Taking temperature: how well do weather models simulate Antarctic sea ice?
3 October 2024
A multi-agency science team has published the first research comparing how well Antarctic sea-ice surface temperature is represented in six commonly used weather models, using 18 years of satellite measurements.
The team comprises researchers from the Institute for Marine and Antarctic Studies, Australian Antarctic Program Partnership, Australian Centre of Excellence for Antarctic Science, Bureau of Meteorology, CSIRO and ARC Special Research Initiative for Antarctic Gateway Partnership.
Atmospheric researcher and lead author Dr Zhaohui Wang said that sea-ice surface temperature simulated by weather models is used as an indicator of ice melt and climate change.
“While Antarctic sea ice plays crucial roles in the global climate system, we have limited large-scale knowledge of key physical surface characteristics of sea ice – like its surface temperature.”
“This is due to a lack of ground-based observations combined with limited satellite validation studies. Therefore, we rely on weather models called ‘reanalyses’ which provide a comprehensive picture of past weather conditions anywhere on the planet.”
“However, it’s not fully understood how well the reanalyses represent sea-ice surface temperature in Antarctica.”
By comparing six reanalyses with satellite observations, the team found there was considerable bias in the simulation of sea-ice surface temperature due to weaknesses in the simulation of clouds and sea ice representations for the Antarctic and Southern Ocean.
“Most were too warm – the largest warm bias was over 4oC due to overestimated cloud cover,” said Dr Wang.
Lead author Dr Zhaohui Wang
Sea-ice scientist and co-author Dr Alex Fraser said the results are important because reanalysis models are considered the ‘gold standard’ record of the state of the atmosphere all around the world, and not enough attention has been paid to ‘data-poor’ regions such as the Antarctic sea ice.
“Our work has shown that oversimplifications in the reanalysis models, combined with model shortcomings in their ability to represent clouds accurately, really limits our ability to know the state of the climate.”
“Currently, the best reanalysis model has an assumption of 1.5-metre-thick sea ice everywhere around Antarctica, with no snow cover. These assumptions were based on Arctic sea ice, but don’t work well for Antarctica,” said Dr Fraser.
Their study ‘Antarctic sea ice surface temperature bias in atmospheric reanalyses induced by the combined effects of sea ice and clouds’ is published in Nature Communications Earth & Environment.
