Skillful Prediction of Indian Monsoon Intraseasonal Precipitation Using Central Indian Ocean Mode and Machine Learning

• Published in: Geophysical research letters Vol. 51; no. 24
• Main Authors: Zhou, Lei, Yu, Yanwei, Yan, Bingqi, Zhao, Xingyu, Qin, Jianhuang, Tan, Wei, Tang, Youmin, Li, Xiaofeng, Li, Xiaojing, Dong, Junyu, Chen, Dake, Murtugudde, Raghu
• Format: Journal Article
• Language: English
• Published: Washington John Wiley & Sons, Inc 28.12.2024; Wiley
• Subjects: Algorithms; Central Indian Ocean mode; Climate; Climate and weather; Climate prediction; Coupled modes; Current prediction; Indian summer monsoon; intraseasonal rainfall forecast; Lead time; Learning algorithms; Machine learning; machine learning algorithm; monsoon intraseasonal oscillation; Monsoon precipitation; Monsoon rainfall; Monsoons; Noise prediction; Oceans; Precipitation; Rain; Rainfall; Summer monsoon; Variability; Verification; Versatility; Weather; Weather forecasting; Wind; Indian Ocean
• ISSN: 0094-8276; 1944-8007; 1944-8007
• DOI: 10.1029/2024GL112308

Monsoonal precipitation is dominated by intraseasonal variabilities, whose skillful prediction lead time is currently less than 5 days and remains a grand challenge. Here we show that an intrinsic variability in the Indian Ocean, the Central Indian Ocean (CIO) mode, when combined with a machine learning (ML) algorithm, can produce skillful predictions of intraseasonal precipitation over the monsoon region with a lead time of over 15 days, which is close to the theoretical predictability limit. This remarkable skill improvement stems from the fact that the CIO mode is dynamically related to the intraseasonal monsoon rainfall, while the data‐driven ML algorithm suppresses unwanted high‐frequency noise. Using the CIO mode and the ML algorithm, the forecast system hybridizes physical fundamentals and versatility of data‐driven algorithms. The identification of CIO mode and the verification of its significant contribution to intraseasonal predictions advance our understanding of the coupled monsoon system and also underscores the great potential of ML techniques in weather forecasts and climate predictions. Plain Language Summary Rainfall during the Indian summer monsoon is dominated by variations with a period of tens of days, which are referred to as intraseasonal variabilities. Current prediction skill of intraseasonal monsoonal rainfall is less than 5 days and it remains a grand challenge in terms of increasing the current prediction skill. Here we show that an intrinsic mode of variability in the Indian Ocean, called the Central Indian Ocean (CIO) mode, when combined with a machine learning (ML) algorithm, can produce skillful predictions of intraseasonal precipitation over the monsoon region with a lead time of over 15 days. This remarkable skill improvement stems from the fact that the CIO mode is dynamically related to intraseasonal monsoon rainfall, while data‐driven ML algorithm suppresses disruptive noise with a period shorter than 10 days. Using the CIO mode and an ML algorithm, the forecast system synergizes physical fundamentals and versatility of data‐driven algorithm. The identification of CIO mode and the verification of its significant contribution to intraseasonal prediction advance our understanding of the coupled monsoon system and also demonstrate the great potential of ML techniques in weather forecasts and climate predictions. Key Points The Central Indian Ocean (CIO) mode provides a dynamical basis for the prediction of monsoon intraseasonal rainfall The machine learning (ML) algorithm suppresses high‐frequency noise while capturing the real‐time CIO mode index The dynamics and ML hybrid forecast system can skillfully predict monsoon intraseasonal rainfall with a lead time of ∼15 days