Optimizing large-eddy simulations for investigating the energy-balance closure problem at typical flux measurement heights

The eddy-covariance method generally underestimates sensible and latent heat fluxes, resulting in an energy-balance gap from 10 % to even 30 % across sites worldwide. In contrast to single-tower eddy-covariance measurements, large-eddy simulations (LES) provide information on a 3D array of grid points and can capture atmospheric processes such as secondary circulations on all relevant scales, which makes them a powerful tool to investigate this problem. In order to compare LES results to field measurements at 20 m height from the CHEESEHEAD (Chequamegon Heterogeneous Ecosystem Energy-balance Study Enabled by a High-density Extensive Array of Detectors) campaign, a LES-setup that provides comparability to the measurements at these low levels is necessary. However, former LES studies have shown that the energy balance is almost closed near the surface, which does not reflect the energy-balance gap in measurements. One possible reason might be the common use of prescribed surface fluxes that cannot adapt to changes in surface temperature and moisture, which would allow for the self-reinforcement of secondary circulations. Therefore, we set up an idealized study, in which we compare the performance of the land-surface and plant-canopy models implemented in PALM to the use of prescribed surface fluxes above homogeneous forest and grassland ecosystems under different atmospheric conditions with respect to realistic energy-balance closure behavior. Furthermore, we evaluate the performance of a dynamic subgrid-scale model, as well as an alternative to the Monin-Obukhov similarity theory (Banerjee et al. 2015, Q. J. R. Met. Soc.).