Estimating PAR absorbed by vegetation from bidirectional reflectance measurements

Satellite remote sensing allows estimation, at a global scale, of the photosynthetically active radiation absorbed (APAR) by the vegetation. Current estimates are based on retrieving the fraction (fAPAR) of PAR absorbed by the canopy from spectral vegetation indices (SVI) derived from combinations of spectral reflectance measurements. We show that currently used SVI are strongly affected by the soil reflectance as well as the sun/view geometry, which yields large uncertainties on the SVI-fAPAR statistical relationships. However, the errors can be reduced using optimal geometries and new SVI less affected by the soil reflectance. Such an approach yields an improvement of the fAPAR assessment down to less than 5% of relative standard error. It is relevant for monitoring of the interannual variations of the net primary productivity (NPP) and its analysis to understand the climate sensitivity of the present ecosystems. These will permit more detailed studies of seasonal and regional biospheric carbon sources and sinks that affect the global distribution of atmospheric carbon dioxide.