1. Introduction
Landfast ice is the defining feature of the Arctic coast. While various definitions of landfast ice exist, Barry et al. (1979) provide perhaps the most comprehensive description, defining landfast ice as ice that (i) “remains relatively immobile near the shore for a specified time interval,” (ii) “extends from the coast as a continuous sheet,” and (iii) “is grounded or forms a continuous sheet which is bounded at the seaward edge by an intermittent or nearly continuous zone of grounded ridges.” Covering the shallow shelves and the narrow channels among high Arctic islands, landfast ice can extend offshore from a few kilometers to several hundred kilometers. The broadest span of about 100–200 km is off the coasts of the East Siberian, Laptev, and Kara Seas, where few shear or pressure ridges exist because the local prevailing wind is offshore (Reimnitz et al. 1994). This broad ice cover is often in sharp contrast to the Beaufort Sea coast where the seaward landfast ice edge is bounded by a narrow strip of highly deformed ridges and rubble ice—the stamukhi zone (Reimnitz et al. 1978). This ice regime forms primarily in response to westward nearshore pack ice motion driven by prevailing easterly winds and currents. The stamukhi zone in winter may expand and become partially grounded. This grounding helps stabilize the landfast ice cover, affecting its evolution and breakup patterns during its growth and melting periods.
The variation of Arctic landfast ice shows a strong seasonal cycle. Each year in fall, ice starts to form close to the coast sometime after the air temperature drops consistently below 0°C. As winter progresses, the ice cover expand offshore and can reach between one and two meters thick at the end of each growing season. As air temperatures rise again above freezing in early summer due to the increased solar radiation and the offshore pack ice gradually retreats northward, landfast ice starts to break up along the coast. At the end of the melting period, most landfast ice disappears from the coastal waters. Typically, landfast ice stays on the shelves for seven to nine months each year, although multiyear fast ice plugs may survive for several years in the Canadian Arctic Islands (Melling 2002).
As noted in many studies, a host of changes are taking place in the Arctic. The changes include rising surface air temperature (Serreze and Barry 2011), increasing Arctic cyclone frequency and intensity (Zhang et al. 2004), retreating ice edge (e.g., Maslanik et al. 2007; Stroeve et al. 2012), increasing coastal erosion and runoff (Jones et al. 2009; Peterson et al. 2006), a shift to younger, thinner sea ice (Kwok and Cunningham 2010; Maslanik et al. 2011), and a less consolidated coastal ice cover and a shortened season as observed directly by the residents of the northern coast of Alaska (Krupnik and Dyanna 2002).
Because of the limited deformation and mobility of ice in the shallow waters, the growth and melt of landfast ice are controlled largely by thermodynamic factors (Barry et al. 1979; Brown and Cote 1992; Flato and Brown 1996; Dumas et al. 2005). This characteristic makes landfast ice a potentially important indicator for the environmental changes related to Arctic warming. However, most studies on landfast ice have so far focused on regional scales. Analyzing gridded sea ice charts and microwave satellite imagery, Divine et al. (2003) reported a 12% decrease in landfast ice in the Kara Sea from 1953 to 1990. Polyakov et al. (2012) examined landfast ice thickness records available along the Siberian coast. They argued that the local surface melting in the eastern Arctic due to atmospheric thermodynamics caused ~0.3 m of thickness loss of landfast ice over the last several decades. Brown and Cote (1992) and Melling (2002) analyzed long-term changes in landfast ice thickness observed in the Canadian Arctic Archipelago (CAA) over the period 1950–89, finding little change. Comparing with observations from the 1970s, Mahoney et al. (2007) found a decreased ice season along the coasts of northern Alaska and northwestern Canada, but they did not notice any significant difference in the maximum extent of landfast ice between 1996 and 2004 and the 1970s. Most recently, Galley et al. (2012) examined the Canadian Ice Service (CIS) ice charts from 1983 to 2009. They observed significant decreases in landfast ice duration for many areas of the Canadian Arctic. While these previous studies have suggested considerable changes in Arctic landfast ice cover, the different time periods and regions studied to date make it difficult to construct a comprehensive picture of large-scale, long-term variability.
In this paper, we report the interannual variability of landfast ice extent and the length of landfast ice duration over the Arctic Basin during a 31-yr period, focusing on observations from winter months between January and May when landfast ice cover is relatively stable. In the following, we will provide in section 2 a general description of the data that are used to derive the time series of landfast ice extent, followed by a discussion on data uncertainty. The results on spatial and temporal variations of landfast ice as well as long-term trends in landfast ice extent around the Arctic coastal regions will be shown in section 3. And, finally, we offer some discussions in section 4 and a summary in section 5.
2. Data
a. National Ice Center sea ice charts
The data used in this study come from the weekly or biweekly ice charts produced by the U.S. National Ice Center (NIC) from 1972 through 2007. These charts are primarily intended to assist operations in ice-covered waters. The early charts were available only in paper format, whereas later charts also come in digital format. Mapped with polygons containing information on sea ice concentration and ice types, a chart identifies areas of landfast ice and their boundaries. Figure 1 is an example of the NIC ice chart for the week of 10–14 May 2004. The polygons corresponding to landfast ice are indicated with color of gray and 100% concentration.
An example of NIC weekly sea ice chart for the Beaufort Sea for 5–9 May 2003. Landfast ice along the coast is delineated by areas of light gray with a concentration of 10 as indicated by the Egg Code. Various sources of data were used to cover the area of the weekly chart.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of NIC weekly sea ice chart for the Beaufort Sea for 5–9 May 2003. Landfast ice along the coast is delineated by areas of light gray with a concentration of 10 as indicated by the Egg Code. Various sources of data were used to cover the area of the weekly chart.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of NIC weekly sea ice chart for the Beaufort Sea for 5–9 May 2003. Landfast ice along the coast is delineated by areas of light gray with a concentration of 10 as indicated by the Egg Code. Various sources of data were used to cover the area of the weekly chart.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
A gridded, digital version of these charts covering 1972 through 2007 is the basis of this study. The collection is titled “National Ice Center Arctic sea ice charts and climatologies in gridded format” (National Ice Center 2006, updated 2009). Documentation for this product includes a history of ice chart digitization and processing (see http://nsidc.org/data/docs/noaa/g02172_nic_charts_climo_grid/). In brief, the digital gridded chart series is composed of two parts: The first set is digital charts made for the Environmental Working Group (EWG) Joint U.S.–Russian Arctic Sea Ice Atlas created under the joint effort of the EWG and NIC (Tanis and Smolyanitsky 2000). The ice atlas, spanning 1972–1994, includes 1199 weekly gridded Arctic-wide sea ice charts that originally were NIC operational paper charts. Figure 2 shows a gridded sea ice chart produced by the EWG. Prior to the EWG effort, the paper charts had been converted to Sea Ice Grid (SIGRID) format at a nominal cell size of about 25 km (15′ in latitude and variable in longitude) (Thompson 1981). Under the EWG project, the digital SIGRID charts were converted to Equal Area Scalable Earth Grid (EASE-Grid) format (Brodzik and Knowles 2002) at 25-km resolution.
