Influences of local weather, large-scale climatic drivers, and the ca. 11 year solar cycle on lake ice breakup dates; 1905-2004

We investigate the temporal patterns in inter-annual variability in ice breakup dates for Lakes Mendota and Monona, Wisconsin, between 1905 and 2004. We analyze the contributions of long-term trends attributed to climate change, local weather, indices of sunspots, and large-scale climatic drivers, such as the North Atlantic Oscillation (NAO) and El NiAo Southern Ocean Index (ENSO) on time series of lake-ice breakup. The relative importance of the aforementioned explanatory variables was assessed using linear regression and variation partitioning models accounting for cyclic temporal dynamics as represented by Moran Eigenvector Maps (MEM). Model results explain an average of 58 % of the variation in ice breakup dates. A combination of the long-term linear trends, rain and snowfall in the month prior to breakup, air temperature in the winter prior to breakup, cyclic dynamics associated with sunspot numbers, ENSO, and for Lake Mendota, NAO, all significantly influence the timing of ice breakup. Significant cycle lengths were 3.5, 9, 11, and 50 years. Despite their proximity, Lakes Mendota and Monona exhibit differences in how and which explanatory variables were incorporated into the models. Our results indicate that lake ice dynamics are complex in both lakes and multiple interacting processes explain the residuals around the linear warming trends that characterize lake ice records.