Renewed methane increase for ﬁve years (2007–2011) observed by solar FTIR spectrometry

Trends of column-averaged methane for the time period [1996, September 2011] are derived from the mid-infrared (mid-IR) solar FTIR time series at the Zugspitze (47.42

resampling for quantifying trend uncertainties. We find a positive trend during [1996, 1998] (9.0 [3.2, 14.7] ppb yr −1 , Zugspitze, 95 % confidence interval), a non-significant growth during [1999, mid 2006 for Garmisch, which is in agreement with 4.8 [3.8, 5.9] ppb yr −1 for Zugspitze. 10 The agreement of methane trends at the two closely neighboring FTIR sites with strongly differing levels of integrated water vapor (min/max = 0.2 mm/12.7 mm for Zugspitze, 1.9 mm/34.9 mm for Garmisch) proves that potentially significant watervapor-methane interference errors do not affect the trend results, if the updated mid-IR retrieval strategy MIR-GBM v1.0 is used. Furthermore, agreement of the trend 15 of 6.6 ppb yr −1 derived from SCIAMACHY (WFMD v2.0) data for the time period [mid 2006, mid 2009] is found within the 95 % confidence interval of the ground-based FTIR result.
While earlier studies using surface network data revealed changes of 8.0±0.6 ppb in 2007 and 6.4±0.6 ppb in 2008 (update from Dlugokencky et al., 2009), our updated 20 result proves that meanwhile, the renewed methane increase has been persisting for >5 yr [mid 2006[mid , September 2011. This is either the longest and largest positive trend anomaly since >25 yr when systematic observations began or the onset of a new period of strongly increasing CH 4 levels in the atmosphere.
The 2007-2008 part of the anomaly was previously attributed to increased nat- been increasing again since 2010, while tropical land temperatures increased only slightly. As recent estimates of anthropogenic emissions are not yet available, it is not possible to finally conclude that the 2009-2011 period of methane increase was dominated by natural wetland emissions, although they probably play a significant role.

5
The molecular symmetry of methane (CH 4 ) allows vibration-rotation excitation by infrared absorption to occur much more actively compared to carbon dioxide (CO 2 ). This is the principal reason why methane is the second important anthropogenic greenhouse gas in spite of its still relatively small abundance in the atmosphere. Methane concentrations in the atmosphere have more than doubled since the beginning of industrialization (Forster et al., 2007). After a period of near-zero growth at the beginning of this century (Dlugokencky et al., 2003;Bousquet et al., 2006), the growth rate of atmospheric methane started to increase strongly again after 2006(Rigby et al., 2008Dlugokencky et al., 2009). Recent publications quantified renewed methane increase up to the year 2009 and discussed the possible causes of the in-15 crease observed in the years (Rigby et al., 2008Duglokencky et al., 2009;Bousquet et al., 2011;Frankenberg et al., 2011;Schneising et al., 2011;Spahni et al., 2011;Montzka et al., 2011). Significant contributions from a reduced OH sink or from biomass burning could not be identified, but evidence was found of an increase in emissions from northern wetlands in 2007 due to unusually high temperatures. In ad-20 dition, an increase in tropical wetland emissions during 2007-2008 was found, which was mainly due to precipitation changes associated with the La Niña event in these years. All in all, the 2007-2008 anomaly was attributed mainly to increased natural emissions from wet ecosystems. The question of how long the 2007-2008 anomaly will 25 persist after 2008 is related to the interpretation of the mechanisms driving the trend until today (2007)(2008)(2009)(2010)(2011). Wetland emissions are widely assumed to drive inter-annual 30759 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | variability (time scale of 1-2 yr), while more gradual changes on the time scale of several years are understood to be due to changing anthropogenic emissions.
It is therefore the goal of this paper to address the question of persistence of the 2007-2008 anomaly via an updated trend analysis including data until today (fall 2011). Contrary to most previous trend studies based on in-situ surface network data, we use 5 ground-based solar absorption Fourier-Transform-Infrared (FTIR) spectrometry. This technique measures methane columns representative of large geographical areas. Measurements at a single FTIR station may be comparable to a set of in-situ measurements (Ohlsen and Randerson, 2004).
Columnar methane retrieved from ground-based FTIR spectrometry in the mid-10 infrared (mid-IR) within the Network for the Detection of Atmospheric Composition Change (NDACC, http://www.ndacc.org) has already been used for trend studies (Zander et al., 1989;Angelbratt et al., 2011) or satellite validation (e.g., Sussmann et al., 2005). However, it became obvious recently that methane-column retrievals at highhumidity (low-altitude) sites may be dominated by water-vapor (H 2 O/HDO)-CH 4 inter-15 ference errors of up to 5 %. Consequently, artifacts dominated the seasonal cycle of methane retrieved with the standard NDACC retrieval strategy. Dry (high-altitude) sites were not affected. Note that a general formulation of the interference problem was given before (Sussmann and Borsdorff, 2007) and water vapor interference also perturbed methane retrievals from SCIAMACHY (Frankenberg et al., 2008). To improve 20 the standard retrieval strategy used within NDACC, we recently developed a new retrieval strategy (MIR-GBM v1.0) using systematically selected spectral micro windows and spectroscopic parameters (Sussmann et al., 2011). MIR-GBM v1.0 was shown to eliminate H 2 O/HDO-CH 4 interference errors down to the ≈0.1 % level at the wettest sites. In return, a good agreement of the retrieved Northern Hemisphere seasonal cy- 25 cle with SCIAMACHY results was achieved for the first time (Sussmann et al., 2011). Therefore, another goal of this paper is to further validate MIR-GBM v1.0 by comparing the methane trend results from two neighboring sites with strongly differing levels of integrated water vapor (IWV), namely, the Zugspitze (2964 m a.s.l.) and Garmisch (743 m a.s.l.). After describing the mid-IR sounding technique (Sect. 2), methane time series and trends (Sect. 3.1) will be presented, the MIR-GBM v1.0 retrieval strategy will be validated by comparing Zugspitze and Garmisch trend results, and the trend results will be related to earlier findings using other techniques (Sect. 3.2). Section 4 will provide a 5 tentative explanation of the trend behavior derived for the 2007-2011 time period along with some conclusions.

