Editor's commentsThis article investigates the conjecture that the Arctic, specifically Arctic ice cover, is more important in determining Northern Hemisphere weather patterns than conventional climate dynamics theory holds. The authors provide support for this hypothesis by correlating summers with abnormal low and high Arctic ice extents with atmospheric variables such as moisture content, surface temperature, pressure and precipitation. It is important to bear in mind that the analysis they provide is statistical only. It is suggestive but cannot taken as proof that the Arctic is important in determining Northern Hemisphere weather patterns. That will require the elucidation of the detailed mechanisms by which the Arctic influences Northern Hemisphere weather. The authors hypothesize possible mechanisms by which the Arctic could influence broader Northern Hemisphere weather patterns. One hypothesis that decreasing ice extents reduces the temperature gradient between high and mid-latitudes resulting in slowing the jet stream (which has important implications for broader weather patterns) has generated much controversy within the climate dynamics community. In 2013 Francis and Stephen Vavrus published a paper that attempts to provide further support for this conjecture and EC will cover this paper in a future article.

The authors investigate the conjecture that Arctic Amplification, which refers to the Arctic warming twice as fast as the rest of the planet, is more important in determining northern hemisphere weather patterns than climatologists have traditionally assumed. As a starting point they take years with extreme summer Arctic ice extents in the period 1979 to 2006 and correlate them with Northern Hemisphere atmospheric conditions such as surface temperature, surface pressure, and precipitation in the following autumn and winter looking for patterns that would support this conjecture. To support the correlations that they find, they suggest possible mechanisms that could lead to the observed patterns.

What is Arctic Amplification ?

The mechanism responsible for Arctic Amplification is the shrinking of Arctic sea ice. Summer sea ice extent has shrunk by more than 11% per decade since 1979. The impact of less sea ice is a greater expanse of dark open water which absorbs more of the sun's energy instead of being reflected by the white sea ice. The extra energy input drives a strong local warming. The result is that surface temperatures are rising twice as fast in the Arctic compared to lower latitudes leading to an increase in surface, autumn air temperatures of 2 to 5 °C over much of the Arctic Ocean during the past decade.

To assess the impact of Arctic Amplification on Northern Hemisphere weather patterns the authors compiled data on summer Arctic sea ice extents for the period 1979 to 2006. They then created two subsets of this data - years when summer sea ice extents exhibited extreme values - either extremely high or extremely low.

Data source for summer ice extents

Observations of sea-ice concentration from the Scanning Multichannel Microwave Radiometer and Special Sensor Microwave/Imager satellite instruments obtained from the National Snow and Ice Data Center were used to calculate summer sea ice extents.

Summer Arctic sea ice extents
Summer Arctic sea ice extents

Figure Actual (solid line) and detrended (dashed line) ice extents north of 65 degrees N.

While summer sea ice extents vary from year to year, there is a multi-year decreasing trend over the investigation period. Mathematically removing this trend from the data allowed the investigators to compare the annual variability in summer ice extents with observed atmospheric conditions and differentiate the yearly changes from the multi-year decreasing trend.

Relationship to Northern Hemisphere weather patterns

To investigate the impact of summer sea ice extremes on Northern Hemisphere weather patterns, the authors compiled data for the entire Northern Hemisphere from different sources for several atmospheric variables such as surface temperature, surface pressure, and precipitation for the period 1979 to 2006 with the final winter extending into 2007. They did this for the autumn months and also for the winter months.

Francis and her colleagues applied statistical analysis to look for correlations between the satellite measurements of summer sea ice extents and the conventional Northern Hemisphere atmospheric weather observations. They then suggested plausible mechanisms by which summer anomalies in the ice cover could cause significant changes in the atmosphere that would persist into the autumn and winter months.

Data sources for atmospheric properties

Monthly mean fields of sea-level pressure, 500-hecto pascal (hPa) geopotential height (the height above sea level where the pressure reaches 500 hPa), surface air temperature, and total-column water vapor content (a measure of total moisture in the atmosphere) were obtained from the National Center for Environmental Prediction/National Center for Atmospheric Research Reanalysis. Monthly mean precipitation data are from the Global Precipitation Climatology Project.

Note: The contours of the 500 hPa surface effectively "determine" our weather - low heights indicate troughs and cyclones while high heights indicate ridges and anticyclones.

Relationship to Autumn atmospheric conditions

Relationship to Autumn surface air temperature

The researchers identified those years in which the sea ice extent was significantly greater and significantly less than the 1979–2006 mean. They then calculated the average air temperatures for autumn months for these years and subtracted the resulting average for the years with high ice extent from the average for the years with low ice extent to create a map showing average temperature differences over the Northern Hemisphere.

a=Surface air temperature
a=Surface air temperature

Figure (a) Surface air temperature differences (low-ice summers minus high ice summers) for Autumns (Oct/Nov) following Septembers with actual ice extents that are less than and greater than one standard deviation from the 1979–2006 mean.

