Improving the potential of red SIF for estimating GPP by downscaling from the canopy level to the photosystem level

摘要

Numerous studies have proved that solar-induced chlorophyll fluorescence (SIF) is a good proxy of gross primary production (GPP). The SIF spectrum covers the spectral window from about 640 nm to 850 nm. Due to the strong chlorophyll absorption effect at the red band, the red SIF is much more influenced by the radiative transfer effect inside the canopy than the near infrared (NIR) SIF is. Therefore, at the canopy level, the NIR SIF shows more potential for use in estimating GPP. However, the red-band SIF contains more information about PS II, which is more sensitive to photosynthesis. So, in theory, if the canopy radiative transfer effect could be corrected for, the potential of the red SIF for estimating GPP would be greatly improved. In this paper, we propose a new simple reflectance-based method for estimating the photosystem-level SIF at both the NIR band and the red band. Longterm observations of SIF and GPP at two crop sites in two years (four datasets in total) were used to investigate the SIF-GPP relationship at the canopy or photosystem level. The SIF downscaled by the simple reflectance-based approach is found to be consistent with that found using the random forest regression method proposed formerly. For the NIR band, the performance of the canopy-level SIF and PS-level SIF for GPP estimation is similar while, for the red band, the correlation between the SIF and GPP is greatly improved after downscaling the canopy-level SIF to the PS level. For the four test datasets, at the red band, the values of R-2 for the SIF-GPP relationship increase from 0.213-0.599 to 0.398-0.759 after the downscaling. At the PS level, the performance of the red SIF for GPP estimation is as good as or even better than that of the NIR-band SIF. The results of this study indicate that estimation of the SIF photosystem level to canopy level escape probability is important, and that more importance should be paid to the red-band SIF in the monitoring of GPP.