Summary: | The evaluation of sea ice models is a necessary step in the course of their development. Evaluation can take many forms, from checks of basic conservation laws to advanced process-oriented diagnostics used to evaluate climate-relevant feedbacks. For this purpose, many diagnostics and metrics have flourished in the past years in the scientific literature, each with their pros and cons. While it is generally agreed that no diagnostic or metric is superior to the others, they should meet a list of minimal requirements such as stability with respect to climate internal variability, ease of physical interpretation and reproducibility among others. In this presentation, I will illustrate with practical examples (historical simulations, and long term projections) the importance, often overlooked, of carefully designing diagnostics and metrics adapted to answer specific scientific questions.
|