Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO.

A7, PAGES 14,315-14,329, JULY 1, 1997

ModelingtheF layerduringspecificgeomagnetic
storms
M. V. Codrescu,
T.J.Fuller-Rowell,
andI.S. Kutiev'
Cooperative
Institute
forResearch
inEnvironmental
Sciences,
University
ofColorado,
andNOAA,
SpaceEnviromnent
Center,Boulder,Colorado

Abstract. Important progress


hasbeenmaderecentlyin developing
anunderstand-
ingof theeffectsof geomagnetic
stormsin thethermosphere
andionosphere.
Numer-
ical simulations
of theoreticalstormswith the coupledthermosphere
ionosphere
model
(CTIM) haveprovided a betterunderstandingof thedynamics of theupperatmosphere
andhavealsopermitted theidentification
of theprocesses responsible for globalstorm
effectsat highlatitude
andmidlatitude. Thetheorydeveloped basedonthemodelsim-
ulationscanexplainmostof theapparent coherence of localtimeandseasonal depen-
dencies andtheapparent randomness in thelongitudinalresponse of theglobaliono-
sphere,uncovered through statistical
analysisof stormobservations. A truetestof the
modelandthetheoryis theirabilityto predictthelarge-scale distributionof stormef-
fectsfor specific
storms.In thispaper,CTIM simulation resultsfor theDecember 7-
9, 1982,periodarepresented. Wecompare modelresults withDE 2 temperature and
plasma
density
data.Wealsocompare
modeled
electron
densities
withionosonde
data
from severalsectorsin bothhemispheres.
The globalcharacteristics
of the response
arereproduced
by themodel,andweareableto explain
thepronounced
longitude
dif-
ferences
in thesummer
hemisphere.
The Australian
sectorpasses
throughmidnight
duringthemaindrivenphase
of thestormandexperiences
thelargest
energy
inputand
thelargest
neutralcomposition
changes.
Thedeepest
ionospheric
negative
phaseseen
in ionosondedatais over Australiaandis consistent
with this interpretation.Given the
largeuncertainties
in ourknowledgeof themagnitude
andspatial distribution
of energy
inputduringa particular
storm,predicting
localchanges
is stilla challenge.

1. Introduction a slightlydeepernegativephase(a decreasein maximum


electrondensityof the F2 layer,nrnF2) in the samelocal
Studiesof ionospheric and thermospheric stormshave timezonesthatextendsa little furtherequatorward.
encountered considerabledifficultiesin the pastdue to the
In contrastto the globalresponse, the localresponseis
apparently chaoticdistributionof globalstormeffects.The "nonlinear." Evensmallchanges in inputfor a givencom-
apparent randomness in the distribution of stormeffects binationmay producedrasticeffectsat somelocations.For
makesit difficultto comparestormsand to uncoverthe our exampleabove, althoughthe global characterof the
responsible mechanisms. Withoutaccepted mechanisms, it response is not altered,therewill be locationsthatchange
is difficultto developa crediblestormscenario. from a positive(increases in nrnF2) to a negativeiono-
On a globalscale,theeffectsof stormsseemto depend sphericstormandviceversa,astheboundarybetweenup-
on manyparameters: thelevelof energyinputandits spa- wellinganddownwellingmovesequatorward or poleward.
tial distribution
andtemporalevolution;season; preexisting The nonlinearlocal effectsaccountfor the poor correla-
conditions;and universaltime (UT) startof the storm. Ul- tion betweensetsof inputsand stormeffectsat individual
traviolet(UV) andextremeultraviolet(EUV) solarradiation locations, and between storm effects at different locations.
and tidal andgravitywavespropagating from belowmay Mechanismsthat accountfor ionosphericstormeffects
alsoplay a role. have beenproposedbefore [e.g., Pr61ss,1993, Rishbeth,
The relativeimportanceof theseparametersin deter- 1986, andreferences therein].They are basedon the close
miningtheglobalcharacter of theresponse variesfor dif- couplingbetweenthe ionosphereand the neutral atmo-
ferentcombinations.For a given combination,the global sphereandaredifficultto verify usingdata,asglobalmea-
response is essentially
"linear,"meaningthatsmallchanges surementsof neutralcomposition,neutralwinds, electric
in theinputwill producesmallandproportional changes in fields, and plasma densitiesare not available simultane-
theglobalpicture.A smallincrease in theenergyinput,for ously.
example,withall otherparameters maintained, will produce The developmentof coupledthermosphere ionosphere
generalcirculationmodelsmakesit possibleto performnu-
mericalexperiments to assess
therelativeimportance of the
•Nowat Geophysical
ln•itute, Sofia.Bulgaria differentprocessesinvolvedin ionospheric storms.Using
one suchmodel, namelythe coupledthermosphere iono-
Copyright1997by theAmericanGeophysical
Union.
spheremodel (CTIM), Fuller-Rowell et al. [1994, 1996a]
performedglobalsimulations of genericstormsstartingat
Papernumber97JA00638. differentUT's andduringdifferentseasons.Analysisof the
0148-0227/97/97JA-00638 $09.00 simulationresultsallowedthe formulationof a new theory
14,315
14,316 CODRESCU
ET AL.: F LAYERMODELING

