Senstivity simulations of the MJO and tropical climate using a superparameterized version of the global WRF model

Abstract:

The method of superparameterization is a well-established approach for (explicitly) representing moist convective-scale processes in coarse-resolution models. While the advantages of this approach include a reduction in the number of uncertain closure assumptions, a corresponding drawback is its much greater computational expense, as compared to traditional parameterization schemes. However, this drawback is becoming less of an issue with recent computing advances, so much so, that superparameterized models can now be regarded as relatively inexpensive tools for developing global cloud resolving models of the future. Indeed, there is growing evidence that even the latter models suffer from the same sorts of climatological biases as much coarser resolution models of the past. The implication is that deficiencies in the treatment of unresolved (microscale) processes are still playing an important role. With these issues in mind, a superparameterized version of the Weather Research and Forecast model has been developed for studying the sensitivities of weather and climate simulations to formulations and parameter settings in microscale physics schemes. Here the primary phenomena of interest include the MJO and climatological pattern of rainfall during boreal summer. Preliminary results show surprisingly strong sensitivities to the choice of turbulent Prandtl number, as well as to the choice of the rain collection efficiency factor.