RadioScience, Volume35, Number6, Pages1437-1444,November-December 2000 IonosphericE region remote sensingwith ELF radio atmospherics Steven A. Cummer ElectricalandComputer Engineering Department, DukeUniversity, Durham,NorthCarolina Umran S. Inan Space,Telecommunications andRadioscience Laboratory, StanfordUniversity,Stanford,California Abstract. Radioremotesensingof the ionospheric E regioncanbe difficult,particularlyin the nighttimeE regionvalleywhichliesbetweentwo regionsof higherelectrondensity.We showthat extremelylow frequency(ELF) electromagnetic waveslaunchedfrombelowpenetratethisregion because of theirlow attenuation andcanbe reflectedfromboththe D regionandthetop of the E regionvalley.This doublereflection,alsocausedby sporadicE layers,produces a measurable effecton the subionospheric propagation of ELF waves,whichwe demonstrate with numerical Earth-ionosphere waveguide simulations. Thissensitivity opensthepossibility of remotelysensing theE regionanddetecting sporadic E layerswithbroadband ELF propagation measurements using lightningdischarges asthesource.Fromnighttimeobservations of ELF lightningradiationoverthe samepropagation pathon differentdays,we extractE regionelectrondensityprofilesthat show significant variabilitynotonlyfromdayto daybutoverthecourseof a singlenight.Thistechnique is inherentlya path-integrated measurement, enablingtomographic large-scale ionospheric remote sensingwith multiplesourcesandreceivers. 1. Introduction Becauseof its relativeinaccessibility the lower ionosphere,consisting of the D andE regions,is amongthe leaststudiedregionsof the Earth'satmosphere. The altitudesinvolved(•70-140 km) aretoo highfor balloons tersareextremelylargeandexpensivebecauseof thelong wavelengthsinvolved[Watt, 1967], making suchsounding difficult in practice. Nevertheless,man-madeVLF transmittershavebeenusedvery successfully to remotely senseperturbationsin this regionof the ionosphere[e.g., Johnsonet al., 1999]. and too low for mostsatellites,makingin situ measureLightningis a powerfulbroadbandsourceof VLF and mentsdifficult. Radio soundingdoesnot alwayswork ELF energyand can be usedas a sourcefor sounding becausethe electrondensitiesin thisregionareoftentoo the lower ionosphere.Cummeret al. [ 1998] showedthat low to reflect radio waves with ionosonde and incoherent broadbandVLF radiationfrom lightningcan be usedto scatterradar,particularlyat night. Rocketexperiments measurean assumedD region electrondensityprofile [e.g.,Mechtlyet al., 1967]havemeasured thisregionbut with an altitudeand scaleheightaccuracyof a few hunonly in limitedspatialandtemporalextents. dred meters. In this work, we show that ELF radiation Very low frequency(VLF, 3-30 kHz) andextremely low frequency(ELF, 3-3000 Hz) electromagnetic waves from lightningis partiallyreflectedfrom E regionaltiELF propagationmeasureare stronglyreflectedby the D and E regionsand can tudesandthat subionospheric ments from lightning can be used to soundthe E rethereforebe usedfor radio soundingof thesealtitudes [Sechrist, 1974]. However, at ELF and VLF, transmit- Copyright2000by theAmericanGeophysical Union. Papernumber2000RS002335. 0048-6604/00/2000RS002335511.00 1437 gion, includingthe E regionvalley, which is difficultto probefrom the ground. We also showtheoreticallythat subionospheric ELF propagationis very sensitiveto the presenceof sporadicE layers(Es). The methodwe describemeasurespath-integratedelectrondensityprofiles and therefore,with multiple sources(i.e., lightning lo- 1438 CUMMER AND INAN: E REGION REMOTE SENSING AT ELF cations)andreceivers, canbe usedto tomographically measure three-dimensional ionospheric structure onlarge spatialscales. Previous theoretical studies have demonstrated