Sea Ice Outlook: August Report

Release Date: 
19 August 2009


The outlook for arctic sea ice in September 2009, based on July data, indicates a continuation of low pan-arctic sea ice extent and no indication that a return to historical levels will occur.

The August Sea Ice Outlook Report is based on a synthesis of 13 pan-arctic contributions (plus nine updates from participating regional groups) utilizing a range of methods.

Figure 1.Figure 1. Distribution of individual outlook values for September 2009 sea ice extent based on July data.

The range of individual outlook values this month is from 4.2 to 5.0 million square kilometers, with more than half of the estimates in a narrow range of 4.4 to 4.6 million square kilometers, representing a near-record minimum. All estimates are well below the 1979–2007 September climatological mean value of 6.7 million square kilometers. The uncertainty / error values, from those groups that provided them, are about 0.4 million square kilometers, thus most of the estimates overlap.

Warm, clear conditions led to significant sea ice melt during the month of July, with some areas of unusually low ice extent and an atmospheric pattern that promotes summer sea ice loss in the Pacific sector of the Arctic. At this point in the sea ice season, the minimum extent will largely be driven by atmospheric conditions, including winds and temperatures.

This August report will be the last formal monthly Sea Ice Outlook this season, though we will provide a short announcement when the sea ice minimum is reached in September. To track the approach to this year's minimum, visit the NSDIC Arctic Sea Ice News and Analysis website, which has daily updates of arctic sea ice.



We received 13 responses for the September Outlook based on July data (Figure 1). Estimates for September sea ice extent are in a narrow range (4.2 to 5.0 million square kilometers), as were the Outlooks based on May and June data. As the submitted uncertainty standard deviations are about 0.4 million square kilometers, most of these Outlook expected value estimates overlap. All sea ice extent estimates for September 2009 are much lower than the past climatological extent of 6.7 million square kilometers.

Figure 1.Figure 1. Distribution of individual outlook values for September 2009 sea ice extent based on July data.

Climate Data for July 2009 and Weather Outlook for August

Also refer to the excellent summary as of 4 August by NSIDC at

The AMSR-E sea ice extent chart for 10 August (Figure 2) shows sea ice extent is unusually low in the Kara Sea, Baffin Bay, north of western Alaska, and along the Russian coast. QuickScat satellite data shows extensive areas of melt across the arctic sea ice cover (provided by Son Nghiem, see also Regional Outlook for figure and access to more images). The time history of extent (NSIDC, Figure 3) indicates that the average pace of ice loss during July 2009 was nearly identical to that of July 2007. Ice loss sped up during the third week of July and slowed again during the last few days of the month. The sea level pressure pattern for July 2009 (Figure 4) continued from that of June, with high pressure on the Beaufort side of the Arctic and lower pressure over Eurasia. The associated wind pattern is similar to the pattern in summer 2007 that promoted extreme melt, although in 2009 the maximum winds are shifted toward the asian side of the Arctic Ocean. The NSIDC report and a contribution by Hori et al., based on MODIS data, document unusually clear skies during July.

As in 2007, an unusually strong high-pressure cell settled over the Beaufort Sea and low pressure was present over Eurasia during June and July 2009, bringing warm air toward the central Arctic and advecting sea ice toward the pole. This high-low pressure pattern has been referred to as the "Arctic Dipole" climate pattern. With the large areal extent of the Beaufort high pressure in 2009, there are also contributions from the negative phase of the Arctic Oscillation and the positive phase of the Pacific North American climate patterns. Historically, the pattern of sea level pressure during the arctic summer is rather flat. The Dipole has been rare, occurring only 20 and 30 years ago, but it is now seen as more common, present in 2005, 2007, and 2009. The cause for the shift to a more frequent Dipole pattern in recent years is an important scientific question, as this pattern promotes summer sea ice loss in the Pacific sector of the Arctic.

Figure 2. AMSR-E ImageFigure 2. AMSR-E Image


Figure 3. Time history of 2009 sea ice extent.Figure 3. Time history of 2009 sea ice extent.


Figure 4. Sea level pressure analysis for July 2009.Figure 4. Sea level pressure analysis for July 2009.


A report from Perovich contrasts the July melt results from two ice mass balance buoys located in the Southern Beaufort Sea and north of Greenland. Through July, the Beaufort buoy had much more surface melt (0.55 m vs. 0.05 m) and bottom melt (0.28 m vs. 0.04 m) than the North Pole buoy. This pattern underscores the difficulty in melting back large stretches of ice at very high latitudes due to the low sun elevation angles, with much less solar heating of ice and surface waters. Looking ahead, surface melt usually begins to wane by mid-August. At the North Pole buoy, large ice concentrations and modest upper ocean heat content indicate that modest additional bottom melting is likely. In contrast, lower ice concentration, significant upper ocean temperature elevation, and a position near the ice edge raised the possibility of significant additional bottom melting at the Beaufort site.

