The IPCC (2007) report considers evidence of whether the annual probability distribution functions for air temperature indices have changed in shape in addition to shifting in means and finds this evidence is still sparse. An increase in the variance of the distribution can lead to an increase in extreme events. We investigated whether the variability in lake ice breakup date has changed over time by examining ice breakup records for the period 1905-2004 for 70 lakes across the N. Hemisphere. The trend of variability was calculated for each lake by generating standard deviation estimates for running ten-year windows.

Results/Conclusions

We found that variability decreased in some lakes during the 100 year period and increased in others, with the proportion of variance (in the standard deviation) being explained by the trend as great as 50 percent and as little as zero. The slope of the trend for standard deviation was significantly related to continent (North America versus Europe) and the interaction of continent and latitude. The mean slope of the trend for standard deviation was negative for the northeast N. America and Europe while positive for the north central United States; thus, in general, variability is increasing in the north central U. S. but decreasing in the northeast N. American and Europe. What might explain such a pattern? We found that the relation of ice breakup date in the United States to the Southern Oscillation index is stronger during the positive phase of the Pacific Decadal Oscillation (PDO). Europe did not have this relationship. The more recent period has been characterized by stronger El Niño’s (positive SOI) and a positive PDO. These changes in the large scale climate drivers coincide with an increase in variability in interannual lake ice breakup dates in the north central United States. We also explored the temporal and spatial patterns of extreme events in lake ice breakup.