4.1. Methane Uptake
 In the current study, we measured CH4 fluxes at a subdaily temporal resolution over an entire year. In addition, we also observed CH4 concentrations in the soil profile at least at weekly intervals. This allowed us to determine the location and timing of processes producing or consuming CH4 in the soil and their exchange with the atmosphere. During the entire observation period, we observed that the investigated steppe soils functioned exclusively as a sink for atmospheric CH4. This observation was also supported by measurements of CH4 concentration in the soil profile. Concentrations in all observation depths were lower than the background atmospheric concentrations (Figures 2c and 2d) (Table 2).
 Despite the relatively weak CH4 uptake rate (average uptake rate of 0.29 mg C m−1 d−1), CH4 oxidation occurred during the freezing period when the average soil temperature was −7°C at both sites. This finding is in agreement with observations by Mosier et al.  for North American prairie soils and Wang et al.  for steppe soils in Inner Mongolia. Methane uptake at subzero temperatures had actually been observed previously for other ecosystems such as forests [Butterbach-Bahl and Papen, 2002; Groffman et al., 2006] and agriculture ecosystems [Dörsch et al., 2004], as well as snow-covered alpine and subalpine regions [Sommerfeld et al., 1993]. A possible explanation for CH4 uptake by frozen soils is most likely a reallocation of the active layer(s) to deeper, unfrozen soil horizons during the winter. Such an interpretation would be in line with Strigel et al. , who suggested that unfavorable condition for CH4 consumption near the soil surface may result in a shift of atmospheric CH4 consumption to deeper soil layers as long as diffusion pathways are available.
 Freeze-thaw events have been shown to affect greenhouse gas exchange significantly in different ecosystems [Mosier et al., 1993; Kammann et al., 1998; Teepe et al., 2001; Butterbach-Bahl and Papen, 2002; Groffman et al., 2006; Mastepanov et al., 2008]. Mosier et al.  found that a subalpine meadow grassland emitted CH4 during the rapid-snow-melt period. Mastepanov et al.  observed a CH4 emission burst following the onset of freezing in tundra. In semiarid Eurasian steppe, Qi et al.  also observed CH4 emissions during the spring thaw period as well as during periods when the soil was continuously frozen. Contrary to the results of Qi et al. , no methane emission was reported in other studies carried out in steppe ecosystems [Mosier et al., 1996; Wang et al., 2005; Holst et al., 2008]. Our high-frequency measurements show that methane was still consumed, rather than emitted, from the investigated sites when snow and soil began to melt and thaw. In contrast to the study of Qi et al. , our findings are in agreement with other reports for prairie [Mosier et al., 1996] and steppe [Wang et al., 2005; Holst et al., 2008] ecosystems.
 A sharp increase in CH4 uptake was observed at UG99 following spring thaw. During this period, the WFPS in the topsoil increased from 13% to 66% and CH4 uptake increased from 0.14 to 0.62 mg C m−2 d−1. This quick onset of increased CH4 uptake was not observed at WG01. However, the increase in soil moisture following spring thaw, with a rise from 10% to 31% WFPS, was less pronounced as compared to UG99. The differences in soil moisture between the sites were mainly due to the differences in snow cover, which was significantly higher at UG99. These differences in snow cover followed differences in surface roughness that led to higher snow accumulation at UG99. Moreover, snow cover also affected soil temperature. Because of the insulating effect of snow cover, the soil temperatures did not drop as much at UG99 during the winter (minimum at 10 cm soil depth: −11°C) as compared to WG01 (minimum at 10 cm soil depth: −17°C). Namely, both higher soil moisture and higher minimum soil temperature most likely contributed to a quick recovery of methanotrophic activity in the topsoil of UG99 following spring thaw, while this was not observed at WG01. This interpretation is in line with our auxiliary measurements of CH4 uptake potentials during spring thaw, which also indicate the highest uptake rates in the top layer (0–2 cm) at UG99 (Figure 5).