An example of a weekly sea ice chart from the Joint U.S.–Russian Arctic Sea Ice Atlas, produced by the Environmental Working Group. Area of landfast ice with 100% of concentration is indicated with the color purple.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of a weekly sea ice chart from the Joint U.S.–Russian Arctic Sea Ice Atlas, produced by the Environmental Working Group. Area of landfast ice with 100% of concentration is indicated with the color purple.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of a weekly sea ice chart from the Joint U.S.–Russian Arctic Sea Ice Atlas, produced by the Environmental Working Group. Area of landfast ice with 100% of concentration is indicated with the color purple.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
The second set of the records, from 1995 through 2007, was digital NIC operational charts in vector interchange format. Because the digital, GIS-compatible format is difficult to use, these charts were converted to EASE-Grid and made compatible with the EWG product by applying the same land mask for the entire time series from 1972 to 2007. Examination of all charts, as well as counts of the number of landfast ice grid cells in the first four years of data, does not show the clearly marked annual cycles of landfast ice of later years, so these early years were discarded. The remaining data cover a 32-yr period from January 1976 to December 2007.
Because a NIC ice chart was produced weekly or biweekly (post-2001), the ice conditions presented in the chart are a composite of ice conditions over the time window of the chart rather than a snapshot. When landfast ice changes within the period of the time window, the new positions are determined by the NIC analyst on the day the chart is produced based on all available information. As noted earlier, one aspect of the landfast ice definition is that the ice remains stationary or quasi-stationary during a set period. For example, Mahoney et al. (2007) used a 20-day period, a minimum time window for three consecutive synthetic aperture radar (SAR) satellite overpasses to cover the entire Alaskan Beaufort coast. In this study, we do not apply such criteria but use NIC’s identification instead, which implicitly assumed that the ice was relatively stationary and fixed to shore within a chart period. Such a definition is similar to the terminology adapted by Reimnitz et al. (1978) and Barry et al. (1979).
Figure 3 shows a 32-yr record of the chart-derived landfast ice extent for the entire Northern Hemisphere from 1976 to 2007. Superimposed on the distinct annual cycle, landfast ice extent displays a large interannual variability. During the period of record, the maximum extent appears to occur during the 1980s, followed by a general decline through the late 1990s. This pattern is consistent with other sea ice observations in the Arctic Basin during that time but does not reflect the continued large decline seen in the pack ice cover since 2003.
Time series of weekly landfast ice extent for the entire Northern Hemisphere from 1976 to 2007. The biweekly sea ice charts after 2001 were interpolated to weekly using linear interpolation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Time series of weekly landfast ice extent for the entire Northern Hemisphere from 1976 to 2007. The biweekly sea ice charts after 2001 were interpolated to weekly using linear interpolation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Time series of weekly landfast ice extent for the entire Northern Hemisphere from 1976 to 2007. The biweekly sea ice charts after 2001 were interpolated to weekly using linear interpolation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
b. Data uncertainty
Chart production incorporates various data that become available over time. The early charts were produced using primarily infrared and visible-band satellite imagery from Advanced Very High Resolution Radiometer (AVHRR) and Operational Linescan System (OLS) instruments with spatial resolutions of about a kilometer. These satellite observations are integrated with a variety of sources including aerial reconnaissance, surface observations, airborne and ship reports, model output, and climatology. Around the early 1990s, SAR images with a resolution as high as 100 m were introduced into ice chart production and computerized analysis techniques were developed, which helped landfast ice detection (Dedrick et al. 2001). A major improvement occurred about 1996, when RADARSAT SAR observations were routinely used in producing ice charts. Therefore, the accuracy of the ice charts improved over the years as the availability and the precision of data increased.
The accuracy of the gridded NIC sea ice charts is affected by two primary factors: the mix of data sources and the transformation of charts into gridded products (Dedrick et al. 2001; Partington et al. 2003). In cloud-covered sea ice regions, AVHRR and OLS imagery were augmented by nowcast, climatology, and microwave satellite imagery introduced in 1972. Because there is no means of revisiting with comparison datasets and the historical paper charts that were processed by EWG, it is difficult if not impossible to evaluate their uncertainty. Likely, the errors in landfast ice position are random and small when ice cover becomes stable in winter. This nature of the ice has allowed NIC analysts to use time sequences of satellite imagery to identify landfast ice that is sometimes separated by flaw leads from unstable, mobile ice that is extensively deformed (Reimnitz et al. 1978). The skill of each analyst also introduces uncertainties that are difficult to quantify, but the human error introduced by different analysts tends to be random in nature and thus to cancel out over time.
Here we attempted to evaluate the uncertainty of the sea ice charts by comparing post-EWG charts with observations derived from RADARSAT SAR imagery along the Beaufort Sea coast between 1997 and 2004. With the same SAR dataset, Mahoney et al. (2007) used the geo-coded landfast ice polygons to define the seaward landfast ice edge (SLIE) at 100-m spatial resolution approximately 30 times per season. Since it required three consecutive RADARSAT overpasses to sample the whole Beaufort coast in roughly 20 days, landfast ice is presumed to remain stationary during this period.
To compare with the SAR-derived SLIE, we obtained the original weekly sea ice charts that are publicly available from the NIC. The original charts that are in e00 ArcInfo vector interchange file format were converted to geo-coded shapefiles, from which we determine the locations of landfast ice boundaries. There are typically three to four weekly ice charts that overlap with a single, 3-week SAR period; thus, we estimated the mean chart-derived SLIEs averaged over the time span of each SAR mosaic. Figure 4 shows an example of landfast ice extent mapped by a SAR mosaic. Between 5 and 28 May 2000, there are three weekly sea ice charts, from which we estimated averaged SLIEs at each 1.25° longitude interval inside of the SAR observation box from about 154° to 136.5°W. We restricted our comparison pairs from January to May, when coastal ice is relatively stationary.
