Title

The Sky in The Limit: Using Convergence Tests for Climate Model Development

Abstract

Earth system models (ESMs) are large, complex codes which enable us to study the climate system in ways that we can not (or should not) explore with experiments. They themselves consist of a zoo of models and discretizations for various, interacting, planetary processes. ESM development means to improve simulation results of the fully coupled system by understanding which individual components are affecting model performance. In this talk, I want to illustrate that simple tools from numerical analysis can help guide model development by exposing components that introduce time-stepping errors or increase model sensitivity. I will present some results from Wan et al. (2015, 2020) on self-convergence tests of CAM5, as well as new results from our research on iterative coupling algorithms for EC-Earth. The former revealed issues in computing water vapor condensation in nonconvective clouds, while we were able to pinpoint the sea ice albedo computation as an error source.