Validation of turbulent heat transfer models against eddy covariance flux measurements over a seasonally ice covered lake

Abstract

Abstract. In this study we analyzed turbulent heat fluxes over a seasonal ice cover on boreal lake located in southern Finland. Eddy covariance (EC) measurements from four ice-on seasons between 2014 and 2019 are compared to three different bulk transfer models: one with a constant transfer coefficient, and two with stability adjusted transfer coefficients: the Lake Heat Flux Analyzer and SEA-ICE. All three models correlate to the EC results well in general, although typically underestimating the magnitude and the variance of the flux in comparison to the EC observations. Differences between the models are small, with the constant transfer coefficient model performing slightly better than the stability adjusted models. Small difference in temperature and humidity between surface and air results in low correlation between models and EC. During melting periods (surface temperature T0 > 0 °C), the model performance for LE decreases when comparing to the freezing periods (T0 < 0 °C), while the opposite is true for H. At low wind speed EC shows relatively high fluxes (±20 W m−2) for H and LE due to non-local effects that the bulk models are not able to reproduce. Finally, the uncertainty in the estimation of the surface temperature and humidity affects the bulk heat fluxes, especially when the difference between surface and air values are small.