Different approaches that are most commonly used for the evaluation and benchmarking of climate models.

Abstract

Different approaches that are most commonly used for the evaluation and benchmarking of climate models. Most of the approaches can be applied to different realms (e.g., atmosphere, ocean, land and land ice, ocean and sea ice), and each approach can include more than one diagnostic or metric. Approach 1 only includes the Portrait Diagram as metric which is very versatile in its application across different domains, analyzed variables and number of included observations or time periods. Approach 2 represents all diagnostics that are based on analyses of biases and variabilities. Approach 3 includes all diagnostics that focus on spatial analyses, e.g. spatial correlations or physical connections between neighboring regions/realms. Approach 4 includes any budget assessments. These diagnostics are commonly applied globally, but can also be applied regionally if boundary conditions and fluxes across boundaries are clearly defined. Approach 5 represents all other statistical approaches for model evaluation, e.g. the analyses of distributions. Approach 6 finally includes all diagnostics that aim for describing Earth system and its interconnections and changes as a whole, e.g. emergent constraints or equilibrium climate sensitivity (ECS).