Improving surface heat flux estimation for a large lake through model optimization and two-point calibration: The case of Lake Geneva

Net Surface Heat Flux (SurHF) was estimated from 2008 to 2014 for Lake Geneva (Switzerland/France), using long‐term temperature depth profiles at two locations, hourly maps of reanalysis meteorological data from a numerical weather model and lake surface water temperatures from calibrated satellite imagery. Existing formulas for different heat flux components were combined into 54 different total SurHF models. The coefficients in these models were calibrated based on SurHF optimization. Four calibration factors characterizing the incoming long‐wave radiation, sensible, and latent heat fluxes were further investigated for the six best performing models. The combination of the modified parameterization of the Brutsaert equation for incoming atmospheric radiation and of similarity theory‐based bulk parameterization algorithms for latent and sensible surface heat fluxes provided the most accurate SurHF estimates. When optimized for one lake temperature profile location, SurHF models failed to predict the temperature profile at the other location due to the spatial variability of meteorological parameters between the two locations. Consequently, the optimal SurHF models were calibrated using two profile locations. The results emphasize that even relatively small changes in calibration factors, particularly in the atmospheric emissivity, significantly modify the estimated long‐term heat content. The lack of calibration can produce changes in the calculated heat content that are much higher than the observed annual climate change‐induced trend. The calibration improved parameterization of bulk transfer coefficients, mainly under low wind regimes.