Elevated uptake of CO 2 over Europe inferred from GOSAT X CO 2 retrievals: a real phenomenon or an artefact of the analysis?

Estimates of the natural CO2 flux over Europe inferred from in situ measurements of atmospheric CO2 mole fraction have been used previously to check top-down flux estimates inferred from space-borne dry-air CO2 column (XCO2 ) retrievals. Recent work has shown that CO2 fluxes inferred from XCO2 5 data from the Japanese Greenhouse gases Observing SATellite (GOSAT) have a larger seasonal amplitude and a more negative annual net CO2 balance than those inferred from the in situ data. The causes of this enhanced European CO2 uptake have since become the focus of recent studies. We show this elevated uptake over Europe could largely be explained by mis-fitting 10 data due to regional biases. We establish a reference in situ inversion that uses an Ensemble Kalman Filter (EnKF) to assimilate surface flask data and the XCO2 data from the surface-based Total Carbon Column Observing Network (TCCON). The same EnKF system is also used to assimilate two, independent versions of GOSAT XCO2 data. We find that the GOSAT-inferred European terrestrial biosphere uptake peaks 15 during the summer, similar to the reference inversion, but the net annual flux is 1.180.1GtCa-1 compared to a value of 0.56-0.1GtCa-1 for our control inversion that uses only in situ data. To reconcile these two estimates, we have performed a series of numerical experiments that assimilate observations with biases or assimilate synthetic observations for which part or all of the GOSAT XCO2 data are replaced with model 20 data. We find that 50-80% of the elevated European uptake in 2010 inferred from GOSAT data is due to retrievals outside the immediate European region, while most of the remainder can be explained by a sub-ppm retrieval bias over Europe. We have used data assimilation techniques to estimate monthly GOSAT XCO2 biases from the joint assimilation of in situ observations and GOSAT XCO2 retrievals. We find a monthly 25 varying bias of up to 0.5 ppm can explain an overestimate of the annual sink of up to 0.18 GtCa-1. Disciplines Medicine and Health Sciences | Social and Behavioral Sciences Publication Details Feng, L., Palmer, P. I., Parker, R. J., Deutscher, N. M., Feist, D. G., Kivi, R., Morino, I. & Sussmann, R. (2015). Elevated uptake of CO2 over Europe inferred from GOSAT XCO2 retrievals: a real phenomenon or an artefact of the analysis?. Atmospheric Chemistry and Physics, 15 (3), 1989-2011. Authors L Feng, Paul I. Palmer, Robert J. Parker, Nicholas M. Deutscher, D Feist, Rigel Kivi, Isamu Morino, and Ralf Sussmann This journal article is available at Research Online: http://ro.uow.edu.au/smhpapers/3740 Edinburgh Research Explorer Elevated uptake of CO2 over Europe inferred from GOSAT XCO2 retrievals: a real phenomenon or an artefact of the analysis? Citation for published version: Feng, L, Palmer, PI, Parker, RJ, Deutscher, NM, Feist, DG, Kivi, R, Morino, I & Sussmann, R 2015, 'Elevated uptake of CO2 over Europe inferred from GOSAT XCO2 retrievals: a real phenomenon or an artefact of the analysis?' Atmospheric Chemistry and Physics Discussions, vol 15, no. 2, pp. 1989-2011., 10.5194/acpd-15-1989-2015 Digital Object Identifier (DOI): 10.5194/acpd-15-1989-2015 Link: Link to publication record in Edinburgh Research Explorer Document Version: Publisher's PDF, also known as Version of record Published In: Atmospheric Chemistry and Physics Discussions General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights. Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please contact openaccess@ed.ac.uk providing details, and we will remove access to the work immediately and investigate your claim. Download date: 23. May. 2016 ACPD 15, 1989–2011, 2015 Elevated uptake of CO2 over Europe inferred from GOSAT XCO2 retrievals L. Feng et al.

