3-3 Ionospheric Storm and Variation of Total Electron Content€¦ · 352 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009 nant

1 Introduction

Technologies based on radio waves makeour lives more convenient and enrich variousaspects of our social system. One major char-acteristic of radio waves is their ability topropagate over long distances. Since HF radiowaves reach as far as the opposite side of theglobe through repeated reflection between theionosphere and ground surface, HF communi-cation and broadcasting used to be the keymeans of distributing and exchanging infor-mation overseas. As the application of satel-lites has become a permanent part of society,radio waves at higher frequencies that propa-gate through the ionosphere have opened up adiverse range of applications. Yet, the ionos-

phere is not a perfect transparent medium forsatellite radio waves. In particular, radiowaves in the UHF band, such as those trans-mitted from GPS satellites, and at lower fre-quencies are significantly affected by propaga-tion delays, fluctuations in signal strength, andother effects. The magnitude of these effectslargely depends on the state of a continuallyvarying ionosphere.

Variations in ionospheric electron densitycan be roughly divided into two kinds: regularand sudden. Regular variations have distinctcharacteristics of periodicity, such as solaractivity having a cycle of approximately 11years, seasonal variations associated with theearth’s revolution around the sun, and dailyvariations, which are therefore somewhat easy

349MARUYAMA Takashi et al.

3-3 Ionospheric Storm and Variation of TotalElectron Content

3-3-1 Observations of TEC Disturbances withGEONET —TEC Storm and SED—

MARUYAMA Takashi, MA Guanyi, and NAKAMURA Maho

Two extreme ionospheric events were analyzed as phenomena in which ionospheric totalelectron content (TEC) was significantly enhanced. The first event occurred on November 6,2001, associated with a magnetic storm. TEC derived from a nationwide GPS receiver network(GEONET) reached 200 TEC units at a lower latitude (27° N) and 100 TEC units at a higher lati-tude (45° N). These values are double the quiet-day averages. This enhanced TEC had beenprimarily caused by the prompt penetration of a magnetospheric electric field into lower latitudesassociated with a geomagnetic disturbance. The second event was storm enhanced density(SED) that occurred on November 8, 2004. TEC became enhanced at higher latitudes after sun-set and it reached 100 TEC units, higher than the daytime maximum on a quiet day at the samelatitude and 20 times higher than at the same local time on a quiet day. In addition to a penetrat-ing magnetospheric electric field, a disturbance dynamo electric field may have played a majorrole in the formation of SED moving in the WNW direction.

KeywordsIonospheric total electron content, TEC storm, SED, Magnetospheric disturbances,Disturbance dynamo

350 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

to model (as reported in References[1][2]). Incontrast, rapid changes in solar activity suchas solar flares could induce a series of distur-bances in the terrestrial upper atmosphere dri-ven by a magnetic storm, resulting in suddenvariations in the ionosphere[3］–［7]. This phe-nomenon is called an “ionospheric storm.”Ionospheric storms in which the electron den-sity diminishes are called “negative storms,”and those in which the electron densityenhances are called “positive storms.” Theway that ionospheric storms behave vary sig-nificantly; for example, a positive storm mightturn into a negative storm as time progresses;either a positive storm or a negative stormmight develop at one time; or a positive stormand a negative storm might develop simulta-neously at different latitude. Generally, ionos-pheric storms are electron density disturbanceson a global scale, but rarely is a sharp rise inelectron density observed in a narrow regionassociated with magnetic storms. This phe-nomenon is called SED (storm enhanced den-sity), whereby the electron density leaps sev-eral times or even 20 times higher than normalat around sunset. SEDs are closely related todaytime positive disturbances. Becausedecreases in critical frequency fOF2 in negativestorms reduce the maximum usable frequencyfor HF communication, active research usingionosonde has long been conducted. Today,positive storms take on growing importance inconnection with radio propagation delays inthe ionosphere, such as those transmitted fromGPS satellites. Ionospheric delays of satelliteradio waves are proportional to the total elec-tron content (TEC), or the integrated value ofelectron density along the propagation pathfrom the satellite to the receiver. Figure 1shows the relation between the TEC and prop-agation delays in the case of GPS L1 and L2

signals, where 1 TEC unit refers to 1×1016

electrons/m2. Regular variations or sudden dis-turbances might cause TEC to vary from sev-eral TEC units up to 200 TEC units.

In recent years, TEC observed by GPSsatellites has been widely used for ionosphericstudies as well as fOF2 observed by ionoson-

des. One reason for this impetus is that GPSreceivers are easier to install than ionosondesand afford more observation stations.Although fOF2 and TEC behave similarly aseach being a measure of an ionospheric varia-tion, both are not identical measures. There-fore, we should use both measures in a com-plementary manner for detailed analysis.Another essential function of ionosondes istheir observation of ionospheric height. Thephoto-chemical process has a significant bear-ing on ionospheric variations, and largelydepends on height. Moreover, since ionos-pheric height variations directly reflect atmos-pheric and plasma dynamics, the ionosphericheight offers a key parameter in elucidatingthe mechanism of ionospheric storm develop-ment, along with fOF2 and TEC.

This paper deals with two extremely sig-nificant events involving positive ionosphericvariations having immense importance tospace weather, with reference to TEC mapsderived from a dual-frequency GPS receivernetwork (GEONET: GPS Earth ObservationNetwork) built by the Geospatial InformationAuthority of Japan (GSI), as well as fOF2 andhmF2 (maximum electron density height)derived from four ionosonde observatories inJapan. The first event is the ionospheric stormthat occurred on November 6, 2001. This

Fig.1 Relation between the ionosphericpropagation delays of GPS L1 andL2 signals and TEC


storm was quite exceptional in exhibiting arecord increase in TEC, but not a particularlylarge event in terms of fOF2. This event there-fore acquired the name “TEC storm.” The sec-ond event is SED observed after sunset onNovember 8, 2004. Ionospheric storms usuallydevelop in an intricate complexity of threefactors; one increasing electron density, thesecond decreasing it, and the third sustainingsuch effects. The next chapter briefly reviewsthe mechanism of ionospheric storms. Afterdescribing the dataset employed, the twoevents are discussed in detail.

2 Mechanism of IonosphericStorm

Negative storms are caused by changes inatmospheric composition driven by energyhaving been injected into polar regions duringa magnetic storm. A thermospheric atmos-phere having its composition changed propa-gates toward lower latitudes in the form oftraveling atmospheric disturbances (TADs)[8].Changes in the composition are represented byan increase in the [N2]/[O] ratio and inducingan increase in the recombination coefficient ofionospheric plasma, thus resulting indecreased electron density. In the initial stageof a TAD, an equatorward wind intensifies(equatorward surge) to push up the plasmaalong inclined geomagnetic field lines. If thisphenomenon occurs in the daytime, the elec-tron density increases since the electron lossrate is lower at higher altitudes. In this way, apositive storm may precede a negative storm,though this largely depends on the time periodduring which the equatorward surge occurred.Another cause of positive ionospheric distur-bances is the plasma having been uplifted(E×B drift) in the direction perpendicular tothe magnetic field (B) by an eastward electricfield (E). The electron loss rate reduces athigher altitudes as in the case of wind effect.When this happens during daylight hours, theelectron density increases. This eastward elec-tric field originates from broadly two sources.One is an intensified magnetospheric convec-

tion electric field penetrating into lower lati-tudes[9]; the other a disturbance dynamoinduced by changes in thermospheric generalcirculation driven by energy having beeninjected into a polar region[10][11].

