Identifying highly skilled ensemble members could boost extreme cold forecasts in East Asia

Seasonal to seasonal forecasts play an important role in the early warning of extreme weather events and disaster risk prevention, but their predictive capabilities are still limited.

Extremely cold events in East Asia cause widespread disasters in the form of cold temperatures and snow, seriously affecting transportation, energy supply and public life. However, current S2S models typically have limited predictive capabilities for extreme cold events over two weeks.

This study evaluated prediction skills for extreme cold events in East Asia using “high-skilled members in the subseason prediction ensemble of extreme cold events in East Asia,” published in the letters of Atmospheric and Marine Science and using ECMWF (European Centre for Medium-Range Weather Prediction) and ERA5 (Fifth major global regenerative analysis generated by ECMWF) data.

The findings reveal that the ensemble averages in the ECMWF model have limited predictive capabilities for extreme cold events after 2 weeks, while some ensemble members show rather high predictive skills.

Members with high prediction skills can accurately predict the rapid changes in surface temperatures and the intensity of minimum temperatures during extreme cold events. This mainly depends on accurate predictions of the Eurasian atmospheric circulation situation (sea surface pressure and high geopotential of 500 hPa).

“Of the ensemble members, at least 10% were highly skilled members who always provide valuable insights,” says Xinli Liu, author of The Paper.

Future efforts should focus on identifying highly skilled members using historical analog or AI and assigning larger weights in ensemble predictions.

“In ensemble predictions, by increasing the number of ensemble members appropriately and assigning large weights to high-skilled members, we improve prediction accuracy and reliability,” added corresponding author Jingzhi Su, pointing to an optimized prediction system.