 Methane oxidation was previously reported to be positively related to soil temperature [Born et al., 1990; Crill, 1991; Dörr et al., 1993; King, 1997], but only a weak temperature dependency was found. In our study, soil temperature, which varied annually within a large range (UG99: from −9°C to 22°C in July; WG01: from −14°C to 23°C in July), played the primary role in regulating the seasonal variation of CH4 uptake. This statement is supported by the fact that the variation in soil temperature (at 5 cm) could explain more than 65% of the temporal variability in CH4 uptake (Figure 3, top). The apparent Q10 (5°C–15°C) was 1.49 (UG99: 1.46; WG01: 1.51), which is in the range of Q10 values of 1.0–2.0 reported in other studies [Born et al., 1990; Crill, 1991; Priemé and Christensen, 1997]. Though temperature dominated the seasonal variation in CH4 uptake rates, no significant diurnal variation of CH4 fluxes could be observed despite clear diurnal variations of topsoil temperatures. On a daily basis, temperature-induced changes in CH4 uptake rates were only observed when soil temperatures crossed zero degree Celsius (i.e., from −5°C to 5°C). Wolf et al. (2010, unpublished data) found that CH4 oxidation at same sites mainly occurred at a soil depth of 5–20 cm. However, the diurnal variation in soil temperature at the depth of 10 cm was small. Thus, the lack of a significant diurnal variation allows for representative, chamber-based CH4 flux measurements at any time of the day.
 Many incubation studies [Dunfield et al., 1995; Van den Pol-van Dasselaar et al., 1998] and field observations [Born et al., 1990; Mosier et al., 1996; Mosier and Delgado, 1997; Liu et al., 2007] show that CH4 uptake is regulated by soil moisture mainly due to the following two aspects: on the one hand, low uptake rates usually occur in conjunction with high soil moisture, which restricts the diffusion of atmospheric CH4 into soil; on the other hand, despite high gas diffusivity, microbial oxidation is often limited, when methanotrophs are subject to significant water stress, especially in extremely dry soils. The effect of soil moisture on the CH4 uptake of steppe soils in our study also followed this pattern when temperature was not the limiting factor. The highest CH4 uptake rates were observed at WFPS values from 13% to 23%, and CH4 uptake rates decreased gradually at WFPS greater than 23% and less than 13%. This result is consistent with a previous report for a semiarid prairie in North America [Mosier et al., 1996]. Only during the spring thaw period were WFPS values of topsoil >50% accompanied by a sharp increase in CH4 uptake (namely at UG99). However, here we assume that CH4 oxidation recovered from hibernation and that this recovery required both sufficient soil moisture and temperatures around or higher than 0°C.
4.2. Nongrowing Season Contribution
 The mean annual CH4 uptake of our investigated sites was 2.9 kg C ha−1 (2.1–3.7 kg C ha−1), which is within the range reported for CH4 uptake of temperate grassland soils [Mosier et al., 1991, 1996; Wang et al., 2005; Qi et al., 2005] and temperate forest soils [Crill, 1991; Ambus and Christensen, 1995; Dobbie and Smith, 1996; Butterbach-Bahl and Papen, 2002]. Nongrowing season (October–April) CH4 uptake at our investigated sites accounted for approximately 30% of the annual total uptake (UG99: 36%; WG01: 25%). For same typical semiarid steppe in Inner Mongolia, Wang et al.  estimated that nongrowing season constitutes to 12%–17% of the annual cumulative CH4 uptake. This estimate, however, was based on measurements collected monthly in the nongrowing season and weekly in the growing season. The contribution of the nongrowing season to the annual total CH4 exchange was approximately −13% (range: −43% to +21%) in the study by Qi et al.  since during this time mainly net CH4 emissions was observed. However, this report lacks sufficient temporal resolution, with measurements performed only every 2 months in the nongrowing season and every 2 weeks in the growing season. The higher values for the contribution of the nongrowing season to total annual CH4 uptake found in our study for this steppe ecosystem were obviously the consequence of a higher density of observations. For a semiarid prairie in North America, a winter (November–February) contribution of 15%–30% to the annual uptake, based upon weekly observations for several years, was reported by Mosier et al. . The winter contribution was calculated to be approximately 10%–20% in our study, which was comparable to the study by Mosier et al. .
 In 2004 and 2005, methane flux measurements were also collected at the same sites, UG99 and WG01, during the growing season by Liu et al. . The values of annual CH4 uptake in 2004 and 2005, respectively, were reestimated at 2.8 and 3.7 kg C ha−1 for UG99 and 1.51 and 1.54 kg C ha−1 for WG01, based on our results for the contribution of the nongrowing season to the total annual CH4 fluxes (36% at UG99 and 25% at WG01). The annual CH4 uptake values in our year of observation, 2007–2008, was 3.7 kg C ha−1 for UG99 and 2.1 kg C ha−1 for WG01; thus, the mean annual CH4 uptake values were 3.4 ± 0.5 kg C ha−1 at UG99 and 1.7 ± 0.4 kg C ha−1 at WG01, combining the estimates for 2004 and 2005 with the measurements for 2007–2008. Hence, the annual CH4 uptake at WG01 was 52% lower than that at UG99, which might indicate that winter grazing practices have a suppressive effect on annual CH4 uptake.