Sea Ice Climatic Atlas for the Northern Canadian Waters 1981-2011: chapter 1

Regional Analyses

Canadian ice charts are produced using imagery from RADARSAT-1 (since 1996) and RADARSAT-2 (since 2008). Other remote sensing data sources include Envisat, NOAA AVHRR and Modis imagery. Where possible, the interpretation of the satellite data is verified using observations from Ice Service Specialists onboard dedicated aircraft and Canadian Coast Guard (CCG) ships. In addition, the United States has exchanged ice data with us for many years: the National Ice Center at Suitland, MD; and the International Ice Patrol (IIP), under the jurisdiction of the United States Coast Guard (East Coast of Canada - sea ice and iceberg information).  In addition, the Danish Meteorological Institute in Copenhagen has shared ice information covering the waters west of Greenland.

The products presented in this atlas were obtained from a statistical compilation of the Regional Ice Charts from the Canadian Ice Service for the years 1981 to 2010. Separate regional charts are produced for Western Arctic, Eastern Arctic and Hudson Bay. These have been merged in this atlas to provide a pictorial representation of climatic sea ice conditions in Northern Canadian waters as a whole. The regions shown in white are outside the boundaries of the above regional charts and no information is provided in this atlas for those regions. Please note that the regions covered by the regional charts have changed slightly over the years and we have limited our analysis to the areas for which data was consistently available over the years. This explains the exclusion of southeast portion of James Bay and the somewhat odd shape of the Beaufort Sea region.

It should be noted that the regional charts were prepared in an operation setting to support shipping activities and are subject to some limitations. They are not available on a weekly basis throughout the year but typically only during the operational season i.e. from June to November. However, since 1980 winter charts have been produced on a monthly basis and in recent years bi-weekly . This reduced winter set has been used to produce winter statistical products.

The Regional Ice Charts are not always done on the same dates each year, so a seven-day period centered on the Historical Dates has been selected for this climatological atlas.  The climate data represents information from charts within three days on either side of the historical date. The historical dates are weekly starting the first of January each year.

It should be noted that the original scale of the Regional Ice Chart was 1:4,000,000 and plotted on paper maps. Although the current analyses are prepared using GIS computer applications the amount of detail and accuracy is still comparable to the original maps.

Methodology

From 1980 to 1995 the Regional ice charts were drawn on paper. These charts were digitized in the late 1990’s for use in climatological studies.  Since 1995 computer technology was used to generate a digital version of the charts. The Regional Ice Chart collection now encompasses over 40 years of sea ice information spanning from 1968 to the present. 

Areas of fast ice in this atlas now contain a stage of development. Since 2004, this information has been added at the time of the chart production; however, for charts before that time, a senior forecaster provided estimated stages of development for areas of fast ice based primarily on Freezing Degree Days (FDDs), amongst other meteorological parameters for the period.

The ice information data is analyzed with GIS software using well-established customized scripts to produce the various statistical outputs. Once the original vector data is assigned a historical date, it is then converted to a raster data format at 1 km resolution. Various algorithms perform operations to statistically summarize the individual ice charts and output the climatological products seen in the atlas.  

In preparing an ice atlas, medians rather than averages are used.  If one considers a single data point near the edge of the fast ice in late spring, the ice conditions can be ten tenths when fast ice is present or open water after the ice breaks up.  Rarely will the four to six tenths range of ice concentrations occur, which is the inevitable result if one averages between no ice and ten tenths.  A median on the other hand will be either zero or ten tenths depending on the relative frequency of break-up before or after the given date.  This is more appropriate for an atlas describing ice conditions.  With a thirty-year time period, an even number of values are used for each particular grid point and the higher of the two middle values is chosen as the median, a policy that has been adopted since the production of the Hudson Bay and Approaches atlas in the early 1980s.

A review of the initial climate products revealed several inconsistencies especially for extreme events and low concentrations of old ice; it was then required to go back to the regional charts to identify the cause of the problem and correct the inconsistency. The following is a list of the types of problems identified and the measures taken to fix the climate products:

