To further examine the effects of climate change on the arctic - more specifically, the Northwest Passage - data from the Canadian Ice Service was examined.
The Canadian Ice Service (CIS) has archived weekly ice thickness and on-ice snow depth measurement for Canadian stations. A list of these CIS ice stations, as well as all archived ice thickness and snow depth data referred here can be found at the following web site:
courtesy of CIS
Data from Resolute, Mould Bay, Alert (Parr Inlet) and Eureka were downloaded from the CIS web site. Resolute lies deep within the Northwest Passage, while Mould Bay is at the Western end of the Passage. Although Alert, and Eureka do not lie along the coast of the Northwest Passage, data ranging back to the late 1940's and early 1950's were available for all four stations. Some stations, Mould Bay for example, have ice thickness records as early as 1947, but only records from 1960 onwards were examined here as both snow and ice records are observed weekly at all fourstations by this time.
Measurements were taken approximately at the same location every year on a weekly basis, from the time the ice is safe to walk on until break-up or when the ice becomes unsafe. For this discussion, it was assumed that ice thickness and snow depth were zero for the period of the year when no data were recorded. The data are presented in a yyyy-mm-dd format, but in order to make time series plots of ice thickness, snow depth, etc., it was necessary to resample the data to a uniform temporal spacing.
Procedure for Resampling Data:
Starting from t=1960.00, start incrementing by dt=0.01. At
each time t, find ti as the maximum of all t subject
to ti<t. Then ice (or snow) at t is:
ice(resampled) = (icei*(ti+1 - t) + icei+1*(t - ti)) / (ti+1-ti)
Once a uniform time series was obtained, the data were used to observe and calculate trends (if any) in the following three categories: maximum ice thickness, the correlation between maximum ice thickness and average snow depth, and open water, or ice-free duration.
There were few records available during 1989 at Alert. Data from this year at Alert were ignored altogether and not included in the statistical analysis. However, records for 1978 at Resolute indicate that measurements were taken all year, and at no time was there an ice-free period. Thus, only in Figure 5.3b were the data ignored for 1978 at Resolute, and not included in the statistical analysis.
Maximum Ice Thickness:
Figure 5.1a - Mould Bay [1960-1996] Figure 5.1b -
Figure 5.1c - Alert [1960-1997] Figure 5.1d - Eureka
Table 5.1 - Results of Linear Regression Analysis for Figures 5.1[a-d]
Increase in Max. Ice Thickness [cm] / year
|95% confidence interval (in slope)||correlation (r)||95% confidence interval (in r)|
|Mould Bay||-0.040||±0.620||-0.022||-0.344 < r < 0.304|
|Resolute||0.624||±0.549||0.358||0.043 < r < 0.608|
|Alert||-0.507||± 0.684||-0.246||-0.528 < r < 0.085|
|Eureka||-0.213||± 0.645||-0.110||-0.415 < r < 0.217|
Figures 5.1[a-d] display the maximum ice thicknesses for the four ice stations. Linear regression analysis was performed on the data. No statistically significant trend was seen at Mould Bay or Eureka. Resolute showed an increase in maximum ice thickness, with 95 % confidence, as shown in Table 5.1. Alert, however, had a decrease in ice thickness but the 95% confidence interval in the slope indicated a small probability that the trend may indeed be positive. There is a strong indication that ice thickness changes from year to year according to corresponding changes in average snow depth, as indicated in the next section.
The Correlation Between Maximum Ice Thickness and Average Snow Depth:
Figure 5.2a - Mould Bay [1960-1996] Figure 5.2b -
Figure 5.2c - Alert [1960-1997] Figure 5.2d - Eureka
Table 5.2: Results of Linear Regression Analysis for Figures 5.2[a-d]
Ice Thickness (cm) / Snow Depth (cm)
|95 % confidence interval (in slope)||correlation (r)||95% confidence interval (in r)|
|Mould Bay||-0.742||± 0.756||-0.319||-0.585 < r < 0.006|
|Resolute||-1.409||± 0.444||-0.737||-0.855 < r < -0.546|
|Alert||-2.265||± 0.942||-0.636||-0.796 < r < -0.393|
|Eureka||-2.430||± 1.175||-0.573||-0.754 < r < -0.310|
Figures 5.2[a-d] show the correlation between the average snow depth and maximum ice thickness for the years indicated above. The regression lines clearly indicate a negative correlation between snow and ice. Correlations, significant to the 95% level, exist for Resolute, Alert, and Eureka, as shown in Table 5.2. It is clear that the increase in ice at Alert and decrease in Resolute, as discussed by Brown and Cote (1992), is associated with the corresponding change in average snow depth.
As discussed in Flato and Brown's (1996) investigation of the sensitivity of landfast ice to potential climatic change, increased snowfall early in the winter would cause a decrease in maximum ice thickness due to the insulating properties of snow. However, a substantial increase in snowfall would cause the ice surface to flood, creating a layer of slush between the ice-snow boundary. Since the thermal conductivity of slush is 10-75 times that of snow (Weeks and Lee, 1958), the presence of slush due to snow would actually cause greater ice growth than snow alone.
It can therefore be inferred that a due to the insulating properties of snow, a thick layer of snow near the end of the winter would cause the ice to melt more slowly, causing break-up to occur later in the season. This increase in the yearly ice-free period is discussed in the next section.
The Increase in Open Water Duration:
Figure 5.3a - Mould Bay [1960-1996] Figure 5.3b -
Figure 5.3c - Alert [1960-1997] Figure 5.3d - Eureka
Table 5.3 - Results of Linear Regression Analysis for Figures 5.3[a-d]
|Slope: Increase in Open Water Duration (days/year)||95% confidence interval
|correlation (r)||95% confidence interval
|Mould Bay||0.940||± 0.554||0.502||0.213 < r < 0.710|
|Resolute||2.75||± 0.939||0.709||0.500 < r < 0.840|
|Alert||1.14||± 0.690||0.494||0.202 < r < 0.705|
|Eureka||1.17||± 0.478||0.638||0.400 < r < 0.795|
The data for Figures 5.3[a-d] were calculated by calculating number of days each year for which there were no data records available (ie: between break-up and freeze-up). These plots of of open-water duration are only estimates, as we do not know for each summer break-up if the ice had simply become unsafe to stand on to take a measurement, or if it had indeed broken up.
For each of the four graphs, the ice-free period is clearly increasing, as shown in Table 5.3, with positive slopes statistically significant to the 95% level. The positive slopes for all three graphs clearly indicate that the open-water duration has increased since 1960 at these four locations. More specifically, this increase is apparent at Resolute and Mould Bay which may have serious consequences for the Northwest Passage, should the entire waterway follow the same trend as two landfast ice stations along its coastline.
[go to Secion 3: Threat - And Opportunity]
[back to main index]
Brown, R.D., and P. Cote, Interannual variability of landfast ice thickness in the Canadian high Arctic, 1950-1989, Arctic, 45, 273-284, 1992.
Flato, G.M., and R.D. Brown, Variability and climate sensitivity of landfast Arctic sea ice, J. Geophys. Res., 101, 25, 767-25, 777, 1996.
Weeks, W.F., and O.S. Lee. Observations on the physical properties of sea-ice at Hopedale, Labrador, Arctic, 11, 134-155, 1958.