Large-scale retrieval of leaf area index and vertical foliage profile fromthe spaceborne waveform lidar (GLAS/ICESat)

摘要

Leaf area index (LAI) and canopy vertical profiles are important descriptors of ecosystem structure. The Geoscience Laser Altimeter System (GLAS) on board ICESat (Ice, Cloud, and land Elevation Satellite) provided three-dimensional observations that can be used to derive these canopy structure parameters globally. While several canopy height products have been produced globally fromGLAS, no comparable data sets for LAI and canopy profiles exist across large areas. In this studywe develop a physically basedmethod of retrieving LAI and vertical foliage profiles (VFPs) from GLAS observations over the entire state of California, USA. This method refines lidar derived LAI and VFP through a recursive analysis of GLAS waveforms using ancillary data obtained from existing remote sensing products. Those supplemental inputs include canopy clumping index derived from POLDER, 500 m land cover type and 1 km LAI data derived from MODIS. Implementation of our method created state-level LAI and VFP data for the existing GLAS transects over California.Wethen analyzed the variability of LAI and VFP data sets across environmental gradients and as a function of land cover type and elevation. Both LAI and VFP showed strong variability across elevational gradients and among land cover types.We compared our results at the scale of GLAS footprints with an LAI map derived from Landsat (at 30 m) and found moderate agreement (r~2 = 0.34, bias= 0.26, RMSD (Root Mean Square Difference) = 1.85) between the two. In particular, Landsat LAI not only appeared to saturate relative to GLAS LAI at around LAI = 5, but also showed an overestimation for LAI less than about 2. Best agreement between the two LAI data sets was shown to occur in areas with slope less than 20°. Results from our study suggest the possibility of retrieving global LAI and VFP data from GLAS data and the potential for synergetic observation of lidar and passive optical remote sensing data such as Landsat.