  • Missing charts from June to November had an impact on week to week consistency of extreme events; adjustments were made to ensure consistency. For the May 15 products, several weeks (charts falling between May 12 and May 25) were combined to provide a full set of charts; this accurately depicts the ice situation because there is very little change in conditions during that time.
  • October 1 is the "ice birthday." Almost half of the charts within three days of this date will bear a September timestamp. First-year ice on these charts was re-coded as second-year ice for compatibility in the charts. In some years, there could be small errors introduced because of this procedure.
  • Procedures for chart preparation allowed the use of multiple eggs per polygon (up to around 1984-85), as long as predominant ice and concentrations category were the same. This introduced errors in old ice distribution during the digitization process and resulted in inconsistencies in the old ice climate products. Appropriate corrections to boundaries and attributes were made directly on the climate products based on a review of the regional charts to ensure consistency from week to week as much as possible. However some inconsistencies may remain, particularly for old ice concentrations less than 4/10. These remaining inconsistencies may be more important in areas of high mobility and dispersion like Baffin Bay, Davis Strait, Foxe Basin, Hudson strait and northern Hudson Bay where it is difficult to track the evolution of low concentrations of old ice embedded in fast moving and rugged first-year ice. For the same reasons even more severe inconsistencies were found for traces of old ice that can occur almost anywhere in northern Canadian waters and it was not considered in the climate products.
  • In the preparation of the regional charts, shore-fast ice can be indicated simply as generic “fast ice” (shown by "blacking in" the area where it is present), or by indicating specific ice attributes to the area (attaching an egg). Blacking in an area is the traditional method for showing fast ice in small bays and inlets and in most cases would represent the first-year stage of development. However, it was also occasionally used on the Arctic regional charts when there were varying concentrations of old ice embedded in the fast ice. This can cause a problem when computing statistics involving old ice as well as the predominant ice type. To minimise adverse effects in the charts resulting from this representation of fast ice some areas were re-analyzed and corrected.
  • The winter series of charts have a very large consolidated ice area. Up until the late 1990s, charts were prepared by hand and when the scale of the digitized charts is considered, even the width of a pencil or pen could be potentially misinterpreted. Month to month differences in the consolidated area thus appeared, but were corrected.

Definition of Sea Ice Climatic Charts

Statistics Described
The ice charts contained within this atlas are derived climatological products representing the "normal" of various ice parameters. Two key statistical terms have been used to derive and describe the charts: median and frequency. The "median" is a statistical value used to examine a dataset and is calculated by ordering the dataset by its values from smallest to largest and selecting the middle value of an odd-numbered dataset or the average of the two middle values in an even-numbered dataset. The median is employed with ice statistics due to the ordinal nature of the ice attributes. For example, 9+/10 ice concentration is greater than 9/10 concentration and first-year ice is greater (thicker) than grey-white ice.

The median is more appropriate than the average or mean when considering ice attributes. As an example, in the following dataset of 5 observations of ice concentration in tenths: (10, 10, 10, 0, 0). The average value would be (10 + 10 + 10 + 0 + 0)/5 = 6/10 which would not represent a "real" ice situation while the median value of 10/10 does.

The "frequency" is another statistical technique used to examine a dataset and is calculated by summing the number of observations of an occurrence or event (e.g. presence of sea ice) and dividing by the total number of observations and expressed as a percent of the total number of observations.

The following is a description of the climate products contained in this atlas. Some products were affected to various degrees by the intrinsic problems described in the methodology section and an indication of the level of confidence is provided. The regions shown in white on the products are outside the boundaries of the regional charts and insufficient information is available for those regions.

Median of Ice Concentration
The "Median of Ice Concentration" charts depict the total concentration of ice for a given date throughout the course of a year. The charts do not represent any specific real ice season but rather a statistical composite of all available years from 1981 to 2010.

The charts represent the statistical "normal" ice concentration for the appropriate date. There is a high level of confidence with this atlas series.

Dates of Freeze-up and Break-up
The "Dates of Freeze-up and Break-up" charts depict the extent of ice on a bi-weekly basis during the freeze-up and break-up periods. They provide a pictorial representation of the evolution of the ice extent during those two periods.

These products are constructed using the Median of Ice Concentration charts and thus the confidence level is high.

Median of Ice Concentration When Ice Is Present
The "Median of Ice Concentration When Ice is Present" charts consider total concentration of ice throughout the course of a year.  The charts are a new addition to the atlas and are meant to assist in interpreting the complementary "Median of Predominant Ice Type When Ice Is Present" charts. The most appropriate way to interpret the charts is to view the median of ice concentration when ice is present in conjunction with the frequency of presence of sea ice charts.  For example, at a particular point, the frequency of presence of sea ice might be in the range of 34-50% and the median of ice concentration when ice is present might be 9/10 to 9+/10.  Thus, at this location, there is a 34-50% chance of encountering sea ice, and when ice is present, it is "normally" 9/10 to 9+/10 concentration. Additional insights may be provided by examining the Predominant Ice Type When Ice Is Present charts.