thatcanexplainthe diurnal,longitudinal,andseasonal vari- The key to the new understanding is the formationand
ationsin globalionosphericstormeffects. evolutionof the neutral "compositionbulge," an area of
Figure1 illustrates the globalresponse of the ionosphere increased mean molecular mass under the influence of neu-
to the December7, 1982, storm,as describedby the six re- tral winds. The bulge is producedby the divergenthori-
gional averagesobtainedfrom midlatitudeionosondedata. zontalwind field over regionsof increasedheating,during
The stormto monthlymedianratiosof nmF•, for threelon- the drivenphaseof the storm,andsubsequently movesun-
gitudesectorsfor thenorthernandsouthernhemispheres are der the influenceof bothstormandbackground circulation
presented.Theseregionalaverageswere computedin an (diurnaland seasonal).It is the movementof the bulge
effort to minimize the local effects. The ionosonde stations [Fuller-Rowell et al., 1994, 1996a] that accountsfor the
usedfor the computationof theregionalaveragesare shown diurnaland seasonalvariationsseenin stormeffects[e.g.,
in Tables 1 to 6. Notice the predominantlypositivestorm Rodger et al., 1989].
effectsin the winter hemisphere(north) and negativeef- In thispaperwe testthe new theoryby comparingsimu-
fectsin the summerhemisphere (south)aswell asthe large lation results with DE 2 measurements and with ionosonde
longitudinaldifferencesin the southernhemisphere.We data, for the storm of December 7, 1982. The model and
interpretthe differences basedon the new understanding of the simulationsare briefly describedin section2, the data
the distribution of storm effects. in section3, and the geomagneticconditionsfor the inter-

NmF2 Ratios for Europe NmF2 Ratios for S. Africa

3 .4

.3

2.5 .2
.1

2 1

.9

1.5 .8
.7

1 .6

.5

0.5
0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24

NmF2 Ratios for Northeast Asia NmF2 Ratios for Australia

2.5

1.5

0.5 .......................
0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24

NmF2 Ratios for America NmF2 Ratios for S. America

.4

2.5

.9
1.5 .8
.7
.6
.5

0.5 ::::::::::::::::::::::: ß4 : : : : : ; ', ', ; ; ; ; : '. : : : :


0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24

UT from 12 UT 07 December 1982 [h] UT from 12 UT 07 December 1982 [h]

Figure 1. The globalresponseof the ionosphereto the December7-8, 1982, storm,as illustratedby
the six regionalaveragesobtainedfrom midlatitudeionosondedata. The ratiosof stormto monthly
mediannmF2 are plotted. Notice the hemisphericdifferencesand the longitudinalasymmetriesin the
southern(summer)hemisphere.
CODRESCU
ET AL.: /v'LAYERMODELING 14,317

Table1. Ionosonde
Stations
Usedto Compute
theAustralian
Regional Table3. Ionosonde
Stations
Usedto Compute
theNorthAmerican
StormResponse Regional
StormResponse

Station code LatitudeLongitude


Dlat Station Code LatitudeLongitude
Dlat
Brisbane BR52P -27.5 152.9 -35.4 Bermuda BJJ32 32.2 -64.5 43.5
Camden CN53L -34.0 150.7 -42.4 Boulder BC840 40.0 -105.3 48.8
Canbera CB53N -35.3 149.0 -43.7 CapeKennedy CC929 28.4 -80.7 39.5
Christchurch GH64L -43.4 172.3 -47.7 Fort Monmouth FMJ40 40.4 -74.1 51.7
Hobart HO54K -42.9 147.3 -51.4 Grand Bahama GB926 26.6 -78.2 37.8
Mundaring MU43K -32.0 116.2 -43.2 PortArguello PA836 35.6 -120.6 42.3
Norfolk NI63! -29.0 168.0 -34.5 San Salvador SSJ25 24.1 -74.5 35.4
Stanford ST837 37.4 -122.2 43.8