The 8-14 day 500 mb geopotential height and anomaly forecast, valid for 17-21 August (Figure 5), shows a shift from the June-July Dipole pattern to a positive Arctic Oscillation pattern, with low geopotential heights over the central Arctic Basin, bringing to an end the southerly wind flow over the Pacific side of the Arctic Basin. There has been extensive sea ice loss in the marginal seas of the Arctic Ocean. This loss is a recurring feature of the last few years and is seen in the buoy measurements from the Beaufort Sea, supporting the premise that it will be difficult to return to 1990 conditions. The decreases in the NSDIC sea ice timeseries and the presence of the Dipole pattern in June and July, but not much melt activity in the central part of the Arctic Basin, supports a persistence projection from 2008 for September 2009, as represented by the Sea Ice Outlook participants.

Figure 5. 8-14 day forecast of 500 mb geopotential heights valid 13-19 August 09Figure 5. 8-14 day forecast of 500 mb geopotential heights valid 13-19 August 2009.


Key Statements from Individual Outlooks

(Name; Estimate in million square kilometers; Method)

Ordered from lowest to highest

Arbetter et al.; 4.2, Heuristic
As was the case last year, the charts represent the parcels of ice that we believe will survive the summer. It does not, however, represent their final location. Drift due to wind and water will transport along the Beaufort Gyre out of the Beaufort and Chukchi Seas. Some ice in the Amundsen Basin will be transported out into the Barents Sea. The distribution of the ice in September 2009 will be very different than the current August 1 conditions. From the spread of prognostications, we believe the moderate case (4.151 million square kilometers) is the most likely.

Rigor et al.; 4.3, Statistical, age estimates
In comparison to 2007 and 2008, there is much more first-year ice in the Beaufort and Chukchi seas in 2009, which we expect to precondition this area for more extensive retreat than in 2007 and 2008. The age of sea ice in the Transpolar Drift Stream is also younger in the areas north of the East Siberian Sea (~80°N 150°E), which also preconditions this area for more retreat compared to previous years. Although there is some first-year ice in the area of the North Pole, this area also gets much less sunlight, thus is less likely to melt out.

Nguyen et al.; 4.4, Model
With JRA25 July 2009 forcings, the model produces less ice loss than those predicted using May and June data. However, there is a systematically higher sea-ice extent of approximately 0.9 million square kilometers in our model results when compared with that from SSMI for both June and July of this year. A closer look shows that the model overestimates sea ice extent in the Laptev, Kara, and Barents seas, consistent with what Kauker et al. reported last month in their report. We correct this month's estimate by 0.9 million square kilometers to arrive at 4.4 ± 0.5 million square kilometers.

Kauker et al.; Ensemble I - 5.0 with a 3% chance to fall below 2007; Ensemble II - 4.4 with a 36 % chance of falling below 2007, Model
Ensemble I starts from the state of ocean and sea ice as it is calculated by a forward run of NAOSIM driven with NCEP atmospheric data from January 1948 to 11 July 2009. Ensemble II starts from an optimized state derived by applying the variational assimilation system NAOSIMDAS (Kauker et al., 2009) for April and May 2009, followed by a one month forward integration (driven with NCEP June 2009 surface data) until 11 July 2009. NAOSIMDAS is being developed in the EU FP6 project DAMOCLES (

Wang; 4.5, Heuristic
Uses the Dipole Anomaly (DA) index to predict arctic summer ice minima. Using 2009 winter-spring (+0.61) and summer (+1.06) DA indices, now we can project that 2009 summer ice will reach another minimum or at least stay similar to the 2008 level.

Zhang; 4.5, Model
No change from last month's outlook.

Pemberton et al.; 4.6, Model
The probability that the 2009 September mean total sea ice extent will fall below:

  • 2007 satellite derived all-time minimum (4.28 million square kilometers) is 21%
  • 2008 second lowest satellite derived (4.67 million square kilometers) is 56%
  • 2005 third satellite derived (5.57 million square kilometers) is 99%.

Hori et al.; 4.6, Heuristic-remote sensing
Sunny weather was seen over the Arctic Ocean in the former half of this July. It seems to promote sea-ice melting to some extent. Sea level pressure pattern in this June-July period also seems to be getting similar to that in 2007 when the historic sea-ice reduction occurred. However, a lot of sea-ice still remains in East Siberian Sea this year.

