Applications Of Deep Learning In The Study On The Features Of Ionospheric F-Layer Irregularities In Low-Latitude Region

Ionospheric F-layer irregularities often occur in low-latitude regions and have a higher occurrence rate during high solar activities years.Ionospheric scintillation caused by ionospheric F-layer irregularities in low latitudes can have a significant negative impact on the quality of satellite signals propagating through the ionosphere,and in severe cases can lead to signal loss of lock.In order to reduce the negative impact of ionospheric scintillation in low-latitude regions on satellite navigation and positioning systems and communication systems,it is necessary to study the characteristics of ionospheric F-layer irregularities in low latitudes.The zonal drift and zonal width of ionospheric F-layer irregularities are two important characteristics of ionospheric F-layer irregularities in low latitudes.The carrier-to-noise density ratio（C/N₀）observations with a 1-s sampling interval of B1I signal of Beidou GEO（Geostationary Earth Orbit）satellites and TEC（Total Electron Content）sequence calculated by using carrier phase measurements of B1I/B2I signals of Beidou GEO satellites from April 2014 to April 2015（i.e.,during a period of high solar activities）,both of which were provided by HKCL station and HKQT station in Hong Kong located at region with a high occurrence rate of ionospheric scintillation,were used to study the zonal drift characteristic and zonal width characteristic of ionospheric F-layer irregularities in low-latitude region of South China.The main work and research results of this paper are summarized as follows.Firstly,in order to quickly and accurately find out all the observation events of appearing ionospheric F-layer irregularities from Beidou GEO satellites observations provided by HKCL station and HKQT station in Hong Kong between April 2014 and April2015,long short-term memory（LSTM）network of deep learning was applied to recognize the observation events of low-latitude ionospheric F-layer irregularities,and an LSTM-based low-latitude ionospheric F-layer irregularities recognition software was developed in this paper.This software consists of two models,namely the amplitude scintillation recognition model and the TEC depletion recognition model.The function of the amplitude scintillation recognition model is to automatically recognize whether amplitude scintillation occurs and give the start and end moment information of amplitude scintillation according to the B1I signal C/N₀ sequence of Beidou GEO satellites.The function of the TEC depletion recognition model is to automatically recognize whether TEC depletion occurs and give the start and end moment information of TEC depletion according to the TEC sequence of Beidou GEO satellites.In this paper,through multiple test experiments on the architecture of the above two models,after comprehensively considering the recognition accuracy and recognition time-consuming,it was determined that the two models use the same architecture,that is,both are composed of Sequence Input Layer,LSTM Layer with 120 hidden units,Fully Connected Layer,Softmax Layer,Classification Output Layer.In each model,the C/N₀ sequence（or TEC sequence）is first input into the LSTM Layer with 120 hidden units through the Sequence Input Layer,and the long-term dependencies between the time steps of the sequence data are extracted to obtain high-level features.Then,these advanced features are linearly combined through the Fully Connected Layer and sent to the Softmax Layer to obtain a probability distribution model,and then the output of the probability distribution model is sent to the Classification Output Layer to infer the specific category of the sequence to complete amplitude scintillation recognition of the C/N₀ sequence（or depletion recognition of the TEC sequence）.The test results of the amplitude scintillation recognition model show that the recognition accuracy of the model reached 98.997%,and the time required to recognize one day’s observations was about 0.56 seconds.The test results of the TEC depletion recognition model show that the recognition accuracy of the model reached 99.259%,and the time required to recognize one day’s observations was about 0.61 seconds.The above test results show that the LSTM-based low-latitude ionospheric F-layer irregularities recognition software developed in this paper is practical and can be used to analyze long-term observations.Secondly,by using LSTM-based low-latitude ionospheric F-layer irregularities recognition software developed in this paper,all observation events of appearing ionospheric F-layer irregularities were found out from the C/N₀ observations with a 1-s sampling interval of BI1 signal and TEC sequences of Beidou GEO satellite C03 between April 2014 and April 2015.These observations were provided by HKCL station and HKQT station in Hong Kong.The ionospheric F-layer irregularities corresponding to all these observation events were plasma bubble structures.Then,based on these observation events of appearing ionospheric F-layer irregularities,according to the method of dividing the distance between ionospheric piercing point（IPP）IPP_CL and IPP_QT by the time difference of amplitude scintillation at IPP_CL and IPP_QT caused by the same ionospheric F-layer irregularities,the zonal drift feature of ionospheric F-layer irregularities in low-latitude region of South China during a period of high solar activities was studied.The research results of this paper show that（1）from April 2014 to April 2015,the ionospheric F-layer irregularities in low-latitude region of South China appeared from 11:00 to 20:00UT（i.e.,19:00 local time to 04:00 the next morning）.（2）The ionospheric F-layer irregularities in low-latitude region of South China occurred more frequently in the vernal equinox months（i.e.,March and April）and autumnal equinox months（i.e.,September and October）,and in the summer solstice months（i.e.,June and July）and winter solstice months（i.e.,December and January）rarely occurred.（3）The ionospheric F-layer irregularities in low-latitude region of South China always occurred zonal drift,all of them were eastward drift,and the eastward drift velocity varied from 20 m/s to 290 m/s,in most cases（the occurrence rate was 86.62%,and the total number of samples was 299）the eastward drift velocity was between 50 m/s and 140 m/s.Thirdly,by using the LSTM-based low-latitude ionospheric F-layer irregularities recognition software developed in this paper,all observation events of appearing ionospheric F-layer irregularities were found out from the C/N₀ observations with a 1-s sampling interval of BI1 signal and TEC sequences of Beidou GEO satellites（i.e.,C01–C04 satellites）between April 2014 and April 2015.These observations were provided by HKCL station and HKQT station in Hong Kong.The ionospheric F-layer irregularities corresponding to all these observation events were plasma bubble structures.Then,based on these observation events of appearing ionospheric F-layer irregularities,according to the method of studying the zonal width of ionospheric F-layer irregularities（i.e.,if the overlapping time of amplitude scintillation of adjacent IPPs in the observation area was more than 5 minutes,it indicates that these IPPs were covered by the same F-layer irregularities,the zonal width range of F-layer irregularities can be estimated by the distance between these IPPs）,the zonal width（i.e.,ZW）feature of ionospheric F-layer irregularities in low-latitude region of South China during a period of high solar activities was studied.The research results of this paper show that from April 2014 to April 2015,the occurrence rate of events in which ZW（27）27 km,27?ZW（27）284 km,284?ZW（27）568 km,568?ZW（27）852 km,ZW?852 km of the ionospheric F-layer irregularities in low-latitude region of South China was 2.83%,29.25%,38.68%,13.21%,16.04%,respectively（note that the total number of samples was 106）,that is,the zonal width of the ionospheric F-layer irregularities in low-latitude region of South China was mainly（the occurrence rate was 81.13%）ranging from 27 km to 852 km.The research results of the zonal drift velocity and zonal width of the ionospheric F-layer irregularities in low-latitude region of South China obtained in this paper are basically consistent with the previous research results.This indicates that the new methods used in this paper to study the zonal drift and zonal width characteristics of ionospheric F-layer irregularities in low-latitude region are feasible.