Abstract
A global, monthly averaged time series of Sun-induced Fluorescence (SiF), spanning January 2007 to June 2015, was derived from Metop-A Global Ozone Monitoring Experiment 2 (GOME-2) spectral measurements. Far-red SiF was retrieved using the filling-in of deep solar Fraunhofer lines and atmospheric absorption bands based on the general methodology described by Joiner et al, AMT, 2013. A Principal Component (PC) analysis of spectra over non-vegetated areas was performed to describe the effects of atmospheric absorption. Our implementation (SiF KNMI) is an independent algorithm and differs from the latest implementation of Joiner et al, AMT, 2013 (SiF NASA, v26), because we used desert reference areas for determining PCs (as opposed to cloudy ocean and some desert) and a wider fit window that covers water vapour and oxygen absorption bands (as opposed to only Fraunhofer lines). As a consequence, more PCs were needed (35 as opposed to 12). The two time series (SiF KNMI and SiF NASA, v26) correlate well (overall R of 0.78) except for tropical rain forests. Sensitivity experiments suggest the strong impact of the water vapour absorption band on retrieved SiF values. Furthermore, we evaluated the SiF time series with Gross Primary Productivity (GPP) derived from twelve flux towers in Australia. Correlations for individual towers range from 0.37 to 0.84. They are particularly high for managed biome types. In the de-seasonalized Australian SiF time series, the break of the Millennium Drought during local summer of 2010/2011 is clearly observed.
Highlights
The Gross Primary Production (GPP) of the terrestrial biosphere is a key quantity in the understanding of the global carbon cycle
We investigate the effect of the number of Principal Component (PC), the fit window, and the PC reference area on retrieval and use the four months of January, April, July and October 2013 as test months for reasons of computational time
Sun-induced Fluorescence (SiF) maps are most noisy for the shortest fit window, which is understandable as there are less spectral points and because there is no filling-in of atmospheric absorption bands that can provide information
Summary
The Gross Primary Production (GPP) of the terrestrial biosphere is a key quantity in the understanding of the global carbon cycle. GOSAT’s coarse temporal and spatial resolution, requires considerable averaging of the individual fluorescence retrievals into spatiotemporal composites to produce global maps, which may introduce sampling or representativeness errors It was shown by Joiner et al [21,33] that the retrieval approaches developed for GOSAT could be applied to the imaging spectrometers SCIAMACHY and GOME-2. Despite the complexity of the underlying mechanisms, satellite-retrieved far-red SiF has been directly compared with GPP from data-driven and global dynamic vegetation models as well as tower-based flux estimates [34,53,54].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