Lukovich and Barber; 4.7, Heuristic
An update to differences between surface winds and sea level pressure, and vortex splitting and minimum sea ice extent composites for June 2009 illustrates distinct patterns in the Beaufort, East Siberian, Kara, and Barents seas, and exhibits conditions in June 2009 that are favorable to ice export through Fram Strait, in contrast to those for May 2009. The presence of a sea level pressure high also establishes conditions conducive to ice convergence to the north of the Canadian Archipelago.

Meier et al.; 4.7, Statistical, Heuristic
NSIDC's original assessment of 4.6 million square kilometers, based on the initial amount and distribution of ice age types, remains unchanged. An alternative statistically-based method, using projected rate of sea ice extent decline from previous years' data, was implemented for an updated, complementary estimate. A weighted average, subjectively weighting more recent years higher, was calculated. This approach yields a best estimate of 4.69 million square kilometers, with a range of 4.38-4.91 million square kilometers.

Stern; 4.7, Statistical
No change from last month's outlook.

Kaleschke and Halfmann; 4.9, Statistical
No change from last month's outlook.

Lindsay; 4.9, Model
The anomalous thin ice in the Beaufort Sea and north of the Canadian Archipelago is most influential in making the prediction, but widespread anomalies in this measure contribute to the prediction.

Pan-Arctic Individual PDFs: 
PDF icon Arbetter et al.2.08 MB
PDF icon Rigor et al.1.84 MB
PDF icon Nugyen et al.543.38 KB
PDF icon Kauker et al.718.53 KB
PDF icon Wang786.94 KB
PDF icon Zhang828.32 KB
PDF icon Pemberton et al.844.41 KB
PDF icon Hori et al.6.53 MB
PDF icon Lukovich and Barber728.38 KB
PDF icon Meier et al1.34 MB
PDF icon Stern588.86 KB
PDF icon Kaleschke and Halfmann671.6 KB
PDF icon Lindsay667.27 KB



Regional Arctic MapRegional Arctic Map

For the August 2009 Regional Outlook report, we received six updates from participating groups. Most contributors reported little change from trends identified in the July report.

The outlook for the Northwest Passage is still divided between a high probability of opening along one of the routes as anticipated by ensemble simulations with a coupled ice-ocean model, and observations of later-than-normal break-up of landfast ice, suggesting a small likelihood for clearing of ice.

In the Chukchi and Beaufort seas, ice retreat continues to exceed normal rates, driven both by rapid northward advection of ice in the western Chukchi and significant bottom and surface melt. It is intriguing to follow ice development under atmospheric circulation conditions quite similar to 2007, but with a number of interesting twists on how the summer plays out (see pan-arctic Outlook). At the regional level, there seem to be some distinct differences as well (e.g., with regards to distribution of cloud cover and melt progression). Hence, there is significant potential to obtain valuable insights into how the large scale patterns play out at the local level once the summer season comes to a close. In particular, the record low extent of multiyear ice this year presents somewhat of a wild card, since only very late in the season will it become apparent how much of the first-year ice will be completely removed over the different arctic marginal seas.

Bering, Chukchi, and Beaufort Seas

Contributions from:

  • Eicken, Petrich, and Kaufman
  • Pokrovsky
  • Perovich
  • Fowler, Drobot, and Maslanik

During July, for the Chukchi and Beaufort Sea region, much of the ice evolution patterns persisted from those identified during June. Buoy drift showed substantial northward ice motion superimposed on the ice retreat pattern in the Chukchi Sea, and less so in the Beaufort. As a result, compared to ice conditions in late July 2008, this year there is less ice in the Chukchi and more ice in the eastern Beaufort Sea. Ice retreat in the Chukchi Sea is also promoted by the absence of multiyear ice, which suggests that rapid retreat will continue through August as there is little thick ice present that is likely to survive through September.

Ice conditions at Barrow illustrate how local weather patterns can result in very different ice melt patterns from what is observed at the pan-arctic level. At Barrow, 2009 had the lowest average shortwave flux at the time of break-up during the past decade. Break-up of the landfast ice, governed by a combination of cumulative solar heating and grounding of pressure ridges, was later (July 11) than during any other year since 2000 (Figure 1.). This is in contrast with the large scale picture of cloudiness as presented by Hori et al. in the pan-arctic Outlook, indicating that this summer has comparatively clear skies if you consider the Arctic as a whole.

