Skip to content Skip to navigation
García M, Gajardo J, Riaño D, Zhao K, Martín P, Ustin S. 2015. Canopy clumping appraisal using terrestrial and airborne laser scanning. Remote Sensing of Environment 161:78-88. doi:
Year Published: 2015

Accurate spatial information of canopy clumping degree (Ω) contributes to better understanding of the light regime within the canopy and the physiological processes associated with it. This paper evaluates the potential of terrestrial (TLS) and airborne laser scanning (ALS) to estimate Ω in different vegetation types after converting the point cloud into a 3-dimensional (3D) voxel-based model. Three methods are presented based on the spatial distribution of the returns (Standardized Morisita's Index — SMI); the gap distribution (Pielou's coefficient of segregation — PCS) and the gap size distribution (Chen & Cihlar's clumping index — CCI). Compared to Ω values derived from hemispherical photographs (HPs), the CCI method outperformed PCS and SMI for both instruments, with a correlation value of 0.93 (vs. 0.79 — PCS and 0.65 — SMI) for oak trees using TLS; 0.83 (vs. 0.78 — PCS and 0.73 — SMI) for a shrub chaparral using ALS data; and 0.84 (vs. 0.81 — PCS and 0.50 — SMI) for a mixed Mediterranean forest using ALS data. Voxel size was an important parameter to estimate Ω showing statistically significant differences for the different resolutions tested. Voxel size had an opposite effect on SMI than that on PCS and CCI, with SMI providing better results for coarser voxel sizes, and PCS and CCI yielding higher accuracies for finer voxels. In the case of the TLS, the influence of the zenith angle was also evaluated by means of a Kruskal–Wallis test. CCI and PCS did not show significant differences among the zenith angles tested, but SMI did. The radius of the plot used to analyze ALS data significantly affected the correlations with HP, with the best results found at 13, 7 and 15 m for mixed Mediterranean forest and at 11, 10 and 5 m for shrubs for CCI, PCS and SMI, respectively. The methods presented have the potential to be operationally applied to other areas using TLS and ALS data, since they are not based on an empirical fit but on the analysis of the gap size in the canopy and the distribution of returns after voxelization.  link to publication