In the case of actual ionospheric storms, anumber of complicated factors may interactwith one another to enhance disturbances oreven cancel one another[3]. Forbes et al.[12]and Forbes[13] attempted to separate multipledisturbance sources by using data collectedfrom a set of ionosondes arranged in themeridian. However, the principal sources ofdisturbance may generally change with thelapse of time, with the significance of eachdisturbance process also varying from event toevent, thus making it a laborious task to fullyidentify the sources of disturbances.

fOF2 is the most fundamental of all para-meters that characterize an ionospheric storm,but inadequate by itself to gain completeinsight into the complexities of disturbances.Even if the ionospheric F-layer is uplifted tohigher altitudes by an eastward electric fieldor equatorward thermospheric neutral wind, itmight cause fOF2 to increase or decreasedepending on the local time or the rate of ris-ing height. As explained by Rishbeth[14], theoutflow of plasma from the ionosphere to theplasmasphere could reduce NmF2 (NmF2 [m-3] =1.24×1010 (fOF2 [MHz])2 ) under certain cir-cumstances, but the degree of this effect variesgreatly depending on the neutral atmosphericwind and E×B drift. The fall in fOF2 causedby the rising maximum electron density height(hmF2) of the F-layer is a consequence of plas-ma being redistributed along the geomagneticfield line and changes in the ionosphericheight profile. The plasmasphere has a func-tion of drawing out ionospheric plasma in thedaytime and then replenishing the ionospherewith plasma at nighttime. The effect of risingfOF2 caused by rising hmF2 is associated with adecrease in the electron recombination rateand the progress of ionization in the bottom-side ionosphere. The increase or decrease infOF2 is ultimately determined depending onwhich of the two conflicting effects is domi-


nant. Likewise, fOF2 may rise or fall evenwhen the F-layer height has been lowered by adownward E×B drift driven by a westwardelectric field or a poleward thermosphericneutral wind. Rising fOF2 results from adeformed ionospheric height profile, whereassagging fOF2 stems from an increase in theelectron recombination rate. The rise in fOF2 istransient and soon overcome by the effect ofrecombination in the bottom-side ionosphere,with fOF2 beginning to decrease.

The method of observing TEC using radiowaves transmitted from GPS satellites hasrecently been widely adopted, thereby encour-aging research on ionospheric storms by usingTEC. Essentially, TEC behaves in a mannersimilar to fOF2 in the course of an ionosphericstorm, though not necessarily matched. WhenfOF2 is varied upon the redistribution of plasmaalong the geomagnetic field line, for example,the integral value of TEC does not show amajor change, because redistribution aloneneither recombines nor generates plasma. Onthe other hand, when fOF2 falls as a result of anincrease in the recombination rate, TEC willfollow it. Since TEC is an integrated parame-ter over the ionosphere to the plasmasphere,TEC is slower to respond to changes in theelectron loss rate in the lower ionospherewhen compared to fOF2, and TEC and fOF2

exhibit different temporal changes. Thus, ana-lyzing two observed values that differ inbehavior should greatly aid in interpretingionospheric storms.

3 Data

GEONET built by GSI consists of morethan 1,200 GPS receivers, allowing TECbetween a satellite and receivers to be estimat-ed using a signal transmitted on two frequen-cies. Among various methods of estimationproposed to date, we have employed themethod developed by Ma and Maruyama[15].According to this method, the coverage of thereceiver network (piercing points at whichsatellite radio waves pass an ionosphericheight) was divided into 32 cells (2×2° in lati-

tude/longitude), and with TEC being estimat-ed using the least square fitting method on theassumption of constant vertical TEC in eachof these cells. While this process of TEC esti-mation is based on the fact that radio wavespropagating at two different frequenciesundergo different propagation delays in theionosphere, it simultaneously solves inter-fre-quency biases (or differences in instrumentaldelay between two frequencies within theelectronic circuit) unique to satellites andreceivers. TEC for the 32 cells was evaluatedevery 15 min from a 24-hr data set, assumingthat the satellite and receiver biases remainunchanged during 24 hours. Based on theresultant TEC data, a time (LT) and latitudeTEC map was produced from the 24-hr TECdata (24×4×32) by the spherical harmonicfunctional fitting method. A map of differ-ences between quiet and disturbed days creat-ed from this TEC map visualizes ionosphericstorms as seen by TEC.

As shown in Fig. 2, ionosondes were inoperation at the four observatories located atWakkanai, Kokubunji, Yamagawa, and Oki-nawa, with each offering an ionogram every15 min. Table 1 summarizes the locations ofthese observatories. Besides fOF2, transmission

Fig.2 Distribution of GEONET receiversused to determine TEC


factor M(3000)F2 was scaled from the iono-grams. Using M(3000)F2, ionospheric heighthmF2 was calculated by the following relation-al expression[17] produced by correcting anequation based on work done by Shimaza-ki[16] to allow for delays in the bottom-sideionosphere:

Berkey and Stonehocker[18] confirmedthat this expression is sufficiently accuratewhen compared to the method of determiningactual heights from whole echo traces by itera-tive computation. M(3000)F2 as used here isone of the standard ionogram scaling parame-ters and can be established with relative ease,while the scaling whole echo traces involveshuge amounts of labor and time. Because fOEappearing in Equation (3) sometimes isunavailable under the influence of the spo-radic E-layer or interference, the empiricalequation developed by Muggleton[19] wasused instead. Parameters of the E layer can bewell defined in terms of the solar zenith angleand the solar activity index, and are less vul-nerable to the effects of ionospheric storms.

4 Space Weather Event in EarlyNovember 2001

4.1 Overview of disturbancesFigure 3 shows the interplanetary magnet-

ic field and geomagnetic variation parametersfor November 5 to 8, 2001. The top panel des-ignates the southward component of the inter-

planetary magnetic field as observed by theACE satellite at the Lagrange (L1) point.Here, the trace is shifted by one hour to allowfor the propagation time of solar wind fromthe L1 point to the earth. The two panels in themiddle indicate the magnetic disturbanceindices of ASYM-H and SYM-H[20]. TheASY-H index (an asymmetric disturbanceindex) is considered a good indicator of anauroral substorm. The SYM-H index (a sym-metric disturbance index) is essentially identi-cal to the Dst index, except that it has a hightime resolution and uses somewhat differentobservatories to derive the index. The bottompanel indicates a third magnetic disturbanceindex—the Kp index. A magnetic storm desig-nated as SC (storm sudden commencement)started at 01:51 UT on November 6, withasymmetric ring current growing immediately.The Kp index reached 9-maximum. During the7-hr period prior to the sudden commence-

Table 1 Location of Ionosondes

Fig.3 Magnetic disturbance index inearly November 2001


ment of the magnetic storm, IMF Bz wassouthward, with the ring current growinggradually from 19:00 UT on November 5. Inresponse to this, the Kp index increased from 3to 5.

