Increasing computational resources are allowing climate research centers to increase the resolution of their atmospheric models. This increase in resolution has a host of benefits for studies of many aspects of climate variability and change, ranging from the effects of global warming on hurricanes to the feedbacks on climate change associated with cirrus clouds. In this proposal, we describe research aimed at accelerating our development of a set of atmospheric models with grid sizes ranging from 50 km down to 12 km. Working at this resolution is rapidly becoming both practical and an exciting source of new insights into the climate system.

We propose a strategy involving both realistically-configured models that represent our best attempts at simulation and a particular set of idealized models. Realistic configurations are obviously required to evaluate how one's model compares to nature, but in isolation they are difficult to analyze due to their complexity. Idealized models are invaluable for clarifying many of the consequences of differing sub-grid closures, in particular the sources of resolution- dependence.

Our focus is on the tropical upper troposphere, including the Tropical Tropospheric Layer between the cirrus convective outflows and the tropopause. We choose this focus based on the conviction that the tropical upper troposphere is a region where models of 12-50 km resolution are likely to be particularly efficacious.

The idealized models we consider are non-rotating and rotating radiative convective equilibria in planar, doubly periodic, horizontally homogeneous domains, and aqua-planet models in which otherwise realistically-configured models are placed over a prescribed temperature ocean surface with zonally symmetric temperatures. The doubly periodic models have the additional advantage that we can generate cloud-resolving non-hydrostatic simulations to compare with mesoscale-resolution models using GCM column physics closures. This is more difficult for the global aqua planet, but in this case we propose to use a deformed mesh model over aqua-planet conditions that directly tests the resolution-insensitivity of our closure schemes.

Complementing our idealized simulations, we also propose a comparison of mesoscale and cloud-resolving resolutions of a test case provided by the Tropical Warm Pool - International Cloud Experiment (TWP/ICE) field experiment, a case that is particularly useful for testing upper tropospheric simulations in a convective region. We also propose to study a model in which meteorology is restored to reanalysis observations to clarify sources of biases in water vapor and ice distributions in the tropical upper troposphere, as well as the sources of water entering the stratosphere, in our global models with different resolutions.