the sen- sitivityof ELF propagation to E regionelectrondensities. Pappertand Moler [1974, 1978] showedwith full 160t '/ 6Oh"' / /; wavemodelcalculations thatEs andsharpelectrondensitygradients around250km altitudeproduce significant attenuationmaximabelow 100 Hz. Barr [1977] showed, againwithfull wavecalculations, thatEs drastically affectsELF attenuation ratesby generating a seriesof resonancesfrom 10 to 1000 Hz andthatthe Es characteristicsaffectthefrequencies anddepthof theseresonances. In thispaper,we showthatthenighttimeE regionvalley generates weakerbutmeasurable ELF resonances that depend strongly onthedetailsof theelectron density pro- '-• 80' 1 '0 80 "" '' 40 40 00 ......... 1000 101 10 210 310 410 5 realpartofn localskindepth(km) Figure1. Indexof refraction n andskindepthforvertically incidentX- andO-modewavesin a modelnighttime ionosphere ELF propagation experiments havebeenperformed at 5 kHz and500Hz andforthe500Hz X modein a daytime with man-made,single-frequency transmitters [Bannis- ionosphere. Thelowerlosses andsharper n gradientat 180km a sigter, 1975]andwithradioatmospherics [Hughesand Gel- altitudefor thenighttimeX modeat 500Hz canproduce file. lenberger,1974]. Theseauthorsobserved significant nificantE-F regionreflection. variability in ELF attenuationrates, but the limited fre- quencyrangesandspectralsmoothing fromtemporalaveragingtendedto obscurethe resonances which enable theremotesensingdescribed in thiswork. cursundertheseconditions at thehigherof the altitudes where fe2 = ffa andfe2 = f v aresatisfied. The extraordinary modeis only weaklyreflectedexceptby extremelysharpionospheric gradients andthere2. ELF PenetrationInto theE Region forepenetrates moredeeplyintotheionosphere. It is this The nominalreflectionaltitudeof electromagnetic modethatcanbe reflectedby E or F regiongradients wavesin theionosphere varieswithfrequency, withlower substantiallyabovethe O-mode reflectionaltitudeif the frequencies usuallyreflectedby lowerelectrondensities frequencyis sufficiently low. Figure 1 showsthe Oandtherefore at loweraltitudes.In the F region,where and X-mode indexes of refraction n at 5 kHz and 500 electroncollisions areinfrequent enoughto berelatively Hz for vertical incidence as a function of altitude in a unimportant, verticallyincidentordinaryO andextraor- modelnighttimeionosphere andtheX-moden at 500 Hz dinaryX wavesarereflected atthealtitudewherefe = f for a daytimeionosphere.Expressions for the indexof andfe2 = f2_ ffB, respectively, where f isthewave refraction in a coldplasma aregivenbyBudden [1985,p. frequency, fe is theelectron plasmafrequency, andfB is 74]. The left panelof Figure1 showstherealpartof n theelectrongyrofrequency [Budden,1985,p. 357]. The ona nonlinear scaleto showbothmodes,whiletheright O and X modes are the two characteristic waves that repanelshowsthelocalskindepth(or amplitude e-folding sultfrompropagation in an anisotropic plasma.Each depth)andreflectsthelossesof thewaves.At night,no modehasits ownpolarization andindexof refraction, dropsaround80 km for bothfrequencies, andabovethis andthepolarization, whilefrequency dependent, is of- altitudethe lossesare veryhigh (skindepthsof 10 km tencircular.As thewavefrequency is reduced, the O and less). Thus the O modeis only reflectednear 80 andX reflection altitudes movedownward intoregions km. In contrast,thenighttimenx changesgradually,the whereelectron collisions canbesignificant andthesere- lossesare significantlylower,and thereforethe X mode flection heights arenotnecessarily valid.In thisregime, significantlypenetrates the nighttimeionosphere.HowBudden[1985, p. 361] describes the reflectionasoccur- ever, at 500 Hz the losses are much lower between 70 and ring at a