WallopsIs WP937 37.9 -75.5 49.2


val in section4. The temperature
comparison
is presented
insection
5,theneutral
composition
isdiscussed
insection Washington
WA938
38.7 -77.149.9
6, therim?2comparisons
aredescribed
in section
7, and White
SandsWS83232.3 -106.541.1
our conclusions
are presentedin section8.
normalmodeof operation,and offersthe bestmatchingof
2. Model Simulations steadystateconditions
asdescribed
by themassspectrom-
eter and incoherentscatter (MSIS) [IIedin, 1987] neutral
The coupledthermosphere ionospheremodel has been atmospheric model.
presentedin previouspublicationsby Fuller-Rowell and The December7-9, 1982, period was first simulatedin
Rees[1980, 1983];and Fuller-Rowell et al. [1987, 1996b] an operationalmode. The sequenceof statisticalE field and
and usedextensivelyover the last 15 years. precipitationpatternswas specified,basedon actual mea-
The magnetospheric inputto the modelcanbe specified surements for the period,by an automatedprocessusing
in severalways. We normallyusetime sequences of sta- the SpaceEnvironmentCenterDatabaseof TIROS/NOAA
tisticalpatternsof auroralprecipitationand electricfields satellitedata. The crosspolar cap potentialwas derived
describedby Fuller-Rowell and Evans [1987] and Foster from solar wind data using the formula by Reiff et al.
et al. [1986], respectively.The hemispheric powerindex [ 1981] andimposedas an additionalinputto scalethe Fos-
(P 1), derivedfrom TIROS/NOAA auroralparticlemeasure- ter et al. [1986] E field patterns.Simulationsusingan
ments[Evans et al., 1988] for December6-9, 1982, is used artificialdoublingof theJouleheatingwerealsoperformed
to prescribethe time sequenceof statisticalinputpatterns to ascertainthe dependenceof storm effects on the total
usedin the simulations presentedin thispaper. energyinput.
The statisticalpatternsused as input are somewhat Simulationswere then performedusing a substormE
broadenedin latitudecomparedwith instantaneous obser- field distribution[Kamide e! al., 1996], insteadof the usual
vationsof auroralprecipitationand E fields. The use of FosterE field patterns.However,the changeto the sub-
the statisticalpatternsalso meansusing averageE field stormE field patternwas imposedonly whenthe activity
valuesfor the computation of Jouleheatingand may re- level was greaterthan 7. This allowed an assessment of
ducethe magnetospheric energydeposited at highlatitudes the sensitivityof stormeffectsto the spatialdistributionof
[ Codrescuet al., 1995]. For this reasons,the electricfields energyinput. The sametime historyof crosspolar cap
usedin the simulations describedin thispaperareincreased potentialdrop was usedfor thesesimulations.
abovethe originalmodel valuesof Foster et al. [1986], Plate 1 showsthe height-integrated Jouleheatingdistri-
by 30 percent.Furthermore, the winterhemisphere (north- butionsfor the Foster model (Plate l a), and for the sub-
ern hemisphere in this study)Jouleheatinggenerationis stormpattern(Plate lb), for the sameUT. The changeof
increasedby a factorof 2 for activitylevel 5, by 20 per- patternproducednot only a spatialredistributionof Joule
centfor activitylevel 10, andlinearlybetween.This is our heatingbut alsoan increasein the totalenergydissipated in

Table2. Ionosonde
Stations
Usedto ComputetheEuropean
Regional Table4. Ionosonde
Stations
Usedto Compute
theNortheast
Asian
StormResponse
RegionalStormResponse

Station Code Latitude LongitudeDlat


i
Station Code LatitudeLongitude
Dlat
Dourbes DB049 50.1 4.6 51.7 Akita AK539 39.7 -64.5 43.5
Juliusruh JR055 54.6 13.4 54.2 Beijing BP440 40.0 116.3 28.7
Kaliningrad KL154 54.7 20.6 53.0 Chita CX452 52.0 113.5 40.7
Kiev KV 151 50.5 30.5 47.1 Khabarovsk KB548 48.5 135.1 38.1
Lannion LN047 48.7 -3.4 52.0 Kokubunji 25535 35.7 139.5 25.7
Leningrad LD 160 60.0 30.7 56.0 Petropavlovsk PK553 53.0 158.6 44.9
Poitiers PT046 46.6 0.4 49.2 Seul SU437 37.2 126.6 26.3
Rome RO041 41.9 12.5 42.3 Wakkanai Wk545 45.4 141.7 35.5
Uppsala UP158 59.8 17.6 58.3 Yamagawa YG431 31.2 130.6 20.6
14,318 CODRESCUET AL.: F LAYERMODELING

Table 5. IonosondeStationsUsed to Computethe SouthAmerican electrondensitycomparisons.Error analysisor approxi-


RegionalStormResponse mateaccuracyfor the RPA instrumentwasnot presentedin
the preflightinstrumentdescription[Hanson el al., 1981].
Station Code Latitude LongitudeDlat
Ionosondedata (foF2) for the Australianstationsshown
in Table1 wereutilizedto computetheAustralianregional
ArgentinaIs AIJ6N -65.2 -64.3 -54.0
averages.The stationspresentedin Table 2 were usedfor
Conception CPJ30 -36.8 -73.1 25.5 the Europeansector. The stationsshownin Tables 3 to
Port Stanley PSJ5J -51.7 -57.8 40.6 6 wereusedin the computation of the respectiveregional
SouthGeorgia SGA5M -57.3 -36.5 -44.5 averages. All ionosondedata was obtainedfrom World
Trelew TwJ4L -43.2 -65.3 -31.9 Data Center A, Boulder, Colorado.

the thermosphere-ionosphere althoughthe samecrosspolar


4. Geomagnetic Conditions
cappotentialwas used.
Simulationswhere the Jouleheatingwas doubledwhile The geomagnetic indicesfor theperiodof December6-9,
usingthe samesubstormpatternwere also performedfor 1982arepresentedin Figure2. December6 is thereference
completeness. quiettime for the model. The day wasvery quiet,andclose
The model does not include dynamo electric field in- to the monthlymedianvaluesfor nmF2. The averageKp
fluences. The absenceof dynamofields is not expected was less than 1, averageA E under 50 and the average
to changethe characterof the globalresponseat high and hemispheric power input was closeto 10 GW.
middle latitudeswhere winds and compositiondriven by Major geomagneticactivity occurredafter 1200 UT on
magnetospheric inputsdominate.However,at low latitudes December7; Kp reached6*, AE spikedabove700 nT,
the electricfields are by far the most dominantfactor for andthe precipitatingparticlepowerpeakwas in excessof
the ionosphere. 150 GW. A partial recoveryfollowed toward the end of
No interhemisphericfluxes are explicitly included in the day and duringthe beginningof the next, with activity
the model, but the ionosphericcalculationsare extended occurringagain around 0800 UT on December8. This
to 10,000 km along the field line, forming a reservoirof time, the maximum K p value was lower (5-), the A E
plasma,andallowinga morerealisticplasmaflux thanany valueswere generallyhigher (peak > 1000 nT), and the
top boundaryconditionwe couldimpose.The modeldoes powerinput was similarto the previousday. December9
not includepenetrationelectricfieldsdue to the breakdown wasratherquiet, with one spikeof activitystartingat 1200
of thehigh-latitudeshieldingduringstorms.This is because [IT andlastingabout5 hours.The peakKp value(5*) was
the morphologyof the penetrationfieldsis not well known, betweenthe maximumvaluesfor the previous2 days,A E
and their relative shortdurationis not expectedto change peakis higherthan both, and the poweris lower.
the characterof the globalresponseof the ionosphere. It is obvious that the indices presentedhere reflect
In the current version of CTIM, the self-consistentiono- differentcombinationsof processesin the thermosphere-
sphereis computed polewardof 230 latitudein bothhemi- ionosphere-magnetosphere system. It is not clear at this
spheres.The empiricalmodelof Chiu [1975] is still used time, however, which index or combination would best be
at equatoriallatitudes.For thisreasonswe do not discuss suitedfor orderingionosphericstormeffects.
stormeffectsat low latitudesin this paper.

