Chlorophyll fluorescence observed by OCO-2 is strongly related to gross primary productivity estimated from flux towers in temperate forests

Solar-induced chlorophyll fluorescence (SIF) opens a new perspective on the monitoring of vegetation photosynthesis from space, and has been recently used to estimate gross primary productivity (GPP). However, previous studies on SIF were mainly based on satellite observations from the Greenhouse Gases Observing Satellite (GOSAT) and Global Ozone Monitoring Experiment-2 (GOME-2), and the evaluation of these coarse-resolution SIF measurements using GPP derived from eddy covariance (EC) flux towers has been hindered by the scale mismatch between satellite and tower footprints. We use new far-red SIF observations from the Orbiting Carbon Observatory-2 (OCO-2) with much finer spatial resolution and GPP data from EC flux towers from 2014 to 2016 to examine the relationship between GPP and SIF for temperate forests. The OCO-2 SIF tracked tower GPP well, and had strong correlation with tower GPP at both retrieval bands (757 nm and 771 nm) and both instantaneous (mid-day) and daily timescales, with the strongest relationship for the 757 nm band (SIF757) at the daily timescale. Daily SIF757 exhibited much stronger correlation with tower GPP compared to the enhanced vegetation index (EVI) and normalized difference vegetation index (NDVI) based on the surface reflectance product derived from the moderate resolution imaging spectroradiometer (MODIS) onboard either Terra or Aqua satellite and had a similarly strong relationship as EVI based on the bidirectional reflectance distribution function (BRDF) corrected reflectance product (MCD43A4) based on data from both Terra and Aqua. Absorbed photosynthetically active radiation (APAR) explained 85% of the variance in SIF757, while the product of APAR and two environmental scalars - fTmin and fVPD (representing minimum temperature stress and water stress) explained higher variance (92%) in SIF757. This suggests that SIF mainly depends on APAR and is also affected by environmental stresses that determine photosynthetic light use efficiency. The BRDF-corrected EVI also has advantages over SIF in that EVI has spatially and temporally continuous coverage and has a much longer record (2000-present). The OCO-2 SIF757 estimated GPP well (R2 = 0.81, p \u3c 0.0001; RMSE = 1.11 gC m− 2 d− 1), and its performance was comparable to or slightly better than that of the MODIS GPP algorithm, a GPP model based on the light use efficiency logic. With intensive OCO-2 observations acquired in the Target mode around Park Falls, a calibration/validation site for OCO-2, we also generated gridded SIF and then estimated GPP on a per-pixel basis at the landscape scale. Our findings demonstrate the strong ability of chlorophyll fluorescence observed by OCO-2 in estimating GPP for temperate forests and reveal its potential and limitations in future ecosystem functioning and carbon cycling studies