Abstract
Current knowledge about the European terrestrial biospheric carbon sink, from
the Atlantic to the Urals, relies upon bottom-up inventory and surface flux inverse
model estimates (e.g., 0.27±0.16 GtCa−1 for 2000–2005 (Schulze et al., 2009),
0.17±0.44 GtCa−1 for 2001–2007 (Peters et al., 2010), 0.45±0.40 GtCa−1 5 for 2010
(Chevallier et al., 2014), 0.40±0.42 GtCa−1 for 2001–2004 Peylin et al., 2013). Inverse
models assimilate in situ CO2 atmospheric concentrations measured by surface-based
air sampling networks. The intrinsic sparseness of these networks is one reason for the
relatively large flux uncertainties (Peters et al., 2010; Bruhwiler et al., 2011). Satellite10
based CO2 measurements have the potential to reduce these uncertainties (Miller
et al., 2007; Chevallier et al., 2007). Global inversion experiments using independent
models and independent GOSAT satellite data products consistently derived a considerably
larger European sink (0.9–1.2 GtCa−1 for September 2009–August 2010 (Basu
et al., 2013), 1.2–1.8 GtCa−1 in 2010 Chevallier et al., 2014). However, these results
15 have been considered unrealistic due to potential large scale retrieval biases and/or
long-range transport errors (Chevallier et al., 2014) or have not been discussed at
all (Basu et al., 2013; Takagi et al., 2014). Here we show that the satellite-derived
European terrestrial carbon sink is indeed much larger (1.02±0.30 GtCa−1 in 2010)
than previously expected. Our analysis comprises a regional inversion approach using
20 STILT (Gerbig et al., 2003; Lin et al., 2003) short range (days) particle dispersion modelling,
rendering it insensitive to large scale retrieval biases and less sensitive to longrange
transport errors. The highest gain in information is obtained during the growing
season when satellite observation conditions are advantageous and a priori uncertainties
are largest. The consistency among an ensemble of five different inversion set-ups
25 and five independent satellite retrievals (BESD (Reuter et al., 2011) 2003–2010, ACOS
(O’Dell et al., 2012) 2010, UoL-FP (Cogan et al., 2012) 2010, RemoTeC (Butz et al.,
2011) 2010, and NIES (Yoshida et al., 2013) 2010) using data from two different instruments (SCIAMACHY, Bovensmann et al., 1999 and GOSAT, Kuze et al., 2009) provides evidence that our current understanding of the European carbon sink has to be revisited.