3. Data 5. Neutral Temperature


The thermospheric neutraltemperaturestructureis deter-
The temperatureandplasmadensitymeasurements used
minedby a seriesof heatingand coolingprocesses, where
for modelcomparisons arefrom the DE 2 UnifiedAbstract
bothlocal andremotedynamicalandchemicalmechanisms
Files at the NationalSpaceScienceData Center(NSSDC).
are important. Feedbackloops connectmost of the pro-
The wind and temperaturespectrometer(WATS) mea-
cessesin nonlinearways. Locally, the temperatureis the
suredneutraltemperature[Spencerel al., 1981] is used
resultof a time integralover heatingand coolingratesthat
for the temperaturecomparison.Note that the temperature
makesit a smoothfunction,and a prime candidatefor en-
accuracy,statedbeforeflight,for this instrumentis approx-
ergy inputdiagnostics.
imately5øK [Spencerel al., 1981] and thereforethe error
The primary driver of temperaturevariationsduring
bars are too small to be representedon the temperature
stormsis the magnetospheric energyinput. Jouleheating,
comparisonplots.
particleprecipitationheating, and viscousdissipationare
Plasmadensitymeasurements from the retardedpoten-
the main heating mechanisms. Vertical heat conduction,
tial analyzer (RPA) on DE 2 are used for orbit by orbit
atmospheric expansion,stormcirculation,andNO infrared
radiationare the main coolingmechanisms.
Table 6. IonosondeStationsUsed to Computethe South African The spatialdistributionof stormheatingdependson the
RegionalStormResponse global distributionof electricpotential,particleprecipita-
tion, and conductivity.In the caseof our numericalsim-
Station Code Latitude LongitudeDlat ulations,the distributionsof electricpotentialand particle
Capetown CT13M -34.1 18.3 -33.1
precipitation are inputsbasedon statisticalpatterns.The
conductivity,althoughcomputedself-consistently, is dom-
JohanesburgJO120 -26.1 28.1 -27.2
La Reunion LR22J -21.1 55.8 -26.0
inated at high-latitudes
by the statistical
patternof particle
precipitation.
CODRESCUET AL.: F LAYER MODELING !4,3 ! 9
14,320 CODRESCU
ETAL.: F LAYERMODELING
CODRESCUET AL.: /7 LAYER MODELING 14,321

.• •:/' '.'•:: ..3 ..

,.
fNI
ß . •.•,.•. ..... ... ,

........... .... . : ."• .. •..,•


'? • . • : ., --.• ..:

• ..--"' ...'..•'•E.'. "---... •' .-• •


• • . .... .' . , •" ..•'

• -.•.- • • ß •. •• :. ...
..•.:-'
• • ,. • • , :: .

• • ,. • •. .-,-'. .•....
I • ' -• L, ,.... ...:...
• ..... .
14,322 CODRESCUET AL.: F LAYER MODELING

+ d- + + -+ + + •

•D
CODRESCUET AL.: 17LAYER MODELING 14,323

7-- Kp a

6•

5•

4 --

0 I • • • I • • • I I I I I I
0 24 48 72 96

200 -- " ' Power Index

100 --

0 24 48 72 96

Hours from 00 UT on 6 Dec 1982

Figure2. Geomagnetic
indices
for December
6-9, 1982,versus
UT. (a) Kp as a barchart. (b)
Hemispheric
power(GW) fromNOAA/TIROS satelliteseries(thin line) andA/• (nT) (thickline). The
AE valueshave beendividedby 10 to allow plottingon the samegraph.

