R. Houborg, M.F. McCabe
Remote Sensing of Environment, Volume 186, pp. 105-120, (2016)
LAI, Leaf chlorophyll, REGFLEC, RapidEye, Red-edge, Aerosols, Foliar dust, Adjacency effects, Precision agriculture
A regularized canopy reflectance model (REGFLEC) is applied over a dryland irrigated agricultural system in Saudi Arabia for the purpose of retrieving leaf area index (LAI) and leaf chlorophyll content (Chll). To improve the robustness of the retrieved properties, REGFLEC was modified to 1) correct for aerosol and adjacency effects, 2) consider foliar dust effects on modeled canopy reflectances, 3) include spectral information in the red-edge wavelength region, and 4) exploit empirical LAI estimates in the model inversion. Using multi-spectral RapidEye imagery allowed Chll to be retrieved with a Mean Absolute Deviation (MAD) of 7.9 μg cm− 2 (16%), based upon in-situ measurements conducted in fields of alfalfa, Rhodes grass and maize over the course of a growing season. LAI and Chll compensation effects on canopy reflectance were largely avoided by informing the inversion process with ancillary LAI inputs established empirically on the basis of a statistical machine learning technique. As a result, LAI was reproduced with good accuracy, with an overall MAD of 0.42 m2 m− 2 (12.5%). Results highlighted the considerable challenges associated with the translation of at-sensor radiance observations to surface bidirectional reflectances in dryland environments, where issues such as high aerosol loadings and large spatial gradients in surface reflectance from bright desert soils to dark vegetated fields are often present. Indeed, surface reflectances in the visible bands were reduced by up to 60% after correction for such adjacency effects. In addition, dust deposition on leaves required explicit modification of the reflectance sub-model to account for its influence. By implementing these model refinements, REGFLEC demonstrated its utility for within-field characterization of vegetation conditions over the challenging landscapes typical of dryland agricultural regions, offering a means through which improvements can be made in the management of these globally important systems.