Sounding technique
Time series of column-averaged dry-air mole fractions of methane (XCH 4 ) were retrieved from long-term FTIR solar absorption measurements at two northern mid-10 latitude sites, namely, the Zugspitze high-altitude site (47.42 • N, 10.98 • E, 2964 m a.s.l.) and Garmisch (47.48 • N, 11.06 • E, 743 m a.s.l.). The two sites are located only a few kilometers apart in horizontal distance. However, the IWV levels differ strongly (min/max = 0.2 mm/12.7 mm for Zugspitze, 1.9 mm/34.9 mm for Garmisch) because of the altitude difference. This helps to validate the methane retrievals and in particular 15 to exclude a potential impact of water-vapor interference errors on the methane trend results. The Zugspitze FTIR system has been conducting continuous measurements since 1995 as part of NDACC. It is operated by the Group "Variability and Trends" of IMK-IFU 1 , Karlsruhe Institute of Technology, at the Zugspitze 2 . The FTIR system is based 20 on a Bruker IFS125HR interferometer; details can be found in Sussmann and Schäfer (1997). The interferograms for the methane retrievals were recorded with an InSb detector using an optical path difference of typically 175 cm. 6 scans were averaged (≈7-30761 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | min integration time). The pressure-temperature profiles necessary for the inversion were taken from the National Center for Environmental Prediction (NCEP) automailer. The Garmisch solar FTIR system was set up in 2004 at the Garmisch site 3 and is part of the Total Carbon Column Observing Network (TCCON, http://www.tccon.caltech.edu/) operating in the near-infrared for high-precision retrieval of column-averaged mixing 5 ratios of carbon dioxide and methane. The system performs mid-IR NDACC-type measurements in parallel (in alternating mode on the time scale of several minutes). The latter are utilized for this study. The measurement settings for the Garmisch mid-IR methane measurements are the same as for the Zugspitze. For inversion of XCH 4 from the mid-IR solar spectra, the MIR-GBM v1.0 retrieval strategy is applied, which minimizes interference errors from water vapor and HDO down to the ≈0.1 % level. The basic features of this strategy are given in Table 1, more details can be found in Sussmann et al. (2011).

Time series and trends
15 Figure 1a shows the time series of monthly-mean column-averaged methane above Zugspitze and Garmisch. Typically, n = 40-60 individual column measurements per month are recorded on 6-8 clear sky days. Shaded bars in Fig. 1a indicate the statistical uncertainties of the monthly means calculated from the individual measurements (±3σ/sqrt(n)). These suggest that statistically stable monthly means are obtained for 20 the trend analysis. The de-seasonalized time series (Fig. 1b) shows a significant increase during the 3yr time interval [1996,1998], followed by a period with no significant growth [1999, mid 2006]. Afterwards, a renewed strong increase can be observed, which has been persisting for 5 yr now [mid 2006[mid , September 2011. This behavior is quantified by a trend analysis based on the approach described in Gardiner et al. (2008). See Table 2 for the resulting figures. Briefly, the approach augments a basic linear trend model applied to the indicated parts of the time series after subtraction of a fitted intra-annual function and uses least squares regression in conjunction with a bootstrap resampling of the residuals in order to determine confidence limits associated with the trend estimates.