complexaltitude,but it is not clear how this 120km altitude,andthegradientin n at 180km is much complexaltituderelatesto the real altitudebelow which sharper thanat 5 kHz. ThustheX modeundergoes some mostof thewaveenergy is confined. Ratcliffe [1959,p. reflectionin thenighttimeE-F region,andthecombina140]showsthatsignificant ordinarymodereflection oc- tionof gradients andlosses makesthisreflection stronger CUMMERAND INAN: E REGIONREMOTESENSINGAT ELF at lower frequencies.During the day, however,X-mode lossesare larger at 500 Hz, the sharpn gradientmoves downto 90 km altitude,and the X modedoesnot penetrateto the E-F region.It is the nighttimeX modeE-F regionreflection thataffectssubi6nospheric ELF propagation and forms the basisfor the remote sensingtechniquedescribedin this work. 3. E RegionEffect on Subionospheric ELF 1439 300 250 .\ - •lect•'ons] •' 200 • 150 Lq • 100 50 Propagation O' . ' . , . , .... ". ...... 012 102 100 102 104 106 108 10TM 1 Becauseof the secondary ELF reflectionsfrom the E collisionfrequency(s-]) region,subionospheric ELF propagation depends on the electron densityat thesealtitudes.Thisdependence can Figure3. Electronandioncollisionfrequency profiles usedin bedemonstrated numerically withtheLongWaveProp- thesimulations throughout thispaper. agationCode (LWPC) [Pappertand Ferguson,1986], whichis commonlyusedfor VLF calculations butis also valid and has beenusedat ELF [Pappertand Moler, presentas requiredfor chargeneutrality. The electron 1974].Figure2 showstwomidlatitude electron density andioncollisionfrequency profilesusedthroughout this profiles fromthe1995international reference ionosphere workarea concatenation of D regionprofilesof Morfitt [Bilitza,1997]anda thirdwitha modelEs layer.Of the andShellman[1976]andE andF regionvaluesof Rishtwonon-Esprofiles,onecontains a significant E region bethandGarriott[1969,p. 131]andareplotted inFigure valleydepletion, andtheotherdoesnot. All of thepro- 3. filesareidentical below90 km.Theionospheric magnetic In the Earth-ionosphere waveguidethe unconfinedO fieldin allthesimulations istypicalofeast-to-west propa- and X plasmawavescombineinto waveguidemodes gationoverthecontinental UnitedStates.Ionsplaya sigthatcontainbothwaves. The numberof propagating nificantrolein subionospheric ELF propagation [Pappert modesdependson the frequencyrelativeto the waveandMoler, 1974],andwe assume throughout thiswork guide cutoff frequencies.We use the LWPC to calcu- thatthepositive iondensity Nil-= Newhere Ne> 102 latethewaveguide propagation constants (phasevelocandthatN•- = 102where Ne< 102,withnegative ions ity andattenuation) of the singlequasi-transverse elec- 220 tromagnetic (QTEM) modethatpropagates at ELF (the othermodes areevanescent). Figure4 shows thesequantitiesasa function of frequency for thethreeionospheres shownin Figure2 withthecollisionfrequency profiles in Figure3. The significant variations with ionosphere demonstrate theimportance oftheE regionforELFprop- : agation. .a. 60 I 10ø ......... ß : :':'::':'::: '•'::-:':::::': :.:.i, .............................. :: i: 101 102 i_iii!!iii: 103 104 105 ' 10 electrondensity(cm-3) Figure2, Threenighttimeelectrondensityprofileswith differentE regioncharacteristics andonewithan Ez layer.The profilesare identicalbelow 90 km. ThisspecificEs layercreatesa strongattenuation maximumat 400 Hz. The depthandfrequency of thismaximum dependon the altitudeand maximumNe of the Es layer[Barr, 1977]. However,eventhe E regionvalley profilewithoutEs produces distinctattenuation maxima at approximately 250, 520, and750 Hz and attenuation minima at 400, 630, and 900 Hz. In both cases,a resonance phenomenon appearsto be responsible for the variations, asexpectedfroman ionosphere