Forest canopy structure and reflectance are fundamentally coupled, which allows for the measurement of vegetation structural indices with remote sensing. Landsat spectral indices provides one-dimensional measurements of forest canopies, whereas LIDAR measures multidimensional structure, including the complex arrangement of vegetation within the canopy (canopy structural complexity). Canopy structural complexity may affect canopy reflectance through its influence on light distribution, so that Landsat spectral indices will be correlated with LIDAR metrics of canopy structural complexity. Terrestrial LIDAR is restricted to a spatial extent of sub-hectares, but Landsat is globally available; thus linking Landsat to terrestrial LIDAR estimates of multidimensional canopy structure could increase the spatial scale at which these measurements can be made. Here, we examined associations between Landsat spectral indices and terrestrial LIDAR measurements of canopy structural complexity. Canopy structural complexity measurements were obtained from plots within eight NEON forested sites across eastern North America, while spectral indices (NDVI, EVI, tasseled cap indices) were calculated for the corresponding locations from Landsat 8 satellite imagery.
Results/Conclusions
Results showed that canopy greenness and brightness are correlated with spatial variation in several categories of canopy structural complexity. We found that greener canopies were associated with a taller canopy, greater leaf area density and variability, and a more fully covered and less porous canopy. Among greenness indices, NDVI explained the largest fraction of variation in canopy structural complexity metrics (adj. R2 = 0.52 – 0.62 for six metrics), and robustly predicted six metrics with linear models. Additionally, we found that a brighter canopy was associated with greater leaf area density and variability, canopy cover, porosity, and lower leaf clumping. These results demonstrate the potential for the future estimation of canopy structural complexity at sub-continental scales using satellite imagery, and could greatly expand the scale at which these metrics can be used in environmental science.