gases Observing SATellite (GOSAT) have a larger seasonal amplitude and a more negative annual net CO 2 balance than those inferred from the in situ data. The causes of this enhanced European CO 2 uptake have since become the focus of recent studies. We show this elevated uptake over Europe could largely be explained by mis-fitting data due to regional biases. We establish a reference in situ inversion that uses an Ensemble Kalman Filter (EnKF) to assimilate surface flask data and the X CO 2 data from the surface-based Total Carbon Column Observing Network (TCCON). The same EnKF system is also used to assimilate two, independent versions of GOSAT X CO 2 data. We find that the GOSAT-inferred European terrestrial biosphere uptake peaks during the summer, similar to the reference inversion, but the net annual flux is 1.18 ± 15 0.1 GtC a −1 compared to a value of 0.56 ± 0.1 GtC a −1 for our control inversion that uses only in situ data. To reconcile these two estimates, we have performed a series of numerical experiments that assimilate observations with biases or assimilate synthetic observations for which part or all of the GOSAT XCO 2 data are replaced with model data. We find that 50-80 % of the elevated European uptake in 2010 inferred from 20 GOSAT data is due to retrievals outside the immediate European region, while most of the remainder can be explained by a sub-ppm retrieval bias over Europe. We have used data assimilation techniques to estimate monthly GOSAT X CO 2 biases from the joint assimilation of in situ observations and GOSAT X CO 2 retrievals. We find a monthly varying bias of up to 0.5 ppm can explain an overestimate of the annual sink of up to

Introduction
Observed atmospheric variations of carbon dioxide (CO 2 ) are due to atmospheric transport and surface flux processes. Using prior knowledge of the spatial and temporal distribution of these fluxes and atmospheric transport it is possible to infer (or invert for) the a posteriori estimate of surface fluxes from atmospheric concentration data. The 5 geographical scarcity of such observations precludes robust flux estimates for some regions due to large uncertainties associated with meteorology and a priori fluxes. Arguably, our understanding of regional CO 2 fluxes, particularly at tropical and high northern latitudes, has not significantly improved for more than a decade (Gurney et al., 2002;Peylin et al., 2013), reflecting the difficulty of maintaining a surface measure-10 ment programme over vulnerable and inhospitable ecosystems. Atmospheric transport model errors compound errors introduced by poor observation coverage, resulting in significant differences between flux estimates on spatial scales < O(10 000 km) (e.g. Law et al., 2003;Yuen et al., 2005;Stephens et al., 2007). The Greenhouse gases Observing SATellite (GOSAT), a space-borne mission 15 launched in a sun-synchronous orbit in early 2009, was purposefully designed to measure CO 2 columns using short-wave IR wavelengths. Validation of current X CO 2 column retrievals using co-located upward-looking FTS measurements of the Total Carbon Column Observing Network (TCCON) (Wunch et al., 2011) show a standard deviation of 1.6-2.0 ppm (e.g., Parker et al., 2012). Their global biases are typically smaller than 20 0.5 ppm (Oshchepkov et al., 2013). The disadvantage of using the TCCON is that sites are mainly at northern extra-tropical latitudes with little or no coverage where our knowledge of the carbon cycle is weakest. Surface flux estimation algorithms are particularly sensitive to systematic errors so that sub-ppm biases can still significantly change the patterns of regional flux estimates (Chevallier et al., 2010). This is further complicated Introduction Several independent studies have shown that regional flux distributions inferred from GOSAT X CO 2 retrievals are significantly different from those inferred from in situ data (Basu et al., 2013;Deng et al., 2013;Chevallier et al., 2014). In particular, these studies report a larger-than-expected annual net emission over tropical continents and a largerthan-expected net annual uptake over Europe. While the GOSAT inversions suffer from 5 observation errors, atmospheric transport errors and aggregation errors (from the seasonal coverage of higher latitudes), the in situ inversions are also unreliable over many regions due to poor coverage and atmospheric transport errors. Consequently, there is an ongoing debate about whether a recent study that shows a large European uptake of CO 2 (Reuter et al., 2014) reflects a real phenomenon or is an artefact of uncharacterized regional biases.
We report the results from a small set of experiments that show systematic bias can cause a large difference between fluxes over Europe inferred from GOSAT and those inferred from in situ data. In the next section we provide an overview of the inverse model framework used to interpret in situ data (flask and TCCON) and GOSAT X CO 2 15 data. In Sect. 3, we describe and present results from two groups of experiments that characterize the role of systematic bias in regional flux estimates. In Sect. 4, we use a modified version of the inverse model framework to estimate monthly biases by jointly assimilating all data. We conclude the paper in Sect. 5. 20 We use the GEOS-Chem global chemistry transport model to relate surface fluxes to the observed variations of atmospheric CO 2 concentrations (Feng et al., 2009) due to spin-up and edge effects. We estimate monthly fluxes on a spatial distribution that is based on TransCom-3 (Gurney et al., 2002) with each continental region further divided equally into 12 sub-regions and each ocean region further divided equally into 6 sub-regions; we estimate fluxes for 199 regions compared to 144 regions used in previous studies (Feng et al., 2009;Chevallier et al., 2014).
The control in situ inversion experiment (INV_TCCON) includes surface flask observations at 76 sites (Feng et al., 2011) and, to improve observation constraints, to-15 tal column X CO 2 retrievals from all the TCCON sites of the GGG2012 dataset (see https://tccon-wiki.caltech.edu for more details). We use daytime mean TCCON retrievals (09:00 to 15:00 LT) with their errors determined by the standard deviations about that daytime mean. Including TCCON observations with the flask observations increases the annual net uptake over Europe from 0.47 to 0.56 GtC a −1 for 2010, which 20 mainly reflects larger summer uptake. Evaluation of the INV_TCCON a posteriori model concentrations agrees well with the independent HIAPER Pole-to-Pole Observations (HIPPO) aircraft measurements below 5 km (Wofsy et al., 2010), with a small bias of 0.04 ppm, and a sub-ppm standard deviation of 0.87 ppm (not shown). For the two control GOSAT inversions, we use two independent data sets: (1) X  15,2015 Elevated uptake of CO 2 over Europe inferred from GOSAT X CO 2 retrievals L. Feng et al. regions, and apply the bias corrections suggested by the data providers. We double the reported observation errors, as suggested by the retrieval groups. As a performance indicator for our ability to fit fluxes to observed X CO 2 concentrations, we compare a posteriori model concentrations with GOSAT X CO 2 retrievals and show that INV_ACOS and INV_UOL agree much better than INV_TCCON. For example, the bias against ACOS X CO 2 retrievals is −0.39 ppm for INV_TCCON and 0.03 ppm for INV_ACOS with a corresponding reduction in the standard deviation from 1.70 to 1.58 ppm. However, comparison of GOSAT a posteriori concentrations against independent HIPPO measurements is worse than INV_TCCON with a positive bias of 0.42 and 0.62 ppm for INV_ACOS and INV_UOL, respectively, which are mainly caused by 10 the overestimation of CO 2 concentrations (∼ 1.5-2.0 ppm) at low latitudes. Table 1  ACPD 15,2015 Elevated uptake of CO 2 over Europe inferred from GOSAT X CO 2 retrievals L. Feng et al. We calculated monthly mean CONTRAIL measurements during 2010 using data below 3 km, where there is greater sensitivity to local surface fluxes. Our current model resolution precludes small-scale sources (or sinks) so we expect model bias. We find that INV_TCCON agrees best with CONTRAIL observations, in particular at the beginning of the 2010, partially reflecting the poor GOSAT X CO 2 coverage over Europe To understand the differences between the INV_TCCON and GOSAT inversions, we conducted two groups of sensitivity tests (see Table 1). First, we replaced all or part of the GOSAT X CO 2 retrievals assimilated in INV_ACOS with a model forced by the a posteriori fluxes from INV_TCCON indicating an external contribution of over 80 % to the enhanced uptake of 0.60 GtC a −1 .

