Attention-based Deep Tropical Cyclone Rapid Intensification Prediction

Rapid intensification (RI) is when a sudden and considerable increase in tropical cyclone (TC) intensity occurs. Accurate early prediction of RI from TC images is important for preventing the possible damages caused by TCs. The main difficulty of RI prediction is to extract important features that are effective for RI prediction, which is challenging even for experienced meteorologists. Inspired by the success of deep learning models for automatic feature extraction and strong predictive performance, we initiate this study that experiments with multiple domain-knowledge guided deep learning models. The goal is to evaluate the potential use of these models for RI prediction. Furthermore, we examine the internal states of the models to obtain visualizable insights for RI prediction. Our model is efficient in training while achieving state-of-the-art performance on the benchmark dataset on HSS metric. The results showcase the success of adapting deep learning to solve complex meteorology problems.

[1]  David S. Nolan,et al.  Tropical Cyclone Intensification from Asymmetric Convection: Energetics and Efficiency , 2007 .

[2]  Timothy L. Olander,et al.  The Advanced Dvorak Technique: Continued Development of an Objective Scheme to Estimate Tropical Cyclone Intensity Using Geostationary Infrared Satellite Imagery , 2007 .

[3]  J. Knaff,et al.  A Revised Tropical Cyclone Rapid Intensification Index for the Atlantic and Eastern North Pacific Basins , 2010 .

[4]  Eric A. Hendricks,et al.  Quantifying Environmental Control on Tropical Cyclone Intensity Change , 2010 .

[5]  James P. Kossin,et al.  New Probabilistic Forecast Models for the Prediction of Tropical Cyclone Rapid Intensification , 2011 .

[6]  F. Marks,et al.  The Hurricane Forecast Improvement Project , 2013 .

[7]  Christopher D. Manning,et al.  Effective Approaches to Attention-based Neural Machine Translation , 2015, EMNLP.

[8]  Christopher S. Velden,et al.  Improvements in the Probabilistic Prediction of Tropical Cyclone Rapid Intensification with Passive Microwave Observations , 2015 .

[9]  Dit-Yan Yeung,et al.  Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting , 2015, NIPS.

[10]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[11]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[12]  Yoshiaki Miyamoto,et al.  Structural Changes Preceding Rapid Intensification in Tropical Cyclones as Shown in a Large Ensemble of Idealized Simulations , 2017 .

[13]  Andrew E. Mercer,et al.  Atlantic Tropical Cyclone Rapid Intensification Probabilistic Forecasts from an Ensemble of Machine Learning Methods , 2017 .

[14]  Zhuowen Tu,et al.  Aggregated Residual Transformations for Deep Neural Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Daisuke Matsuoka,et al.  Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model , 2018, Progress in Earth and Planetary Science.

[16]  Rahul Ramachandran,et al.  Tropical Cyclone Intensity Estimation Using Deep Convolutional Neural Networks , 2018 .

[17]  Mohammed Bennamoun,et al.  Attention in Convolutional LSTM for Gesture Recognition , 2018, NeurIPS.

[18]  Sa-Kwang Song,et al.  DeepTC: ConvLSTM Network for Trajectory Prediction of Tropical Cyclone using Spatiotemporal Atmospheric Simulation Data , 2018 .

[19]  Hsuan-Tien Lin,et al.  Rotation-blended CNNs on a New Open Dataset for Tropical Cyclone Image-to-intensity Regression , 2018, KDD.

[20]  Hsuan-Tien Lin,et al.  Estimating Tropical Cyclone Intensity by Satellite Imagery Utilizing Convolutional Neural Networks , 2019, Weather and Forecasting.