Identification Method of Typical Dynamic Time Series Weather Scenes in Terminal Area

The weather scenes identification in the terminal area is a data analysis method that can effectively identify the historical operating environment and improve the consistency of the controller's strategy. Aiming at the problem that the existing air transportation meteorological image feature extraction methods are easy to lose the spatial information in the image, and the clustering technology is not high in the identification accuracy of meteorological scenes, this paper proposes a simple and effective dimensionality reduction method-rasterized weather severe index (WSI), and considers the time series of weather information can cluster and identify the weather in the airport terminal area. Rasterized WSI takes into account the spatial structure characteristics of the terminal area and the actual needs of airport control for meteorological information. This dimensionality reduction method not only extracts important spatial information, but also achieves a high degree of dimensionality reduction simply and effectively. Then this paper uses the dynamic time warping (DTW) after optimizing the search path to calculate the time series similarity measure, and uses spectral clustering to identify different weather patterns. In the experiment, we compared different image dimensionality reduction methods. The experimental results proved the effectiveness of our method. At the same time, we visualized the clustering results and analyzed the results.