Figure 4 shows variations in NmF2

observed at the four stations from Wakkanai toOkinawa for November 2 to 7, along with theKp index. Each dot denotes a value of NmF2

observed every 15 min; each solid line denotesan average value of NmF2 on quiet days. Theaverage from November 2 to 4 was used as aquiet day reference. NmF2 was found to havedisturbed at all the stations following the mag-netic disturbance on November 6. Distur-bances began with an increase in NmF2 at sun-rise, soon followed by a reduction therein. Atthe lower latitude station, the earlier a reduc-tion in NmF2 began by a larger amount. TheNmF2 disturbances turned positive at 12:00 JSTat the three lower-latitude stations, followedby Wakkanai turning positive at 13:30 JST.Thereafter, NmF2 remained higher than onquiet days until midnight at higher latitudes(Wakkanai and Kokubunji), while the NmF2

disturbance turned negative again at around15:00 JST at lower latitudes (Yamagawa andOkinawa). In particular, NmF2 reduced notice-ably at Okinawa and the reduction persisteduntil midnight. NmF2 began to change to a pos-itive disturbance at 22:00 JST at Yagagawaand at 01:30 JST (on the following day) at

Okinawa. In this way, the ionospheric distur-bances observed in NmF2 are quite complicatedin terms of both latitude and time, but showedrather small variation for the magnitude ofmagnetic storm, except during the evening atOkinawa.

Figure 5 shows a map of time–latitude dis-tributions of TEC. The top panel shows anaverage on quiet days from November 2 to 4as in the case of NmF2, with each numericvalue accompanying a contour line indicatingTEC units (1016 electrons/m2). TEC startedincreasing at sunrise and reached its maximumaround noon at the northern end (45° N) andaround 14:30 JST at the southern end (27° N).A second peak appeared at latitudes lowerthan 30° N after sunset as a consequence ofthe equatorial anomaly starting to developagain as driven by an increased E×B driftresulting from prereversal enhancement of theeastward electric field. The middle panelshows the distributions of TEC during themagnetic storm on November 6. Note the sig-nificantly marked increase in TEC in the day-time. TEC maximized at 13:45 JST at thesouthern end (27° N) and somewhat later at14:15 to 14:30 JST at the northern end (45° N).

Fig.4 Variations in the F-layer maximumelectron density

Fig.5 TEC disturbances(a) Typical behavior on quiet days (b) TEC storm observed on Novem-

ber 6, 2001(c) Differences from quiet days


The difference (∆TEC) map subtractingthe map on quiet days from the TEC map onNovember 6 is given in the bottom panel toprovide more evidence of the appearance ofdisturbances. The difference map not only evi-dently spots anomalous increases in TEC butalso indicates that the increases startedbetween 11:00 and 12:00 JST at about thesame time at all latitudes. The time at whichthe increase in TEC peaked was found to shifttoward a later time with latitude as denoted bythe dotted line: It was 13:45 JST at the south-ern end and 14:45 JST at the northern end. Atlatitudes higher than about 33° N, a moderatepositive disturbance lasted until midnight afterthe peak increase in TEC. Another factrevealed by the difference map is that theanomalous increase in TEC was preceded by areduction in TEC at 09:00 to 11:30 JST atlower latitudes. The magnitude of this reduc-tion was more pronounced in a lower-latituderegion as in the case of the increase in TECfollowing it.

Since the TEC examined here is the quan-tity integrated along the radio wave propaga-tion path from a GPS satellite to the ground, agreater contribution stems from the F-layerpeak electron density height. Hence, the waysionospheric storms are perceived in terms ofNmF2 and TEC should be similar. The top twopanels in Fig.6 compare the disturbance com-ponents of NmF2 and TEC from 12:00 JST onNovember 5 to 12:00 JST on November 7. AtWakkanai, ∆TEC and ∆NmF2 amid the large-scale disturbance gave fairly different looksagainst expectations. ∆TEC graduallyincreased from 06:00 JST on November 6,accelerated at 11:00 JST, and then peaked at14:45 JST before decreasing at a monotonousrate. On the other hand, ∆NmF2 was found topeak moderately around 10:00 and 15:30 JST;between those peaks, it sagged to a negativevalue. Meanwhile, there was a rapid surge in∆TEC during the same time period.

∆TEC and ∆NmF2 at lower latitudes(Kokubunji, Yamagawa, and Okinawa) exhib-ited similar trends in variation from 09:00 to15:00 JST. Both posted negative values

around 10:30 JST, and positive values around14:00 JST. When compared quantitatively,however, the two characteristics varied con-siderably in their amplitude of positive andnegative variations. ∆TEC had significantlylarge positive variations (of about 100 TECunits observed at the latitude of Okinawa), butits negative variations were not so large (atabout 20 TEC units observed at the latitude ofOkinawa). In contrast, ∆NmF2 showed a posi-tive range of variation equal to or smaller thanthe negative range of variation. Such a differ-ence is more pronounced at lower latitudes.Another point worthy of mention regardingthe difference between ∆TEC and ∆NmF2 is anextremely deep reduction in ∆NmF2 centeringon 21:00 JST at the latitude of Okinawa, whencompared with only a shallow drop in ∆TEC.

The ionospheric disturbance observed onNovember 6, 2001, was a most complex one.The goal of elucidating each of the underlyingphysical processes might not be easily attainedby simply reviewing changes in NmF2 andTEC, but if variations in ionospheric heightare analyzed in conjunction with thosechanges, a variety of processes would emergeinto view. Figure 7 is mass plots of hmF2 dailyvariations observed at Kokubunji fromNovember 2 to 6, during which the quiet daysfrom November 2 to 5 are shown by thin solidlines, while the disturbance day of November6 is designated by a thick solid line. Largeday-to-day variations are found at nighttime,particularly from midnight to sunrise, but con-verged in a similar pattern in the daytime,except for the disturbance day. The ionospher-ic height started rising at about 11:00 onNovember 6 and reached a maximum differ-ence of 100 km from the quiet days, maintain-ing that state until about midnight. Similar toTEC and NmF2, the values of ∆hmF2 calculatedby subtracting the quiet-day level from thedisturbance day values are shown in the bot-tom panel of Fig.6. Figure 8 presents a sketch(O-mode only) of the ionograms (every hour)recorded at this time, with a continuous linedesignating the disturbance day and a dottedline marking the ionogram for the previous


Fig.6 Differences in TEC, F-layer maximum electron density, and F-layer height from their quiet-day averages

Fig.7 Variations in the F-layer height on November 2 to 6, 2001


day for the sake of comparison.

