A red-edge spectral index for remote sensing estimation of green LAI over agroecosystems

Leaf area index (LAI) is a key biophysical parameter for the monitoring of agroecosystems. Conventional two-band vegetation indices based on red and near-infrared relationships such as the normalized difference vegetation index (NDVI) are well known to suffer from saturation at moderate-to-high LAI values (3–5). To bypass this saturation effect, in this work a robust alternative has been proposed for the estimation of green LAI over a wide variety of crop types. By using data from European Space Agency (ESA) campaigns SPARC 2003 and 2004 (Barrax, Spain) experimental LAI values over 9 different crop types have been collected while at the same time spaceborne imagery have been acquired using the hyperspectral CHRIS (Compact High Resolution Imaging Spectrometer) sensor onboard PROBA (Project for On-Board Autonomy) satellite. This extensive dataset allowed us to evaluate the optimal band combination through spectral indices based on normalized differences. The best linear correlation against the experimental LAI dataset was obtained by combining the 674nm and 712nm wavebands. These wavelengths correspond to the maximal chlorophyll absorption and the red-edge position region, respectively, and are known to be sensitive to the physiological status of the plant. Contrary to the NDVI (r2: 0.68), the red-edge NDI correlated strongly (r2: 0.82) with LAI without saturating at larger values. The index has been subsequently validated against field data from the 2009 SEN3EXP campaign (Barrax, Spain) that again spanned a wide variety of crop types. A linear relationship over the full LAI range was confirmed and the regression equation was applied to a CHRIS/PROBA image acquired during the same campaign. A LAI map has been derived with an RMSE accuracy of 0.6. It is concluded that the red-edge spectral index is a powerful alternative for LAI estimation and may provide valuable information for precision agriculture, e.g. when applied to high spatial resolution imagery.