Comparing ice and temperature simulations by four dynamic lake models in Harp Lake: Past performance and future predictions

The physical dynamics of lake temperature and ice phenology are important in the modelling and management of temperate aquatic ecosystems. One-dimensional hydrothermal lake models have not been well evaluated in terms of how they simulate ice dynamics in particular. We chose four models (Hostetler, Minlake, Simple Ice Model or SIM and General Lake Model) to test and compare their performance modelling water temperature and ice dynamics using 16years of field data from Harp Lake, an extensively studied inland lake in south-central Ontario. Each model produced satisfactory water temperature profiles over the simulated period, with small differences in the model performance. Model fits for ice phenology and ice thickness were, however, considerably lower than those for water temperature, with Minlake generating the best agreement with observed ice-on and ice-off dates as well as ice thickness, followed by SIM. The responses of lake ice dynamics to future climate scenarios were simulated by running each of the four models for 91years, from 2010 to 2100. The predicted decrease in ice season length was significantly different among models, varying between 30 and 81days, with an average of 48days. Corresponding decreases in ice thickness varied between 0.11 and 0.20m, averaging 0.17m. This study demonstrates that uncertainty due to model performance and selection is considerable, and further testing and refinement of hydrothermal lake dynamic models are needed to improve predictive abilities for ice dynamics. © 2014 John Wiley & Sons, Ltd.