The maps reveal large areas with temperature differences exceeding 3°C centered over the Northern Pacific and Atlantic zones where ice loss has been most pronounced.

Relationship to autumn cloud cover

Varying summer ice extent has been observed to increase total cloud cover over areas of ice loss during autumn. More ice cover means more long wavelength radiation (heat) is absorbed by the sea.

According to estimates derived from satellite sounder retrievals, long wavelength fluxes over the Arctic and adjacent seas are 10 to 40 Watts per square meter (W/sq meter) larger in autumn and winter after summers with significantly less ice relative to those with extensive ice.

Relationship to autumn sea level air pressure

The authors hypothesize that the huge area of significant warming shown in the temperature difference maps would be expected to affect large-scale atmospheric circulation patterns and significantly influence weather in the northern hemisphere.

To test this hypothesis composite difference maps were calculated for autumn sea-level pressure in a similar way for the reduced-versus-extensive ice years.

b=Sea level pressure
b=Sea level pressure

Figure (b) Sea level pressure differences (low-ice summers minus high ice summers) for autumns (Oct/Nov) following Septembers with actual ice extents that are less than and greater than one standard deviation from the 1979–2006 mean.

It was found that sea level pressure is substantially higher over much of the Arctic Ocean and in the northeast Atlantic after summers with significantly reduced sea ice extent. It was also found that higher sea level pressure in high latitudes is compensated by lower pressure in mid-latitudes.

In the North Pacific it was found that the Aleution low pressure zone is weaker after summers with less ice.

In the North Atlantic a relationship was found with the North Atlantic Oscillation (NAO). The NAO refers to the oscillation in the pressure difference between a semi-permanent high near the Azores and a semi-permanent low near Iceland. The pressure difference is high when the NAO index is positive and low when it is negative. A positive NAO index is associated with increased storminess in the North Atlantic. The authors hypothesize that anomalously small ice cover during summer causes a more negative NAO index during the subsequent autumn and conversely above average summer ice cover causes a more positive NAO index.

Relationship to Winter atmospheric conditions

Relationship to winter poleward temperature gradient

The authors hypothesize that low-ice summers reduce the poleward temperature gradient between mid- and high-latitudes. The gradient is lessened by 10 to 20% after summers with a reduced ice cover.

Poleward pressure gradient
Poleward pressure gradient

Figure Poleward gradient in the geometric thickness of the 1000–500 hPa layer (m/km} in the North Atlantic and the North Pacific Oceans during years with above- (red) and below-normal (blue) sea ice during summer. Data extend from September of the extreme ice year to the following March.

The authors speculate that a reduced gradient would result in a decreased polar jet stream and a tendency for weaker semi-permanent low pressure centers near Iceland and the Aleutians to continue for at least six months later into winter. The polar jetstream is about 7–12 km above sea level and separates Arctic from the middle to northern latitudes of North America, Europe, and Asia and their intervening oceans. The wind speeds vary according to the temperature gradient, but generally exceed 92 km/h.

The authors also calculate the correlation between the Arctic-mean September ice extent and atmospheric variables during the following early winter (November to January) over the northern hemisphere.

Relationship to winter air pressure, surface air temperature and water vapour content

A large area of increased 500 hPa heights (the altitude at which the pressure reaches 500 hPa) occurs in winter over much of the Arctic Ocean after summers with less ice than normal.

The regression analysis reveals that low values of summer ice extent are related to higher winter surface air temperature, not just over the Arctic, but throughout the Northern Hemisphere.

Water vapor content in winter also rises as summer ice declines, with large areas of significant correlation over much of the subtropical latitudes.

Relationship to winter precipitation

Correlation analysis of actual summer sea ice extents with winter precipitation data indicates a significant tendency for increased precipitation over much of the region north of approximately 40N after reduced ice cover summers.

The map of winter-mean precipitation differences after summers when ice extent was less than and greater than one standard deviation from the 1979–2006 mean reveals a coherent area of reduced precipitation resulting from reduced summer ice cover over a large region of the northeast Atlantic Ocean extending into northern Europe as well as over much of the U.S. and Alaska. A zone with wet winters extends from Spain eastward across the northern Mediterranean, over waters southwest and northeast of Greenland, and in the Sea of Okhotsk north of Japan.


Based on the correlations found in this investigation the authors argue that the evidence supports their conjecture that Arctic summer ice extents are associated with broad Northern Hemisphere atmospheric weather features occurring well outside the Arctic region during the following autumn and winter. To further support this conclusion they suggest plausible mechanisms by which Arctic Amplification could influence Northern Hemisphere weather patterns. One hypothesis is that decreasing ice extents reduces the temperature gradient between high and mid-latitudes which results in slowing the jet stream which has important implications for broader weather patterns. A potential practical application of this research is that summer sea extents could be an important input into predicting Northern Hemisphere weather in the following autumn and winter.

Francis, Chan, Leathers, Miller, Veron (2009), Geophys. Res. Lett., 36,L07503, doi:10.1029/2009GL037274.