4.2 Discussions on the November 2001event

The leading factor causing ionosphericdisturbances should change from time to timeduring the storm event summarized in the pre-vious section, which made the ionosphericstorm appear so complex. Thus, we will dis-cuss the storm mechanisms by dividing theperiod according to aspects of the disturbance.

4.2.1 Before 07:00 JST on November 6 According to Fig.6, TEC and NmF2 during

this time period showed virtually no differ-ences from the quiet days, but hmF2 signifi-

cantly varied. Since ∆hmF2 was found to haveits phase shifted toward a later time at lowerlatitudes (as marked by an inclined longdashed line in the diagram), this disturbancemay have been an effect of traveling ionos-pheric disturbances (TIDs). From the inclina-tion of the dashed line, the velocity of propa-gation is estimated at 740 m/s—the velocity ofa typical large-scale TID (LSTID). A similarTID was observed from 18:00 JST on Novem-ber 6 to 06:00 JST on November 7, with anestimated velocity of propagation rangingfrom 350 to 700 m/s. Despite major variationsin hmF2, there had been little disturbance toTEC and NmF2, which is characteristic of anighttime TID. Generally, an ionosonde chainset up in a latitudinal direction, like the oneused here, should provide very useful data tohelp distinguish between the effect of a TID asa source of change in hmF2 and an electric fieldor neutral atmospheric circulation effects.

4.2.2 From 07:00 to 11:00 JST onNovember 6

During this time period, IMF Bz turnedsouthward while the Kp index exhibited aweak magnetic disturbance, as can be seen inFig.3. The ionosphere shown in Fig.6 is char-acterized by a noticeably pronounced reduc-tion in TEC at lower latitudes, with ∆TECturning negative at latitudes lower than 38° Nfrom 09:00 to 11:00 JST. In response to thisdrop in TEC, NmF2 decreased at Okinawa,Yamagawa and Kokubunji approximately 30min later than ∆TEC. There was little heightvariation. ∆hmF2 changed from a weak positivedisturbance to a weak negative disturbance inOkinawa. The higher the latitude, the less dis-tinct this trend appeared. In Fig.8, almostidentical ionogram traces were seen at 08:00on November 5 and 6, but at 09:00 on Novem-ber 6 the height began to fall at lower frequen-cies, with fOF2 declining. One hour later, fOF2

further declined.A summary review of comparisons of

TEC, NmF2 (fOF2), and hmF2 reveals two con-flicting ionospheric storm processes. An equa-torial wind stirred a weak positive storm at

Fig.8 Sketches of the ionogram for theTEC storm day and the previousday (O-mode only)


higher latitudes, while a transient negativestorm driven by a westward electric fieldoccurred at lower latitudes; both effects bal-anced near 38° N.

4.2.3 Extreme enhancement ofTEC from 11:00 to 16:00 JST onNovember 6

Among the sequence of disturbances, theincrease in TEC during this time period wasmost significant. TEC rose by 100 TEC unitsat the latitude of Okinawa during a period of 2hours, and by 50 TEC units at the latitude ofWakkanai. The sharp rise in TEC appeared tostart at the same time at all latitudes, but thetime of their maximum came later at higherlatitudes. Variations in NmF2 shown in the mid-dle panel of Fig.6 were not so large whencompared with noticeably large variations inTEC shown in the top panel. ∆NmF2 even hadits sign changed. Rises in the F-layer heightstarted at 11:00 JST simultaneously at all thestations, and coincided with the start of risingTEC as designated by the short-dashed linepenetrating the middle through to the bottompanels. The rising height reached an equilibri-um state in 2 hours and remained constantuntil the next day. The sources of these distur-bances will be elucidated one by one.

Penetrating magnetospheric electricfield

Simultaneous rises in ∆TEC and ∆hmF2 ateach latitude may well be understood as mani-festations of a magnetospheric electric fieldpenetrating the lower-latitude ionosphere (asreported in References[9][21］–［23]). IMF Bzwas southward for 7 hours before the suddencommencement of the magnetic storm at01:51 UT on November 6, with southward Bzintensifying around 02:00 UT and beingaccompanied by an increase in ASY-H.According to theoretical calculations by Spiroet al.[22], such penetrating electric field distur-bance is eastward, which is consistent with therising ionospheric height caused by the E×Bdrift observed. Low-latitude geomagneticobservation data also confirm the penetrationof an eastward electric field[23].

The electric field strength can be estimatedfrom the rate of rise in F-layer height and thegeomagnetic inclination over Okinawa. Fromthe slope of the curve for the 30-min periodfollowing the start of rising hmF2 in Fig.6, therate of rise can be estimated at 28 m/s, whichis an eastward electric field equivalent of 1.4mV/m. This would be even more pronouncedwhen considering the effect of induceddescent of the F-layer along the geomagneticfield line by gravity[24].

Dissipation to the plasmasphereThe smaller change in NmF2 when com-

pared with the increase in TEC might becaused by the dissipation of ionospheric plas-ma to the plasmasphere, thereby suppressingrises in the maximum electron density. A gen-eral solution that represents the equilibriumstate of the plasma diffusion equation atheights above the F-layer peak can be stated inan equation as:

where, N denotes the electron density atheight h, and H the scale height of the neutralatmosphere. The first term on the right-handside represents a diffusive equilibrium; thesecond term is a flux solution that designatesdeviation from the diffusive equilibrium[14].In the intermediate region between the F-layerpeak and O+－H+ transition height, the plasmaflux along the geomagnetic field line can beapproximated as[14].

where, Dm denotes the diffusion coefficientof plasma at the F-layer peak and I the mag-netic inclination. If the topside ionosphere isin a diffusive equilibrium, N2 would equal 0,but normally there is an upward flux (N2 > 0)in the daytime and a downward flux (N2 < 0)in the nighttime through exchanges of plasmabetween the ionosphere and plasmasphere.

Assume that the ionospheric height is sud-denly elevated 100 km in the daytime when N2

is positive as shown in Figs.6 and 7. Because


this rise in height is greater than the scaleheight (~ 80 km) of the neutral atmosphere atthe F-layer peak, the diffusion coefficient ofthe plasma appreciably increases by thedecrease in ion-neutral collision frequency. Asa result, the upward flux expressed by Equa-tion (5) would increase by more than e times.A quantitative approach to this problem isdetailed in another paper[25] in this specialissue.

TEC enhancementConsider the electron density at a given

height in the bottomside ionosphere. If theionosphere is suddenly uplifted, the electrondensity would initially fall sharply because thedensity gradient is upward on the bottomside.However, the electron density is soon com-pensated by the progress of an ionizationprocess and it reaches a photo-chemical equi-librium. As a result, there is a resultant netincrease in TEC. Compensation takes approxi-mately 1/β to complete (withβbeing therecombination coefficient of electrons), whichis about 103 s at a height of 250 km and 104 sat a height of 350 km. On the other hand, nearthe maximum electron density height, increas-es in density caused by a reduced loss rateconflict with the lowering density caused byan increased upward dissipative flux, whichproduces a delay of 2 to 3 hours (104 s) in therecovery of NmF2 despite a temporally (103 s)increase in TEC.