Mosaic of landfast ice extent off the coast of Alaska mapped by RADARSAT synthetic aperture radar (light gray) for the period between 5 and 28 May 2000. Landfast ice polygons from the NIC sea ice charts (dark gray) are also plotted for comparison. The light gray box marks the sampling boundary for the SAR mosaic. As plotted in the figure, there are three ice charts that correspond to this SAR mosaic period.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Mosaic of landfast ice extent off the coast of Alaska mapped by RADARSAT synthetic aperture radar (light gray) for the period between 5 and 28 May 2000. Landfast ice polygons from the NIC sea ice charts (dark gray) are also plotted for comparison. The light gray box marks the sampling boundary for the SAR mosaic. As plotted in the figure, there are three ice charts that correspond to this SAR mosaic period.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Mosaic of landfast ice extent off the coast of Alaska mapped by RADARSAT synthetic aperture radar (light gray) for the period between 5 and 28 May 2000. Landfast ice polygons from the NIC sea ice charts (dark gray) are also plotted for comparison. The light gray box marks the sampling boundary for the SAR mosaic. As plotted in the figure, there are three ice charts that correspond to this SAR mosaic period.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Figure 5 is a scatterplot of SLIEs expressed as latitudes measured from SAR (LatSAR) and averaged from sea ice charts (
A scatterplot of landfast ice seaward boundaries derived from SAR mosaics and from the NIC sea ice charts for the coast of Alaska over the period 1997 through 2004.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
A scatterplot of landfast ice seaward boundaries derived from SAR mosaics and from the NIC sea ice charts for the coast of Alaska over the period 1997 through 2004.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
A scatterplot of landfast ice seaward boundaries derived from SAR mosaics and from the NIC sea ice charts for the coast of Alaska over the period 1997 through 2004.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
To further quantify the difference between the two, we estimated the distances 
Uncertainty is also introduced when transforming the original ice charts into SIGRID and then EASE-Grid format. During the gridding, a single grid cell along the landfast ice boundary could be shifted in either direction. The uncertainty due to this shifting was discussed in product documentation (http://nsidc.org/data/docs/noaa/g02172_nic_charts_climo_grid/) and is believed to be random in nature and no more than ±25 km (or one EASE-Grid cell). Uncertainty with such a magnitude can be problematic in coastal areas where landfast ice extent is narrow but less critical when averaging over space and time or where landfast ice coverage is broad.
Changes in instrumentation and processing procedures over the years may also affect the constructed time series. Two such transitions are of particular note. The first spans from December 1994 to January 1995. Charts constructed prior to 1995 were digitized by encoding them in SIGRID; these were later converted to EASE-Grid. Charts from 1995 and later were gridded directly from vector format. Another major transition occurred in 1992, when high-resolution (30–240 m) images from the European Remote Sensing Satellite 1 (ERS-1) SAR were incorporated, and then in 1996, when RADARSAT SAR (25–200-m resolution) observations began to be used routinely in ice chart production. The incorporation of SAR data enhances the ability to identify landfast ice under all weather conditions and at a fine resolution, which tends to increase the estimation of sea ice total concentration (http://nsidc.org/data/docs/noaa/g02172_nic_charts_climo_grid/). To test the potential effects of these changes on data consistency, we applied a regime shift scheme (Rodionov 2004) to the landfast ice time series. Using a Student’s t test and at 95% probability, we found no statistically significant shift in means around the periods 1994/95 and 1996/97. For a further test, we also computed the variances of the winter (January–May) means for the pre-1995 EWG charts and the post-1994 EASE-Grid charts. The variances of the two nonoverlapping time series are accepted as being from the same population at a 99% confidence level. Overall, we did not observe any significant shift in the time series at the times of the major changes in source data and processing schemes, and the area of landfast ice is reasonably continuous across the data transition from 1994 to 1995 (Fig. 3). In a few isolated cases, landfast ice was mislabeled as other types, errors that were easily identified using consecutive charts.
Because of the nature of the dataset, it is difficult if not impossible to quantify the exact uncertainty in the derived landfast ice extent from ice charts. Therefore, we keep in mind the limitation of the data, focusing our analysis on the large-scale patterns and the relative change over the long time period. To minimize the uncertainty, we average landfast ice extent over space and time, computing the winter averages over broad coastal regions.
3. Results
a. Seasonal variation
Along the Arctic coasts, landfast ice appears in the fall when the upper ocean cools to the freezing point and the sea condition becomes favorable for ice formation. As the winter progresses, the entire inner shelf will become ice covered. Depending on the location, landfast ice may persist throughout winter or breakup and reform periodically.
The timing of ice return in each year depends largely on local conditions often influenced by the large-scale weather and climate conditions. To examine the spatial occurrence of landfast ice, we computed its frequency from January to May when the ice cover is stable. The higher the frequency, the more persistent the landfast ice is from January to May. Figure 6 shows the mean frequency of occurrence averaged from 1976 to 2007 for the whole Arctic Basin. Landfast ice with a frequency of occurrence greater than 75% is mostly located on the inner shelves around the Arctic Basin, in particular over the eastern Arctic and in the channels of the CAA. Toward the outer shelves of the eastern Arctic, the Amundsen Gulf, and the inner M’Clure Strait, the occurrence decreases to about 40%–75%. It drops further to less than 20% at the ice seaward boundaries and in the regions of Gulf of Boothia and Lancaster Sound where the low occurrence reflects active ice motion in the regions (Agnew et al. 2008). Landfast ice occurrence also displays a strong seasonal variation. The frequency of landfast ice occurrence decreases to less than 50% in most coastal waters during melting season and drops to zero by late summer, except in some areas north of the CAA.
Winter (January–May) landfast ice occurrence in the Arctic (percentage of time when ice is present from January through May), averaged from 1976 to 2007.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Winter (January–May) landfast ice occurrence in the Arctic (percentage of time when ice is present from January through May), averaged from 1976 to 2007.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Winter (January–May) landfast ice occurrence in the Arctic (percentage of time when ice is present from January through May), averaged from 1976 to 2007.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
The annual cycle of total landfast ice extent is shown in Fig. 7 for the whole Arctic Basin. The ice coverage starts to increase during October, reaching its maximum of 20 × 105 km2 in late April, diminishing in July, and completely melting away in most coastal regions by late summer. The seasonal cycle from each year shows considerable variations from the climatological mean averaged over 1976–2007, with substantial decrease in landfast ice extent in recent years. The reduction in winter landfast ice cover since the early 1990s has resulted in a decrease in the amplitude of the annual cycle, peaking toward earlier seasons between 1985 and 2007. The reduced seasonal amplitude is in contrast to that of perennial ice in the deep basin, where the seasonal cycle of Arctic sea ice extent is consistently amplified under the warmer climate simulated in the GCMs (Zhang and Walsh 2006). This could signal a different response of sea ice between the deep water and coastal region to the climate warming. The changes in winter landfast ice and the resulting impact on its seasonal cycle can have a large impact on its interannual variation.
Seasonal cycle of Arctic landfast ice area during 1976–2007 over the Northern Hemisphere. Lines in gray are annual cycles from each year, line in black represents the 31-yr mean, and lines in colors represent the three decadal periods.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Seasonal cycle of Arctic landfast ice area during 1976–2007 over the Northern Hemisphere. Lines in gray are annual cycles from each year, line in black represents the 31-yr mean, and lines in colors represent the three decadal periods.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Seasonal cycle of Arctic landfast ice area during 1976–2007 over the Northern Hemisphere. Lines in gray are annual cycles from each year, line in black represents the 31-yr mean, and lines in colors represent the three decadal periods.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
b. Long-term behavior in landfast ice extent
To examine the regional and long-term changes in landfast ice coverage, we divided the Arctic and subarctic seas into 17 regions (Fig. 8). For each region, we derived a time series of weekly landfast ice extent for the period of 1976–2007, as shown in Fig. 9. On top of the distinct seasonal cycle, the local maxima, typically occurring in late winter to spring, display some distinct long-term variations.
