The weather always affects people’s daily life to a greater or lesser extent, and the accuracy of the weather forecast considerably impacts the way that people deal with the weather. In particular, extreme weather events at hourly scale pose a serious threat to the safety of people’s life and property. However, traditional physical dynamics models are too weak to curb these events. In recent years, deep learning has been showing bright potential and substantial advantages in manifesting the mechanisms and forecasting of short-term extreme weather events (STEWE). The essence of deep learning is built upon massive amounts of different types of data, from which it can learn new laws. Yet how to deeply filter out key information from colossal data and carry out scientific justification still remains to be deciphered.
This Research Topic aims to present and disseminate recent advancements in applications of deep learning in emerging multidisciplinary issues of climate dynamics, especially with regard to STEWE. It shall facilitate a better understanding of (1) the formation mechanisms, development characteristics, and key driving factors of STEWE; (2) the interplay between STEWE and meteorological disasters; and (3) the forecasting accuracy and associated risk assessment of STEWE.
This Research Topic seeks high-quality contributions from meteorologists, physical geographers, earth system scientists, remote sensing scientists, IT engineers, and machine vision experts to address the applications of deep learning in mechanisms and forecasting of STEWE in the following themes that include, but are not limited to:
• Time series information, deep convolutional neural networks, Generative Adversarial Network, fusion of deep learning and physical models
• Heavy rain, tropical cyclone, and flood
• Extreme heat and heat waves
• Remote sensing and radar signal processing
• Solar and photovoltaic, wind energy
• Feature selection, feature decomposition, and multi-model ensemble forecasting
The weather always affects people’s daily life to a greater or lesser extent, and the accuracy of the weather forecast considerably impacts the way that people deal with the weather. In particular, extreme weather events at hourly scale pose a serious threat to the safety of people’s life and property. However, traditional physical dynamics models are too weak to curb these events. In recent years, deep learning has been showing bright potential and substantial advantages in manifesting the mechanisms and forecasting of short-term extreme weather events (STEWE). The essence of deep learning is built upon massive amounts of different types of data, from which it can learn new laws. Yet how to deeply filter out key information from colossal data and carry out scientific justification still remains to be deciphered.
This Research Topic aims to present and disseminate recent advancements in applications of deep learning in emerging multidisciplinary issues of climate dynamics, especially with regard to STEWE. It shall facilitate a better understanding of (1) the formation mechanisms, development characteristics, and key driving factors of STEWE; (2) the interplay between STEWE and meteorological disasters; and (3) the forecasting accuracy and associated risk assessment of STEWE.
This Research Topic seeks high-quality contributions from meteorologists, physical geographers, earth system scientists, remote sensing scientists, IT engineers, and machine vision experts to address the applications of deep learning in mechanisms and forecasting of STEWE in the following themes that include, but are not limited to:
• Time series information, deep convolutional neural networks, Generative Adversarial Network, fusion of deep learning and physical models
• Heavy rain, tropical cyclone, and flood
• Extreme heat and heat waves
• Remote sensing and radar signal processing
• Solar and photovoltaic, wind energy
• Feature selection, feature decomposition, and multi-model ensemble forecasting