The reduction of vertical grid point distance needed to obtain a substantial improvement, results in a massive increase of grid points. The increased computational costs would make standard GCM simulations too costly and impractical.
During winter, most of the IPCC models simulate too warm surface conditions in high northern latitudes. GCM simulations of 20th century climate, with observed forcing, contain the well known warming north of 60° after the mid-sixties. However, the simulated geographic distribution of the warming is quite different from the observed one. The present work, which is a part of Øyvind Byrkjedal’s PhD thesis, has focused on how turbulence representation in GCMs affects the systematic temperature biases.
Turbulence in the surface layer is a key process in determining the temperature near the ground. This process in unrealistically represented with standard vertical resolution in today’s GCMs. The group furthermore pointed out that the most popular parameterisation for such processes is relevant and built on valid assumptions in the Arctic climate. It is, however, the implementation, i.e. the use of the low resolution, which deteriorates the simulation results. An experiment with the ARPEGE GCM demonstrated that the vertical resolution needed to significantly improve the results, will increase the number of grid points to an extent that makes standard GCM simulations too costly on present state-of-the-art computers. A turbulence parameterisation that rests on alternative principles is therefore needed to remove the systematic errors.
Global Climate models simulate too warm surface temperatures in the Arctic
In a new paper in Climate Dynamics, a group of scientists at the Bjerknes Centre for Climate Research has shown that wintertime systematical errors in General Circulation Models (GCM) over the Arctic can largely be attributed to the large vertical distance between of computational (grid) points employed.
Body