Study of the effect of the QNSE-based surface layer parameterization on the warm bias in simulations of stably stratified boundary layers

A warm bias of the 2m temperature is a notorious problem arising in numerical simulations of stably stratified boundary layers. This bias increases with the strength of stratification and hinders predictive skills of various numerical weather prediction (NWP) systems. The bias gives rise to the Nordic Temperature Problem occurring in simulations of very cold Arctic boundary layers where it can reach 25ºC in magnitude. Most NWP systems use the Louis or similar schemes for the parameterization of the momentum, heat and moisture transfer in the near-surface region of the boundary layer. Using the recently developed quasi-normal scale elimination (QNSE) theory of stably stratified turbulent flows, we have derived an alternative surface layer parameterization scheme and tested it in one-dimensional versions of HIRLAM and WRF with various vertical resolutions. While with the Louis scheme, the warm bias is present for any resolution, it is practically eliminated when the QNSE-based parameterization is employed. The improvement is particularly strong when the near-surface resolution is relatively crude, as is the case in state-of-the-art NWP systems. The QNSE-based surface layer parameterization has also been incorporated in a fully operational 3D weather forecast system HIRLAM. Preliminary simulations indicate that this parameterization considerably improves the predictive skills of HIRLAM in 48-hours forecasts. Similar improvements can be expected upon implementing the QNSE-based turbulence model in 3D WRF.