Sea-ice biases and simulation fidelity across multiple CMIP phases
Polar sea ice variability and change is an iconic climate property receiving considerable research attention. Assessing the fidelity of simulated sea ice extent, distribution, variability, and their changes is a continuing climate modeling challenge. This study describes several approaches for improving the objective assessment of sea ice distribution. The analysis compares data from CMIP5 and CMIP6 historical simulations with latest-generation observations. Rather than pursuing a simple quantification of the total hemispheric sea ice area and its equatorward extent, we investigate regional sea ice distribution, considering three sectors across the Arctic and Antarctic regions, respectively. Across these regions, the climatological spatial distribution and the annual cycle are quantified. The results highlight that some models' error compensation across sectors can be substantial. Moving beyond total sea ice extent, our technique can further quantify model improvement across versions and prioritize development opportunities for targeted model improvement. These sea-ice variability metrics are implemented in the PCMDI Metrics Package (PMP), enabling efficient reuse and the rapid evaluation framework aimed at CMIP7.