The statisticalpatternsare not necessarily


truerepresen- heightat the satelliteare presentedin the abscissafor each
tationsfor a particularstorm[t•vans, 1987], and their use orbit. The approximatemid-time of eachorbit is indicated
introduces the largestuncertaintyin the magnitudeanddis- in Figure 2 by a vertical line markedby the appropriate
tributionof ionosphere-thermosphere energyinput. Some orbit number.The polar orbit of the satellitesampledthe
of theuncertainty,
duringspecific
storms,
canberemoved 5.7 LT sector on its northward transit and the 17.6 LT sector
by adjustingthe modelinputsto matchmeasuredtempera- as it movedsouthward,at a heightthat variedbetween500
tureson a globalscale.This is true only if the heatingand and 260 km. The plottedtemperatures are at the satellite
coolingprocesses are appropriatelymodeled. height.
Figure3 showstemperature comparisons betweenDE 2 The DE 2 measured temperatures
arerepresented by the
measurements and CTIM simulations for four consecutive solid line while the simulatedtemperaturesfor the three
orbits on December 7, 1982. The UT, invariant latitude, and forcingsused,are shownby the dot/dashedlines. The one
14,324 CODRESCUET AL.: F LAYER MODELING

2OOO i i i i

1800

16OO

.
., ! .

p
•' ' •• ,.• L!
I' ! •a•'
.,s,'.," .--I
..--' ..... ' •
•,.•
i •t ..
r....----
-'
/ •.
.•"
,.......-.
1200
.•,i '-•_ II

1000

14.1• i ......... 14.188 14.243DE2


i ......... • [hi 14.553 1.4,.147•7
14.1482
! .................. i ....... 15.849 15.944 16.059Dœ2
UT[hi 16.228 16.132516.418
,, i i , i 51.27 72.14 79.41InvLot
i • , •[de9] 5•(91 8?7 21(47
I • • • .i , • • i , , , , , , • , ,
47•.20 460.g2 437.•0 Hei9ht[km] 392.55 56•.44 3'•.1g 410.90 569.2,5 527.41 Hei9ht[km] 266.22 262.45 282.75

2OOO

1800

1600

1200 1200 "•' •'- -

IOO0 IOO0

17;54 17•70 0œ2


• iUT
,[hi 18;05 18;19 18;56
i , , , , • , , , , , , , 19.70
i ..................19;79 19.89
i ......... DE2
UT [h] 20.07
i ......... i ......... 20.17
• ......... 20.26
i

i i , , , , • , , ß,,.,•,, ,".-", , ,'•.-",,,•.t,[•,.,,],


.';-", , ,,;..., , , ;.,o
5•.• 265.• Height[km] 527,95 410.• 476.28 386.21 457.62 480.28 Hei9ht[km] 511.74 502.99 •1.58

Figure 3. Temperature versusUT comparison betweenDE 2 (solidline) and CTIM usingdifferent


inputs(dashedlines). The onedoted-dashed line temperatures are fromthe operational
CTiM run,the
threedotted-dashed line temperatures
are from the artificialdoublingof Jouleheatingin the CTIM
run, and the dashedline temperatures are from the CTIM run that usedthe substormelectricfield
pattern[Kamide el al., 1996]for activitylevelsgreaterthan7. (a) Day 82341,orbit7419 (northern
hemisphere); (b) Day 82341,orbit7420;(c) Day 82341,orbit7421;(d) Day 82341,orbit7422(southern
hemisphere).

dotted-dashed
line temperatures
are fromthe operationaling generatedwhenthe substormE field patternis usedis
CTIM run, the threedoted-dashedline temperatures
are 50 to 70 percenthigherthanfor the operationalsimulation,
fromtheartificialdoubling
of Jouleheatingin theCTIM althoughthe samecrosspolar cap potentialwasapplied.
run, andthe dashedline temperatures are fromthe CTIM There is also considerablymore structurein the modeled
run thatusedthe substorm electricfieldpattern[Kamide temperatures in this case,whichis to be expected,given
et al., 1996]for activitylevelsgreaterthan7. thefactorof 4 increase in the peakJouleheating(Plate1).
We canconclude fromFigure3 thatthe inputfor the Thetemperature is overestimatedin theCTIM runusing
operational CTIM run (dashed line withonedot)doesnot the substormpatternE fields and an artificialfactorof 2
provide enough heating to matchtheDE 2 temperatures. for Jouleheatingin the energyequation(figurenot shown
DoublingtheJouleheatingby introducing an artificial here).
factorof 2 in thecorresponding termin theenergy equation The bestforcingfor this period seemsto be a combi-
(dashedlinewiththreedots)bringsthesimulated temper- nation of statistical and substorm distributions of E fields.
aturescloserto measurements, but doesnot improvethe It is not possibleto determinethe optimumcombination
matching of thetemperaturestructure
seenby DE 2. Dou- withourmodelandtheavailabletemperature data.The as-
blingtheJouleheating seemsto shiftsthetemperature
up similativemappingof ionospheric electrodynamics (AMIE)
ratheruniformly alongtheDE 2 orbit,suggesting
a linear procedure[Richmondand Karaide, 1988] couldbe usedto
relationship
between totalenergyinputandtemperaturefor specifythe inputs,however,AMIE resultsarenotavailable
thisorbitandforcingmode.NotethattheglobalJouleheat- for thisperiod,andfor therestof thispaperthemodelrun
CODRESCUET AL.: F LAYERMODELING 14,325

usingthe substormpatternE fields,which bestreproduces to theareas of increased mmm(theneutral-composition


