This section summarizes the input products and describes how the ice edge is derived. See Divine and Dick (2006) for additional information, or refer to the original data sources.

Arctic Climate System Study (ACSYS) archive (1750-1966)

The original ACSYS archive is a compilation of historic sea ice data from various sources, including newspapers, ship observations, aircraft observations, diaries and more (ACSYS, 2003). The sea ice data from the original ACSYS archive from 1750-1966 are distributed as shapefiles (ACSYS, 2003), with polylines depicting the sea ice extent. A polyline is defined as straight line segments merged together to form a continuous line. The polylines characterize the ice conditions that are bounded by the line. For example, the red line in Figure 1 depicts the edge of ice concentrations that range from 70-80 percent ice concentration.

Figure 2 shows the attribute table that corresponds to Figure 1. Notice the ICE_COND column that is highlighted. These numbers represent different ice conditions. For example, the number "3" corresponds to the red line in Figure 1 that represents 70-80 percent ice concentration. The data in the SOURCE column provides information on how each polyline can be traced back to its original source (for example, the newspaper the sea ice data originated from).

There are usually two files per month in the ACSYS archive from 1750 through 1966. However, only data from one file per month is used in this data set.

Once the shapefile was chosen, the data providers decided which polylines to use to construct the ice edge. Preference was given to lines with 30 percent or more ice concentration. An ASCII file with geographical coordinates of the ice edge (at least 30 percent ice concentration) was produced.

Hunters, primarily consisting of sealers and whalers, collected most of the data prior to 1950 and consistently reported ice concentrations ranging from 30 percent to 60 percent. Therefore, users should understand that the ice edge is not defined exactly at the 30 percent ice concentration contour. If sea ice edge data were available as a set of segments (likely from varying sources), they were lined up in sequence to form a continuous set of coordinates in the ice edge file. See Divine and Dick (2006) for further details.

NMI files (subset of ACSYS archive from 1967-1978)

The ACSYS ice chart archive from 1967-1978 contains ice charts collected by the Norwegian Meteorological Institute (NMI) and then digitized at the Norwegian Polar Institute (NPI). These data are based primarily on satellite imagery. Whereas the earlier ACSYS subset from 1750-1966 are distributed as polylines, the data from 1967-1978 use polygons to denote the different classifications of ice concentration (see Figure 3). The ice conditions are categorized as follows: open water (ice concentration below 10 percent), very open drift ice (10-40 percent), open drift ice (40-70 percent), close drift ice (70-90 percent), very close drift ice (90-100 percent) and fast ice (ACSYS, 2003). The boundary between very open drift ice and open drift ice was used by the data providers to determine the ice edge. Occasionally, the shape of the marginal ice zone made this inappropriate, and a different boundary (eg. between close drift ice and open drift ice) was used. The file selected contained the best spatial coverage, but if both charts were of comparable coverage and quality, the later file was used (Divine and Dick, 2006). The polygons were converted to a gridded (raster) format directly in the GIS and then the grdcontour program of the Generic Mapping Tools (GMT) was used (Wessel and Smith, 1998) to draw the boundary.

Figure 3 shows an example of the original ACSYS file ice19750403.shp containing polygons with the different ICE_TYPE categories.

Soviet reconnaissance files (1953-1965)

Data from Fetterer and Troisi (1997) were used to fill in gaps in the Barents Sea that exist in the ACSYS archive. These data are available as gridded ice concentrations with 12.5 km cell spacing. The data set is a 10-day product and usually only one 10-day file per month falls within the spatial extent of this data set. Therefore, the Soviet reconnaissance files provide ice edge data for approximately one file per month from 1953-1965. To produce contour lines corresponding to a given ice concentration, the grdcontour program was used (Wessel and Smith, 1998) with the 30 percent ice concentration denoting the ice edge.

In some instances, the Soviet reconnaissance data overlapped temporally with the ACSYS files. When this overlap occurred, data from both sources were merged, meaning that a file in the current data set can contain data from more than one input product. For example, there may be segments in an ice edge file with one segment from the ACSYS archive and another segment from the Soviet reconnaissance files. For a complete list of merged files, see the Appendix.

The segments of the ice edges obtained for the Barents Sea were subsequently merged with the respective data for the eastern part of the study area. Figure 4 shows an example of an ice edge file.

SMMR and SSM/I data (1978-2002)

Passive microwave ice concentration data were used from SMMR (1978-1987) and SSM/I (1987-2002). The input data are in an approximate 25 km grid in a polar stereographic projection. The region over the Nordic Seas was extracted and ice concentrations were passed through the grdcontour program of the Generic Mapping Tools (GMT) (Wessel and Smith, 1998). The 30 percent contour line determined the ice edge.

Monthly means

Monthly means were derived using a technique described by Shapiro et al. (2003). With this technique, each ice edge is expressed as a sequence of rectangular coordinates, parameterized by their distance along the ice edge from a fixed location. Points can then be grouped into bins according to their distance along the edge. The monthly mean ice edge is then computed by calculating the mean location of points for each bin for each month of each year.

Multidecadal means

The multidecadal means are calculated from the monthly means described above. They are calculated for each calendar month for these subperiods: 1870-1920, 1921-1961, 1962-1988 and 1989-2002. These subperiods were chosen to correlate with the warming and cooling in the Arctic region described by Polyakov et al (2003).

The GIS shapefiles were created at NSIDC and are distributed in a geographic projection with WGS84 datum. Here is an example projection (.prj) file:

/ice_edge_positions
   /data 
   /browse

/monthly_means
   /data
   /browse
   /shapefiles

/multidecadal_means
   /data
   /browse
   /shapefiles

A sample browse image from iep17510715.jpg:

The first portion of iep17510715.dat: