Research On Mechanisms Of Auroral Radiation And Application Of Multi-band Spectral Data

Aurora,the spectacular phenomenon that occurs in high latitudes,is induced by the collisions of energetic particles from solar wind and magnetosphere with the ambient neutral particles when they precipitate in the polar atmosphere.To unveil its mystery,abundant observations from the most primitive ones using the naked eye to the current spectroscopic measurements are being carried out.As auroral spectra reveal the decisive characteristics of precipitating particles,the observations by spectroscopic instruments play a pivotal role in complementing our knowledge about how the earth responds to solar activities.Generated by sophisticate spectrographs,aurora spectral data are analogous to ultraspectral data that contain the instant high-resolution spectral information besides spatial information about aurorae.However,though having a great value of research,they are not used as widely as other spectral data.This thesis concentrates on putting forward an effective application of these data,as well as dealing with the storage and transmission challenge for their utilization.The main novelties and achievements are as follows:1.Spectrographs configured at the Chinese polar stations like Antarctic Zhongshan Station and Arctic Yellowriver Station have been generating the aurora spectral data for over twenty years to acquire variable auroral spectra.It is of significance to study the generated data for their rich information.However,they lack sufficient utilization because the relevant investigation poses a challenge that requires the researchers to have knowledge of multiple disciplines such as physics,spectroscopy,and computer science.Take these together,models and algorithms proposed in this thesis are pioneer.2.Structural and statistical details of aurora spectral data are discussed to provide a comprehensive understanding of these data that seem new to many researchers.Deviating the acquired optical emissions during data acquisition,most noises that follow the Gaussian distribution can be removed with the aid of mean filters.The correlation is originally higher in the spatial domain than in the spectral domain,but for the spectral correlation,its variation with time is more significant.3.The representative auroras include the optical emissions of 557.7 nm and 630.0 nm that result from the collisions between secondary electrons and oxygen atoms,and appear yellow-green and red at about 150 km and 250 km,respectively.These emissions are treated as the key features for the following derivation because their intensities are directly determined by the precipitating electrons,and in turn,their characteristics can be derived in terms of the aurora spectral data.Three strategies are applicable for this derivation,of which the most computational efficient one is used in practice.Compared to the average calibrated energies from satellite observations,the derived characteristic energies of precipitating electrons are proved to be reasonable.4.Massive auroral spectral data and relevant data obtained from simultaneous observations of satellite are used for deriving the precipitating electron characteristics.Researchers are concerned about the local storage of these data and try to realize their realtime transmission under the available network bandwidth.In this consideration,a predictionbased lossless compression framework that combines hybrid spatial-spectral decorrelation with outlier recognition is developed for the pure auroral spectral data.Besides,with respect to the mixture of the auroral spectral data and relevant satellite data,a generic compression framework facilitates their compact storage and efficient management by drawing help from hierarchical clustering.