Computational Cost for the RDC Scheme

First uploaded on 2022/07/30
Last updated on 2026/03/01
Copyright(C)2022-2026 jos <jos@kaleidoscheme.com> All rights reserved.

Radiatively Driven Circulation (RDC) scheme requires solving a boundary value problem, which is an iterative computation by nature. But since it is a problem in a horizontal two-dimensional plane for each altitude, the computational complexity per calculation is not so large that it is not unrealistic from a computational cost point of view. In addition to this, the RDC scheme calculation only needs to be performed for each time step Δt_R much longer than Δt required by the dynamics calculations (but short enough to resolve the time scale over which the radiation field changes), so the frequency of calculations can be reduced and the overall computational cost should not be too burdensome.

Most importantly, however, the RDC scheme's computational cost advantage over other dynamical methods stems from its ability to impose an upper limit on climate model resolution. Methods based on cumulus dynamics require resolving convective motion, necessitating a grid resolution fine enough to capture internal motion. In contrast, the RDC scheme disregards the detailed internal motion of cumulus clouds and treats it as mixed turbulence instead. Consequently, the computational domain is confined to the cloud's exterior. Thus, once the model achieves sufficient resolution to distinguish the cloud from clear-sky regions, no further improvement in resolution is necessary. This approach puts an end to the endless competition among research organizations to achieve ever-higher climate model resolutions. Researchers can then allocate computational resources to other meaningful physical processes instead of meticulously handling climatologically insignificant turbulence within cumulus clouds.