Quantification of parameterization behavior across scales

Description

The current methodology used to downscale climate simulations is to nest a high-resolution regional model into a coarse global climate model domain. This nesting approach works because parameterizations can be chosen and configured independently for each grid such that the setup is appropriate for the grid spacing used at each scale. This mixed physics-parameterization choice leads to the application of differing assumptions at different scales, which prevents the use of the same parameterization suite for both grids. Complicating matters, many next-generation climate models currently being developed will most likely not use this nesting approach. Instead, they will use a single global domain that has variable resolution. Higher resolution will be used where needed, either for improving model accuracy or where regional details are deemed more valuable, and lower resolution will be used elsewhere. This multi-resolution grid methodology imposes new requirements on model physics parameterizations. The parameterizations will need to be scale-aware and able to adapt to the range of grid spacing used throughout the domain. To prepare for these next-generation models, a methodology is proposed to test the scale adaptability of physics parameterizations. A testbed is being developed that uses separate grids within an atmospheric model for dynamical processes (e.g., transport) versus physics processes (e.g., clouds and radiation). The methodology for this approach will be described, showing its benefits for designing an appropriate parameterization suite for the next-generation multi-resolution climate models.