Using moderate and high spatial resolution imagery to characterize seasonal variation in leaf area in a temperate broadleaf forest.

At the time scale of individual growing seasons, the composition and structural properties of forests are generally static, but leaf area index (LAI), which controls light absorption and primary productivity, varies spatially and seasonally, depending on the leaf type and phenology of trees. The objective of this study is to use multiple sources of data, including airborne, satellite, and field observations, to better understand how remote sensing can be used to characterize seasonal variation in LAI in deciduous broadleaf forests. To do this, we used time series of remotely sensed vegetation indices to estimate variation in LAI across multiple growing seasons at the Harvard Forest long term ecological research site in central Massachusetts. Field measurements of LAI, along with airborne hyperspectral remote sensing data at 1 m resolution collected by the NEON airborne observation platform at the Harvard Forest for 2016-2018, were used to build a model that predicts LAI from hyperspectral vegetation imagery. Based on this model, LAI was mapped for the entire study area for a single date, and then aggregated to 30 m. These data were then used to calibrate a model predicting LAI using data at 30 m spatial resolution from the Harmonized Landsat-Sentinel (HLS) data set. The results of this model were then used to estimate seasonal variation in LAI using time series of HLS data. Accuracy of predicted LAI through time was assessed using field data of LAI collected at 36 ground sampling plots from 2016-2018. Initial results show that the hyperspectral data can be used to estimate LAI at fine spatial resolution, and that the model based on HLS data provides realistic maps of seasonal variation in LAI across the Harvard Forest site. In this framework, HLS imagery provides an excellent source of time-varying information related to phenology, while the hyperspectral data provide fine scale information that supports accurate mapping of LAI at local scale.