Multirate Time Integration Methods for Atmospheric Microphysics

Abstract:

The parameterized atmospheric physics component of global climate models comprises a significant portion of the computational cost of large-scale climate simulations. This aspect of the climate model is composed of numerous physical processes acting on a wide range of temporal scales requiring time integration schemes that capture this variety while providing high accuracy at relatively low computational costs. Currently, many models utilize low order splitting schemes that can suffer from poor accuracy (less than first order) and may lead to instabilities. In this work we explore the application of high order explicit-explicit multirate time integration methods in the Morrison-Gettleman 2 (MG2) microphysics model where sedimentation (e.g., rain) can occur at rates on the order of 100 times faster than other microphysics processes. When evolving sedimentation with a smaller time step relative to the other microphysics processes tests show a third order multirate method is more efficient than the native sequentially split and substepped explicit Euler method when greater accuracy is required.