Evaluation of Cold Wave Impact-based Forecasts in South Korea

In recent years, cold waves have caused the increase in the number of patients, hospitalizations and deaths during the wintertime in South Korea. To mitigate these damages, the cold wave impact-based forecasting can be a measure to alleviate the victims by alarming the warnings to public before it happens.

Since 2020, the Korea Meteorological Administration has been providing operational cold wave impact-based forecasts, categorizing four levels of impact (attention(blue), caution(yellow), warning(amber), and danger(red)) and offering different actions. Thresholds for impact levels are determined by two criteria: when the daily minimum temperature drops sharply from the previous day or when it remains below a certain threshold for more two consecutive days. In addition, these thresholds are classified by region to provide cold wave warnings that considers local climate characteristics.

Through the case studies of winter in 2020(’20-’21) and 2021(’21-’22), we have examined the correlation between cold-related illnesses and levels of impact. The results show a correlation coefficient of 0.51 between the forecasts and cold-related illnesses above the amber. To study the lagged effect of cold waves, we also calculated the correlation coefficient for the lag of 0 to 3 days. The results show that the correlation coefficient for the one-day lag was 0.56 in the red, which is higher by 0.27 than that with 0-day lag(0.29) for the year of 2020 case. The coefficient for 1-day lag is 0.73 in the red, which is also higher by 0.08 than the coefficient(0.65) with 0-day lag for the year of 2021 case. Although further study with longer data sets, this result shows that the likelihood of cold-related illnesses become higher at the one day after a strong cold wave occurs.

Key words: cold wave, impact-based forecast, levels of impact, post-hoc analysis, evaluation.

This study was conducted as part of the project “Advancement of Regionalized Impact-based Forecasting Service” (KMA2018-00723) by the Impact-based Forecasting Promotion Team of the Korea Meteorological Administration.