thatproduces two distinctreflections.An oversimplified but instructivemodelof thisprocess is asfollows.Let therebe two ionospheric reflectionaltitudeshi andh2, with h2 > hi. The regionbetweenthe reflectionaltitudeshasan index 1440 CUMMER AND INAN: E REGION REMOTE tion from lightningdischarges.For the remainderof the paperwe focusprimarilyon non-Esionospheres. • 1'0I ..... •--no-valle• /\,, •- -valley .:.,. '• • SENSING AT ELF / ..... valley w/Es 4. QuantitativeE RegionRemoteSensing o. 0 ' 660 ' 880 lOOO ELF propagationdependenceon the E 26 frequency(Hz) o ß • -10 • novalley I',..,' -15/]......... ...... :"vnaø11• Iley I• / J -20 0 - valley w/EsI !ti.[ i i 200 i ¾ i i 400 600 frequency(Hz) i . i 800 "•• . J i 1000 Figure 4. QTEM modecharacteristics (phasevelocityand attenuationrate) computedwith LWPC for the threeprofilesin Figure2. The differencesdemonstrate the importanceof the E regionon ELF propagation. of refractionnh >> 1, and the regionbelow h] is free space.For any angleof incidencebelowh l the refracted wave aboveh] will have a nearlyverticalwave normal anglebecausenh >> 1. Thus the phasedifferencebetweenthe two reflectedwavesin the free spaceregionis q•h= wnh(h2-- hl)/C, wherec is the speedof light. The two waveswill interferedestructively if q5 h = (2m - 1)zr and constructively if qbh= 2mzr. The nearlyconstant spacingbetweenadjacentminima and maxima (Af m 110) impliesthat nh(h2 -- hi) - c/(2Af) = 1360 km. Reasonable E region valley parameters (Ne= 500cm-3, v = 6000s-], andfb = 1.5MHz)giveanX-modeindexof refractionof --,20 at 500 Hz, givingh2 - h• m 68 km. This simpleestimateis quitecloseto the heightdifferencebetweenthe undersideof the D region(--,80 km) andthe top of the E regionvalley(,-, 160 km), suggesting that the double-reflectionprocessis responsiblefor the subionospheric ELF attenuationrate and phasevelocity variations.The processis similar with an Es layer that generatesan evenstrongersecondaryreflectionandreso- region ionosphereopensthepossibilityof remotelysensingthe electron densityin thisregion.As discussed in section1, the value of this ELF techniqueis that it is sensitiveto E regionvalley parametersthat are difficult to measureby othertechniques. However,we needto quantitatively understandthis sensitivityin orderto probethisregionwith ELF. A broadbandELF sourceis ideal for this application becausetheresonances canbe easilydetectedwith a completeELF spectrum.Naturehasprovidedus with sucha sourcein lightningdischarges.We usethe LWPC model to simulate,for a varietyof ionospheres, the ELF spectra of the signal radiatedby a lightningdischargeand received 2000km away. We assume thatthelightning discharge is an impulsive sourcefor frequencies up to 1 kHz, a valid assumption sincethe risetimesand fall timesof lightningcurrentareusuallylessthana fewhundredmicroseconds. Everylightning discharge generates quasi-static electricfieldsin the mesosphere that drive slowlyvaryingrelaxation electriccurrents thatgenerate theirownelectromagnetic fields[Greifinger andGreifinger, 1976]. Theserelaxation-drivenfields, which are not implicitlycalculated by modetheorysimulations, canbe significant andtherefore canaffectourassumption of an impulsivesource.Recentsimulations haveshown,how- .;' .... ß::..:!..5../:' . • ß. ;.!.-•.i.i'i.• 'valieY:"11': : ,,,.'"51:5 ' '::.. ß [ '?':' !'•a:iiey i--;(;•'::.D__r_e_g!gni":ill ! 5::'.... nance. The abovesimulationsshow that Es and ambient vari- ationsaffectnighttimesubionospheric ELF propagation. On the basisof the simplifiedmultiple reflectionanalysis, the observablequantitiesthat containinformation aboutthe E regionare the frequenciesand depthof the amplitudeminima and maxima. We explorehow these quantitiesdependon quantitativeE regioncharacteristics and whether these characteristics can be extracted from measurements of subionospheric broadbandELF radia- h;½-11'!'