Harlaß, Jan (2016) Tropical Atlantic bias dependence on atmospheric model resolution. (PhD/ Doctoral thesis), Christian-Albrechts-Universität Kiel, Kiel, Germany, 111 pp.

Text diss-jh-2016.pdf - Published Version Restricted to Registered users only Available under License German copyright act UrhG. Download (13MB) | Contact

Abstract

In this thesis the influence of atmospheric model resolution in a global coupled general circulation model on the simulation of the tropical Atlantic (TA) climate is investigated. Special emphasis is given to the vertical resolution, which is shown to be the agent of choice to substantially reduce model biases in the TA. As essentially all state-of-the-art coupled models fail to reproduce the TA climate in a satisfying manner, there is need for a more general strategy outside of model specific changes. For this purpose a range of atmospheric horizontal resolutions from 2.8◦ to 0.47◦ and vertical resolutions from 31 to 62 levels have been employed in the Kiel Climate Model, while keeping the ocean configuration fixed at a relatively coarse resolution. The results suggest that the latter is of secondary importance, whereas enhanced atmosphere resolution is indispensable for a realistic simulation of the TA mean state as well as interannual variability. A strong warm sea surface temperature (SST) bias in the eastern and south eastern TA alleviates just slightly at enhanced horizontal resolution alone and can only be substantially reduced at higher vertical resolution. Thus, the zonal SST gradient along the equator is no longer reversed. Moreover, subsurface temperature biases along the equator are removed when enhanced wind stress at higher vertical resolution shoals (deepens) the thermocline in the east (west). A large westerly wind bias in the western TA extends from the surface to a height of approx. 600 hPa and is already present in atmosphere-only configurations. A higher number of vertical levels leads to a spatial redistribution of rainfall, a northward shift of convective activity as well as enhanced atmospheric vertical mixing that both impact entrainment of zonal momentum and altogether the zonal pressure gradient. Subsequently, the westerly bias is essentially removed. As a consequence, erroneous equatorial wave activity excited by the westerly bias impacting the upper ocean temperature distribution at the equator and southward along the African coast is shut down. Together with an enhanced meridional current structure in the Benguela upwelling system, resulting from improved wind stress and wind stress curl pattern at higher resolution, the large temperature bias reduces considerably. Improved upper ocean structure along with a correct latitudinal position of maximum rainfall north of the equator and associated strong cross-equatorial winds in the eastern TA during late boreal spring allow a seasonal cold tongue development. Seasonal phase locking of SST in that region and the complete Bjerknes feedback can only be resolved at increased vertical atmospheric resolution. Contrary to intuition in the model runs that have a realistic interannual variability the Atlantic is more damped. In the Bjerknes index dynamical and thermodynamical damping outweigh the destabilising thermocline feedback resulting in a heavily damped system. The seasonal phase locking of SST originates from the interaction of windinduced changes of thermocline slope and subsurface to surface ocean temperatures.

Actions (login required)

View Item