Map of landfast ice regions. See Table 1 for region names.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Map of landfast ice regions. See Table 1 for region names.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Map of landfast ice regions. See Table 1 for region names.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Time series of weekly landfast ice extent for the 17 regions shown in Fig. 8.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Time series of weekly landfast ice extent for the 17 regions shown in Fig. 8.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Time series of weekly landfast ice extent for the 17 regions shown in Fig. 8.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
To best represent the interannual variation, we calculated the winter mean averaged from January through May of each year using the time series from Fig. 9. Averaging over large regions and multiple charts helps minimize the effect of local errors in a single weekly chart. The statistical results and trends for each region are given in Table 1. Over the 32-yr period of record, the chart data indicate a widespread decrease in winter-averaged landfast ice extent within the Arctic Basin. The significant changes occurred along the coast of Svalbard, in the Laptev and Chukchi Seas, and on the north coast of the CAA. Negative but statistically insignificant changes are also found in the Kara, East Siberian, and Beaufort Seas, as well as the CAA and in subarctic areas such as Baffin and Hudson Bays, East Greenland, and the Labrador Sea. In the CAA, most ice is classified as landfast and accounts for 80%–100% of the total ice cover from January to May (Fig. 6). In an early study using passive microwave data, a negative but not statistically significant trend was also found in the CAA in the winter total sea ice extent between 1979 and 2006 (Parkinson and Cavalieri 2008).
Mean winter landfast ice area (103 km2) for 17 coastal regions (see Fig. 8) and the whole Northern Hemisphere for the period 1976–2007. Annual change (103 km2 yr−1) is the slope of the least squares fit, and the percentage is change per decade relative to the mean; R is the ratio of the magnitude of the slope to the standard deviation of the slope. In the last column, values in boldface are significantly different from zero with 95% probability or above according to the F test. Values in italic boldface are significant with 99% probability.
In spite of a nonsignificant increase in winter landfast ice area in the Barents, Baltic, and Bering Seas and Sea of Okhotsk, the overall change for the Northern Hemisphere is significantly negative between 1976 and 2007. Figure 10 shows the long-term variation in the winter mean landfast ice extent for the Northern Hemisphere. During the 32-yr period, the ice coverage was most extensive in the early to mid-1980s, followed by a decrease starting around 1990. A linear fit of the winter landfast ice extent shows a generally downward trend over the record period at −12.3 (±2.8) × 103 km2 yr−1 with a significance of 99% based on an F test. It is clear, however, that landfast ice extent exhibits decadal variability rather than a strict linear decline. Compared to the peak ice coverage in 1979–87, landfast extent after 1990 was reduced by about 15%; the overall downward trend is about −7 (±1.5)% decade−1 relative to the mean. This percent change is between the value of −3.7% decade−1 reported by Parkinson and Cavalieri (2008) for the total sea ice extent in the Northern Hemisphere and the value of −17.2% decade−1 reported by Comiso (2012) for perennial sea ice extent. The significant decline in winter mean landfast ice extent occurred during the same period when there were widespread decreases in sea ice thickness and multiyear ice population (e.g., Rothrock et al. 1999; Tucker et al. 2001; Yu et al. 2004; Maslanik et al. 2011). It is noted that the decrease of pack ice is primarily in summer, as opposed to landfast ice extent that is decreasing during winter.
Long-term variation of landfast ice extent for the Northern Hemisphere.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Long-term variation of landfast ice extent for the Northern Hemisphere.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Long-term variation of landfast ice extent for the Northern Hemisphere.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
A number of studies have shown that the long-term declines in sea ice thickness and extent are accompanied by changes in the length of the sea ice season (e.g., Smith 1998; Markus et al. 2009). Similar changes in landfast ice seasonality were observed with ice chart data over the 31-yr period. As indicated in Fig. 11, the onset of the annual ice growth, approximated by weeks when landfast ice extent exceeds 15% of the annual maximum, gradually delayed by about two weeks from week 42 (mid-October) in the 1980s to week 44 since the early 1990s. As a result, there was an overall slowdown in landfast ice annual expansion well into the winter season. The changes in spring are relatively smaller when compared to those in fall. The end of spring melt, approximated by the weeks when ice extent drops below the 15% of the annual maximum, has remained steady over the years. However, the melt onset corresponding to the initial decrease of annual maximum in mid-May shows a noticeable long-term change, starting around week 19 during the 1980s to a week or so later in the 1990s and early 2000s. The net result is a shortened ice growth season.
Interannual variations of weekly landfast ice extent for the Northern Hemisphere as indicated by the color bar on top (units 105 km2). Lines in white, indicating 15% of maximums for the Northern Hemisphere, were plotted to highlight the gradual changes in the length of landfast ice duration.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Interannual variations of weekly landfast ice extent for the Northern Hemisphere as indicated by the color bar on top (units 105 km2). Lines in white, indicating 15% of maximums for the Northern Hemisphere, were plotted to highlight the gradual changes in the length of landfast ice duration.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Interannual variations of weekly landfast ice extent for the Northern Hemisphere as indicated by the color bar on top (units 105 km2). Lines in white, indicating 15% of maximums for the Northern Hemisphere, were plotted to highlight the gradual changes in the length of landfast ice duration.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
c. Changes in season duration
To quantify the observed changes in the ice growth and melt patterns shown in Fig. 11, we examine the length of landfast ice duration. Again, it is defined as the time between annual breakup and freeze-up, identified by the points at which the landfast ice extent exceeds or drops below 15% of its local maximum. Although arbitrary, the 15% threshold helps avoid the relatively large uncertainty in the charts during the early part of freeze-up or near the end of the breakup. Figure 12 is an example of how such a threshold is calculated for a typical annual cycle. The difference between these two dates is then defined as the landfast ice duration, from which we compute the means and changes for each of the17 regions given in Table 2. On average, the annual mean duration of landfast ice is about 30 weeks or 7–8 months, ranging from the shortest duration of 12 weeks in the Baltic Sea to the longest of 43 weeks along the north coast of the CAA.
An example of how landfast ice duration is defined for a typical annual cycle. The horizontal dashed line represents the threshold of 15% of the annual maximum, and the two solid black dots indicate the first week (or the last week) when landfast ice extent exceeds (or drops below) 15% of its local maximum. The length of the landfast ice season is defined as the time period between the two dots.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of how landfast ice duration is defined for a typical annual cycle. The horizontal dashed line represents the threshold of 15% of the annual maximum, and the two solid black dots indicate the first week (or the last week) when landfast ice extent exceeds (or drops below) 15% of its local maximum. The length of the landfast ice season is defined as the time period between the two dots.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
An example of how landfast ice duration is defined for a typical annual cycle. The horizontal dashed line represents the threshold of 15% of the annual maximum, and the two solid black dots indicate the first week (or the last week) when landfast ice extent exceeds (or drops below) 15% of its local maximum. The length of the landfast ice season is defined as the time period between the two dots.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
Averaged length of landfast ice season (weeks) for 17 regions (see Fig. 8) and the Northern Hemisphere for the period 1977–2007. Decadal change (week decade−1) is estimated from the least squares fit, and the percentage is change per decade relative to the mean; R is the ratio of the magnitude of the slope to the standard deviation of the slope. In the last column, values in boldface are significantly different from zero with 95% probability or above according to the F test. Values in italic boldface are significant with 99% probability.
