Evaluation of global leaf area index and fraction of absorbed photosynthetically active radiation products over North America using Copernicus Ground Based Observations for Validation data

Abstract With a growing number of Earth observation (EO) products available through operational programmes such as the European Union’s Copernicus, there is increasing emphasis on product accuracy and uncertainty, necessitating evaluation against in situ reference measurements. Whilst existing reference datasets have proven a valuable resource, they incorporate little data with which products from recent EO instruments can be assessed. A reliance on individual field campaigns has also led to several inconsistencies, whilst limiting the extent to which temporal variations in EO product performance can be captured. Recently established environmental monitoring networks such as the National Ecological Observatory Network (NEON), which collect routine in situ measurements using standardised instruments and protocols, provide a promising opportunity in this respect. The Copernicus Ground Based Observations for Validation (GBOV) service was initiated in recognition of this fact. In the first component of the project, raw observations from existing networks have been collected and processed to provide reference data for a range of EO land products. In this study, we focus on leaf area index (LAI) and the fraction of absorbed photosynthetically active radiation (FAPAR). Raw digital hemispherical photography (DHP) from twenty NEON sites was processed to derive in situ reference measurements, which were then upscaled to provide high spatial resolution reference maps. Using these data, we assess the recently released Copernicus Global Land Service (CGLS) 300 m Version 1 (V1) products derived from PROBA-V, in addition to existing products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Radiometer Suite (VIIRS). When evaluated against reference data, the CGLS 300 m V1 product demonstrated the best agreement (RMSD = 0.57 for LAI and 0.08 for FAPAR), followed by the Collection 6 VNP15A2H and MOD15A2H products (RMSD = 0.81 to 0.89 for LAI and 0.12 for FAPAR). Differing assumptions of the products and in situ reference measurements, which cause them to be sensitive to slightly different quantities, are thought to explain apparent biases over sparse vegetation and forest environments. To ensure their continued utility, future work should focus on updating the GBOV in situ reference measurements, implementing additional corrections, and improving their geographical representativeness.