A Novel Anticyclone Objective Recognition Method Based on Mask R-CNN

Abstract

The present invention proposes an objective anticyclone identification method based on the Mask R-CNN deep learning model to improve the accuracy of anticyclone identification and improve the ability to objectively describe the two-dimensional shape characteristics of the anticyclone system. The anticyclone objective identification method proposed by the present invention uses sea level air pressure data and artificially identified anticyclone data to train the Mask R-CNN deep learning model, and obtains machine-identified anticyclone data through the trained model. The objective recognition method of the present invention can perform relatively accurate individual position recognition on anticyclones, and at the same time, the objective recognition method has a relatively good ability to express two-dimensional shape characteristics of actually existing anticyclone systems.

Description

Technical field: [0001] The invention relates to the technical field of automatic identification of weather systems, in particular to a novel anticyclone objective identification method based on the Mask R-CNN deep learning model. Background technique: [0002] The activity of cold high pressure / anticyclone on the surface of Eurasia in winter is often accompanied by large-scale cold air outbreaks. At the same time, strong anticyclones can cause disastrous weather / climate such as local blizzards, freezing rain, and strong winds. In the context of global warming, the frequent occurrence of extreme cold events in winter in Eurasia in the northern hemisphere has attracted widespread attention in recent years. The near-surface anticyclonic activities in Eurasia in winter are closely related to the change of cold air, but the research on it, especially the automatic recognition of the shape, is still insufficient. The location, intensity, and shape changes of winter anticyclones ...

AI Technical Summary

Problems solved by technology

[0003] Due to the complex shape of the anticyclone and the generally weak internal pressure gradient, it is difficult to use algorithms to identify objectively

The anticyclone objective automatic identification algorithm proposed in the past has insufficient ability to identify the two-dimensional activity characteristics of the system's influence range and shape variation. The identification and tracking of unclosed systems is accompanied by large uncertainties, and it is easy to identify multi-center systems. There is an error

Images