Ground-and Space-based GNSS Ionosphere Inversion Technique And Its Application In Space Weather Monitoring

The ionosphere is a kind of dispersion medium,and can produce a variety of ionospheric effects,such as reflection,scattering,absorption and refraction,when the radio waves passing through it.As a result,it represents a natural hazard that is known to interrupt and damage technologies critical to modern society such as electric power grids,airlines,trains,pipelines,and global navigation satellite system(GNSS).Therefore,it is of great importance to probe the ionosphere.Ionosphere is an integral element of space weather,closely linked to the intensity and the energetic spectrum of electromagnetic and corpuscular radiation emitted from the Sun.The utilization of ground-and space-based GNSS signals is attractive for monitoring space-weather-driven effects in the ionosphere in a twofold manner.Firstly,space-weatherinitiated processes can effectively be investigated by probing the ionosphere with high temporal and spatial resolution.Secondly,monitoring results help to develop correction models,mitigation techniques and ionospheric threat models to further reduce the ionospheric impact in numerous GNSS applications.Based on the above fact,this paper aims at the accurate extraction of the GNSS ionospheric observations,the characteristics of the topside ionosphere,the optimization of the empirical international reference ionosphere(IRI)model,the multi-scale ionosphere structure monitoring during a magnetic storm,and the ionospheric shortterm prediction.The main contributions are as follows:(1)In recent years,low earth orbit(LEO)onboard GNSS receiver has gradually become a new means for ionosphere detection,and has played an important role in space weather monitoring.Based on the spherical symmetry assumption,we estimated topside ionosphere TEC,Jason-2 receiver differential code bias(DCB)and GPS satellite DCB parameters through carrier-to-code leveling(CCL)method.Results show that the STD of GPS DCB is better than 0.15 ns,and agrees well with the monthly DCB products of CODE center.Jason-2 receiver DCB distribution is stable,and its monthly mean value and std are-2.65 ns and 0.34 ns,respectively,but its stability is slightly weaker than GPS DCB.The topside ionosphere TEC above the Jason-2 orbit height varies from 0 to 6 TECU,which is in good agreement with the general plasmaspheric distribution investigated by other measurements and methods.The maximum v TEC appears near the low latitudes of 15:00 LT and 20:00 LT,which indicates that the maximum value of VTEC is the representation of the equatorial ionization anomaly(EIA)structure of the topside ionosphere.(2)The plasmasphere,which is located above the ionosphere,is a significant component of Earth’s atmosphere,and the plasmasphere electron content(PEC)distribution is determined by different physical mechanisms to those of the ionosphere electron content(IEC).However,the observation for the PEC is very limited.In this study,we introduced a methodology(called zero assumption method,which is based on the assumption that PEC can reach zero)to extract the PEC over TOPEX/JASON(T/J)and global navigation satellite system(GNSS)overlapping areas.Results show that the daily systematic bias(T/J vertical TEC >GNSS-derived vertical TEC)for both low(2009)and high(2011)solar activity condition is consistent,and the systematic bias for JASON2 and JASON1 is different.We suggest that systematic biases predominantly arise from the sea state bias(SSB)and instrumental bias.After removing the systematic bias,we extracted reliable PEC inferred from differences between GNSS-derived vertical TEC and T/J vertical TEC data.Finally,the characteristics of the plasmaspheric component distribution for different local times,latitudes,and seasons were investigated.(3)To adapt the IRI model from ionospheric climatological model to near realtime weather predictions,TEC data from global ionosphere maps(GIM)are ingested into the IRI-2016 model through retrieving the optimal ionospheric index(IG)over Europe on an hourly basis.When the retrieved hourly effective IG indices are used to drive the IRI-2016 model,the resulting ionospheric parameters are externally evaluated with respect to multiple sources,including the COSMIC/ionosonde electron density(Ne)profiles,ionosonde F2 layer critical frequency(fo F2),and individual GNSS-derived TEC for both quiet and storm conditions.Results show that:(1)The updated IG indices for different latitudinal zones tend to follow a similar trend under quiet conditions,but vary much more significantly during storm days.(2)The retrieved Ne profiles from the updated IRI-2016 agree better with those from the COSMIC Ne profiles,especially for the F2 layer maximum electron density(Nm F2)values.Furthermore,the updated IRI-2016 Ne profiles show improved agreement with ionosonde measurements under quiet conditions,particularly for the bottom-side Ne profiles and Nm F2 as well as for the storm-time Ne profiles.(3))Comparing the IRI-updated TEC with the GNSS-derived TEC,IRI-updated TEC improved approximately 19% for both quiet and storm days,and the nighttime TEC improvement is better than that during daytime.When compared to the ionosonde fo F2 measurements,the daytime IRI-updated fo F2 improvement during quiet time is better than that during storm condition,while the performance for nighttime fo F2 drops during quiet time.Discussions about possible reasons for the nighttime fo F2 degradation are included.(4)Ionospheric TEC from GIM is widely applied to scientific research about space weather impacts,so there is great interest in the community in short-term ionosphere forecasting.In this study,the long short-term memory(LSTM)neural network(NN)is applied to forecast the 256 spherical harmonic(SH)coefficients used to construct GIM based on multiple input data,including historical time series of the SH coefficients,solar extreme ultraviolet(EUV)flux,disturbance storm time(Dst)index,and hour of the day.The above datasets from Jan 1,2015 to May26,2016 are used to train the LSTM model,and those from May 26 to Dec 31,2016 are selected as the Testing set to evaluate the model performance.Comparing to those models with no external solar radiation or geomagnetic activity drivers concatenated into the LSTM layer,our developed LSTM model is able to improve the forecast of the SH coefficients by including the solar EUV flux and Dst index.After using the developed LSTM model,the global hourly TEC maps are reproduced from 256 predicted SH coefficients by using the SH function,and a comprehensive evaluation is carried out with respect to the CODE GIM TEC.Results show that the global residuals for both geomagnetic quiet and storm days are very small,mostly within 1 TECU.Moreover,typical ionospheric structures,such as EIA and storm-enhanced density(SED),are well reproduced from the predicted TEC maps during storm time.Besides,a long-term evaluation shows that the RMSE/ME errors of the hourly predicted TEC are mostly within 1.5/±0.5and 2.5/±1 TECU for quiet and storm times,respectively.Therefore,our developed LSTM model performs well during both quiet and storm times.It is also important to note that the predicted TEC for the 1st hour consistently outperforms the 2nd hour results,though the degradation with time is very minor.(5)We investigate multi-scale ionospheric responses to the May 27,2017 geomagnetic storm over the Asian sector by using multi-instrumental observations,including ground-based GNSS network,Constellation Observing System for Meteorology,Ionosphere and Climate(COSMIC)Radio Occultation(RO),the Feng Yun-3C(FY-3C)GNSS Occultation Sounder(GNOS)electron density profiles(EDP),and in situ plasma density observations provided by both Swarm and Defense Meteorological Satellite Program(DMSP)missions.This geomagnetic storm was an intense storm with the minimum symmetric horizontal(SYM-H)component reaching-150 n T and was caused by a coronal mass ejection(CME)released on May 23.The main observations are summarized below:(1)Two ionospheric positive storm periods were observed.The first one was observed in the noon-afternoon sector during the main phase of the storm on May 28,with nearly 120% TEC enhancement.The second one was of a smaller scale and occurred on the night side during the recovery phase of the storm on May 29.The first dayside positive storm was initiated by the Interplanetary Magnetic Field(IMF)Bz southward turning and eastward penetration electric field,while the second nightside one was terminated by a later southward turning of the IMF Bz since the Asian sector was on the nightside and the penetration electric field changed westward.(2)A negative storm occurred from 00:00-12:00 UT on May30 over the Asian sector,nearly two days after the main phase,which was due to the thermospheric composition change,i.e.,decrease of the O/N2 ratio,as shown in the TIMED/GUVI measurements.(3)A band-like TEC enhancement was observed aligning in the northwest-southeast direction and propagated slowly southwestward from 15:00-20:00 UT(23:00-04:00 LT,near midnight)on May 28 during the recovery phase of the storm.In situ density observations from the Swarm B and DMSP F15&16 satellites confirmed the density enhancement at 460 km and 850 km,respectively,and revealed that this band-like TEC enhancement structure resembles the so-called plasma blob.The similarities of the observed plasma blob characteristics in terms of spatial structure,propagation trend and temporal evolution with the nighttime traveling ionospheric disturbance(TID)are consistent with the TID-blob theory.(6)We investigate an equatorial plasma irregularities(EPI)event in Japan near the early morning sector(04:30-7:30 LT)during the recovery phase of a geomagnetic storm occurred on Memorial weekend(May 28)2017 by using multi-instrumental measurements,including ground-based GNSS network,ionosonde stations,and space-based Swarm and DMSP missions.A stream-like EPI structure appeared in the northeast-southwest direction and extended to northwestward during 19:50-20:30 UT(04:50-05:30 LT)on May 28.Interestingly,this irregularity structure evolved into bifurcated branches at 20:40 UT(05:40 LT)on May 28,and continued to strengthen and then drifted to the northwest direction until subsided during 20:40-21:10 UT(05:40-06:10 LT).This unusual EPI extended northward in a maximum latitude of ～37°N(magnetic latitude: 30.3°N),which corresponds to an apex height of 2171 km over the magnetic equator.The dawn EPI signature was also observed at the topside ionosphere by in situ electron density(Ne)measurements and ROTI results by space-born Swarm A,C and DMSP F15 satellites,indicating that the equatorial irregularities arose to higher altitudes.Furthermore,the uplift of the bottom F layer was confirmed by the ionosonde virtual height(h’F)variations.We suggest that both the eastward disturbance dynamo electric field(DDEF)and over-shielding electric field(associated with the northward turning of IMF Bz)near the early morning sector created a favorable condition for the growth of the Rayleigh-Taylor(R-T)instability via upward E×B drifts,and thus result in the generation of the dawn EPI.