Improving Leaf Area Index Estimation With Chlorophyll Insensitive Multispectral Red-Edge Vegetation Indices

As an essential vegetation biophysical trait that determines the plant's structure and photosynthetic capacity, characterizing of leaf area index (LAI) is important for vegetation growth and health monitoring. The empirical models based on vegetation indices (VIs) from remote sensing images is an effective method for deriving LAI. However, due to the coupled impacts of LAI and leaf chlorophyll content (LCC) on canopy reflectance and saturation effect, most VIs cannot achieve a good accuracy of LAI estimation. The remotely sensed red-edge reflectance can provide valuable information to delineate the LAI, therefore a series of leaf chlorophyll insensitive red-edge VIs by using the Sentinel-2 and GF-6 multispectral images are developed in this work to improve the LAI estimation accuracy. The potentials of reflecting LAI variations and sensitivity to LCC changes for each Sentinel-2 and GF-6 red-edge band are comprehensively analyzed based on the PROSAIL model to select the optimal band in VIs design. The proposed VIs are then evaluated in multiple ways, including with PROSAIL simulated datasets, ground measured LAI with canopy spectra and real satellite images. The evaluation results based on field LAI measurements indicate that the proposed red-edge VIs can effectively improve crop LAI estimation accuracy with the best regression coefficient (R2=0.81 for Sentinel-2 andR2=0.65 for GF-6) among all comparative VIs. Our work showcases incorporating red-edge bands with suitable formula is promising for improving VI-based LAI retrieval, and they offer a practicable solution to fast achieve decameter LAI maps by using the Sentinel-2 or GF-6 images.