The process explained above is consistentwith the hourly changes in the ionogram plot-ted in Fig.8. The F-layer height started risingat 11:00 JST and the trace began shiftingupward. At this time, there was no increase inNmF2 (fOF2) yet. The trace continued to risefrom 12:00 to 13:00 JST, while an intenseeffect of maintaining a photo-chemical equi-librium (compensating for the density loss dueto upwelling) was found at work at lower alti-tudes (low frequency part of the ionogram).Folded segments below 7 MHz in the tracesrepresent an F1 layer, which normally does notappear distinctly at this time of the year. fOF2

also showed a distinct rise at 12:00 and thenprogressed at a high level. At 15:00 JST, com-

pensation from the progress of the new ioniza-tion virtually pervaded the entire frequencyrange (at all altitudes), with fOF2 remaining ata high level. Since fOF2 starts declining duringthis time period on quiet days, the differencefrom the quiet days ∆NmF2 plotted in Fig.6 (forKokubunji in the middle panel) is maximized.

Delay in the enhancement peak of TEC The extreme enhancement in TEC has

been attributed to the penetration of a magne-tospheric electric field into lower latitudes.Such an electric field should work simultane-ously at all latitudes. In fact, rises in the F-layer driven by the E×B drift began simulta-neously as shown in the bottom panel of Fig.6.Yet, maximum of the enhancement in TECtends to come later at higher latitudes as indi-cated by a dotted line in the bottom panel ofFig.5. At higher latitudes, the geomagneticfield line lengthens to penetrate a greater pro-portion of the plasmasphere. Consequently,the plasma takes longer to dissipate from theionosphere to the plasmasphere, thereby possi-bly slowing down the rate of the increase inTEC.

4.2.4 Period of recovery from 16:00 JSTto midnight

The aspect of recovery from the extremeenhancement of TEC to the quiet day level isclearly different at higher and lower latitudesacross 33° N as can be seen in Fig.6. At lowerlatitudes, ∆TEC continued to decrease until21:45 JST with a higher rate at lower latitude.Conversely, at higher latitudes, ∆TEC plum-meted from 50 to 30 TEC units and kept ahigh level from 17:00 to 23:00 JST. Theionospheric height remained at a higher levelthan on quiet days at all latitudes as can beseen from ∆hmF2 shown in the bottom panel inFig.6, except for a superposition of a verticalfluctuation of TID. Therefore, at higher lati-tudes, the uplifting of the ionosphere by anequatorward wind may have suppressed elec-tron loss along with the downward diffusionof plasmaspheric plasma stored in the day-time, allowing a high TEC level to be main-tained.


Negative ∆TEC after sunset at lower lati-tudes was observed at 21:00－23:00 JST (bot-tom panel of Fig.5), somewhat later than thesecond peak as observed on quiet days causedby the prereversal enhancement of the electricfield (top panel of Fig.5) (regrowth of theequatorial anomaly) in day-to-day TEC varia-tions. The occurrence of negative ∆TEC mightbe caused by the development of a disturbancedynamo electric field[10] driven by a sequenceof geomagnetic/thermospheric disturbances,which modulated the electric field during thetime of the prereversal enhancement[26] to has-ten the westward reversal. As the effect of theelectric field around the prereversal enhance-ment is pronounced at lower latitudes[27], thenegative storm effect of the disturbancedynamo electric field might overcome the pos-itive storm effect of the equatorward neutralwind (which weakens with lowering lati-tudes). The more prominent negative storm inNmF2 rather than TEC reflects the latitudestructure as explained in Fig.9. In the figure,negative disturbance is confined in the shadedregion. The quantity NmF2 is a local value atthe maximum electron density height (F-peak)

schematically represented in the figure, whilethe value of TEC is an integrated quantityalong the vertical line. Magnetic field lineshigher above the F-peak at low latitudes areconnected to the F-peak at higher latitudes toextend the effect of positive storms in thoseregions.

4.2.5 After midnightBoth TEC and NmF2 no longer showed a

major difference from their quiet day levelsafter midnight. As a general trend, the lowerthe latitude, the higher the value of ∆TEC,evidencing the effect of a disturbance dynamoelectric field turning eastward at nighttime.Despite a larger value of ∆hmF2 at a higher lat-itude, TEC continued to recover (dampen) tothe quiet day level, probably under the influ-ence of changes in the neutral atmosphericcomposition caused by a continuing south-ward circulation. In either case, the series oflarge disturbances came to an end.

5 Space Weather Event onNovember 8, 2004

5.1 Overview of disturbancesAn intense magnetic storm occurred on

November 7, 2004 during a declining phase ofthe solar activity cycle. Figure 10 plots IMFBz (top panel) observed by the ACE satellite,the AE index (middle panel), and the Dst index(bottom panel). Dst started declining at 21:00UT on November 7, when IMF turned south-ward and peaked at －373 nT at 06:00 UT.The vertical dotted line in the storm recoveryphase designates the time at which the abnor-mal increase in TEC being discussed here wasobserved.

Figure 11 shows a map of time－latitudedistribution of TEC variations. Dotted linesdenote the times of sunrise and sunset at aheight of 200 km. Here, Fig.11a represents themeans for November 1 to 6 (JST) as typicalquiet day variations; Fig.11b shows TEC varia-tions on November 8 (when the disturbancewas observed). The differences in ∆TECbetween quiet and disturbance days are depict-

Fig.9 Positive and negative phases of anionospheric storm varying with lati-tudeThe positive storm at higher latitudes is causedby an uplift of ionospheric height due to anequatorward neutral wind, the negative stormat lower latitudes is caused by a receding foun-tain effect resulting from decay in the eastwardelectric field.


ed in Fig.11c. ∆TEC began to increase slowly ashort while after sunrise and exhibited threemajor enhancements thereafter. The first ofthese enhancements was from 14:00 to 15:00JST over the entire range of latitudes, with aslight delay in time at higher latitudes. The sec-ond one was observed from 17:00 to 19:00 JSTaround sunset and, unlike the first enhance-ment, was limited to a region lower than 37° Nin latitude. The last one took place around21:00 JST at latitudes higher than 35° N.