measuredtemperatures,will be used for further compar- bulge),and positiveeffectsto regionsof decreasedmmm.
isons. Negativeionosphericstormeffectsdominatein the South-
ern, (summer)hemisphere,while positiveeffectsare more
prevalentin the Northern (winter) hemisphere(Plate 5).
6. Neutral Composition
The effectsare more pronouncedwhen the bulge is on the
Geomagneticstormsproducelarge changesin neutral daysidewhere both productionand lossare affectedby the
compositionwhich can be examinedat equal pressure, compositionchange.
equal height, or at hmF2. In this paper,in the absence Figure 4 illustratesthe comparisonbetweenDE 2 (solid
of suitablecompositionmeasurements, we show modeled line) andCTIM (dashedline) plasmadensitiesfor the same
differencesin mean molecularmass(mmm) on pressure four orbits of Figure 3. The modeledand measuredden-
level 12 (about300 km), betweenthe stormand a quiet, sitiesare at the sameneutralpressurelevel. Note that the
steadystaterun. The quiet run approximates the monthly plotsrepresentlocal conditionsand that relativelysmaller-
median ionosphericconditionsfor December 1982. Lati- rorsin modelinputscanproducelargediscrepancies in this
tudesabove10ø are shownin bothhemispheres in Plates comparison.
2-5, usinga scalefrom -0.6 to 2.4 amu. The plasmadensityis flat duringorbit 7419 (Figure4a).
The summerhemisphere(south)changesin neutralcom- This is explainedby the low power flux (Figure 2) and the
position,as simulatedby CTIM, are presentedin the left fact that the orbit is entirely on the night side. There is a
panelsin Plates2 to 4. The snapshots are at 15.6, 22.8, hint of the presenceof the auroraloval in the measurement
and 06.0 LIT, and illustrate the creation and evolution of around550 InvariantLatitudewhich is not reproducedby
the neutralcompositionbulge. Plate5 left panelshowsthe the model.
winter (north) hemisphereat 06.0 UT. Orbit 7420 showsmuch more structure. The position
In winter, the bulge developsearlier and has a more of the auroral oval at about 570 Invariant Latitude, on the
pronouncedpeak (Plate 5) but it occupiesa muchsmaller northbound partof the orbitis reproduced by the model,but
area (confinedto the polar regions)than in summer(Plate the magnitudeis not matched.On the southbound part of
4). This is becausethe seasonalcirculation (summer to the orbit,the magnitude is matched but the positionis off by
winter) preventsthe latitudinalpenetrationof the bulge in more than 50. The largestructureobservedbetween15.99
winter and helps it in summer. This effect was described and 16.06 UT is a polar cap featureand is not reproduced.
in detail by Fuller-Rowell et al. [1996a]. The rotationof The steepincreasesin plasmadensityafter 16.14 UT in
the bulgewith the Earth is evidentin Plates2 to 4. Space bothmodelanddataare causedby the terminatorcrossing.
limitationsdo not allow us to showthe rotationof the bulge The plasmadensityis againflat, andwell reproduced by
in the winter hemisphereor the equatorwardmovementat the modelduringorbit7421, with the exceptionof the polar
night and polewardby day, but the resultsare similar to capstructure above71o, whichis notmatchedby the model.
the those in Fuller-Rowell et al. [1996a]. Note also the The situationis similar during orbit 7422. The sudden
extent and the biggeramplitudeof the decreasesin mmm drop in modeleddensityat about20.15 UT is due to the
in winter comparedto summer(Plate 4 and 5). crossingof the boundaryof the self-consistent ionosphere
calculation.
The midlatitude agreementbetween the modeled and
observedplasmadensitiesis good while the matchingof
7. F 2 Electron Density
structurein the polar cap and auroralzone is poor. This is
The difficulty of explainingthe global distributionof consistent with the uncertainties associated with the use of
ionosphericstormeffectsis not in findinga possibleexpla- statisticalpatternsfor E field and particleprecipitation.
nation for the effectsof a particularstorm,but in finding Figure5 showsthe modelednmF2 stormto quietratios
the consistencybetweendifferent stormsand uncovering for the samesix regionalsectorsasFigure 1. Theseregional
the responsiblemechanisms. averageswere computedin an effort to minimize the local
The ionospheric plasmadensityis the resultof a dynamic nonlineareffects due to forcing uncertainties.The model
equilibriumof competingprocesses. Many mechanisms can resultsdisplay the same tendencytoward positive storm
affect electrondensitiesat F regionheightsduringstorms. effectsin winter and negativestormeffectsin summer,and
Fuller-Rowell et al., [1994, 1996a] extendedthe theory showpronouncedlongitudinaldifferences,especiallyin the
of Prb'lss [1993] and showed that the formation and evo- southernhemisphere.The trendspresentin the data(Figure
lution of the neutralcompositionbulge during a storm is 1) arewell reproduced by the modelbutthepeakamplitudes
sufficientto explain the local time and seasonaleffectsin and the variability are not matched.The amplitudesof the
the nmF2 negativephase.The magnitudeand distribution positivephasein winter are overestimated by the model
of the negativeionosphericphasedependson the magni- while the amplitudesof the negativephasein summerare
tude and positionof the neutralcompositionbulge, which underestimated.
is determinedby the time history and spatialstructureof In the data (Figure 1), the Australiansectorshowsthe
the high-latitudeinputs.This is not surprisingsinceneutral most pronouncednegativephaseionosphericstorm (over
compositionaffectsboth productionand loss mechanisms 60 percentdecreasein electrondensity)of all longitude
to producelarge effects. sectors. For the model runs, Australia is the sectormost ex-
The global changesin mean molecular mass (mmm), posedto the composition bulge(Plates2-4) andalsoshows
modeledby CTIM, arereflectedin the stormto quietnmF2 the largestnegativeionosphericstormeffectsof all sectors
simulatedratiospresented in the rightpanelson Plates2 to (Figure5). However,the magnitudeof the plasmadensity
5. Negativeionosphericstormeffectscorrespond roughly depletionis underestimated by the model.
14,326 CODRESCUET AL.: /• LAYER MODELING

