A data driven approach towards addressing satellite geometry dependent variations in solar-induced fluorescence (SIF)

While solar-induced fluorescence (SIF) shows promise as a remotely sensed measurement directly related to photosynthesis, interpretation remains a challenge. SIF is largely driven by the fraction of absorbed photosynthetically-active radiative at the canopy level that depends upon illumination geometry. A SIF measurement from any platform also depends upon the viewing geometry. To decipher SIF variations related to changes in photosynthesis therefore requires disentangling sun-sensor geometrical effects from variations related to plant physiology. Several approaches to estimate the effects of sun-sensor geometry on satellite-based SIF have been proposed, and some have been implemented; however, no standard method has yet emerged for global observations. Here, we examine in detail how current SIF measurements from satellites in low Earth orbit vary systematically due to orbital features. We directly compare SIF measurements from different satellites at similar times of day and at similar observational geometries. Specifically, we compare SIF retrievals from the Global Ozone Mapping Experiment 2 (GOME-2) on the MetOp-B platform with those from the TROPOspheric Monitoring Instrument (TROPOMI) on the Sentinel 5 Precursor satellite at high northern latitudes in summer where overlap is possible. We examine whether observed sun-sensor SIF dependences agree with those computed using the Soil-Canopy Observation Photosynthesis and Energy fluxes (SCOPE) canopy radiative transfer model. We then explore potential advantages and short-comings of machine learning methods, in particular artificial neural networks trained with reflectance data, to predict SIF including the sun-sunsor geometry dependence in order to help interpret satellite-based SIF data sets.