!'!'; i i i!;.i.i!-:;. ; 101 102 103 104 105 electrondensity(cm'3) Figure 5. Five-parameter ionospheric modelusedto calculate thesensitivity of ELF propagation to ionospheric variability. CUMMER AND INAN: E REGION ever, that theserelaxation-drivenfields are insignificant 700 km from the source above 50 Hz [Cummer, 2000]. Thus,for our propagationdistances(> 1000 km) andfrequencies(>50 Hz), the lightning sourcecan be treated as impulsive.Furtherdetailson modelingELF radiation from lightningare givenby Cummerand lnan [2000]. Figure5 showsa five-parameterionospherewhichis a slightlysmoothedpiecewiseexponentialelectrondensity profilethat we useto quantitativelyinvestigatethe effects of ionospheric changeson theELF spectrum.The parametersHval, Emin,and Drnaxdescribethe heightof the E regionvalley,the minimumNe in the valley,andthe peak D regionelectrondensitybelow the valley,respectively. REMOTE SENSING AT ELF 1441 •, 4xld5 •• '<'•"...... "•?>-,,......... / [ Emin=200 / Emin=400 •,,,, Hval=125 •.......... Hva,=1! .....;• Dmax=1000 ,....• .fir'./, '- ................ Dmax=2000 The D regionparameters h' andfl arethealtitudeandinversescaleheight of an exponentiallyincreasingmodel electrondensityprofile originallydescribedby Waitand Spies[1964] and commonlyusedin D regionmodeling. The D regionelectrondensityprofileis givenby Ne(h) = 1.43 x 107exp(-0.15h')exp[(• - 0.15)(h- h')]cm-3. We simulatetheeffectof perturbations froma basicnight- ,.,:.."•... /.•./"- / and maxima. ,.. h=85 ............... I I I I ./'•,. timeionosphere (h' = 85km,fl = 0.5km-1, Hval= 130 km, Emin= 400 cm-3, andDmax= 2000cm-3) in all parameters individuallyto understand whichmightbe measurablefrom broadbandELF spectra. Figure 6 showsthe effect of variationsin the five ionosphericparametersdescribedaboveon the broadband ELF spectrumobserved2000 km away from the lightningdischarge(or any impulsivesource). The top panel showsthat reducingEminincreasesboth the frequencyand depth of the ELF resonances.This is precisely what is predictedby the approximateanalysisin section3. DecreasingEminreducesthe averageindex of refractionnh betweenthe two reflectinglayers,whichincreasesthe resonancefrequencies.DecreasingEminalso reducesthe attenuationof the fields which penetratethe firstreflectingaltitudeandreachthe secondreflectingaltitude, which increasesthe amplitudeof the secondreflectionandthereforeleadsto deeperinterferenceminima h'=82 '"-'•.•........ / 0 • 200 ................... 400 600 800 .... ----•-m ::-- .... 1•=0.4 • 1•=o.5 1000 1200 frequency(Hz) Figure 6. Variationin ELF spectrumwith the ionospheric parametersin Figure 5. increasesthe index of refraction nh betweenthe two re- flectionaltitudes,whichreducesthe resonance frequenciesjustasdoesincreasing Ah. The D regionaltitudeh' canbe inferredindependently usingVLF lightningpropagationmeasurements[Cummeret al., 1998], but the simi- larityof theeffectsof Hva•andDrnaxonELF propagation makes it difficult to measure both with this method. The lastpanelof Figure6 showsthatthe D regioninversescaleheightfi doesnot changethe resonancefrequencies,but doesstronglyaffectELF attenuationabove •,300 Hz. The D regioninversescaleheightfi canalsobe measuredfrom VLF observations[Cummeret al., 1998], but the strongELF dependence providesan independent andperhapsmoresensitivemeasurement of thisparame- The middlethreepanelsof Figure6 showthat variationsin Drnax,Hval,andh' all havevery similareffects on theELF spectrum.IncreasingHva•,increasingDrnax, anddecreasing h' all reducetheresonant ELF frequencies withoutchanging theresonance