Simultaneous Estimation of LAI, PAR, FAPAR, and Surface Albedo at Multiple Spatial Scales From Top-of-Atmosphere Satellite Observations With Different Spatial Resolutions

Current global land-surface parameter products are generally derived from surface reflectance data acquired by a single satellite sensor, which requires atmospheric corrections for top-of-atmosphere (TOA) reflectance data. And the retrieval methods for different parameters are commonly based on different assumptions, as a result of which the estimated parameters are inconsistent in terms of physical detail. Furthermore, the currently available land-surface parameter products only have a limited number of spatial resolutions. These problems existing in the current land-surface parameter products greatly limit their applications in earth sciences. This paper proposes a novel approach to simultaneously estimate leaf area index (LAI), photosynthetically active radiation (PAR), fraction of absorbed photosynthetically active radiation (FAPAR) and surface albedo at multiple spatial scales from satellite TOA reflectance data with different spatial resolutions. Based on the average of multi-year Global Land Surface Satellite (GLASS) LAI product, an Ensemble Multiscale Tree (EnMsT) was constructed to describe the initial conversion relationships among LAI values at different spatial scales. The EnMsT was coupled with a land surface-atmosphere radiative transfer model to simulate TOA reflectance data with different spatial resolutions, which were compared with the corresponding satellite TOA reflectance data to update the LAI values of each node in the EnMsT. Finally, the estimated LAI values were input into the land surface-atmosphere radiative transfer model to calculate PAR, FAPAR and surface albedo values at different spatial resolutions. In this paper, the method was applied to estimate LAI, PAR, FAPAR and surface albedo at 28.9, 57.8, 115.6, 231.2, 462.4, and 924.8 m from Thematic Mapper (TM) or Enhanced Thematic Mapper Plus (ETM+) and Moderate Resolution Imaging Spectroradiometer (MODIS) TOA reflectance data at four sites with cropland and grassland biomes. The retrieved parameter values were compared with the corresponding MODIS, GLASS, the second version of the Geoland2 (GEOV2) products, and available LAI and FAPAR ground measurements at these sites. The results demonstrate that the retrieved parameter values agree with the corresponding parameter products and the retrieved LAI and FAPAR values at different resolutions are in good agreement with the aggregated LAI and FAPAR values from the high-resolution reference maps.