一种基于深度学习的热带气旋路径集成预报方法

本文提出了一种基于深度学习的热带气旋(tropical cyclone,TC)路径集成预报方法.该方法以长短期记忆深度网络为模型构架,利用前4个时刻(24 h,间隔6 h)及当前时刻的TC路径记录,以及由不同环境因素所计算的方向预报因子作为输入,分别直接预报和间接(通过预报移动速度)预报路径,集成两者预报结果实现时效为...

Consensus forecast method for a tropical cyclone track based on deep learning

This paper proposes a Tropical Cyclone (TC) track forecast method. The method is based on Long Short-Term Memory (LSTM) networks as the backbone. The inputs include the TC historical data (24h, 6h interval) and the current track, as well as the Direction Predictor (DP) calculated by different environmental factors. The model TC-track forecasts track directly, while the model TC-velocity forecasts track indirectly by velocity. The models are integrated based on the idea of Consensus Forecast. The consensus model forecasts the TC track after 24 hours. In the experiments, this paper compares the models driven by different DPs to explore the influence of environmental factors on TC track. The results show that the DP of wind field has the more obvious effect on the forecast model performance, while the DPs of temperature fields and height fields at sea surface have relatively small effect. In addition, the forecast accuracy can be further improved by integrating the selected forecast models of different DPs. The above results verify the feasibility and effectiveness of the proposed DP and consensus forecast method for the TC track forecast.