Functional Representation of Great Lakes Surface Temperatures 1

Abstract

We have fitted a continuous parametric function to daily averages of lake surface (depth 1 m) temperature measurements made at buoys deployed since 1979 in the Great Lakes by the National Data Buoy Center. The function, which is based on the Johnson S u probability distribution, successfully reproduces the major features of the observed annual surface temperature cycle including gradual warming from winter minimum to the onset of spring stratification, rapid warming to mid-summer peak, and slow cooling to winter minimum. Best fit parameters describing the observed average annual temperature cycles vary in a reasonable way with latitude and water depth. Although year-to-year variations in the temperature cycles can be large, the root-mean-squared differences between the fitted functions and the averaged buoy observations are on the order of a few tenths of a degree C. While no trend in lake surface temperatures is evident over the past 10 years, the effects of climatologically abnormal years can be seen in the observed data. The functional representations of lake surface temperature will be useful for computer simulations of bio-geochemical processes in the Great Lakes and for evaluations of climatological departures from typical conditions.