The TEC enhancement was associatedwith variations in F-layer height as in the pre-vious event. Figure 12 shows variations inhmF2 observed at the four ionosonde stationsfor this event in 2004. Dots connected with aline denote a variation on the disturbance day(November 8) and the solid line is a meanvariation on November 1 to 6, 2004 for refer-ence. The letter "G" in the plots for Wakkanaiand Kokubunji denotes that hmF2 is not calcu-lated because the upper part of the traceexceeded the range of ionogram display due toa large propagation delay. A height rise tookplace around sunrise time (shown by aninverted open triangle) and remained highuntil past sunset (inverted filled triangle). Therise in the F-layer height at sunrise resulted ina moderate increase in ∆TEC; a large solarzenith angle right after sunrise might inhibit asharp increase in TEC as noted in the 2001event. The rising height began over Wakkanaiat 04:00 JST and its onset time delayed with

decrease in latitude came at 06:00 JST overOkinawa. The leading edge of the riseappeared stepwise more distinctively at lowerlatitudes. This uplifting is not caused by aneastward electric field as described in the pre-vious section, but by the effect of traveling

Fig.10 Magnetic disturbance indices inearly November 2004

Fig.11 TEC variations indicative of a TECstorm and SED(a) Typical behavior on quiet days(b) TEC storm and SED observed

on November 8, 2004 (c) Differences from quiet days

Fig.12 Variations in the F-layer heightobserved at the four ionosondestationsDots connected by a line denote a distur-bance observed on November 8, 2004. Thesolid line designates a mean on quiet daysfrom November 1 to 6, 2004.


atmospheric disturbances (TADs) originatingfrom a higher latitude or an equatorwardsurge[28].

The pronounced increase in TEC at14:00－15:00 JST began sharply at about thesame time, though the TEC peak at higher lati-tudes tended to delay. This is quite similar tothe case discussed in the previous section,which is suggestive of the effect of E×B driftdriven by a penetrating magnetospheric elec-tric field, except for the different behavior ofhmF2. The rise in hmF2 was not necessarily dis-tinct in response to the rapid increase in TECunlike the previous event. This may be due toan already intensified equatorward thermos-pheric wind having appreciably uplifted the F-layer, as well as a superposition of non-sys-tematic and unidentifiable height fluctuation.

A gradual enhancement peak in hmF2

(shaded in Fig.12) was observed in Okinawafrom 15:00 to 18:00 UT. This peak was barelynoticeable in Yamagawa as well, but not inKokubunji and Wakkanai. Obviously, this cor-responds to the second upsurge in TEC at18:00 JST. In Okinawa, TEC took about 3hours to peak after hmF2 began to rise.

Lastly there was a prominent increase inTEC at 21:00 JST. A rising height characteris-tic of TID (a phase of variation delayed withlowering latitude) was observed during aslightly earlier time period as marked by thedotted line in Fig. 12. Height variationsobserved at Wakkanai reverted to the quiet-day level by 20:15 JST, followed by a sharpincrease in TEC. Because the rise in the F-layer height had occurred in a mid-latituderegion after sunset, the source of TEC varia-tions should obviously differ from the mecha-nism of positive storm occurrence as dis-cussed in the previous section. This is ascribedto SED (storm enhanced density), which israrely observed in Japan. The next sectiongives a detailed analysis of SED .

5.2 Storm Enhanced Density (SED)Although Fig.11 gives the perspective of

latitude distribution and temporal changes inionospheric disturbances, TECs are average in

the longitudinal direction across a number ofcells with the same latitude. To examine spa-tial structure of SED in further depth, wechose cells aligned in the north-south directionat 141° E (cells 22, 23, 24, and 25 shown inFig.13) and those aligned in the east-westdirection at 41° N (cells 19, 24, and 28 shownin the same figure). The temporal variations inTEC for individual cells are plotted in Fig.14.A closer look at Fig.14a, which compares thenorth-south direction, reveals that TEC beganrising after sunset as indicated by the dashedline, and then peaked at 20:30 JST at 43° N(cell 25). The increment in TEC during thisperiod was 75 TEC units and the peak valuelarger than the value in the daytime at thesame latitude. The lower the latitude, the laterTEC began increasing and peaked, with adelay of 45 min at 37° N (cell 22). The lowerthe latitude, the less TEC increased, with 20TEC units observed at 37° N. The apparentpropagation velocity of the disturbance can beestimated at 8°/hr equatorward. A closer lookat Fig.14b, which compares the east-westdirection, reveals that the increase in TECafter sunset began earlier eastward, with anapparent propagation velocity of the distur-

Fig.13 Cells for which TECs were evaluat-ed


bance estimated at 8°/hr westward, which isroughly equivalent to half the solar terminatormovement.

The TEC disturbance appearing to propa-gate equatorward strongly invokes a TADoriginating from the auroral zone. Variationsin the F-layer height observed around 18:00JST at Wakkanai and Kokubunji exhibit acharacteristic of a TAD. These variations areimmediately followed by an increase in TECof 53 TEC units an hour (cell 25). This is aflux equivalent of 1014 m－2s－1. At mid-lati-tudes with L ~ 1.5, the electron densities at themagnetic conjugate points are coupled witheach other by H+ flux along the magnetic fieldline[29]. If the increase in TEC observed herewas caused by plasma flux from the oppositehemisphere, under the condition of the ionos-phere being supported at a high altitude by aTAD, the H+ field-aligned flux at the top ofthe ionosphere would be larger than the limit-ing flux[30] by two orders. Hence, it is diffi-cult to ascribe the large increase in TEC to aTID induced by a TAD, and the effect of hori-zontal advection should be considered instead.

Figure 15a plots the total ion density at aheight of 850 km as observed by the DMSP-

F15 satellite. The solid line shows data record-ed on November 8 (on the orbit shown inFig.13) and the dashed line shows data record-ed on November 6, a quiet day (on about thesame orbit as for November 8) for the sake ofcomparison. The diagram should depict lati-tude changes, since this satellite is in a north-south sun-synchronous orbit (orbital inclina-tion 98.4°). The ion density on November 8was elevated from the level on quiet days overthe entire latitude range, showing a density“bump” (at 41.1° N, 143.4° E; 11:00 UT;20:00 JST) as marked by the long dashed line.If the increase in electron density (total iondensity) had been induced by advection ofE×B drift, the high-electron density structureshould be aligned with the geomagnetic fieldline. When the position of the densityenhancement at a height of 850 km observedby the DMSP-F15 satellite is projected ontothe thin-shell height[15] of 400 km along thegeomagnetic field line, it would be 43.5° N,143.1° E, corresponding to cell number 29.TEC at 20:00 JST for cells 27 to 30 alignedsouth to north at 143° E was 54.7, 74.2, 95.7,

Fig.14 (a) TEC temporal variations in thenorth-south aligned cells

(b) TEC temporal variations in theeast-west aligned cells

Fig.15 Total ion density and westward iondrift velocity observed by theDMSP satellite


and 85.2 TEC units, respectively. Therefore,the density peak observed by the DMSP satel-lite and the TEC peak by the ground GPS net-work agree. In Fig. 15a, another densityenhanced region is noted at 19° N (12° Nmagnetic latitude) near the magnetic equator,corresponding to the location of the northerncrest of the equatorial anomaly. This was notexperienced on a quiet day and is suggestiveof an intensified eastward electric field to pro-mote the growth of the equatorial anomaly onNovember 8.