.-- 20

E
20

.'-. 15 '
_>. '

• 10 [ 1o

a:
i

0 , I i

14.298 14.325 14.353 DE2LIT[11] 14.407 14.455 14.462 15.849 15.944 16.059 DE:2UT[hi 16.228 16.323 16.418
I ......... I ......... ! ......... I ......... , ......... I ......... I
52.91................
i .........
58.68
i .........
64.64
i .........
InvLot ,[d,

ecji
, , i .... .....
75.69
i .........
80.23,
i .........
,
85.00

51.27 72.14 79.41 InvLot[deg] 34.91 8.27 21.47

' . . 379.35
410.90 . ' . . 347.08
ß J , ,Height
, I[k•11]
, 287.g5
, . i , , ,269.31
I , , '2611.28

I ! I I I I
ß

• 20

lo
.,-#11
,.-' i

I I I I I o
7.58 17.40 17.42 OF.2UT[hi 17.46 17.48 17.50 19.7o 19.78 19.85 0[2 UT[hi 20.01 20.09 20.17
, .... I .... I , , , , I , , I , I , , , , I .... I i ......... ! ......... i ......... i ......... ! .........
4.•.77 49.70 54.84 InvLot[de(j] 64.24 69.78 75.10 4•.52 61.74 79.01 InvLot[de(J] 64.00 45.95 26.44
15.61..........................
405.94 396.17 '1•'
lit' il•i...............
578.22 ,...........
368.1 ' , . 428.57
586.21 . , . . 468.42
ß , . .Height
, I[k•]' 508.97
' ' ' ' $11.39
' I , , ,502.99

Figure4. PlasmadensityversusUT for DE 2 (solidline)andCTIM runusingthepeakepochsubstorm


patternfor electricfields(dashed
line),for thesame4 orbitsof Figure3. The comparison
is on equal
neutralpressurelevels.

In the caseof doublingthe substorm energyinput(not The largestuncertainty


in themodeldatacomparison is
shownhere),theamplitude of the negativephaseis closer causedby our impreciseknowledgeof the inputsfor this
to measurement but the ionospheredoes not start to re- period.Boththe magnitude andthe spatialdistributionof
cover before the next disturbance and the modeled values the JouleHeatingcalculated by the modelhave an uncer-
becometoolow duringandfollowingthenextburstof ac- taintyontheorderof 50 percent.Thisis theprobablecause
tivity.In thewinterhemisphere, doubling theJouleheating for theunderestimationof the negativeionospheric effects
increased theinitialrateof increase in electrondensitybut overAustraliaandtheoverestimation of thepositiveeffects
decreased the amplitudeof the peak. This is an illustration over Europe.
of nonlineareffectson regionalscales. The new understanding of the distribution
of geomag-
neticstormeffectsopensthepossibilityof ionosphericand
8. Conclusions spaceweather forecasts. However, a better characteriza-
tion of the magnitude, spatialdistribution and variability
Thefirstextensive
comparison betweenCTIM, andsatel- associated with the inputsusedfor forcingthermosphere-
lite and ionosondedata, for a specificstorm(December ionosphere modelsis necessary beforemoreprogress can
be- be made.
7, 1982), confirmsthe causalqualitativerelationship
tweenneutralcomposition andnegativeionospheric storm
effects(bulgetheory).The modelreproduces qualitatively Acknowledgments. Support of thefirsttwoauthors forthis
theglobaldistribution
of stormeffects,matching thelarge- workwasby NASA grantNAGW-3530to theUniversityof Colorado.
scalelongitudinaland seasonal dependencies observedin The Editor thanks J. H. A. Sobral and E. P. Szuszczewicz for their
the thermosphereandionosphere. assistance
in evaluatingthispaper.
CODRESCU
ET AL.: F LAYERMODELING 14,327
14,328 CODRESCUET AL.: F LAYER MODELING

NmF2 Ratios for Europe


,
NmF2 Ratios for S Africa
..... , ..... , ..... , .....

5.0
..... , ..... , ..... , .....

1.4

2.5 1.2

2.0 1.0

1.5 0.8

1.0
0.6

0.5 0.4
0 6 12 18 24 0 6 12 18 24

NmF2 Ratios for Northeast Asia NmF2 Ratios for Australia


5.0 1.4

2.5 1.2

2.0 1.0

1.5 0.8

1.0
0.6

0 6 12 18 24 0 6 12 18 24

NmF2 Ratios for North America NmF2 Ratios for S America


5.0 1.4 ' ' '

2,5 1.2

2.0
1o0 -- •..•.•
1.5 0.8

1.0
0,6

0.4 , , ,
0 6. 12 18 24 0 6 12 18 24
UT from 12 UT 07 December1982 [hi UT from 12 UT 07 December 1982 [h]

Figure 5. Sameas Figure 1 but for CTIM results.