The change in landfast ice duration is spatially variable. Except for the Barents, Baltic, and Bering Seas, the coastal regions in the eastern Arctic all have experienced a shortened ice season. A decrease in landfast ice duration was found to be significant in Svalbard and Franz Josef Land and farther west in the Laptev, East Siberian, and Chukchi Seas. Here, the landfast ice season is shortened by about 2–4 weeks decade−1 over the 32-yr period. The marked changes in the East Siberian and Chukchi seas are associated with the later start dates and earlier end dates of the landfast ice season in those regions. The shortened landfast ice season may be explained by the significant spring warming in the Siberian sector from the Kara Sea to the Chukchi Sea from 1979 to 1997 (Rigor et al. 2002).
The shortened landfast ice duration is also observed along the coast of the western Arctic, such as the Beaufort Sea, the CAA, Baffin and Hudson Bays, and the Labrador Sea. The season is shortened by about 0.3 week decade−1 in the Beaufort Sea, compared to nearly 2 week decade−1 in the CAA and the Labrador Sea, although the trends in most of these areas are not statistically significant except in the CAA. Howell et al. (2009) also observed an increased melt season duration within the CAA, with a significant increase of 7 day decade−1 attributed to both an earlier melt (−3.1 day decade−1) and later freeze (3.9 day decade−1). Although statistically insignificant, landfast ice duration shows an increase along the coast of the Barents, Baltic, and Bering Seas. Despite these positive changes, however, the length of the landfast ice season on average has been shortened by about 1.2 ± 0.5 week decade−1 in the Northern Hemisphere at a significance of 95% based on an F test.
4. Discussion
a. Warming Arctic and associated freezing patterns
Along the Arctic coasts, thermodynamics tends to be the dominant factor that controls local ice growth and melt. Therefore, the observed long-term changes in landfast ice extent and the growth and melt patterns indicate some changes in the thermodynamic processes. The decrease in landfast ice coverage is indeed consistent with the widespread warming in the Arctic. A report shows that the warming trend in the Arctic is the greatest in winter and fall of 1989–2008, with a weaker temperature increase in spring and summer (Screen and Simmonds 2010). In the Canadian Arctic, the trends in surface air temperature follow the general pattern of the Arctic but are consistently greater than the pan-Arctic trend, with the greatest change occurring during fall/winter and the least during summer (Tivy et al. 2011). The positive warming trends between 1990 and 2008 are also shown to be the strongest in the coastal regions flanking the Eurasian Basin (Polyakov et al. 2012). This observation is consistent with Overland et al. (2008), who noted widespread positive winter surface air temperature anomalies over the continental and coast areas of the Eurasian Arctic from 1985 to 1995, when the major Northern Hemisphere atmospheric circulation pattern, the Arctic Oscillation (AO), was in strong positive phase. This was also the same period when the ice drift patterns shifted in response to the North Atlantic Oscillation (NAO) (Serreze et al. 2007). Rigor and Wallace (2004) found that this shift favored an enhanced ice transport away from the Siberian and Alaskan coasts, leaving coastal areas with more areas of open water for thin ice production and for heat to escape from the ocean to the atmospheric. This ice–ocean feedback arguably contributes to the continued warming of the Arctic since 2000 (Screen and Simmonds 2010), when the Arctic atmospheric dipole anomaly (DA) played a critical role in producing anomalous meridional wind advection, increasing ice outflow out of the Arctic Basin, and further reducing sea ice extent in later summer/fall (Wang et al. 2009).
An FDD value can be either positive or negative, depending on whether the mean daily air temperature is below or above the freezing point of seawater. To count only positive effects of air temperatures, we set the negative FDD values to zero. At each grid cell, a FDD was computed and then integrated from the previous 1 September to the following 31 May. The resulting FDD is thus a measure of both duration and magnitude of air temperature below freezing and thus indicates a net growth of sea ice during the cold season.
Figure 13a shows the spatial variation of cumulative FDD averaged over 1949–2009. Within the Arctic Basin, the value varies between 4000 and 6000 along the coasts of the eastern Arctic and the Beaufort Sea. The largest FDD occurs toward Greenland and the CAA. When the three decadal periods, 1978–87, 1988–97, and 1998–2007, are compared, the largest and most widespread decrease in FDD occurred between the first two decades, in particular over the Eurasian Arctic (not shown). We note that there is a smaller change in FDD between the second and third decades compared to the first and second decades. This could explain why the loss of winter landfast ice extent slowed down since the late 1990s. Overall, the trends in FDD are negative (i.e., decreased potential for ice growth) in much of the Arctic Basin, indicated by the slope of least squares regression calculated over the 31-yr period (Fig. 13b). These negative trends in cumulative FDD are consistent with the overall reduction of landfast ice extent since the early 1990s. As shown in Fig. 13c, the areas with significant reduction in winter-averaged landfast ice correspond well with the largest decrease in FDD.
(a) Mean freezing degree days (FDD) averaged over 1977–2007, (b) estimated trends in FDD, and (c) trends in landfast ice extent relative to mean for the 17 regions defined in Fig. 8. The stars mark the regions where trends are significant above 95% probability. The NCEP daily 2-m air temperatures at a 25-km resolution were used for the calculation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
(a) Mean freezing degree days (FDD) averaged over 1977–2007, (b) estimated trends in FDD, and (c) trends in landfast ice extent relative to mean for the 17 regions defined in Fig. 8. The stars mark the regions where trends are significant above 95% probability. The NCEP daily 2-m air temperatures at a 25-km resolution were used for the calculation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
(a) Mean freezing degree days (FDD) averaged over 1977–2007, (b) estimated trends in FDD, and (c) trends in landfast ice extent relative to mean for the 17 regions defined in Fig. 8. The stars mark the regions where trends are significant above 95% probability. The NCEP daily 2-m air temperatures at a 25-km resolution were used for the calculation.
Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-13-00178.1
b. Influences from local winds
Besides the thermodynamic impacts on both local and large scales, changes in surface wind patterns can also affect local landfast ice conditions. For instance, in southeastern Hudson Bay, wind is a major factor in determining the western extent of landfast ice (Larouche and Galbraith 1989). Off the coast of Alaska, the wind may affect the local ice grounding, a critical process in stabilizing the local landfast ice cover, or cause partial breakups at the landfast ice edge (Druckenmiller et al. 2009). Similar break-up events were also observed in the Kara Sea during passing storms (Divine et al. 2003). Storm climates have been noted to be changing in recent years. According to a study by Dickson et al. (2000), there was an intensified storm track across the Eurasia Arctic in the late 1980s and the early 1990s, in response to a phase of extreme high NAO index. This is also the period when landfast ice extent declined considerably along the coasts of the eastern Arctic, although the long-term influences of cyclones on landfast ice are largely unknown.
