Lossless compression codec of aurora spectral data using hybrid spatial-spectral decorrelation with outlier recognition

摘要

Aurora spectral data, the particular hyperspectral data of auroral spectra, have an irreplaceable research value in bridging the gap between solar activity and terrestrial evolution. Their requirement for high volume storage and real-time transmission had been a great challenge until we proposed a CPU paralleled and online-biprediction-based method. As it is no longer applicable because of the recent spectrograph reassembling, this paper presents a replacement strategy combines the unidirectional predictor with the entropy coder of the other dimension, as well as with using smoothing and outlier recognition. The hybrid encoders of Spat-SPCC and Spec-SPCC are developed distinguished by their respective prediction direction. Spat-SPCC tuned on one-day trial data is used for its better capability for compression, in the further comparison with various classical algorithms, it achieves the top-ranking compression performance and has the average processing time of 1 s per file so that its availability for the practical applications is validated. (C) 2019 Published by Elsevier Inc.