Forecasting ENSO using convolutional LSTM network with improved attention mechanism and models recombined by genetic algorithm in CMIP5/6

• Published in: Information sciences Vol. 642; p. 119106
• Main Authors: Wang, Yong, Zhang, Yiming, Wang, Gai-Ge
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.09.2023
• Subjects: Deep learning; ENSO; Genetic algorithms; LSTM; Self-attention; SST anomalies; Deep learning; LSTM; SST anomalies; ENSO; Self-attention; Genetic algorithms
• ISSN: 0020-0255; 1872-6291
• DOI: 10.1016/j.ins.2023.119106

•Forecasting ENSO using 12 consecutive months of ocean data.•Recombining CMIP datasets using a genetic algorithm.•Improving attention mechanisms for regression problems.•Learning spatio-temporal data using a multi-module combined deep learning model.•Effective forecast ENSO can be up to 20 months. El Niño-Southern Oscillation (ENSO) has a profound impact on global climate, and the ability to forecast it effectively over the long term is essential. In recent years, deep learning methods have demonstrated superior prediction outcomes compared to conventional numerical models. However, due to limited observational data, most of these deep learning methods learn information from simulation data derived from physical models, and ensuring the quality of such simulation data can prove challenging. As a result, the models in CMIP5/6 were recombined using genetic algorithms (GAs) to create our training dataset. A deep learning model was then used to learn features from the output of these combined CMIP models so that these physical numerical models can complement each other. To address the issue of inadequate spatiotemporal feature extraction present in many deep learning methods, we devised an ENSO deep learning regression model. An improved self-attention mechanism was introduced into the convolutional LSTM (long short-term memory) network to enable our model to better extract local and global spatiotemporal features. With our method and model, we were able to obtain less erroneous forecast results, and our method surpassed other state-of-the-art deep learning methods in terms of its capability for long-term forecasting.