Corresponding Author: M. Ogi, Japan Agency for Marine-Earth Science and Technology, 3173-25 Showa-machi, Kanazawa-ku, Yokohama, Kanagawa 236-0001, Japan. (email@example.com)
 Strong summertime anticyclonic wind anomalies over the Arctic Ocean, with anomalous flow toward the Fram Strait, during summer months of 2007 contributed to the record-low the Arctic sea-ice extent observed in September of that year. Had the summer winds over the Arctic during the summers of 2010 and 2011 been the same as those in 2007, September sea ice extent would have reached new record lows in those years as well. By regulating the flow of ice toward and through the Fram Strait, variations in low-level winds over the Arctic have contributed to the month-to-month, year-to-year, and decade-to-decade variability of sea ice extent.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The seasonal evolutions of Arctic sea ice extent (SIE) during the summers of 2010 and 2011 are contrasted with that in 2007 in Figure 1. Daily time series of Arctic SIE from June to early October are shown based on data from the National Snow and Ice Data Center (NSIDC) website (http://nsidc.org/index.html). The June SIE in 2010 (Figure 1, top, blue line) was lower than that in 2007 and was the lowest for that calendar month in the 32-year (1979–2010) record. The September SIE in 2010 would have set a new record low had it not been for the fact that the ice retreated more slowly during the summer months in that year than it did in 2007. Hence from early July onward, the SIE in 2010 remained at levels above those observed in 2007. The SIE minimum in September 2010 proved to be the third lowest on record, eclipsed by values in both 2007 and 2008.
Figure 1 (bottom) shows the daily time series of Arctic SIE from June to early October during the summers of 2007 (dashed green line) and 2011 (blue line). In spring and summer of 2011, the Arctic SIE was as low as it was in 2007, but the SIE in September 2011 did not reach record low levels. The SIE minimum in 2011 proved to be the second lowest on record for the period of 1979–2011.
 A number of different mechanisms have been proposed to explain the recent Arctic sea-ice decline. Reduced cloudiness contributed to the unprecedented retreat of sea ice during the summer of 2007 [e.g.,Kay et al., 2008]. The warming of intermediate-depth Atlantic water on the Russian side of the Arctic in recent decades has helped precondition the polar ice cap for the ice loss [e.g.,Polyakov et al., 2010].
 Year-to-year variations in wintertime winds, such as variations associated with the Arctic Oscillation (AO), influence the SIE in subsequent summers by mediating the transport of the thicker multi-year ice toward the Fram Strait, where it exits the Arctic [Rigor et al., 2002; Stroeve et al., 2011]. Such ice preconditioning might have contributed to the low SIE observed in winter and spring, but it cannot explain the different rates of decline for the September SIE of 2010 and 2011 compared with that of 2007.
 Summertime atmospheric conditions play an important role in controlling the variations in Arctic SIE [Ogi and Wallace, 2007; L'Heureux et al., 2008; Screen et al., 2011]. In a previous study based on statistical analysis of data collected prior to 2007 [Ogi and Wallace, 2007], we showed that anticyclonic summertime circulation anomalies over the Arctic Ocean during the summer months favor low September SIE. We also found that the record-low ice summer year 2007 was characterized by a strong anticyclonic circulation anomaly, accompanied by an Ekman drift of ice out of the marginal seas toward the central Arctic and eventually toward the Fram Strait, as evidenced by the tracks of drifting buoys [Ogi et al., 2008]. Here we assess the extent to which year-to-year differences in summer winds over the Arctic might have contributed to the differing rates of retreat of ice during the summers of 2007, 2010, and 2011.
 In this section we show the pattern of 925-hPa wind anomalies in summer 2010 and 2011 in the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis dataset [Kistler et al., 2001] and compare them with the patterns in the corresponding months in 2007.
 In the time series shown in Figure 1 (top) the different rates of retreat of SIE in 2007 and 2010 first become apparent in late June. Figure 2shows the patterns of 925-hPa wind anomalies averaged over (a) May-June (MJ) and (b) July-August-September (JAS) in 2010. The MJ pattern (Figure 2a) is characterized by strong anticyclonic wind anomalies over the Arctic Ocean. The corresponding pattern for JAS is dominated by a cyclonic gyre centered over the Kara Sea. The lower panels in Figure 2 show the corresponding patterns for 2007, which are weak in MJ and strongly anticyclonic in JAS.
