Satellite monitoring of lake ice breakup on the Laurentian shield as a robust climate indicator
The overall objective of this study was to explore satellite monitoring of lake ice breakup as a spatially explicit climate indicator in the mid- to high latitudes in late winter/early spring, the spatiotemporal domain within which temperature increases due to greenhouse warming are often predicted to be greatest. Lake ice breakup dates from 1980-1994 for 81 selected lakes and reservoirs in the U.S. upper Midwest and portions of Canada (60°N, 105°W to 40°N, 85°W) were determined via analysis of 1,830 archival images from the visible band (0.54-0.70 JLm) of the GOES-VIS SR. Comparison of satellite-derived breakup dates to available ground reference data revealed the remarkable accuracy of the methodology, as the mean absolute difference and mean difference between methods were 3.2 and -0.4 days respectively, well within the natural variability in lake ice breakup dates (o 12 days) for a single lake over time. A numerical thermodynamic process model of lake ice was collaboratively developed and validated to enable sensitivity analyses of factors influencing ice phenology. These analyses revealed that the ice-off date is more sensitive than the ice-on date to air temperature changes and increased snowfall produces a monotonic delay in the simulated breakup date. Temporal coherence analysis revealed the general lack of coherence in lake ice-off dates and the significant correlation of coherence with differences in latitude and mean ice-off date. The predominant spatial trends on a mean basis can be attributed to latitude (itself a surrogate for temperature and solar radiation) and snowfall (R2 = 93\% ). However, the relative importance of both latitude and snowfall varies annually. Contrary to expectations, the only significant trends toward earlier ice-off dates occurred in southern Wisconsin. There were also statistically significant trends toward later ice-off dates in northwestern Wisconsin. Analysis of the pooled 1980-1991 data set for all 81 lakes revealed a significant (p = 0.0005) overall trend toward earlier ice-off dates. These results indicate the continued need for monitoring the various anticipated fingerprints of climate variability and change. The date of lake ice breakup as determined from metsat data is a robust climate change indicator inherently amenable to operational monitoring in this context.