Surface and bottom melt obtained at a drifting buoy in the eastern Beaufort Sea support this picture with significant surface and bottom melt in July (contribution by Perovich).

While overall agreement between different satellite data sets (in particular active and passive microwave data) is good on the pan-arctic scale, potential discrepancies at the local or regional level provide an additional challenge to tracking ice evolution. This is illustrated in Figure 2, indicating that tongues of sea ice identified by QuikSCAT radar imagery in the eastern and western Chukchi Sea are not apparent in SSM/I passive microwave data. This is likely due to the presence of water at the ice surface masking ice types, and adds considerable uncertainty to projections of regional ice evolution.

Figure 1.Figure 1. Break-up timing and solar shortwave energy incident at the surface (mean and cumulative shown on bottom and left axis, respectively) for 2009 (thick red line) and other recent years. Curves terminate at observed break-up. The shortwave flux is used as an indicator for radiative forcings. The grey area at the top corresponds to the seasonal stage at which ice break-up is imminent and determined by local sea level and winds.

Figure 2.Figure 2. QuikSCAT radar remote sensing scene for 24 July 2009 (courtesy of Son Nghiem, JPL), overlain on an SSM/I passive microwave scene (courtesy of NSIDC) for the same day (shown as black circle with a view through the overlying QuikSCAT data).

Note the good correspondence between SSM/I and QuikSCAT data in the top part of the circle and the lack of ice in the SSM/I scene in regions that show tongues of ice extending along the western and eastern margin of the region in QuikSCAT imagery. To explore images and related data further, visit the swath viewer site here and select the following images (by checking boxes):

Quikscat Ice Types—2009—Quikscat.Icetypes.2009.07.24
SizoNet Misc—IceSSMI.F17.09205.Composite.V1
SizoNet Misc—NIC_MIZ2009205NC_PL_A

The top tool bar then provides access to an "x-ray tool" under the Image Information icon.

Northwest Passage (NWP)

Contributions from:

  • Zhang
  • Gudmandsen et al.
  • Howell and Duguay

Ensemble simulations with a coupled ice ocean model continue to indicate that one route of the Northwest Passage (NWP) is likely to open up (Zhang, see last month's contribution for details).

In contrast, Howell and Duguay find that in late July ice in the NWP Parry Route is still landfast, two weeks later than in 2007 and 2008 (see Figure 3). They assume that opening of the channels is hence very unlikely. Similar to the pattern observed for break-up of landfast ice at Barrow, Howell and Duguay also discuss the fact that early onset of melt in and of itself is not a good indicator of break-up time.

For the Nares Strait region between Greenland and the Canadian Archipelago, Gudmandsen reports on enhanced ice flux and clearing of ice due to failure of an ice barrier earlier in the season. While Nares Strait is not a navigational thoroughfare, the flushing of this ice into adjacent channels and bays may impact ship traffic and scientific operations in other areas.

Figure3.Figure 3. Sea ice concentration (tenths) in the Western Parry Channel region of the Northwest Passage on 27 July 2009. From contribution by Howell and Duguay, with data from the Canadian Ice Service.

High Arctic

Contributions from:

  • Perovich
  • Fowler, Drobot, and Maslanik

Previous observations and anticipation of sluggish melt in the High Arctic continue to hold through July. As indicated by Perovich's report of mass-balance buoy data, snow cover was comparatively deep at North Pole Observatory buoys and hence, by late July, only 5 cm of surface ice melt had been observed. Bottom ice melt was also low (4 cm) compared to previous years. It is not clear whether solar heating of the upper ocean has built up a heat reservoir that may promote bottom melt later in the season.

Northeast Passage / Northern Sea Route

No change or no new information from last month's report.

Greenland / Barents Sea

Contributions from:

  • Gerland and Goodwin

Ice extent in the Greenland and Barents seas for July ranges from below to well below average values. In the eastern Barents, July ice extent is at a record low for this month, surpassing July 2007, while it is roughly average just to the east in the eastern and southern Kara Sea, illustrating how regional variability and different impacts of ice drift, ocean/atmosphere warming, and ice composition on ice retreat complicate assessments of large scale anomalies. (see Figure 4)

Figure4.Figure 4. Ice extent (monthly means, July) southern border of 30% ice concentration, in the Greenland Sea / Fram Strait and Barents Sea, based on passive microwave satellite data (red = July 2009, orange = mean July 1999-2008, green = mean July 1979-2008, purple = mean July 1980-1999). The thin blue line indicates the ice extent for June 2007. From contribution by Gerland and Goodwin.

Regional Individual PDFs: