General Circulation Model Errors Are Variable across Exoclimate Parameter Spaces

Abstract General circulation models (GCMs) are often used to explore exoclimate parameter spaces and classify atmospheric regimes. Models tuned give reasonable climate states for standard test cases, such as the Held–Suarez test, then simulate diverse exoclimates by varying input parameters rotation rates, instellation, optical properties, frictional timescales, so on. In studies, there is an implicit assumption that model works reasonably well case will be credible at all points in arbitrarily wide space. Here, we this using open-source GCM THOR on tidally locked Earth-like planets with periods of 0.1–100 days. We find error, quantified ratio between physical spurious numerical contributions angular momentum balance, extremely variable across range some cases where errors dominant component. Increasing grid resolution does improve errors, but a higher-order diffusion scheme can sometimes magnify finite-volume dynamical solvers. further show minimize error make balance more within our model, surface friction timescale must smaller than rotational timescale.