Convergences and divergences between data-driven GPP estimates and high-resolution SIF measurements across vegetation and climatic gradients

Sun-induced chlorophyll fluorescence (SIF) retrieved from satellites has shown potential as a remote sensing proxy for gross primary productivity (GPP). However, current studies have generally been limited by the spatial resolution of datasets with a sufficiently long archive. For example, while available since 2007, the commonly used GOME-2 SIF data has a spatial resolution in the order of 0.5° (~50km), too coarse to effectively separate the competing effects of different types of vegetation from the overall ecosystem dynamics, or to draw general conclusions relevant to the land cover of a region. While finer SIF retrievals are becoming available, such as from the TROPOMI instrument on-board of the Sentinal-5P platform, several years will be needed before their archives reach a sufficient temporal depth.

 

Using GOME-2 SIF retrievals, downscaled to a resolution of 0.05° (~5km) via a proven methodology [1], comparisons are made with the data-driven FLUXCOM GPP dataset and divergences and convergences explored to see where high-resolution SIF can enhance our understanding of GPP. This includes an exploration of the spatial and temporal relationships between estimates of GPP and SIF at a global scale. The high resolution of the SIF data allows the relationships to be broken down by plant functional type (PFT) for separate climate zones, thus enabling a confrontation between FLUXCOM GPP and SIF at fine granularity and eventually a future integration of SIF in the estimation of data-driven GPP products.

 

Whilst a linear relationship is generally observed between SIF and GPP in all vegetation categories, areas of non-linearity suggest where SIF could potentially provide more information about ecosystem dynamics that are not represented in the GPP dataset. For example some vegetation types experience saturation in the seasonal GPP measurements (likely driven by the saturation of the fraction of absorbed PAR), that are not emerging from the SIF signal. In addition, in highly productive ecosystems like tropical rainforests, a wide range of spatio-temporal variation in SIF is observed, while only a considerably smaller variability is reproduced in the modelled GPP. Further studies are conducted on how SIF and GPP behave differently in anomalies of air temperature and soil moisture. Overall, the study suggests there is room to improve global land-climate models by incorporating information from SIF.

 

 

[1] Duveiller, G., Filipponi, F., Walther, S., Köhler, P., Frankenberg, C., Guanter, L., and Cescatti, A.: A spatially downscaled sun-induced fluorescence global product for enhanced monitoring of vegetation productivity, Earth Syst. Sci. Data Discuss., , in review, 2019.

How to cite: Pickering, M., Cescatti, A., and Duveiller, G.: Convergences and divergences between data-driven GPP estimates and high-resolution SIF measurements across vegetation and climatic gradients, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20596, https://doi.org/10.5194/egusphere-egu2020-20596, 2020