Modeling of Lake Ice Characteristics in North America Using Climate, Geography, and Lake Bathymetry

This is the authors' abstract. We don't add key points for this paper.

Ice cover records from 128 freshwater lakes in the United States and Canada were analyzed. Multivariable linear regression models, log-transform models, and a combination of the two (the “hybrid” form) were used to express ice-in date, ice-out date, and maximum ice thickness as functions of mean air temperature, latitude, average depth, elevation, and surface area of each lake. Mean air temperatures are for periods from September 1 to December 31 for ice-in dates, February 1 to June 30 for ice-out dates, and September 1 to June 30 for maximum ice thickness. Data for individual years as well as averages (over the record length) for each lake were analyzed. The log-transform formulas proved best for estimating ice-in date, while the hybrid form provided the best models of maximum ice thickness. The linear regression model estimated the ice-out date best. In most cases, mean air temperature and/or latitude were the most influential parameters, followed by elevation. Lake surface area and depth had a small or no influence. R2 values of all equations ranged from 0.50 to 0.92.