Turbulence intermittency and parameterisations for stably stratified conditions in the ICON-WarmWorld model

Numerical weather prediction and climate models typically rely on parameterisations developed for mid-latitude conditions and apply those also over the Arctic. These parameterisations do not adequately capture the intermittent and non-stationary nature of turbulence under stable stratification, as they frequently occur over the Arctic or during nighttime over land. Yet adapting the physical representation of Arctic air–sea–ice–ocean interactions may not only affect the variability and long-term changes of Arctic atmospheric circulation, but also have the potential to influence mid-latitude atmospheric circulation.

As part of the WarmWorld project consortium, which develops a storm- and eddy-resolving Earth system model based on the German community model ICON, we aim to improve the representation of turbulence under stably stratified conditions. Motivated by previous work by Boyko and Vercauteren (2023), who developed a data-driven stochastic generalization of traditional Monin–Obukhov similarity theory, we aim to develop and implement an improved parameterisation of turbulent surface fluxes that captures variability and the net effects of non-turbulent, small-scale processes on the mean flow.

As a first step, we run the ICON model in a limited-area setup at a horizontal resolution of 5km, centred on the research vessel Polarstern during the MOSAiC expedition in winter 2019/20. This setup allows for an evaluation of the model’s default performance under Arctic winter conditions and facilitates the identification of pronounced yet common model biases, such as cold surface temperatures and excessive near-surface stability. In particular, we investigate how changes in model resolution and adaptations to the turbulent surface-flux parameterization over sea ice under stably stratified Arctic conditions affect the lower Arctic boundary layer and may help to mitigate model biases.

Alongside this, and based on MOSAiC turbulence measurements, initial attempts are made to develop a data-driven stochastic model that accounts for unsteady mixing and uncertainty around classical surface stability functions under stable conditions. This is achieved by identifying a scaling of the parameters of a stochastic differential equation (SDE) with a flow stability parameter, such as the bulk Richardson number. The resulting stochastic formulation of the surface stability functions can then be incorporated into a bulk parameterization of turbulent surface fluxes within the model.