sharpness. All of these effectsarepredicted by theapproximate analysisin sec- ter. Givenhowionospheric parameters affecttheELF spection 3. Decreasingh' reducesthe firstreflectionaltitude hi, and increasingHva•increases the secondreflection trum, the followingproceduremay be usedto determine altitudeh2, which in both casesincreasesthe reflection the probableionospheregivenan ELF spectrum.AssumVLF altitude difference Ah = h2 - hi and thereforereduces ing that h' is alreadyknownfrom the broadband the resonance frequencies.IncreasingDrnaxeffectively spectrum[Cummer et al., 1998], select Emin to match 1442 CUMMER AND INAN: E REGION REMOTE SENSING AT ELF thedepthof the observed resonances, selecteitherDmax The generallygood agreementbetweenthe simulations or Hva]to matchthe resonancefrequencies,and select andobservations meansthatthe inferredionospheres do explaintheobservations, subjectto thenonuniqueness is/5 as neededto matchthe upperELF (500-1500 Hz) in section4. The ELF spectrum of thelast fieldamplitude.Thereis inherentnonuniqueness in the suesdiscussed deducedionosphere becauseof the similarityDmaxand time periodcouldnotbe reproducedwith anyreasonable parameterization in Figure5, Hva]. This resultis not surprising,sincewe are only variationof the ionospheric matching a three-parameter measurement (resonance frequency, resonance depth,andupperELF amplitude) to a .0 . ..• .... ://..'•i..... i.... !" Jul• 22, (•700-07S0 UT four-parameter model. To resolvethis,in section5, we 1.51 ß fix Dmax= 2000cm-3 andonlyvaryHva!tomatchthe 1.2 observations. -':•., '.•,,•.'•.' ' '--'i•','X"i ..... i" ":•'"...,.."i" '//' •- observed -..........modeled 0. 5. Comparisonof Observationsand Simulations 1.2 [ Lightning-generated broadband ELF spectrado show significant variabilitythatcanbeattributed toionospheric E regionvariations.We usenighttimehorizontalmagnetic field measurements recordedby the StanfordUni- versityELF-VLF Radiometer Experiment[Chrissan and Fraser-Smith,1996] in July 1996. Usinga procedureessentiallyidenticalto that of Cummeret al. [1998], we temporallyaveragetransientradioatmospherics (sferics) launchedby lightningdischarges in a smallgeographic region(asmeasured by theNationalLightningDetection [ 3 o.} •1 • •' [- '"i .... ". ;'-'' "'/'•'•'-i .... '"// -'•' •' "• 1'2I F - 0.8 ?" ?" IJuly 26,0400-0430 UT ---- observed iii'•'";: ""i'"I ........... mødeled I Network [Cumminset al., 1998]) in a 30 min time period, whichis usuallylongenoughto accumulate enough sfericsfor a goodsignal-to-noise ratio. We implicitlyas0.4 .... sumethatthe ionospheredoesnot vary significantlyover 0 thisperiod,andwe do notincludewaveforms fromlarge 0 400 800 1200 1600 positivedischarges thattendto havesignificant continua. frequency (Hz) ing currents[Uman, 1987, p. 200] andare thereforenot an impulsiveELF source.The ELF portionof the spec200 trum of the resultingaveragewaveform,whichincludes .... July 24,0400-0430 UT 180 the effect of a single-pole,high-passfilter at 420 Hz in ......... July 26,0400-0430 UTI -the receiver,is the measuredELF spectrum. 160 The solidlinesin Figure7a are measuredELF spectra from July22 (0700-0730 UT), July24 (0400-0430 UT), and July 26 (0400-0430 UT and 0845-0915 UT). The •140 propagation distancefrom sourceto receiverfor these 120 spectra is •2000 km. Themeanamplitudes differslightly betweenperiodsbecausethe averagelightningdischarge lOO ' ß ::'i'i'i:i - '":':' : '":':' x ':'i'!':' strengthis differentin eachperiod. The detailsof the spectra,includingthe resonance frequencies, vary more 8o thancouldbe causedby just the simultaneous D region variationsmeasured usingthetechniqueof Cummeret al. [1998], whichwe appliedto the simultaneous VLF data. b. electron density (cm '3) This indicatessignificantE regionchangesbetweenthe four periods.For the firstthreetime periodsthe dashed Figure 7. (a) ObservedbroadbandELF spectraand, in three linesin Figure7a aremodelELF spectrafoundby using cases,model ELF spectrathat agreewell. (b) The threeionothe iterativematchingproceduredescribedin section4. spheresinferredfrom the observations. ... CUMMER AND INAN: E REGION REMOTE SENSING AT ELF 1443 servations.In general,the agreementwas good, and we foundthatin mostcasesthevariationsin thespectracould behighlyvariable,andmatchingall of thedatarequiresa be explainedby normalE regionvariability. The utility of the single propagationpath technique broaderionospheric parameterization. that we havedemonstrated can be extendedsignificantly Figure7b showsthe three D and E regionelectron with multiple receivers and multiple source locations. densityprofilesthatreproduced the observations in the Since only a single mode propagates in the frequency first threetime periods. The ionospheric variabilityis significant but not extraordinary, highlighting the sensi- range of interest,a WKB-type solutionis valid in the tivity of ELF spectrato the D andE regionionosphere. presenceof evenrelativelysharp(with respectto waveThusthe Thoughnoneof theobserved spectracontained evidence length)horizontalionosphericinhomogeneities. of Es, the theoreticallyhigh sensitivityshownin section observedELF spectrumis essentiallythat which would ionospherethat is the 3 shouldmakeEs layerseasilydetectable with thistech- be producedby a homogeneous meanionospherealongthe entirepath. With multiplerenique. It shouldbe mentionedthat the valuesof/5 extracted ceiversand sourcelocations,suchpath-integratedmeafromtheVLF technique of Cummeret al. [1998]andthe surementscan be invertedusingtomographictechniques andit doesnot havean unusuallysharpresonance indicative of an Es layer. Clearly,the nighttimeE regioncan ELF techniquedescribed here,bothof whichassumea singlescaleheightbelow•90 km, differby asmuchas to deduce the two-dimensional horizontal structure of the ionosphere.Sucha techniquewouldbe especiallyuseful in trackingthe size and motionof largeinhomogeneities 0.15km-1. TheactualD regionionosphere mayhave like sporadicE. multiplescaleheightsandmay,for example,containa sharpledgeasdoestheprofileof Hale [1994].TheELF and VLF techniques are probablysensitiveto the scale References heightfor differentaltituderanges,whichmay explain Bannister,P.R., Variationsin extremelylow frequencypropagathis discrepancy. 6. Summaryand Conclusions tion parameters,J. Atmos.Terr.Phys.,37, 1203-1210, 1975. Barr,R., The effectof sporadic-Eon thenocturnalpropagation of ELF radio waves, J. Atmos. Terr. Phys., 39, 1379-1387, 1977. Althoughtheionospheric reflectionaltitudeof electro- Bilitza, D., International Reference Ionosphere--Status 1995/96,Adv.SpaceRes.,20(9), 1751-1754, 1997. magnetic wavesgenerally decreases withdecreasing fre- Budden,K. G., The Propagationof Radio Waves,Cambridge quency, lowerlosses andsharper gradients in theindexof Univ. Press, New York, 1985. refraction cangenerate E-F regionreflections atELF that Chrissan, D. A., and A. C. Fraser-Smith, Seasonal variations of globallymeasuredELF/VLF radio noise,Radio Sci., 31, donotoccurfor VLF. Thishigh-altitude reflectionmakes 1141-1152, 1996. subionospheric ELF propagation dependondetailsof the Cummer,S. A., Modeling electromagnetic propagationin the ionospheric E regionandtherebyprovidesa technique Earth-ionosphere waveguide,IEEE Trans.AntennasPropag., for remotelysensingthis region. Usingthe full wave in press,2000. LWPC model, we showedthat signatures of both spo- Cummer, S. A., and U.S. Inan, Modeling ELF radio atmosphericpropagationand