Intercomparison of fraction of absorbed photosynthetically active radiation products derived from satellite data over Europe

- Boston FAPAR product estimates exceed those by other products over boreal biomes.
- RTM assumptions (structure, optical properties) largely influence FAPAR retrievals.
- Independent estimation methods provide more consistent uncertainty representation.
- All three FAPAR products show agreement in seasonal and interannual variations.

The Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) is recognized as an essential climate variable (ECVs), playing a critical role in the estimation of the global energy and carbon balance. With multiple space-borne remote sensing FAPAR global products available from several sources the need for continual comparison and validation has become imperative. In this study, the performance of three global FAPAR algorithms (JRC-TIP, ESA/JRC MGVI and Boston University FAPAR) was evaluated over Europe for the year 2011. Results show an overall agreement among FAPAR products on sites having high and low FAPAR values, except for the north-eastern region of Europe characterized by boreal forest and the transition region with tundra biomes, where the Boston product exceeds values in other products by up to 0.5. Differences in FAPAR estimates over forest biomes suggest that assumptions on structure and optical properties of land surfaces in the different radiative transfer models play an important role in remote-sensing-derived FAPAR products. Uncertainty assessments were carried out using both quality indicators as proposed by the individual product teams as well as independent theoretical uncertainty estimates obtained with the triple collocation error model. The former revealed consistent spatial patterns but large differences in magnitudes (up to 0.1) with systematically lower uncertainties for the Boston product. The latter instead suggests similar uncertainty ranges among the three products. Finally, a comparison with ground estimates for the 2009-2011 period over four European flux tower sites showed consistent, plausible seasonal variations of remote-sensing-derived FAPAR products. Findings suggest that differences in absolute values and inconsistency in uncertainty representation among FAPAR products are still considerable. Standardization frameworks quantifying the impact of different radiative transfer formulations on the estimation of biophysical variables, independent uncertainty estimation methods and well-defined ground measurement protocols need to be put in place before FAPAR products can be reliably fed into existing biogeochemical process models.