深度学习研究现状及在延伸期预报上的潜在应用初探
Research Status and Potential Application of Deep Learning in Extended Range Forecast
中文摘要：
分别从深度学习技术和延伸期预报业务应用两个方向展开，一方面简要说明时间序列数据处理和局部特征提取采用的主要深度学习模型，以及为克服单模型局限而发展的集合学习采用的主要策略，同时介绍了深度学习在信号搜索和气象预测中的最新成果；另一方面搜集延伸期动力模式的预报水平和国内在延伸期预报业务中采用的主要技术路线；最后讨论了深度学习技术在延伸期预报上的潜在应用方向，以期为国内延伸期预报业务的发展提供一个有益的思路。
中文关键词：
深度学习，持续性强降水过程，延伸期预报
KeyWords：
deep learning, persistent heavy rainfall events, extended range forecast
Abstract：
The application of major deep learning models in time series data processing and local feature extraction is introduced. And some strategies for ensemble learning are also put forward so as to overcome the limitations of single models. In addition, the latest progress of deep learning in the fields of signal search and weather forecast is discussed. On the other hand, an overview of the forecast capacity of seasonal to sub-seasonal dynamic models as well as the technological routes employed in operational extended-range weather forecast in China is presented. And the potential application of deep learning technology in extendedrange forecast is discussed in a bid to offer some enlightenment for the development of extended-range forecast in China.