Experimental assessment of the Sentinel-2 band setting for RTM-based LAI retrieval of sugar beet and maize

The present work aimed at testing the potential of the upcoming Earth Observation satellite Sentinel-2 (European Global Monitoring for Environment and Security (GMES) programme) for the operational estimation of the leaf area index (LAI) of two contrasting agricultural crops (sugar beet and maize). Mapping of LAI was achieved using a look-up table (LUT) based inversion of a physically based radiative transfer model (SAILH + PROSPECT). In addition to the Sentinel-2 spectral sampling, another band set described as ideal for vegetation studies has been evaluated in a comparative way. Analyses were mainly carried out using hyperspectral data acquired by the optical airborne instrument Compact Airborne Spectrographic Imager (CASI) during the European Space Agency (ESA) AgriSAR 2006 campaign. Additionally, data from two other experiments were tested to extend the validation database. Alternative inversion methods, i.e., an iterative optimization technique (SQP) and a neural network (NN), have been evaluated for comparison purposes. The GMES defined precision of 10% for LAI estimation, evaluated with in situ LAI measurements, was met for sugar beet (8%-9%) but not for maize (16%-22%). The inversion approach and band setting had only a minor influence on the retrieval accuracy, with the only exception being the iterative optimization technique, which failed to give reliable results. The results demonstrate the importance of using an appropriate radiative transfer model for each crop. For row crops with strong leaf clumping and not completely covering the soil surface, such as maize in the early stage, the standard SAILH + PROSPECT model does not appear to be suitable.