Phenological variation in vegetation indices, leaf area index, and fraction of absorbed photosynthetically active radiation from Harmonized Landsat 8 and Sentinel-2 Data

Abstract:

Leaf area index (LAI) and the fraction of absorbed photosynthetically active radiation (fAPAR) are essential parameters in many ecological and earth system models. However, these variables often have large uncertainties that propagate through to modeled water, energy, and carbon budgets. In this research, we evaluated the biophysical relationship between forest canopy phenology measured via remotely sensed vegetation indices and phenology measured via in-situ measurements of leaf area and fAPAR at the Harvard Forest, a mixed forest dominated by deciduous broadleaf species located in central Massachusetts. Using the Harmonized Landsat Sentinel-2 (HLS) dataset, we developed and evaluated several semi-empirical methods to estimate LAI and fAPAR from HLS-derived vegetation indices. As part of this analysis, we used a two-stream radiative transfer model to explore how leaf optics, leaf area index, leaf angle distribution, and the angle of incident radiation individually and jointly affect vegetation indices, thereby impacting remotely sensed LAI and fAPAR estimates, over the course of the growing season. The phenology of the vegetation indices was dominated by changes in near infrared (NIR) reflectance, which is primarily controlled by leaf area index, but with important secondary controls imposed by changes in leaf angle distribution, leaf optical properties, and solar geometry. Using a simple semi-empirical approached based on a modified version of Beer’s Law, we found that inclusion of the solar zenith angle improved prediction of LAI and fAPAR. This result is consistent with two-stream model results, which show that independent of changes in LAI, solar zenith angle is the main contributor to changes in NIR reflectance, which therefore influence remotely sensed estimates of LAI and fAPAR.