong correlation (R2 = 0.93) was found between the shadow fraction and the PRI. The second application was a tree species classification problem. A measure for classification uncertainty (CU) was introduced, based on the Shannon entropy. It was shown that the majority of pixels with a low shadow fraction were classified with a lower uncertainty.