Chlorophyll Fluorescence Is a Better Proxy for Sunlit Leaf Than Total Canopy Photosynthesis

Abstract:

Chlorophyll fluorescence (CF) results from non-photochemical quenching during plant photosynthesis under excessive radiation. We explore the relationship between gross primary productivity (GPP) and CF using a process ecosystem model, which separates a vegetation canopy into sunlit and shaded leaf groups and simulates the total canopy GPP as the sum of sunlit and shaded leaf GPP. Using GOME-2 and GOSAT data acquired in 2010 over the global land surface, we found that measured CF signals gridded in 1 degree resolution are well correlated with simulated total GPP and its sunlit and shaded components, but the correlation coefficients (R) are largest for the sunlit GPP and smallest for shaded GPP. The seasonal R² values vary from 0.57 to 0.74, 0.58 to 0.71, and 0.48 to 0.56 for sunlit, total and shaded GPP, respectively. The significance levels for these correlations are all greater than p<0.01. Averaged over the globe, the total simulated shaded GPP is 39% of the total GPP. Theoretically, CF from vegetation comes mostly from sunlit leaves. The significant correlation between measured canopy-level CF and the shaded GPP is likely due to the correlation between shaded and sunlit GPP as both increase with leaf area index. Our simulation confirms the validity of using canopy-level CF measurements to assess the total GPP as the first approximation, although these measurements are a consistently better indicator of sunlit GPP than total GPP. In previous studies, the R² values for the correlation between CF and total GPP were found to range from 0.76 to 0.88, 0.56 to 0.78, and 0.57 to 0.77 for MPI-BGC, MODIS and CASA model results, respectively. These values are similar or larger than those for sunlit GPP simulated in our study, but are considerably larger than those for total GPP in our study because the correlation for total GPP is contaminated by the inclusion of shaded GPP. All these three models use canopy total light use efficiency without considering the differences between sunlit and shaded leaves, and therefore they mostly capture spatio-temporal variations in sunlit GPP. We therefore argue that solar-induced CF measured from vegetation is a better proxy of sunlit GPP than the total GPP, and the use of CF data for assessing the terrestrial carbon cycle can be improved when sunlit and shaded GPP are modelled separately.