10
We have also investigated the contribution from regions outside of Europe to the European flux estimate using quasi-regional inversions where only observations over Europe have been assimilated (see Appendix A). Second, we crudely demonstrate how regional bias could explain the remaining discrepancy of 0.33 GtC a −1 between GOSAT and INV_TCCON inversions over Eu-15 rope. In our experiment INV_ACOS_SPR_0.5ppm, we add 0.5 ppm to the GOSAT ACOS retrievals within Europe taken in February-April, inclusively, which effectively reduces the uptake by 0.09 GtC a −1 from 1.20 to 1.11 GtC a −1 . Similarly, when a 0.5 ppm bias is added to the GOSAT data taken in June-August we find a larger reduction of 0.14 GtC a −1 for the summer peak uptake (INV_ACOS_SUM_0.5ppm). These results 20 emphasize the importance of characterizing sub-ppm regional bias to avoid erroneous flux estimates.

Bias estimation
Here we demonstrate a simple approach to quantify systematic bias using an online data assimilation approach. We assimilate the GOSAT X CO 2 retrievals with the in situ ACPD Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | regional biases from 11 TransCom land regions (Gurney et al., 2002). Compared to the off-line comparisons between GOSAT X CO 2 retrieval and model concentrations, the main advantage of our online bias estimation is that the uncertainties associated with errors in flux estimates can be taken into account. To investigate the spatial pattern of the X CO 2 biases within Europe, we split Europe into West Europe (west of 20 • E) and

