Detection Of Arc Aurora Event And Research On Its Physical Mechanism

Aurora is the luminous atmosphere that occurs in the high latitudes regions in the earth, caused by the collision of high-energy charged particles of the magnetosphere with neutral particles, it is also a physical activity caused by coupling of the solar wind-magnetosphere-ionosphere. So people found that different forms of aurora images reflect the different physical processes by studying and observing the change of aurora morphology and evolution of its system. Arc aurora is an important form of dayside auroral, its occurrence and distribution reveals magnetospheric pulsation of geomagnetic activity. Based on this, the paper systematically studied the detection of arc aurora event and research of its physical mechanism.Firstly, the existing detection methods which contain redundancy region of no shooting in a part of the aurora when dealing with the whole arc aurora image, a classification method of arc aurora based on Ring Local Directional Pattern(RLDP) is proposed. First, using method of Local Directional Pattern encode the auroral image and get LDP image, and the LDP image is divided many ring regions by polar coordinate system, then by ring region image feature extraction and method of self-tuning spectral clustering. Experimental results show that, compared with other arc aurora feature extraction method, method of the RLDP feature extraction can more accurately detect the starting and ending time of aurora sequence.Secondly, according to the arc aurora sequence detection problem, we propose a ring sector blocking SRLDP arc aurora sequence detection method. The method use the most prominent local directional value of the aurora sequence to represent the Sequence Local Directional Pattern(SLDP) image according to the temporal relationship between auroral images.Then it is divided into many ring areas in the polar coordinate system, and feature extraction in ring sector blocking region and method of self-tuning spectral clustering is used to detect arc aurora sequence. The experiment results show that this feature extraction method which is used to represent arc aurora sequence feature have a better detection results.Finally, according to the problem of the study of the physical mechanism of arc aurora,the methods using mean of Histogram of Oriented Gradient(HOG) feature and mean brightness of each sector of auroral images is proposed to detect the arc aurora. The former way is that the method of Histogram of Oriented Gradient is used to extract auroral image feature and its feature mean, then adjusting the threshold value to detect arc aurora; The latter way is that the method that image is divided into the sector region and then get its mean brightness, and using the mean difference to detect arc aurora. By using the existing DMSP satellite data, the results show that the physical parameters energy flux of the precipitating electrons and the average energy of occurrence period of aurora will have different trends as the change of the auroral morphology. By analyzing the results of the occurrence of arc aurora and the relationship with IMF components, the experiment results show that the distribution of arc aurora presents feature of double-peak.