5
For the intra-annual model, a 3rd-order Fourier series is used.
As a main result from

Validation and consistency with other results
Development of MIR-GBM v1.0 (Sussmann et al., 2011) was motivated by the finding that the standard retrieval strategy used within the NDACC network until then was subject to H 2 O-CH 4 interference errors ranging up to 5 % at high-humidity (low-altitude) 15 sites like Garmisch. This error may even affect the retrieved methane trends, e.g., if there is a trend in columnar water vapor at one site and no trend at another. A particular result from   (Table 3 in Schneising et al., 2011).

Discussion and conclusion
The methane increase persisting for 5 yr [mid 2006[mid , September 2011 as documented in Fig. 1 and Table 2 is either the longest and largest positive trend anomaly since more 15 than 25 yr when systematic observations began or marks the onset of a new period of strongly increasing CH 4 levels in the atmosphere. Our study cannot identify the mechanisms behind this increase, but our findings shall be discussed briefly in relation to previous studies on this subject and most reasonable causes shall be proposed. Previous studies discussed possible causes of the increase in the years 2007 and 20 2008 (Rigby et al., 2008;Duglokencky et al., 2009;Bousquet et al., 2011;Frankenberg et al., 2011;Spahni et al., 2011;Montzka et al., 2011;van der Werf et al., 2011 OH loss may be assumed to play a minor role in the recent methane increase. The biomass burning contribution also turned out to be insignificant, because no large CO anomaly is observed in (van der Werf et al., 2010. Bousquet et al. (2011) invoke an increase in emissions from natural, mostly tropical wetlands in 2007-2008, but with a significant contribution from northern wetlands in 2007. The land surface 5 models ORCHIDEE (Ringeval et al., 2010) and LPJ (Spahni et la., 2011) both show a positive trend in natural wetlands emissions between 2005 and 2008, which is mainly due to tropical precipitation changes associated with the La Niña event in these years and to boreal positive temperature anomalies in 2007. The question is whether the continued increase shown in Fig. 1  However, a persisting natural wetland anomaly over more than a few years may be in conflict with the understanding that wetland emissions typically cause inter-annual vari-25 ability (i.e., 1-2 yr time scale), while more gradual changes on the time scale of several years would be due to changing anthropogenic emissions. In this context, it is worth considering the causes of the near-zero growth during 1999-2006(Duglokencky, et al., 2003. While one might be tempted to attribute this trend stagnation to the collapse 30765 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | of industry in the former USSR, Bousquet et al. (2006) pointed out that global emissions continued to increase during this period in spite of this negative trend component. According to the EDGAR 4.1 database (European Commission 2009), anthropogenic emissions between [1999,2005] totaled 31 Tg and were dominated by strongly increasing emissions in China, which is more than sufficient to explain the post-2006 anomaly 5 in atmospheric concentrations. This leads to the question why no continued increase of methane had been observed after 1999 already, as was pointed out by Frankenberg et al. (2011). In contrast to the EDGAR inventory, Aydin et al. (2011) recently proposed on the basis of an ethane-based fossil fuel emission history that decreasing fossil fuel emissions of methane may explain the observed atmospheric slowdown. The answer 10 by Bousquet et al. (2006) was that the expected increase had been masked from 1999 to 2005 by continuously decreased wetland emissions due to long-lasting drier conditions encountered in various regions of the Northern Hemisphere. According to this finding, Bousquet et al. stated in 2006 (just before the renewed methane increase was detected by measurements) that a renewed increase of global CH 4 emissions would 15 have to be expected as soon as wetland emissions rise back to a normal level. This explanation is in line with recent simulations by the ORCHIDEE global dynamical vegetation model (Ringeval et al., 2010) and by the LPJ-WHyMe global dynamical vegetation model using model parameters derived from 2003-2005 inversions constrained by SCIAMACHY and surface network data (Spahni et al., 2011). The latter found a 20 strong increase in ecosystem CH 4 emissions (3.62 Tg yr −1 ) for the period [2005,2008] compared to emissions of 1.1 Tg yr −1 during the [1900,2008] period, although they pointed out that their simulations can only partly explain the long-term decline of the atmospheric growth rate during [1990,2006]. According to the different possible interpretations, the renewed CH 4 increase after 25 2006 is expected to continue on the decadal time scale as long as no large-scale and long-term droughts will follow. Anyway, there is no need for an explanation of the actual methane trend by strongly increased emissions from stores of carbon in melting permafrost and from marine hydrates as a reaction to climate change. However, both remain to be potential sources of severely increasing methane emissions in the future, which should be closely monitored by remote sensing at the surface or from space.

30769
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Table 1. Characteristics of the strategy for retrieval of column-averaged methane from mid-IR solar spectra (MIR-GBM v1.0). For more details, see Sussmann et al. (2011 Table 2 for trend magnitudes and significance.