Accuracy assessment on the number of flux terms needed to estimate in situ fAPAR

The fraction of Absorbed Photosynthetically Active Radiation (fAPAR) is a crucial variable for assessing global carbon balances and currently, there is an urgent need for reference data to validate satellite-derived fAPAR products. However, it is well-known that fAPAR ground measurements are associated with considerable uncertainties. Generally, fAPAR measurements can be carried out with two-, three- and four-flux approaches, depending on the number of flux terms measured. Currently, not much is known about the number of flux terms needed to satisfactorily reduce systematic errors. This study investigates the accuracy of different fAPAR estimates based on permanent, 10-min PAR measurements using Wireless Sensor Networks (WSNs) at three forest sites, located in Central Europe (mixed-coniferous forest), North America (boreal-deciduous forest) and Central America (tropical dry forest). All fAPAR estimates reflect the seasonal course of fAPAR. The highest average biases of different fAPAR estimates account to 0.02 at the temperate, 0.08 at the boreal and -0.05 at the tropical site, respectively, thereby generally fulfilling the uncertainty threshold of a maximum of 10 % or 0.05 fAPAR units set by the Global Climate Observing System (GCOS, 2016). During high wind speed conditions at the boreal site, the bias of the two-flux fAPAR estimate exceeded the 0.05-uncertainty threshold. Three-flux fAPAR estimates were not found to be advantageous, especially at the tropical site. Our findings are beneficial for the development of sampling protocols that are needed to validate global satellite-derived fAPAR products.