fnctId=thesis,fnctNo=426
Deep-learning-based Prediction of Tropical Intraseasonal Oscillations and Associated Heatwaves
- 작성자
- 기후시스템전공
- 저자
- Vazhaparambil Sasi Arya
- 발행사항
- 발행일
- 2026-02
- 저널명
- 국문초록
- 영문초록
- Subseasonal prediction continues to show very limited forecast skill due to the challenges in understanding and representing the interactions between major predictability sources such as characteristics of Intraseasonal Oscillations, evolution of ENSO, tropical-extratropical teleconnections, etc. Madden Julian Oscillation (MJO) is the key factor of subseasonal predictability that influences temperature and precipitation extremes like heatwaves, floods globally. Heat extremes (heat waves) have profound impacts on human health, socioeconomic stability, and the environment worldwide. Recent studies suggest that the heat waves are becoming more frequent and intense over parts of the globe in the present climate. Accurate and advanced prediction of these heat waves is crucial for enhancing preparedness and mitigating future risks. However, state-of-the-art numerical weather prediction models often struggle with significant biases, particularly in forecasting heat extremes associated with intraseasonal oscillations. The present study investigates the prediction of MJO and its associated heatwave events, using an advanced AI-based global weather forecast model, KIST’s Atmospheric Rhythm with Integrated Neural Algorithms (KARINA). Trained on daily-mean atmospheric variables from the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 reanalysis dataset at a 250-km horizontal resolution, KARINA provides forecasts up to 30 days in advance. KARINA outperforms traditional models, providing skillful MJO forecasts up to 30 days in advance. This performance is especially evident in boreal summer and winter, with the model consistently capturing MJO dynamics and their link to tropical and mid-latitude weather extremes. The 2018 summer heatwave events over Northeast Asia are investigated to analyze the model’s ability to capture heatwave events associated with Tropical Intraseasonal Oscillations. KARINA’s innovative algorithms captured tropical dynamics, which are highly promising for a data-driven model. The model’s superior representation of the MJO, along with its advanced ensemble forecasting techniques, enhances subseasonal prediction skill and offers valuable insights into extreme weather preparedness. These results highlight KARINA’s potential as a powerful AI-driven tool for improving heat extreme predictions, ultimately enhancing preparedness and mitigation strategies across affected regions.
- 일반텍스트