Correct representation of physical processes, the parametrizations, and their interaction with the resolved circulation is crucial for the performance of numerical models. Here, focus is put on understanding model biases and developing tools to alleviate existing biases. Atmospheric blocking can divert the typical atmospheric flow for several days up to weeks and thereby impacts the mean climate of the region experiencing blocking. Models typically underestimate the frequency of atmospheric blocking. Based on results from the global climate model EC-Earth, it is found that the atmospheric model resolution is not strongly influencing the representation of atmospheric blocking once the grid reaches about 80 km grid length in the horizontal. Updating several physical parametrizations, and thereby the model version, is the largest contributor to advancements in simulating atmospheric blocking. The importance of the topography for the large-scale atmospheric flow is further investigated with the reanalysis ERA-Interim by applying a simplified theoretical analysis. It is found that the idealized topographic forcing theory can explain some part of the observed large-scale properties of the flow, though the method does mainly produce relative results. The explained part of the large-scale structure is increased during periods of northwesterly flow and when the flow impinges the mountain ridge almost orthogonally.

Small-scale processes acting in air masses transported from midlatitudes to the Arctic are also discussed. Numerical models often struggle with representing the stable conditions in the Arctic and tend to underestimate the downward longwave impact during cloudy conditions. A comparison of single-column models (SCMs) indicates that most models can capture the bimodal longwave distribution which develops from alternating cloudy and clear-sky conditions. SCMs are often used for model development as they allow to decouple the parametrized physical processes from the large-scale environment and enable many parameter sensitivity tests. A new tool is presented which can be used for the development of physical parametrizations in marine and polar conditions. It combines one-dimensional models of the atmosphere and ocean, including sea-ice, into a coupled atmosphere-ocean SCM (AOSCM). The presented setup constitutes an advantage compared to SCMs of one component because the coupling is directly modelled and the interaction between the respective boundary layers does not dependent on prescribed boundary conditions.