A sub-grid parameterization scheme for topographic vertical motion in CAM5-SE

Abstract

Overestimation of precipitation over steep mountains is always a common bias of atmospheric general circulation models (AGCMs). One basic reason is the imperfection of parameterization schemes. Sub-grid topography has a non-negligible role in the dynamics of the actual atmosphere, and therefore the sub-grid topographic parameterization schemes have been the focus of model development. This study proposes a sub-grid parameterization scheme for topographic vertical motion in CAM5-SE (Community Atmospheric Model version 5 with spectral element dynamical core) to revise the original vertical velocity by adding the topographic vertical motion, resulting in a significant improvement in simulations in precipitation over steep mountains. The results show a better improvement in precipitation simulation in steep mountains, such as the steep edge of the Tibetan Plateau and the Andes. The positive deviations of the precipitation on the mountain tops and the negative deviations in the windward slope are revised. The improved scheme of topographic vertical motion reduces the model biases of summer mean precipitation simulations by up to 48 % (6.23 mm d−1) on the mountain tops. The improvement in convective precipitation (4.83 mm d−1) contributes the most to the improvement in the total precipitation simulation. In addition, we extend the dynamic lifting effect of topography from the lowest layer (Single experiment) to multiple layers, approaching the bottom model layers (Multi experiment). Moreover, the water vapor transport in low-altitude regions in front of the windward slope is also considerably improved, leading to simulations of much more realistic circulation patterns in the multi-layer scheme. Since the sub-grid parameterization scheme addresses the more detailed problem caused by topography, the water vapor is transported further to the northwest in the multi-layer scheme. The topographic vertical motion schemes in both the Single and Multi experiments can improve the model performance in simulating precipitation in all regions with complex terrain.