Figure 15b plots the westward ion driftvelocity (a horizontal component perpendicu-lar to the orbit) on the same orbit as inFig.15a. The density enhanced region hadvelocity of ~ 250 m/s (or ~ 220 m/s when pro-jected on the ground surface)—slightly fasterthan the velocity of the TEC frontal structureat 8°/hr (200 m/s). Since a northward driftvelocity component of 40 m/s perpendicular tothe geomagnetic field line was observed at thesame time, the southward movement of thefrontal structure of the TEC enhanced regionwas apparent, such that the actual movementof the density enhanced region is presumed tobe a flow in the direction from ESE to WNW.Considering the fact that the apparent velocityof propagation of disturbances estimated fromthe TEC variations among the cells aligned inthe north-south and east-west directions is8°/hr both westward and equatorward, thebehavior of the TEC enhanced region can beschematically depicted as shown in Fig.16.Such an extreme and localized densityenhancement and the dynamics of the density-enhanced region are characteristic of SED.Furthermore, during this event, a westwardhigh-speed plasma flow of the so-called SAPS(subauroral polarization stream) was observedat higher latitude by the DMSP satellite asshown in Fig.15b. SAPS is another characteris-tic closely related to SED/TEC plumes[31][32].

Previous studies [32］–［34] have revealedthat SED/TEC plumes involve positive ionos-pheric storms at low latitudes resulting fromdevelopment of the equatorial anomaly, butthe equatorial anomaly and the density

enhanced region of SED appear to be separatedby the relatively low density region betweenthem. However, given the fact that high elec-tron density at low latitudes caused by devel-opment of the equatorial anomaly accounts forthe high electron density source of SED/TECplumes, both regions are linked[33]. From thisstandpoint, the disturbances exemplified heremay well be representative of a SED/TECplume resulting from an intense TEC stormduring the daytime.

Since the ion drift velocity of 250 m/s (or220 m/s when projected on the ground sur-face) observed by the DMSP satellite in thedensity enhanced region was lower than thevelocity of the solar terminator on the groundsurface (400 m/s at 41° N), it follows that thedensity enhanced region had been formed tothe east (and to the south) and at an early LT(local solar time). What should be noted aboutFig. 11 is that an increase in TEC wasobserved at lower latitudes at 18:00 JST,which corresponded to the uplift of the ionos-phere at 15:00－18:00 JST. This is considereda consequence of the equatorial anomalyexpanding to higher latitudes. The F-layerheight was maximized at 16:00 JST, with

Fig.16 Schematic illustration of SEDobserved on November 8, 2004


upward E×B drift being halted. The increasein TEC was maximized 2 hours later, which isin agreement with model calculations ofheight response to electric field changes[35].Although the origin of the electric field distur-bance occurring during this time periodremains unclear, the equatorial anomalyexpanding higher latitudes in the wake of theintense TEC storm in the daytime may have aclose bearing on the formation of SED.

Another key factor involved in the forma-tion of SED is the driving source for a sus-tained westward/poleward advection at mid-to low latitudes. The F-layer height remainedhigh throughout the daytime as stated earlier,offering evidence of intensified equatorwardthermospheric circulation. Such thermosphericwind disturbances set up a disturbancedynamo electric field. According to model cal-culations[10], a poleward electric field of 5mV/m is produced all day long, along with aneastward electric field of 1 to 3 mV/m at amid-latitude (41°) in the nighttime after 19:00LT. These electric fields may provide a drivingsource for SED.

6 Conclusions

This article described two events in whichextreme enhancement of ionospheric totalelectron content (TEC) —a key element ofionospheric space weather— was observed.The physical processes underlying the occur-rence of these events were analyzed. Oneevent is a TEC storm that occurred during thedaytime; the other is SED (storm enhanceddensity) observed after sunset. Both events arethe last stage of a sequence of space weatherdisturbances as schematically illustrated inFig.17 and the largest TEC events ever record-ed in the last solar cycle of 11 years (cycle 23)at Japan’s longitudes. The reason for mention-ing “Japan’s longitudes” in particular is thatthe time at which magnetic disturbancesoccurred is very important as a source of dri-ving ionospheric storms: Even a magnetic dis-turbance inducing a massive ionospheric dis-turbance in East Asia does not necessarily pro-duce strong ionospheric disturbances at longi-tudes of the U.S. and vice versa. While ionos-pheric disturbances have a close bearing on

Fig.17 Occurrence of radio wave propagation difficulty and origin of ionospheric disturbances


longitude and time, whereas magnetic distur-bances are a matter of the relation between thesolar wind and the earth, with solar wind dis-turbances occurring independently of earth’sexistence. To be able to predict ionosphericdisturbances occurring in a given longitudinalregion on the earth from disturbances on thesolar disk, a chain of “causes” and “effects” ofdisturbances as summarized in Fig.17 must bepredicted with a time accuracy on the order ofhours across the entire process flow. Simplybeing able to predict the occurrence of a mag-netic storm would be far from reassuring as ameans to forecasting radio wave propagationfailures, because the ionospheric disturbancecould become a positive or negative storm

depending on the timing of its occurrence (seeReference[36] for the occurrence conditions ofintense negative storms). As a practical solu-tion to space weather prediction, developing atechnique for predicting the passage of anionospheric disturbance from its initial stagesof occurrence onward would be valuable. Tothis end, in-depth post analyses as described inthis paper should be indispensable.

Acknowledgments

GEONET was constructed and is main-tained by the Geospatial Information Authori-ty of Japan (GSI). We would like to thank GSIfor the use of GEONET data.

References01 D. Bilitza and B. W. Reinisch, “International Reference Ionosphere 2007: Improvements and new para-

meters,” Adv. Space Res., Vol. 42, pp. 599–609, 2008.

02 T. Maruyama, “Regional reference total electron content model over Japan using solar EUV proxies,”

Special Issue of This NICT Journal, 3-3-5, 2009.

03 T. J. Fuller-Rowell, M. V. Codrescu, R. J. Moffett, and S. Quegan, “Response of the thermosphere and

ionosphere to geomagnetic storms,” J. Geophys. Res., Vol. 99, pp. 3893–3914, 1994.

04 T. J. Fuller-Rowell, M. V. Codrescu, H. Rishbeth, R. J. Moffett, and S. Quegan, “On the seasonal

response of the thermosphere and ionosphere to geomagnetic storms,” J. Geophys. Res., Vol. 101, pp.

2343–2353, 1996.

05 G. W. Prölss, “Ionospheric F-region storms,” in Handbook of Atmospheric Electrodynamics, Vol. 2, edit-

ed by H. Volland, pp. 195–248, CRC Press, Boca Raton, Fla., 1995.

06 G. W. Prölss, “Magnetic storm associated perturbations of the upper atmosphere,” in Magnetic Storms,

Geophysical Monograph 98, edited by B. T. Tsurutani, W. D. Gonzales, Y. Kamide, and J. H. Arballo,

American Geophysical Union, Washington, D.C., 1997.

07 M. J. Buonsanto, “Ionospheric storms – Review,” Space Sci. Rev., Vol. 88, pp. 563–601, 1999.

08 G. W. Prölss, “Storm-induced changes in the thermospheric composition at middle latitudes,” Planet.

Space Sci., Vol. 35, pp. 807–811, 1987.

09 M. C. Kelley, B. G. Fejer, and C. A. Gonzales, “An explanation for anomalous equatorial ionospheric

electric fields associated with a northward turning of the interplanetary magnetic field,” Geophys. Res.

Lett., Vol. 6, pp. 301–304, 1979.

10 M. Blanc and A. D. Richmond, “The ionospheric disturbance dynamo,” J. Geophys. Res., Vol. 85, pp.

1669–1686, 1980.

11 B. G. Fejer, J. T. Emmert, G. G. Shepherd, and B. H. Solheim, “Average daytime F region disturbance

neutral winds measured by UARS: Initial results,” Geophys. Res. Lett., Vol. 27, pp. 1859–1862, 2000.

12 J. M. Forbes, M. Codrescu, and T. J. Hall, “On the utilization of ionosonde data to analyze the latitudinal

penetration of ionospheric storm effects,” Geophys. Res. Lett., Vol. 15, pp. 249–252, 1988.


13 J. M. Forbes, “Evidence for the equatorward penetration of electric fields, winds, and compositional

effects in the Asian/Pacific sector during the September 17-24, 1984, ETS interval,” J. Geophys. Res.,

Vol. 94, pp. 16,999–17,007, 1989.

14 H. Rishbeth, “On the F2-layer continuity equation,” J. Atmos. Terr. Phys., Vol. 48, pp. 511–519, 1986.

15 G. Ma and T. Maruyama, “Derivation of TEC and estimation of instrumental biases from GEONET in

Japan,” Ann. Geophys., Vol. 21, pp. 2083–2093, 2003.

16 T. Shimazaki, “World-wide daily variations in the height of the maximum electron density of the ionos-

pheric F2 layer,” J. Radio Res. Labs., Vol. 2, pp. 85–97, 1955.

17 P. A. Bradley and J. R. Dudeney, “A simple model of the vertical distribution of electron concentration in

the ionosphere,” J. Atmos. Terr. Phys., Vol. 35, pp. 2131–2146, 1973.

18 F. T. Berkey and G. H. Stonehocker, “A comparison of the height of the maximum electron density of the

F2-layer from real height analysis and estimates based on M (3000)F 2,” J. Atmos. Terr. Phys., Vol. 51,

pp. 873–877, 1989.

19 L. M. Muggleton, “A method of predicting fOE at any time and place,” Telecommun. J., Vol. 42, pp.

413–418, 1975.

20 T. Iyemori and D. R. K. Rao, “Decay of the Dst field of geomagnetic disturbance after substorm onset

and its implication to storm-substorm relation,” Ann. Geophys., Vol. 14, pp. 608–618, 1996.

21 B. G. Fejer, C. A. Gonzales, D. T. Farley, M. C. Kelley, and R. F. Woodman, “Equatorial electric fields

during magnetically disturbed conditions, 1. The effect of the interplanetary magnetic field,” J. Geophys.

Res., Vol. 84, pp. 5797–5802, 1979.

22 R. W. Spiro, R. A. Wolf, and B. G. Fejer, “Penetration of high-latitude-electric-field effects to low latitudes

during SUNDIAL 1984,” Ann. Geophys., Vol. 6, pp. 39–50, 1988.

23 T. Kikuchi, K. Hashimoto, and K. Nozaki, “Penetration of magnetospheric electric fields to the equator

during a geomagnetic storm,” J. Geophys. Res., Vol. 113, A06214, doi:10.1029/2007JA012028, 2008.

24 T. Maruyama, “E×B instability in the F-region at low- to midlatitudes,” Planet. Space Sci., Vol. 38, pp.

273–285, 1990.

25 H. Jin and T. Maruyama, “Different behaviors of TEC and NmF2 observed during large geomagnetic

storms,” Special Issue of This NICT Journal, 3-3-2, 2009.

26 L. Scherliess and B. G. Fejer, “Storm time dependence of equatorial disturbance dynamo zonal electric

fields,” J. Geophys. Res., Vol. 102, pp. 24,037–24,046, 1997.

27 J. A. Murphy and R. A. Heelis, “Implications of the relationship between electromagnetic drift compo-

nents at mid and low latitudes,” Planet. Space Sci., Vol. 34, pp. 645–652, 1986.

28 G. W. Prölss, “On explaining the local time variation of ionospheric storm effects,” Ann. Geophys., Vol.

11, pp. 1–9, 1993.

29 G. J. Bailey, P. A. Simmons, and R. J. Moffett, “Topside and interhemispheric ion flows in the mid-lati-

tude plasmasphere,” J. Atmos. Terr. Phys., Vol. 49, pp. 503–519, 1987.

30 P. G. Richards and D. G. Torr, “Seasonal, diurnal, and solar cyclical variations of the limiting H+ flux in

the earth’s topside ionosphere,” J. Geophys. Res., Vol. 90, No. A6, pp. 5261–5268, 1985.

31 J. C. Foster, P. J. Erickson, A. J. Coster, J. Goldstein, and F. J. Rich, “Ionospheric signatures of plasma-

spheric tails,” Geophys. Res. Lett., Vol. 29, doi:10.1029/2002GL015067, 2002.

32 J. C. Foster and W. Rideout, “Midlatitude TEC enhancements during the October 2003 superstorm,”

Geophys. Res. Lett., Vol. 32, L12S04, doi: 10.1029/2004GL021719, 2005.


33 J. C. Foster, “Storm time plasma transport at middle and high latitudes,” J. Geophys. Res., Vol. 98, No.

A2, pp. 1675–1689, 1993.

34 M. Vlasov, M. C. Kelley, and H. Kil, “Analysis of ground-based and satellite observations of F-region

behavior during the great magnetic storm of July 15, 2000,” J. Atmos. Solar-Terr. Phys., Vol. 65, pp.

1223–1234, 2003.

35 C. H. Lin, A. D. Richmond, R. A. Heelis, G. J. Bailey, G. Lu, J. Y. Liu, H. C. Yeh, and S. -Y. Su, “Theoreti-

cal study of the low- and midlatitude ionospheric electron density enhancement during the October

2003 superstorm: Relative importance of the neutral wind and the electric field,” J. Geophys. Res., Vol.

110, A12312, doi:10.1029/2005JA011304, 2005.

36 T. Maruyama and M. Nakamura, “Conditions for intense ionospheric storms expanding to lower-midlati-

tudes,” J. Geophys. Res., Vol. 112, A05310, doi:10.1029/2006JA012226, 2007.

MA Guanyi, Ph.D.

Professor, National AstronomicalObservatories, Chinese Academy ofSciences.

Upper Atmospheric Physics, SatelliteCommunication and Navigation

MARUYAMA Takashi, Ph.D. (Eng.) Executive Researcher

Upper Atmospheric Physics

NAKAMURA Maho, Ph.D.

Expert Researcher, Space-Time Standards Group, New Generation Network Research Center

Ionosphere, Information Engineering

Documents

3-3 Ionospheric Storm and Variation of Total Electron Content€¦ · 352 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009 nant