 The SIE in 2007 was lower from July to October than in the same months of 2010 (Figure 1, top). The September SIE in 2007 was the lowest of any year in the period 1979–2011. The retreat of sea ice during summer 2007 showed evidence of a substantial Ekman drift, based on the tracks of drifting buoys relative to the monthly mean sea level pressure contours [Ogi et al., 2008]. The anomalous cross-isobar drift acted to enhance the sea ice transport across the Arctic Ocean toward and out of the Fram Strait. The MJ pattern in 2010 (Figure 2a) is dominated by strong anticyclonic wind anomalies with the flow directed toward the Fram Strait, consistent with the rapidly decreasing SIE during these months. In contrast, the JAS pattern in 2010 (Figure 2b) is not characterized by anticyclonic anomalies that would have favored a strong transport of sea ice toward and out through the Fram Strait.
 In the time series shown in Figure 1 (bottom) the September SIE in 2011 did not reach a value as low as in 2007 even though it was running lower than 2007 in JJ. The upper panels in Figure 3show the patterns of 925-hPa wind anomalies averaged over (a) June-July-August (JJA) and (b) September in 2011. The JJA pattern in 2011 is characterized by anticyclonic wind anomalies over the Arctic directed toward the Fram Strait, whereas the September pattern exhibits wind anomalies directed away from the Fram Strait across the central Arctic Ocean toward the Chukchi Sea. The lower panels inFigure 3 show the corresponding patterns for 2007, which are strongly anticyclonic and directed toward the Fram Strait in both JJA and September. In the absence of the late season push by the winds, the ice did not retreat quite as far in 2011 as it did in 2007.
 To investigate whether the degree to which low level wind anomalies regulate the rate of retreat of Arctic sea ice during the summer months, we show in Figure 4a scatter plot of the rate of retreat of SIE, as inferred from a centered two-month difference, and a monthly 925 hPa wind index that provides a quantitative measure of the degree to which the wind anomaly field for that month resembles the patterns for months with large sea ice retreat. The wind index was created by projecting the 925-hPa wind anomaly field for each month from June 1979 onward upon the pattern formed by correlating the anomalous the wind field for each month with the monthly time series of the rate of retreat of sea ice. Positive values of the index are indicative of an anomalous 925 hPa wind pattern characteristic of summer months with an anomalously large retreat of sea ice and vice versa. The domain used in these projections is the oceanic region north of 65°N. As in our previous study [Ogi et al., 2010], we weight each grid point by the cosine of its latitude so that equal areas receive equal weights.
Figure 4shows a scatter plot of the rate of change of SIE versus the 925-hPa wind index as inferred from 2-month centered differences, based on wind data for June-September for the period of record 1979–2011. The two variables are correlated at a level of −0.65 for all years and −0.54 for years other than 2007–2011: both correlations are statistically significant at the 99.9% level. Hence, month-to-month variations in summer winds over the Arctic are reflected in the variability of the rate of decline of SIE, not only during the three years emphasized in this study, but more generally. Our previous paper [Ogi et al., 2010] includes a similar analysis but on a year-by-year basis.
 We have shown evidence that low level winds over the Arctic, play an important role in mediating the rate of retreat of sea ice during summer. Anomalous anticyclonic flow over the interior of the Arctic directed toward the Fram Strait favors rapid retreat and vice versa. We have argued that the relative rankings of the September SIE for the years 2007, 2010 and 2011 are largely attributable to the differing rates of decrease of SIE during these summers, which are a consequence of year-to-year differences in the seasonal evolution of summertime winds over the Arctic.
 Patterns of wind and cloudiness over the Arctic are related. Hence, the empirical relationships that we have described between the 925 hPa wind field and the rate of decrease of SIE reflect not only variations in sea ice motion but also differences in sky cover, which mediates both the flux of solar radiation reaching the surface [Screen et al., 2011] and the downward flux of infrared radiation from atmosphere [Francis and Hunter, 2006].
 In the summers from 2007 onward, the low level circulation over the Arctic has been much more anticyclonic than in prior years, as documented in Figure 5. The change in the circulation has contributed to the record lows in recent September SIE but the thinning of the ice [Rigor and Wallace, 2004; Zhang et al., 2008; Kwok and Rothrock, 2009] has been an important factor as well. It is not clear why anticyclonic wind anomalies have been prevalent in recent years.
 We would like to thank J. Walsh, C. Deser, and J. Screen for helpful comments and suggestions. We thank the NSIDC for providing sea ice data and maps. M. Ogi is supported by a Grant-in-Aid for Young Scientists (B) (KAKENHI, 22740317) and a Grant-in-Aid for Scientific Research on Innovative Areas (22106010). J.M. Wallace is supported by the U.S. National Science Foundation under grant AGS-1122989.
 The Editor thanks Clara Deser and James Screen for assisting with the evaluation of this paper.