Thunderstorm and gale automatic identification method and system based on YOLOv3 model

Abstract

The invention discloses a thunderstorm and gale automatic identification method and system based on a YOLOv3 model, and relates to a deep learning technology. The scheme is provided for solving the problems that in the prior art, the workload of manual prediction is large and the like, and the method comprises the steps of establishing a historical thunderstorm and gale standard data set and the standard image data; forming a YOLO label data set; constructing the YOLOv3 model by utilizing the YOLO label data set; inputting the real-time Doppler radar reflectivity jigsaw data to the YOLOv3 model for identification; and after an identification result is obtained, collecting the real-time thunderstorm and gale data of an automatic station for contrastive analysis and verification. The thunderstorm and gale identification method has the advantages of objectiveness, refinement and gridding, the labor intensity can be greatly reduced, and the thunderstorm and gale identification refinement degree is improved. According to the present invention, multilayer radar data is adopted, the vertical structure of radar echoes is fully considered, and thunderstorm and gale and other severe convection types can be effectively distinguished; the thunderstorm and gale identification efficiency can be improved through the training, testing and checking of massive radar data and by adopting the most advanced deep learning technology at present.

Description

technical field [0001] The invention relates to deep learning technology, in particular to a method and system for automatic identification of thunderstorms and strong winds based on the YOLOv3 model. Background technique [0002] Thunderstorms and strong winds are one of the strong convective weather produced by small and medium-scale systems, which often cause major disasters such as sheds, billboards, and tree collapses, including major property losses and even casualties. Because of its suddenness, locality and small scale, Doppler weather radar is the most important monitoring equipment for thunderstorms and strong winds. The identification and early warning of thunderstorms and strong winds is mainly based on the Doppler radar reflectivity factor intensity echo characteristics, radial velocity characteristics, and automatic station live monitoring, including reflectivity factor intensity, echo moving speed, etc. The identification of these radar features related to th...

AI Technical Summary

Problems solved by technology

The identification of these radar features related to thunderstorms and strong winds currently mainly relies on manual monitoring and identification by on-duty forecasters; there are many problems in such monitoring and identification methods: (1) Pure manual monitoring and identification is heavy and time-consuming, which is difficult to deal with Large-scale, everywhere-blossoming or local rapid occurrence and disappearance of thunderstorms and strong winds; (2) The experience level of forecasters is uneven, and inexperienced personnel are likely to miss thunderstorms and strong winds

Images