Improvement of spatially continuous forest LAI retrieval by integration of discrete airborne LiDAR and remote sensing multi-angle optical data

Abstract

Forest leaf area index (LAI) is a critical variable in modeling climates and ecosystems, and is required on regional and global scales for models. However, forest LAI has proven to be difficult to obtain. In this study, we sought to improve forest LAI retrieval in a large study area in the Dayekou forest, Gansu province, by combining airborne discrete LiDAR, MODIS, and MISR data. In our retrieval scheme, canopy height is the key parameter, and the canopy height precision is of great importance when estimating LAI. To address this issue, we introduced LiDAR data and combined it with the MODIS and MISR products. First, the canopy height for the LiDAR data coverage was calculated using a local maximum filtering algorithm with a variable window size. Then, a multivariate linear regression model was developed to extrapolate the LiDAR-derived canopy height to the whole study area using the MODIS BRDF/Albedo product. In addition, the bi-directional reflectances from MODIS and MISR were used to invert the geometric-optical mutual-shadowing (GOMS) model structural parameters (nR2, b/R, h/b, Δh/b) of the forest. These structural parameters were then combined with the forest canopy height and field measurements to retrieve the LAI of the continuous forest area at a 500-m resolution. After comparison with the true LAI measured by LAI-2000 combined with TRAC, and by TRAC alone, the highest R2 values of the estimated LAI were 0.73 and 0.69, respectively. The results indicate that the LiDAR canopy height derived from the optical multi-angle remote sensing data can be used to retrieve the large-scale forest LAI when combined with the canopy structure information derived from GOMS model.