Using the Landscape Evolution and Forecasting Toolbox for Global On-Demand Mapping of Surface Biophysical Products derived from Sentinel 2 and Landsat 8 data

International panels have identified the need for global mapping of a range of surface biophysical products at medium (<1ha) resolution. Algorithms, such as the Simplified Level 2 Prototype Processor (SL2), exist for deriving products such as albedo, fAPAR and LAI from analysis ready satellite data records from Sentinel 2 (S2) and Landsat 8 (L8) imagers but require implementation in a manner that facilitates free, open and efficient product generation and evaluation of spatiotemporal coverage and thematic uncertainty at global scale.

The Landscape Evolution and Forecasting (LEAF) toolbox, implemented in Google Earth Engine https://github.com/rfernand387/LEAF-Toolbox, is used to produce a global assessment of monthly coverage during the 2019 growing season of both S2 and L8 products and to intercompare products with corresponding MODIS Land products at 500m resolution. Medium resolution LAI products are also directly validated using in-situ data from 12 North American sites including the National Ecological Observatary Network and Canada Centre for Remote Sensing validation network. The probability of meeting temporal requirements for either monthly or 10d coverage based on using one or both of S2 and L8 products is globally assessed. The agreement between derived products and reference MODIS products is reported as a function of biome, land cover and MODIS product estimates. Validation results for LAI are reported in terms of scatter plots and summary statistics for homogenous canopies and forests as a function of canopy clumping. Conclusions are drawn regarding the potential for S2 and L8 data to satisfy spatiotemporal coverage requirements and for using SL2P with these data to meet thematic requirements. The LEAF Toolbox is recommended to facilitate current product generation and for implementation of revised retrieval algorithms customized based on local calibration data.