The charts represent the statistical "normal" ice concentration when ice is present for the appropriate date. The level of confidence is generally high except in areas where only a few occurrences of ice are found for a chart date.

Median of Predominant Ice Type When Ice Is Present
The "Median of Predominant Ice Type When Ice Is Present" charts consider the predominant ice type (ice type of the greatest concentration) throughout the course of a year.

The charts involve more interpretation than any of the other ice charts. The most appropriate way to interpret the charts is to view the Median Of Predominant Ice Type in conjunction with the Frequency Of Presence Of Sea Ice chart. For example, at a particular point, the frequency of presence of sea ice might be in the range of 34-50% and the median of predominant ice type when ice is present might be first-year ice. Thus, at the point, there is a 34-50% chance of encountering sea ice, and when ice is present, it is "normally" first-year ice. Additional insights may be provided by examining the Ice Concentration When Ice Is Present charts.

The charts represent the statistical "normal" predominant ice type when ice is present for the appropriate date. The level of confidence is generally high except in areas where only a few occurrences of ice are found for a chart date.

Frequency of Presence of Sea Ice (%)
The "Frequency of Presence of Sea Ice (%)" charts provide the likelihood of total concentration of ice greater than or equal to 1/10 throughout the course of a year and are anticipated to give the reader an idea of the likelihood that ice will occur at a particular location for the appropriate date.

The charts can be interpreted as the probability of encountering sea ice for the appropriate date. The charts depict above normal extent (1 to 33%), near normal extent (34 to 66%) and below normal extent (67 to 99%). The 0% line represents the maximum extent of sea ice, beyond it no ice was reported in the dataset; the 100% line represents the minimum extent of sea ice, within it there has always been ice reported in the period. There is a high level of confidence throughout this atlas series.

Median of Old Ice Concentration
The "Median of Old Ice Concentration" charts consider the concentration of old ice throughout the course of a year. The charts do not represent any real ice season but rather a statistical composite of all available years.

The charts represent the statistical "normal" concentration of old ice for the appropriate date. The level of confidence is not as high as the median of total ice but efforts have been made to ensure a reliable output product.

Frequency of Presence of Old Ice: 1 to 10/10 (%)
The "Frequency of Presence of Old Ice: 1 to 10/10 (%)" charts provide the likelihood of old ice greater than or equal to 1/10 throughout the course of a year and are anticipated to give the reader an idea of the likelihood that old ice will occur at a particular location for the appropriate date.

The charts can be interpreted as the probability of encountering old ice in concentrations of 1/10 or more for the appropriate date. The charts depict above normal extent (1 to 33%), near normal extent (34 to 66%) and below normal extent (67 to 99%). The 0% line represents the maximum extent of old ice, beyond it no old ice was reported in the dataset; the 100% line represents the minimum extent of old ice, within it there has always been old ice in concentrations of 1/10 or more reported in the dataset.

The level of confidence is generally good but is lower in areas where old ice is seldom observed or in areas of high ice mobility such as Baffin Bay, Davis Strait, Foxe Basin, Hudson strait and Northern Hudson Bay.

Frequency of Presence of Old Ice: 4 to 10/10 (%)
The "Frequency of Presence of Old Ice: 4 to 10/10 (%)" charts consider the likelihood of old ice greater than or equal to 4/10 throughout the course of a year and are anticipated to give an idea of the likelihood that old ice greater than or equal to 4/10 will occur at a particular location for the appropriate date.

The charts can be interpreted as the probability of encountering old ice in concentrations greater than or equal to 4/10 for the appropriate date. The charts depict above normal extent (1 to 33%), near normal extent (34 to 66%) and below normal extent (67 to 99%). The 0% line represents the maximum extent of 4/10 or greater old ice, beyond it no 4/10 or greater old ice was reported in the dataset; the 100% line represents the minimum extent of 4/10 or greater old ice, within it there has always been old ice in concentrations of 4/10 or greater reported in the dataset.

The level of confidence is generally good (and higher than the Frequency of Old Ice: 1 to 10/10 product) but is lower in areas where old ice is seldom observed or in areas of high ice mobility such as Baffin Bay, Davis Strait, Foxe Basin, Hudson strait and Northern Hudson Bay.

Report a problem or mistake on this page
Please select all that apply:

Thank you for your help!

You will not receive a reply. For enquiries, contact us.

Date modified: