Assessment of the Performance of Ionospheric Models with NavIC Observations during Geomagnetic Storms
UURSI RCRS 2020, IIT (BHU), Varanasi, India, 12 - 14 February, 2020
Assessment of the Performance of Ionospheric Models with NavIC Observations duringGeomagnetic Storms
Sumanjit Chakraborty* ( ) and Abhirup Datta ( ) (1) Discipline of Astronomy, Astrophysics and Space Engineering, Indian Institute of Technology Indore. Simrol Campus.Indore- 453552, Madhya Pradesh, India Abstract
The paper presents an assessment of the performances ofthe global empirical models: International Reference Iono-sphere (IRI)-2016 and the NeQuick2 model derived iono-spheric Total Electron Content (TEC) with respect to theNavigation with Indian Constellation (NavIC)/ Indian Re-gional Navigation Satellite System(IRNSS) estimated TECunder geomagnetic storm conditions. The present study iscarried out over Indore (Geographic: 22.52 ◦ N 75.92 ◦ E andMagnetic Dip: 32.23 ◦ N, located close to the northern crestof the Equatorial Ionization Anomaly (EIA) region of theIndian sector). Analysis has been performed for an intensestorm (September 6-10, 2017), a moderate storm (Septem-ber 26-30, 2017) and a mild storm (January 17-21, 2018)that fall in the declining phase of the present solar cycle.It is observed that both IRI-2016 and NeQuick2 derivedTEC are underestimates when compared with the observedTEC from NavIC and therefore fail to predict storm timechanges in TEC over this region and requires real data in-clusion from NavIC for better prediction over the variableIndian longitude sector.
The low latitude ionosphere consists of several features,such as the equatorial ionization anomaly (EIA), equatorialelectrojet, equatorial plasma fountain, spread F and plasmabubbles, as a result of the horizontal orientation of the ge-omagnetic field at the geomagnetic equator [2]. It is ex-pected that as the sun shines over the geographic equator,the ion and electron density should be maximum aroundthat region and will go on decreasing towards the poles.Measured values show that this density has peculiar crestsaround ± ◦ magnetic latitude and trough around the mag-netic equator [1]. The ionosphere over central India fallsunder this anomaly region where sharp latitudinal gradientin the ionization is observed. The latitudes which fall inthe EIA has the highest concentration of electron densityand is nearly about 70% of the global density distribution.The ionospheric total electron content (TEC) is a vital pa-rameter of the ionosphere and is defined as total numberof electrons integrated between two points, along a tube ofunit cross sectional area and is expressed in TECU, where1 TECU = 10 electrons/m . It gets enhanced or depleted as a result of positive or negative geomagnetic storms.Geomagnetic storms are temporary disturbances of themagnetosphere of earth. They are caused by the solarwind shock wave which interacts with the geomagneticfield. The increase in the solar wind pressure initially com-presses the magnetosphere [5]. Whenever there are periodsof such magnetic disturbance, the horizontal component ofthe Earth’s magnetic field (H) gets depressed. The recov-ery to its average value is gradual. Earlier studies showthat at the mid-latitudes and the latitudes at the equatorialregion, the decrease in H can be represented by a uniformmagnetic field parallel to the geomagnetic dipole axis andthat it is directed southward. The magnitude of this dis-turbance field which is axially symmetric in nature, varieswith the storm-time or the time measured from the on-set of the storm. This onset can be understood to be asa sudden increase in the value of H globally, this is wellknown in literature as the storm sudden commencement(SSC). Following this SSC, H remains above its averagelevel for a few hours, this is known as the initial phaseof the storm. It is followed by a very large decrease in Hwhich is globally observed and it indicates the main phaseof the storm. The magnitude of this decrease in H indicateshow severe the storm is and the variation changes fromstorm to storm. The disturbance field which is representedby Disturbance storm time (Dst) index, is symmetric axi-ally with respect to the dipole axis (http://wdc.kugi.kyoto-u.ac.jp/dstdir/dst2/onDstindex.html). Severity of geomag-netic storms can be classified [8] as: Dst > -50 nT signi-fying a mild storm; -50 nT ≤ Dst < -100 nT signifying amoderate storm and -100 nT ≤ Dst < -200 nT signifyingan intense storm. As a result it is essential to study modelperformances during disturbed ionospheric conditions dueto the geomagnetic storms in order to verify the predictionpotentials of such models.The ionosphere over Indore (22.52 ◦ N and 75.92 ◦ E geo-graphic; magnetic dip: 32.23 ◦ N) falls near the anomalycrest in the Indian longitude sector. In this paper, for thefirst time to the best of our knowledge, results of the iono-spheric model derived TEC from the IRI and the NeQuickmodels have been compared with the NavIC estimated TECover Indore, under intense, moderate and mild geomagneticstorms during 2017 and 2018, falling in the declining phase a r X i v : . [ phy s i c s . s p ace - ph ] M a y f the present solar cycle. The International Reference Ionosphere (IRI) is an empir-ical model of the ionosphere. The sources of data to thismodel are the incoherent scatter radars and the dense world-wide network of ionosondes along with the Alouette top-side sounders in-situ instruments on board satellites. Themodel output provides the electron temperature and den-sity, ion temperature and composition and the TEC from 50km to 2000 km altitude range [13].NeQuick2 model is an upgraded version of the NeQuickmodel. This model uses, a modified DGR profile for-mulation [4] that consists five semi-Epstein layers [12]with modelled thickness parameters [11], for describingthe ionospheric electron density from 90 km to peak of F2layer. The model topside is represented by a semi-Epsteinlayer with a height-dependent thickness parameter that isdetermined empirically [6,3]. The inputs to this modelare the position(latitude, longitude) and either solar flux orsunspot number. Specific routines are present in NeQuickto evaluate electron density and the corresponding TEC bythe method of numerical integration [10].Navigation with Indian Constellation (NavIC), a regionalsatellite navigation system developed by ISRO, has a spacesegment which consists of Geostationary Earth Orbit(GEO)and Geosynchronous Orbit(GSO) satellites. The primarytarget of developing the NavIC is to provide information onpositional accuracy not only to the Indian users but also toregions of 1500 km from its boundary, designated as its pri-mary service area. It also has provisions for an extendedservice area that lies between the primary service area andarea enclosed by the rectangular grid having latitudinal ex-tent of 30 ◦ S to 50 ◦ N and longitudinal extent of 30 ◦ E to130 ◦ E. Three of the satellites are GEO while the remain-ing are GSO. The sub-satellite positions of the satellites aresuch that all of them have continuous radio visibility withthe Indian control stations. The GSO have an orbital incli-nation of 29 ◦ . These satellites broadcast signals in 24 MHzbandwidth of spectrum in the L5 and S band having carrierfrequencies 1176.45 and 2492.03 MHz respectively [9]. A NavIC receiver, provided by the Space Applications Cen-tre (SAC), ISRO, capable of receiving NavIC L5 and S1signals along with GPS L1 signal, is operational in the Dis-cipline of Astronomy, Astrophysics and Space Engineering,Indian Institute of Technology, Indore. An elevation anglehigher than 20 ◦ has been chosen for the NavIC values in or-der to avoid multipath error. The receiver provides the iono-delay at its output with a 1 Hz resolution. This iono-delayis converted to the slant TEC (STEC) [7] by the formula: ν = . f . T EC (1) where v is the iono-delay, f is the operational frequency ofthe signal emitted by satellites in Hz. This sTEC is con-verted to the equivalent vertical TEC (VTEC) [9] by themapping factor: M ( E ) = (cid:34)(cid:20) − (cid:104) R e . cos ( E ) R e + h I (cid:105) (cid:21)(cid:35) − / (2)where R e is the radius of the Earth (6371 km), h I denotesthe altitude of the thin shell model of the ionosphere (350km) and ( E ) is the elevation angle of the space vehicle. In this section, the diurnal variations of VTEC from the IRIand NeQuick models have been compared with the NavICestimated VTEC and the deviations on the disturbed daysfrom the selected periods of the geomagnetic storms arepresented.Figure 1 shows the variation of Dst(nT) as a function of UT(h) for the 3 storms selected based on their severity. Theintense storm period (September 6-10, 2017) is depicted byred, the moderate storm (September 26-30, 2017) by greenand the mild storm (January 17-21, 2018) by blue. The mid-dle day (48-72 UT) out of 5 days for all the three storms hadbeen the disturbed day with Dst reaching the minimum. Ta-ble 1 shows the minimum Dst (nT) values for the selectedstorms along with the corresponding time at which mini-mum Dst values were observed.
Table 1.
Dst values for the selected storms
Date
Time (UT) Minimum Dst (nT)
September 08, 2017 02:00 -124September 28, 2017 07:00 -55January 19, 2018 09:00 -27The diurnal variations IRI-2016 and NeQuick2 model de-rived VTEC are compared with the NavIC estimated VTECduring the storm period of September 6-10, 2017 is de-picted in Figure 2. It is observed that although NavIC showsenhancements in VTEC on September 7 and 8, there is novariation at all in IRI while the values go on decreasingfrom September 6 onward in the NeQuick.Similarly, Figures 3 and 4 show diurnal VTEC variation forthe moderate and the mild storms of September 28, 2017and January 19, 2018 respectively. In Figure 3 NavIC esti-mated values show enhancements on September 28,29 and30 which is not at all captured by both the models whilein Figure 4, higher VTEC values observed from NavIC onJanuary 17, 2018 is not captured by these models. Thissuggests the poor prediction capability of the two modelsduring moderate to quiet conditions of the ionosphere. igure 1.
Variation of Dst(nT) with UT(h) for the in-tense storm of September 8, 2017(red), moderate storm ofSeptember 28, 2017(green) and mild storm of January 19,2018(blue). The Dst dropping below -100 nT on September8 in the top panel signifies intense, while Dst values in themiddle and bottom panel on September 28 and January 19signify moderate and mild storm respectively.
Figure 2.
Diurnal variation of NavIC estimated VTECalong with IRI and NeQuick derived VTEC during Septem-ber 6-10, 2017Finally, on the days when Dst dropped to a minimum fromall the 3 storms, the diurnal maximum VTEC obtained fromthe two models and the observed VTEC from NavIC aresummarised in Table 2. Table 3 shows the deviations ofmodel derived values with the observed ones under varying
Figure 3.
Diurnal variation of NavIC estimated VTECalong with IRI and NeQuick derived VTEC during Septem-ber 26-30, 2017
Figure 4.
Diurnal variation of NavIC estimated VTECalong with IRI and NeQuick derived VTEC during January17-21, 2018storm conditions. For the intense storm, IRI values showhigher deviation from the observed VTEC while NeQuickpresents greater deviations during the moderate and themild storms. It can be observed that for all the three stormsNeQuick and IRI derived VTEC underestimates the ob-served NavIC VTEC and the deviation from NavIC ob-served values is highest during the intense storm.The studypoints out to the fact that during disturbed ionospheric con-ditions, both IRI-2016 and NeQuick2 derived TEC are notery reliable and require modifications for a more realisticpredictions in and around the anomaly region. Inclusion ofreal time data from the NavIC receivers could help in betterprediction over the Indian longitude sector.
Table 2.
Disturbed Day Peak TEC from NeQuick2, IRI-2016 and NavIC
Day
NeQuick-TEC IRI-TEC NavIC-TEC
Sep 08, 2017 55.80 43.60 77.43Sep 28, 2017 45.72 47.80 79.13Jan 19, 2018 23.99 28.20 35.63
Table 3.
Deviation between Observed and Model derivedTEC
Day
NeQuick-TEC IRI-TEC
Sep 08, 2017 21.63 33.83Sep 28, 2017 33.41 31.33Jan 19, 2018 11.63 7.43
The paper, for the first time to the best of our knowledge,presents an analysis of the performances of the empiricalionospheric models IRI 2016 and NeQuick during the ge-omagnetic storms of varying severity over Indore, locatednear to the ionization anomaly crest. It is observed that forall the three storms, the models are unable to capture theenhancement or depletion of ionization caused as a resultof the geomagnetic storms. The deviations are of highermagnitude during the intense and moderate storm while it iscomparatively lower for the mild storm. These observationssignify that the prediction capabilities of these models un-der storm conditions are not reliable and therefore requiremodifications and inclusion of data from NavIC, which isconceived for accurate analysis of the ionosphere, to the ex-isting database in order deliver a more realistic values andlesser deviations from observations of the ionosphere nearan anomaly region like Indore.
SC acknowledges Space Applications Centre (SAC), ISROfor providing fellowship to pursue his research. The au-thors acknowledge SAC, ISRO for providing the NavICdata (ACCORD receiver) under the project number: NGP-17 to the Discipline of Astronomy, Astrophysics and SpaceEngineering, IIT Indore. Further acknowledgements go toMs. Deepthi Ayyagari and Dr. Saurabh Das for helpfuldiscussions.