Wind impacts on local landfast ice conditions were also observed in the CAA. For instance, Agnew et al. (2008) found a positive correlation between the AO/NAO indices and the monthly ice area fluxes through the gates of the Amundsen Gulf (AG), M’Clure Strait, and Lancaster Sound (LS). This finding suggests that the strong Beaufort Sea high pressure and gyre may help move ice out of the AG and M’Clure Strait and increase ice drift along the northern coast of the Archipelago. The NIC charts show that the loss of landfast ice in CAA in the early 1990s occurred primarily in AG and LS as well as along the northern coast of the Canadian high Arctic. A similar reduction of landfast ice is also noted in the areas of Cape Bathurst polynya and the Gulf of Boothia, where decreased ice concentration and increased ice mobility were also observed (Barber and Hanesiak 2004; Barber and Iacozza 2004). Therefore, the decreased landfast ice population observed in CAA may be explained by the increased ice mobility and thus less ice being classified as landfast, as opposed to the loss of total sea ice in the region. Certainly, ice mobility is affected not only by wind but also by currents and ice internal stress, which can be large within the Archipelago, and thus deviates from free ice drift that is mainly wind driven (Agnew et al. 2008). More studies are needed to understand the changes in landfast ice population in this region.
c. Potential impacts on the coastal ecosystem
As a defining feature of the Arctic shelves, a decrease in the extent of winter landfast ice and its early spring melt onset as well as the shortened ice season may have profound impacts on the Arctic coasts. For instance, in the Arctic estuaries that receive a large amount of river discharge, such as the Mackenzie Delta, the east Siberia coast, and the Lena Delta, the winter landfast ice cover typically isolates the river plume from winds, slowing the spreading and mixing of outflow water and leading to a well-stratified upper layer (Ingram and Larouche 1987). However, when landfast ice develops late in fall and winter and when the ice season is shortened, we may see a greater influence of wind on the spreading of outflow plume, altering the intensity of upwelling, mixing, and brine-driven convection at the landfast ice seaward edge, particularly during freeze-up (Carmack and Chapman 2003). The altered coastal circulation pattern and wind forcing will further affect the terrestrial discharges of freshwater and the redistribution and transport of sediments, organic matter in Arctic coastal food webs (Dunton et al. 2006), and pollutants that are fed to the rest of the ocean. Estimates of coastal nutrient inventory showed that the delayed landfast ice development in late fall and winter may also help increase nutrient inventories on the shelf, setting the stage for greater spring productivity in estuaries (Macdonald and Yu 2005), whereas variations in landfast ice duration may have a large impact on fish as well as marine birds and mammals that feed in coastal waters (Carmack and Macdonald 2002).
The decreased winter landfast ice extent may also affect the coastal freshwater processes through seasonal ice growth and melting. Ice formation reduces buoyancy by injecting salt into the underlying water column whereas melting increases buoyancy by adding freshwater. During freezing, landfast ice can store a significant amount of freshwater from river discharge in late season (Macdonald 2000). As a majority of landfast ice accumulated in winter melts close to shore and reenters coastal waters (Macdonald and Yu 2005), the released freshwater will be added back to the shelves during the summer. If we assume that an average 2 m of landfast ice forms on the Arctic shelves in each winter with a salinity of 4 psu, the annual production of landfast ice from the Barents Sea to the Beaufort Sea would be ~1500 km3. This value corresponds to an equivalent freshwater storage of roughly 1358 km3, which is about 8% of the total annual freshwater volume (17 300 km3) stored in sea ice estimated for the whole Arctic Ocean (Aagaard and Carmack 1989). The freshwater stored in landfast ice alone is about 68% of the total annual terrestrial discharge from the four major Arctic rivers, that is, the Ob (530 km3 yr−1), Yenisey (603 km3 yr−1), Lena (530 km3 yr−1), and Mackenzie (340 km3 yr−1). Of climatic significance are the year-to-year changes in the storage and the timing of the released freshwater by the landfast ice. We estimate that the freshwater storage by landfast ice on the Arctic shelves has decreased by about 16% since the 1990s compared to 1978–86 due to the reduction of winter landfast ice cover. Such a reduction may further affect the release of terrestrial sediment as well as organic and inorganic material to the Arctic coastal waters.
Storms and high winds have been linked to considerable coastal erosion to the Beaufort and Chukchi coasts, in particular during the fall season when the shoreline is not protected by landfast ice (Manley et al. 2003). Therefore, a late development of coastal landfast ice in fall will make the coasts more vulnerable to erosion, as coasts, when free of ice protection, are more impacted by ocean waves, which can be greatly influenced by local passing storms. Studies suggest some changes in Arctic cyclone patterns in all seasons (Zhang et al. 2004), and storms are the most frequent and vigorous in fall (Simmonds and Keay 2009). Continued sea level rise together with larger open-water fetch extending later in fall will lead to increasingly vigorous assault on the coastlines, which will add additional sediment to the coasts, altering the coastlines and impacting humans and marine life. As most climate models show, the intensity of Arctic cyclones is predicted to increase in several Arctic regions under anthropogenic climate change (Ulbrich et al. 2009). A change in coastal landfast ice cover will most certainly change the dynamic balance of the Arctic coasts, making them more vulnerable, if the Arctic warms further. More studies are needed to quantify and predict the impacts of changing Arctic landfast ice on the costal ecosystem.
5. Summary
The NIC sea ice charts provide new insight into the spatial and temporal behavior of Arctic landfast ice extent from 1976 to 2007. Using the weekly and biweekly charts, we find that the winter averaged landfast ice extent in the Arctic Basin displays large interannual variations. The total landfast ice extent was the broadest from the early to mid-1980s but declined during the early 1990s in many coastal regions. The reductions are relatively greater in the eastern Arctic compared to the western Arctic. Change for the Northern Hemisphere is significantly negative during the 32-yr period of the record. Despite the relatively stable total winter sea ice extent in the Canadian Arctic, the NIC charts show a decrease in landfast ice extent in this region in recent years. Such a reduction is thought to be contributed in part by the changes in ice compactness and mobility affected by the large-scale atmospheric circulation patterns are thought to contribute to this reduction.
The observed decrease in winter landfast ice extent is consistent the changes over the same period in the seasonal patterns of ice growth and melt. We find a gradually delayed landfast ice freeze-up starting from the late 1980s to the early 1990s, as well as an early spring melt onset in recent years. This slowdown in growth is more pronounced from late fall to early winter, likely a key factor that may retard the landfast ice development in fall. As a result, the length of landfast ice duration is shortened, especially in the East Siberian and Chukchi Seas where we note significant decline in winter landfast ice extent. On average, the length of the landfast ice season is about three weeks shorter compared to 31 years ago, presumably a result of a warmer Arctic especially in autumn and winter.
The decline of winter landfast ice extent in the early 1990s is consistent with some other large-scale changes observed in the Arctic Basin, such as the reduced sea ice thickness and extent, a weakened Beaufort Gyre and atmospheric anticyclonic circulation, and warmer air temperature. Commensurate with the decline in the total sea ice extent over the Arctic Ocean, the decrease in the coastal winter landfast ice extent and the shortened ice season will have major impacts. As the surface air temperature continues to rise with an Arctic amplification influenced by global warming and ice/ocean interactions, we may expect more changes in landfast ice extent along the Arctic coastal regions in the future. The challenge is to understand and project the responses of the whole coastal ecosystem to changing ice cover and Arctic warming.