extractinglightningcurrentsfrom radicE andordinarynighttimeE regionvariabilityare ELF observations,Radio Sci., 35, 385-394, 2000. present in broadband spectra of subionospherically prop- agated ELF waves.Resonances fromtheinterference of multipleionospheric reflections createamplitude minima Cummer,S. A., U.S. Inan,andT. F. Bell, IonosphericD region remotesensingusingVLF radioatmospherics, RadioSci.,33, 1781-1792, 1998. and maximawhosefrequencyand depthare controlled Cummins, K. L., E. P. Krider, and M.D. Malone, The US NationalLightningDetectionNetwork and applicationsof by differentparameters of a five-parameter modelionocloud-to-groundlightning data by electric power utilities, sphere.Threeof theseparameters produceverysimilar IEEE Trans.Electromagn.Compat.,40, 465-480, 1998. effectsin thesimulatedELF spectra(oneof whichcanbe Greifinger,C., and P. Greifinger,TransientULF electricand measured independently), leavingthreeparameters that magneticfieldsfollowinga lightningdischarge,J. Geophys. canbe measuredeffectivelywith thistechnique. Res., 81, 2237-2247, 1976. To demonstrate the technique,we analyzedmeasured Hale, L. C., Couplingof ELF/ULF energyfrom lightningand MeV particlesto the middle atmosphere,ionosphere,and ELF spectraproducedby lightningdischarges, a natu- globalcircuit,J. Geophys.Res.,99, 21,089-21,096, 1994. Hughes,H. G., andR. J. Gellenberger, Propagation of extremely pathon3 different daysin 1996.Averaging multiplesferlow-frequency(ELF) atmospherics over a mixedday-night icsto generate cleanbroadband ELF spectra, we useditpath,J. Atmos.Terr.Phys.,36, 1643-1661, 1974. ral broadbandELF source,on a 2000 km source-receiver erativesimulations to find the ionospheres thatfit theob- Johnson,M.P., U.S. Inan, S. J. Lev-Tov, and T. F. Bell, Scat- 1444 CUMMER AND INAN: E REGION REMOTE SENSING AT ELF teringpatternof lightning-induced ionospheric disturbances Rishbeth,H., and O.K. Garriott,Introductionto Ionospheric Physics,Academic,SanDiego, Calif., 1969. associated with early/fastVLF events,Geophys. Res.Lett., 26, 2363-2366, 1999. Sechrist,C. F., Jr.,Comparisons of techniques for measurement Mechtly,E. A., S. A. Bowhill,L. G. Smith,andH. W. Knoebel, of D-regionelectrondensities, RadioSci.,9, 137-149, 1974. Lower ionosphereelectronconcentration and collisionfreUman,M. A., TheLightningDischarge,Academic,SanDiego, Calif., 1987. quencyfromrocketmeasurements of Faradayrotation,differentialabsorption, andprobecurrent,J. Geophys. Res.,72, Wait, J. R., and K. P. Spies, Characteristicsof the Earth5239-5245, 1967. Morfitt,D. G., andC. H. Shellman, MODESRCH,animproved computerprogramfor obtainingELF/VLF/LF mode con- stantsin anEarth-ionosphere waveguide, InterimRep.77T, Nay.Electr.Lab.Cent.,SanDiego,Calif., 1976. ionospherewaveguidefor VLF radio waves,NBS Tech.Note 300, 1964. Watt, A.D., VLF RadioEngineering,Pergamon,Tarrytown,N. Y., 1967. Pappert,R. A., and J. A. Ferguson,VLF/LF modeconversion model calculations for air to air transmission in the earth- ionospherewaveguide,Radio Sci.,21,551-558, 1986. Pappert, R. A., andW. F. Moler,Propagation theoryandcalculationsat lowerextremelylow frequencies (ELF), IEEE Trans. Commun.,22, 438-451, 1974. Pappert,R. A., andW. F. Moler, A theoreticalstudyof ELF S. A. Cummer,Electricaland ComputerEngineering Department, Duke University, Durham, NC 27708. ([email protected]) U.S. Inan, STAR Laboratory, StanfordUniversity,Durand 324, Stanford, CA 94305-9515. normalmodereflection andabsorption produced by nighttimeionospheres, J.Atmos.Terr.Phys.,40, 1031-1045,1978. Ratcliffe,J. A., TheMagneto-lonic Theory,Cambridge Univ. (Received February12,2000; revisedAugust23, 2000; Press,New York, 1959. acceptedAugust24, 2000.)
© Copyright 2026 Paperzz