Correcting Crown-Level Clumping Effect for Improving Leaf Area Index Retrieval From Large-Footprint LiDAR: A Study Based on the Simulated Waveform and GLAS Data

The demand for leaf area index (LAI) retrieval from spaceborne full-waveform LiDAR increases due to its direct sampling of the three-dimensional forest structure at a near-global scale. However, the nonrandomness (i.e., clumping effect) of canopy composition limits the reliability of LAI derived from two common methods. They either assume a homogeneous scene in the footprint or just correct for the large gaps-induced between-crown clumping. The clumping in the crown is still an unaddressed issue. We proposed a method to compensate occlusion (i.e., lower canopy layers are occluded by the upper canopy in the process of LiDAR measurement), through which the vertical canopy profile can be resolved from the waveform. Further, we developed a method of deriving relative path length distribution that can reflect the heterogeneity of the canopy from the occlusion-corrected waveform. In addition to correcting the between-crown clumping, we corrected the within-crown clumping further using the derived relative path length distribution, based on path length distribution (PATH) theory. We used simulated waveform data with known LAI and GLAS data with corresponding field-measured LAI to test the performance of our and the other two common LAI retrieval methods. Results show that the errors of our approach are the lowest (with an error generally below 10% and the maximum error below 20%, compared with up to 69% and 47% for the other two methods), and it is relatively stable in various scenes. This study demonstrated the potential of improving LAI retrieval through full utilization of full-waveform data.

Automated summary

This study introduced a method to improve LAI retrieval by addressing occlusion in LiDAR measurements, resulting in lower errors compared to other methods and relative stability across various scenes.