Acknowledgments
We thank the analysts and staff at the U.S. National Ice Center for their assistance and Hajo Eicken for providing RADARSAT data. We also thank Andy Mahoney, Stephen Howell, Igor Polyakov, and Bruno Tremblay for their careful reviews and helpful discussions on the manuscript. This work was supported by the National Science Foundation, Office of Polar Programs, Grants OPP-0229473, ARC-0454912, and ARC-0714078. Fetterer’s contributions were supported by the NOAA/NESDIS National Geophysical Data Center.
REFERENCES
Aagaard, K., and E. C. Carmack, 1989: The role of sea ice and other fresh water in the Arctic circulation. J. Geophys. Res., 94 (C10), 14 485–14 498.
Agnew, T. A., A. Lambe, and D. Long, 2008: Estimating sea-ice area flux across the Canadian Arctic archipelago using enhanced AMSE-E. J. Geophys. Res., 113, C10011, doi:10.1029/2007JC004582.
Barber, D. G., and J. M. Hanesiak, 2004: Meteorological forcing of sea ice concentrations in the southern Beaufort Sea over the period 1979 to 2000. J. Geophys. Res., 109, C06014, doi:10.1029/2003JC002027.
Barber, D. G., and J. Iacozza, 2004: Historical analysis of sea ice conditions in M’Clintock Channel and the Gulf of Boothia, Nunavut: Implications for ringed seal and polar bear habitat. Arctic, 57, 1–14.
Barry, R. G., R. E. Moritz, and J. C. Rogers, 1979: The fast ice regimes of the Beaufort and Chuckhi Sea coasts, Alaska. Cold Reg. Sci. Technol., 1, 129–152.
Brodzik, M. J., and K. W. Knowles, 2002: EASE-Grid: A versatile set of equal-area projections and grids. Discrete Global Grids, M. Goodchild, Ed., National Center for Geographic Information and Analysis. [Available online at http://www.ncgia.ucsb.edu/globalgrids-book/ease_grid/.]
Brown, R. D., and P. Cote, 1992: Interannual variability of landfast ice thickness in the Canadian high Arctic, 1950–89. Arctic, 45, 273–284.
Carmack, E. C., and R. W. Macdonald, 2002: Oceanography of the Canadian shelf of the Beaufort Sea: A setting for marine life. Arctic, 55 (Supp. 1), 29–45.
Carmack, E. C., and D. C. Chapman, 2003: Wind-driven shelf/basin exchange on an Arctic shelf: The joint roles of ice cover extent and shelf-break bathymetry. Geophys. Res. Lett.,30, 1778, doi:10.1029/2003GL017526.
Comiso, J. C., 2012: Large decadal decline of the Arctic multiyear ice cover. J. Climate, 25, 1176–1193.
Dedrick, K. R., K. Partington, M. Van Woert, C. A. Bertoia, and D. Benner, 2001: U.S. National/Naval Ice Center digital sea ice data and climatology. Can. J. Remote Sens., 27, 457–475.
Dickson, R. R., and Coauthors, 2000: The Arctic Ocean response to the North Atlantic oscillation. J. Climate,13, 2671–2696.
Divine, D., R. Korsnes, and A. P. Makshtas, 2003: Variability and climate sensitivity of fast ice extent in the north-eastern Kara Sea. Polar Res., 22, 27–34.
Druckenmiller, M. L., H. Eicken, M. A. Johnson, D. J. Pringle, and C. C. Williams, 2009: Towards an integrated coastal sea-ice observatory: System components and a case study at Barrow, Alaska. Cold Reg. Sci. Technol., 56, 61–72.
Dumas, J., E. Carmack, and H. Melling, 2005: Climate change impacts on the Beaufort shelf landfast ice. Cold Reg. Sci. Technol., 42, 41–51.
Dunton, K. H., T. Weingartner, and E. Carmack, 2006: The nearshore western Beaufort Sea ecosystem: Circulation and importance of terrestrial carbon in Arctic coastal food webs. Prog. Oceanogr., 71, 362–378.
Flato, G. M., and R. D. Brown, 1996: Variability and climate sensitivity of landfast Arctic sea ice. J. Geophys. Res., 101 (C10), 25 767–25 777.
Galley, R. J., B. G. T. Else, S. E. L. Howell, J. V. Lukovich, and D. G. Barber, 2012: Landfast sea ice conditions in the Canadian Arctic: 1983–2009. Arctic, 65, 133–144.
Howell, S. E. L., C. R. Duguay, and T. Markus, 2009: Sea ice conditions and melt season duration variability within the Canadian Arctic Archipelago: 1979–2008. Geophys. Res. Lett., 36, L10502, doi:10.1029/2009GL037681.
Ingram, R. G., and P. Larouche, 1987: Variability of an under-ice river plume in Hudson Bay. J. Geophys. Res., 92 (C9), 9541–9547.
Jones, B. M., C. D. Arp, M. T. Jorgenson, K. M. Hinkel, J. A. Schmutz, and P. L. Flint, 2009: Increase in the rate and uniformity of coastline erosion in Arctic Alaska. Geophys. Res. Lett., 36, L03503, doi:10.1029/2008GL036205.
Krupnik, I., and J. Dyanna, Eds., 2002: The Earth is Faster Now: Indigenous Observations of Arctic Environmental Change. Arctic Research Consortium of the United States, 384 pp.
Kwok, R., and G. F. Cunningham, 2010: Contribution of melt in the Beaufort Sea to the decline in Arctic multiyear sea ice coverage: 1993–2009. Geophys. Res. Lett., 37, L20501, doi:10.1029/2010GL044678.
Larouche, P., and P. S. Galbraith, 1989: Factors affecting fast-ice consolidation in southeastern Hudson Bay, Canada. Atmos.–Ocean, 27, 367–375.
Macdonald, R. W., 2000: Arctic estuaries and ice: A positive–negative estuarine couple. The Freshwater Budget of the Arctic Ocean, E. L. Lewis, Ed., Kluwer, 383–407.
Macdonald, R. W., and Y. Yu, 2005: The Mackenzie estuary of the Arctic Ocean. Handbook of Environmental Chemistry, P. J. Wangersky, Ed., Springer-Verlag, 91–120.
Mahoney, A. R., H. Eicken, A. G. Gaylord, and L. Shapiro, 2007: Alaska landfast sea ice: Links with bathymetry and atmospheric circulation. J. Geophys. Res., 112, C02001, doi:10.1029/2006JC003559.
Manley, W. F., L. R. Lestak, and J. A. Maslanik, 2003: Photogrammetric analysis of coastal erosion along the Chukchi coast at Barrow, Alaska. Ber. Polar Meeresforsch., 443, 66–68.
Markus, T., J. C. Stroeve, and J. Miller, 2009: Recent changes in Arctic sea ice melt onset, freezeup, and melt season length. J. Geophys. Res., 114, C12024, doi:10.1029/2009JC005436.
