A prediction model for heatwaves and tropical nights based on apparent temperature

초록

According to the Korea Meteorological Administration (KMA), a heatwave is defined as two or more consecutive days with maximum temperatures above 33${}^{\circ }\mathrm{C}$. Due to severe health risks, the KMA revised its criteria in 2020 from daily maximum temperature to apparent maximum temperature, emphasizing the need for predictive models based on apparent temperature. This study proposes and compares two models for forecasting heatwaves and tropical nights: a logistic regression model using explanatory variables of apparent temperature, and XGBoost, a machine learning algorithm known for robust predictive performance. Both models were trained on ensemble meteorological data from the KMA, and their performance was evaluated using mean squared error (MSE), logarithmic loss, and accuracy. Results demonstrate the strengths and limitations of each method in forecasting extreme temperature events, offering insights to improve early warning systems and public health responses.

키워드