Simulating ice cover in lakes as a function of air temperature: preliminary results from Lake Tovel

Ice formation on lake surfaces is a complex process that is attracting increasing interest. Following a recently proposed approach that exploits hybrid models (physically rooted, but with calibrated parameters, see the air2water model proposed by Piccolroaz et al., HESS 2013), we developed a simple formulation to predict lake surface temperature (LST) and ice thickness in winter as a function of air temperature alone. Daily averaged LST is simulated by means of an ordinary differential equation where the effect of summer stratification is suitably taken into account. The new element is the sub-model for ice formation, whose thickness is modelled by an additional differential equation solely depending on air temperature. Importantly, no ice thickness measurements are needed for calibration of the model parameters. The model was applied to Lake Tovel (Trentino, Italy), a LTER site (IT09-005-A). The lake (altitude = 1178 m asl; area = 39 ha, volume = 7.4x106 m3, max depth = 39 m) is ice-covered from December to April. Air temperature was measured by a meteorological station located at the lake, while water temperature was provided by a central lake platform equipped with temperature sensors. The dataset covered 6 years (2010-2015), when LST was recorded. Model outputs satisfactorily agree with the available LST measurements, showing that accounting for ice formation is necessary to correctly predict LST in spring. Moreover, a qualitative comparison of modelled ice thickness with occasional measurements and visual observations suggests that this simple ice model has the potential to be used to reconstruct ice dynamics in other lakes where only air temperature and LST data are known.