Maslanik, J. A., S. Drobot, C. Fowler, W. Emery, and R. Barry, 2007: On the Arctic Climate paradox and the continuing role of atmospheric circulation in affecting sea ice conditions. Geophys. Res. Lett., 34, L03711, doi:10.1029/2006GL028269.
Maslanik, J. A., J. Stroeve, C. Fowler, and W. Emery, 2011: Distribution and trends in Arctic sea ice age through spring 2011. Geophys. Res. Lett., 38, L13502, doi:10.1029/2011GL047735.
Maykut, G. A., 1986: The surface heat and mass balance. The Geophysics of Sea Ice, N. Untersteiner, Ed., Plenum, 395–463.
Melling, H., 2002: Sea ice of the northern Canadian Arctic Archipelago. J. Geophys. Res., 107, 3181, doi:10.1029/2001JC001102.
National Ice Center, 2006 (updated 2009): National Ice Center Arctic sea ice charts and climatologies in gridded format. F. Fetterer and C. Fowler, Eds., National Snow and Ice Data Center, Boulder, CO, digital media. [Available online at http://nsidc.org/data/g02172.html.]
Overland, J. E., M. Wang, and S. Salo, 2008: The recent Arctic warm period. Tellus, 60A, 589–597, doi:10.1111/j.1600-0870.2008.00327.x.
Parkinson, C. L., and D. J. Cavalieri, 2008: Arctic sea ice variability and trends, 1979–2006. J. Geophys. Res., 113, C07003, doi:10.1029/2007JC004558.
Partington, K., T. Flynn, D. Lamb, C. Bertoia, and K. Dedrick, 2003: Late twentieth century Northern Hemisphere sea-ice record from U.S. National Ice Center ice charts. J. Geophys. Res., 108, 3343, doi:10.1029/2002JC001623.
Peterson, B. J., J. W. McClelland, R. Curry, R. M. Holmes, J. E. Walsh, and K. Aagaard, 2006: Trajectory shifts in the Arctic and subarctic freshwater cycle. Science, 313, 1061–1066, doi:10.1126/science.1122593.
Polyakov, I., J. Walsh, and R. Kwok, 2012: Recent changes of Arctic multiyear sea ice coverage and the likely causes. Bull. Amer. Meteor. Soc., 93, 145–151.
Reimnitz, E., L. Toimil, and P. Barnes, 1978: Arctic continental shelf morphology related to sea-ice zonation, Beaufort Sea, Alaska. Mar. Geol., 28, 179–210.
Reimnitz, E., D. Dethleff, and D. Nurnberg, 1994: Contrasts in Arctic shelf sea-ice regimes and some implications: Beaufort Sea and Laptev Sea. Mar. Geol., 119, 215–225.
Rigor, I. G., and J. M. Wallace, 2004: Variations in the age of Arctic sea-ice and summer sea-ice extent. Geophys. Res. Lett., 31, L09401, doi:10.1029/2004GL019492.
Rigor, I. G., J. M. Wallace, and R. L. Colony, 2002: On the response of sea ice to the Arctic Oscillation. J. Climate, 15, 2648–2668.
Rodionov, S. N., 2004: A sequential algorithm for testing climate regime shifts. Geophys. Res. Lett., 31, L09204, doi:10.1029/2004GL019448.
Rothrock, D. A., Y. Yu, and G. A. Maykut, 1999: Thinning of the Arctic Sea ice cover. Geophys. Res. Lett., 26, 3469–3472.
Screen, J. A., and I. Simmonds, 2010: Increasing fall-winter energy loss from the Arctic Ocean and its role in Arctic temperature amplification. Geophys. Res. Lett., 37, L16707, doi:10.1029/2010GL044136.
Serreze, M. C., and R. G. Barry, 2011: Processes and impacts of Arctic amplification: A research synthesis. Global Planet. Change, 77, 85–96.
Serreze, M. C., M. M. Holland, and J. Stroeve, 2007: Perspectives on the Arctic’s rapidly shrinking sea-ice cover. Science, 315, 1533–1536.
Simmonds, I., and K. Keay, 2009: Extraordinary September Arctic sea ice reductions and their relationships with storm behavior over 1979–2008. Geophys. Res. Lett., 36, L19715, doi:10.1029/2009GL039810.
Smith, D. M., 1998: Recent increase in the length of the melt season of perennial Arctic sea ice. Geophys. Res. Lett., 25 (5), 655–658.
Stroeve, J., M. Serreze, M. Holland, J. Kay, J. Malanik, and A. Barrett, 2012: The Arctic’s rapidly shrinking sea ice cover: A research synthesis. Climatic Change, 110, 1005–1027, doi:10.1007/s10584-011-0101-1.
Tanis, F., and V. Smolyanitsky, Eds., 2000: Arctic Climatology Project, Environmental Working Group Joint U.S.–Russian Arctic Sea Ice Atlas. National Snow and Ice Data Center, Boulder, CO, digital media. [Available online at http://nsidc.org/data/G01962.html.]
Thompson, T., 1981: Proposed Format for Gridded Sea Ice Information (SIGRID). National Snow and Ice Data Center, Boulder, CO, digital media. [Available online at http://nsidc.org/noaa/gdsidb/formats.html.]
Tivy, A., S. E. L. Howell, B. Alt, S. McCourt, R. Chagnon, G. Crocker, T. Carrieres, and J. J. Yackel, 2011: Trends and variability in summer sea ice cover in the Canadian Arctic based on the Canadian Ice Service Digital Archive, 1960–2008 and 1968–2008. J. Geophys. Res., 116, C03007, doi:10.1029/2009JC005855.
Tucker, W. B., III, J. W. Weatherly, D. T. Eppler, D. Farmer, and D. L. Bentley, 2001: Evidence for the rapid thinning of sea ice in the western Arctic Ocean at the end of the 1980s. Geophys. Res. Lett., 28, 2851–2854.
Ulbrich, U., G. C. Leckebusch, and J. G. Pinto, 2009: Extra-tropical cyclones in the present and future climate: A review. Theor. Appl. Climatol., 96, 117–131, doi:10.1007/s00704-008-0083-8.
Wang, J., J. Zhang, E. Watanabe, M. Ikeda, K. Mizobata, J. E. Walsh, X. Bai, and B. Wu, 2009: Is the dipole anomaly a major driver to record lows in Arctic summer sea ice extent? Geophys. Res. Lett., 36, L05706, doi:10.1029/2008GL036706.
Yu, Y., G. A. Maykut, and D. A. Rothrock, 2004: Changes in the thickness distribution of Arctic sea ice between 1958–1970 and 1993–1997. J. Geophys. Res., 109, C08004, doi:10.1029/2003JC001982.
Zhang, X., and J. Walsh, 2006: Toward a seasonally ice-covered Arctic Ocean: Scenarios from the IPCC AR4 model simulations. J. Climate, 19, 1730–1747.
Zhang, X., J. Walsh, J. Zhang, U. Bhatt, and M. Ikeda, 2004: Climatology and interannual variability of Arctic cyclone activity: 1948–2002. J. Climate, 17, 2300–2317.