5
East Europe (east of 20 • E). We assume a priori that monthly biases are 0.0 ± 0.5 ppm.
The annual European carbon flux estimates in 2010 for INV_ACOS_INS and INV_UOL_INS are 0.61 ± 0.08 and 0.66 ± 0.08 GtC a −1 , respectively, which are close to the in situ control estimate of 0.56 ± 0.1 GtC a −1 . This suggests that the coarse coverage of in situ observations is unlikely to be the main reason for the associated smaller 10 estimate of European uptake. Figure 4 shows the estimated monthly biases in ACOS and UOL X CO 2 retrievals over East and West Europe during 2010. Monthly biases are typically smaller than 0.5 ppm, but have different seasonal cycles over East and West Europe. We find that after correcting for these biases the annual European uptake estimate from INV_ACOS 15 is reduced by 0.18 GtC a −1 , representing more than half of the contribution from GOSAT observations within Europe. This result is consistent with the sensitivity tests. The effect of bias correction is much smaller for INV_UOL (0.03 GtC a −1 ), which is mainly due to the positive biases in the summer time over West Europe. The differences in GOSAT X CO 2 retrievals and their effects on regional flux estimates have also been investigated 20 in previous studies (e.g., Takagi et al., 2014).

Discussion and Conclusions
We used an ensemble Kalman Filter to infer regional CO 2 fluxes using three different data sets: (1) surface in situ flask and TCCON observations; (2) GOSAT X CO 2 retrievals by JPL ACOS team; and (3)  are consistent with a significantly larger European uptake than inferred from in situ data during 2010. We showed using sensitivity experiments that 50-80 % of the enhanced European uptake of CO 2 is determined by a positive model bias of CO 2 being transported into Europe, due to the assimilation of GOSAT X CO 2 data outside of Europe. This model 5 bias is supported by a comparison of GOSAT a posteriori X CO 2 concentrations with independent aircraft observations. The main consequence of the elevated CO 2 inflow into the European domain is that mass balance dictates that the European uptake must be increased, even if GOSAT X CO 2 retrievals within the European domain are not biased. Adding an additional 0.5 ppm to INV_ACOS X CO 2 data outside the European region increases European annual net uptake from 1.20 to 1.53 GtC a −1 , while the same increase to the INV_TCCON in situ observations outside Europe only increases the annual net uptake by 0.14 GtC from 0.56 to 0.70 GtC a −1 . Erroneous interpretation of X CO 2 data can result from analyses if unbiased boundary conditions are not addressed because biases in the model inflow can affect both the background concentrations and 15 the internal X CO 2 gradients. We showed using sensitivity tests that sub-ppm bias can explain the remaining 0.33 GtC a −1 flux difference between INV_ACOS and the in situ inversion after accounting for biased boundary conditions. By assimilating the in situ and GOSAT observations to estimate surface fluxes and monthly X CO 2 biases, we infer a monthly bias that is typi-20 cally less than 0.5 ppm over East and West Europe corresponding to an overestimated sink of 0.18 GtC a −1 . The inferred monthly biases for UOL X CO 2 are also not the same as the ACOS X CO 2 data, particularly over West Europe during the summer months. This level of sensitivity of regional flux estimates to time-varying sub-ppm biases highlights the challenges we face as a community when evaluating X CO 2 retrievals using 25 current observation networks.
Flux estimates are also sensitive to a priori assumptions, idiosyncrasies of applied inversion algorithms, and the underlying model atmospheric transport (Chevallier et al., 2014;Peylin et al., 2014;Reuter et al., 2014). The presence of regional biases further complicates the inter-comparisons of flux estimates based on different inversion approaches. In our assimilation of ACOS retrievals, we find that doubling the a priori flux error (INV_ACOS_DBL_ERR) increases the estimated European uptake from 1.20 to 1.55 GtC a −1 , reflecting the increased vulnerability to the X CO 2 biases both within and outside Europe particularly when using weak a priori constraints. In contrast, doubling 5 the a priori flux errors only increases the uptake by 0.09 to 0.7 GtC a −1 for the joint data assimilation (INV_ACOS_INS_DBL_ERR), with very little changes in the estimated biases (not shown). This study highlights the adverse effects of regional biases in current GOSAT X CO 2 retrievals that can attract erroneous interpretation of resulting regional flux estimates.

10
A thorough evaluation of the X CO 2 retrievals using independent and sufficiently accurate/precise observations is urgently required to increase the confidence of regional CO 2 flux estimates inferred from space-based observations. Our study suggests that inferring flux estimates in a limited spatial domain like Europe, the observational density outside this domain is crucial. International networks such as TCCON have focused 15 on establishing observation sites in remote regions and reducing inter-station biases, which represent an important activity within the broader carbon cycle science community.