4.3.2 Changes in Freeze-up and Breakup Dates
Freeze-up is defined conceptually as the time at which a continuous and immobile ice cover forms; however, operational definitions range from local observations of the presence or absence of ice to inferences drawn from river discharge measurements. Breakup is typically the time when the ice cover begins to move downstream in a river or when open water becomes extensive at the measurement location for lakes. Here again, there is some ambiguity in the specific date, and in the extent to which local observations reflect conditions elsewhere on a large lake or in a large river basin.
Selected time series from a recent compilation of river and lake freeze-up and breakup records by Magnuson et al. (2000) are shown in Figure 4.5. They limited consideration to records spanning at least 150 years. Eleven out of 15 records showed significant trends towards later freeze-up and 17 out of 25 records showed significant trends towards earlier breakup. When averaged together, the freeze-up date has become later at a rate of 5.8 ± 1.6 days per century, while the breakup date has occurred earlier at a rate of 6.5 ± 1.2 days per century.
A larger sample of Canadian rivers spanning the last 30 to 50 years was analysed by Zhang et al. (2001). These freeze-up and breakup estimates (based on inferences from streamflow data) exhibit considerable variability, with a trend towards earlier freeze-up and breakup over much of the country. The earlier freeze-up dominates, however, leading to a significant decrease in open water duration at many locations as shown in Figure 4.6. A recent analysis of Russian river data by Smith (2000) revealed a trend towards earlier freeze-up of western Russian rivers and later freeze-up in rivers of eastern Siberia over the last 50 to 70 years. Breakup dates did not exhibit statistically significant trends.
A comparable analysis of freeze-up and breakup dates for Canadian lakes has recently been completed by Duguay et al. (2006). These results (shown in Figure 4.7) indicate a fairly general trend towards earlier breakup (particularly in western Canada), while freeze-up exhibited a mix of early and later dates.
There are insufficient published data on river and lake ice thickness to allow assessment of trends. Modelling studies (e.g., Duguay et al., 2003) indicate that, as with the landfast sea ice case, much of the variability in maximum ice thickness and breakup date is driven by variations in snowfall.