References convection
associated
withdiscrete levelsof particleprecipitation,
GeophiIs.Res. œett., 18, 656-659,1986.
Fuller-Rowell,
T.J.,andD.S. Evans,Heightintegrated Pedersen and
Carignan,G.R. et al., The neutralmassspectrometer
on Dynamics Hall conductivity
patternsinferredfrom theTIROS-NOAA satellite
Explorer
B, SpaceSci. Instrum., $(•), 429,1981. data,J. GeophiIs. Res., 9œ,7606-7618,1987.
Chiu,T., An improved phenomenological modelof ionospheric
density, Fuller-Rowell,
T.J.,andD. Rees,A three-dimensional,time-dependent,
J. Atmos. Terr. PhiIs., 87, 1563-1570,1975. globalmodelof the thermosphere, J. Atmos. Sci., 0ø7,2545-
Codrescu, M.V., T.J.Fuller-Rowell,
andJ.C.Foster,On theimportance 2567, 1980.
of E-fieldvariabilityfor Jouleheatingin thehigh-latitude
thermo- Fuller-Rowell,T.J.,andD. Rees,Derivation of a conservative
equation
sphere,GeophiIs. Res. œett., œœ, 2393-2396,1995. for meanmolecular weightfor a twoconstituent gaswithina three-
Evans,D.S., T.J. Fuller-Rowell,S. Maeda,andJ. Foster,Specifica- dimensional,
.
time-dependentmodelof thethermosphere, Planet.
tionof the heatinputto the thermosphere frommagnetospheric SpaceSci., 81, 1209-1222,1983.
processes usingTIROS/NOAAauroralparticleobservations, Adv. Fuller-Rowell,T.J., D. Rees,S. Quegan,R.J. Moffett,andG.J. Bai-
Astronaut. Sci., 65, 1649-1667,1988. ley,Interactions
between neutralthermospheric composition
andthe
Evans,D.S., Globalstatistical patternsof auroralphenomena, in polarthermosphere usinga coupledglobalmodel,J. GeophiIs.
Proceedings of QuantitativeModelingof Mailnetøsphere- Res., •œ,7744-7748,1987.
IonosphereCouplinilProcesses, pp. 325-330, KyotoUniver- Fuller-Rowell, T.J.,M.V. Codrescu, R.J.Moffett,andS. Quegan, Re-
sity,Kyoto,Japan1987. sponse of thethermosphereandionosphere to geomagnetic
storms,
Foster,J.C.,J.M. Holt, R.G.Musgrove, andD.S. Evans,Ionospheric J. GeophiIs.Res., •, 3893-3914,1994.
CODR1}SCU
ETAL.:F LAYERMODELING 14,329

Fuller-Rowell, T.J., M.V. Codrescu,H. Rishbeth, R.J. Moffett, and S. Richmond,A.D., and Y. Kamide,Mappingelectrodynamic featuresof
Quegan,On the seasonal
responseof the thermosphere
and iono- thehigh-latitude
ionosphere
fromlocalizedobservations:
technique,
sphereto geomagnetic
storms,J. Geophgts.Res., 101, 2343- J. Geophgts.Res., 93, 5741-5759,1988.
2353, 1996a. Reiff, P.H.,R.W. Spiro,andT.W. Hill, Dependehce
of polar-cap
po-
Fuller-Rowell,T.J., D. Rees,S. Quegan,R.J. Moffett, M.V. Codrescu, tentialdropon interplanetary
parameters,J. Geoph!ts.Res., 86,
and G.H. Millward, A coupledthermosphere-ionosphere model 7639-7648, 1981.
(CTIM), STEPreport,editedby R.W. Schunk,ScientificCommittee Rishbeth,H., On theF2-1ayer
continuity
equation,
J. Atmos. Terr.
on Solar TerrestrialPhysics(SCOSTEP), NOAA/NGDC, Boulder, Phlts.,48,511-51•,1986.
Colorado, 1996b. Rishbeth,H. and R. Edwards,The isobaricF2-1ayer,J. Atmos.
Hanson,W.B., et al., The retardingpotentialanalyzerfor Dynamics Terr. Ph!ts.,51,321-338, 1989.
ExplorerB, SpaceSci. Instrum. $(•), 503,1981. Rodger,A.S., G.L. Wrenn,andH. Rishbeth,Geomagneticstormsin the
Hedin,A.E., MSIS-86thermospheric model,J. Geophgts.Res., 9œ, AntarcticF-region,II, Physical
interpretation,
J. Atmos. Terr.
4649-4662, 1987. Ph!ts., 51, 851-866,1989.
elec- Spencer
Kamide,Y, W. Sun,and S.-I. Akasofu,The averageionospheric N. W., et al., The DynamicsExplorerwind andtemperature
trodynamicsfor thedifferentsubstorm
phases,J. Geophgts.Res., spectrometer,SpaceSci. Instrum. 5(•), 417,1981.
101, 99-109, 1996.
Pr61ss,G.W., Seasonalvariationof atmospheric-ionospheric distur-
bances,J. Geophgts.Res., 8œ,1635-1640,1977.
M.V.Codrescu, andT.J.Fuller-Rowe!l, CIRES.UniversityofColorado,
Pr61ss,G.W., Magneticstormassociated perturbations
of the upperat-
andNOAA/SECR/E/SE.325 Broadway, BoulderCO. 80303.(e-mail:
mosphere:Recentresultsobtainedby satellite-bornegasanalyzers, codrescu•sec. noaa.gov;tjfr•sec.noaa.gov)
Rev. Geophgts.,18, 183-202,1980. I.S. Kutiev,Geophysical Institute,Acad.G. Bonchev.b13. Sofia 1113.
Pr61ss,G.W.,Storm-inducedchangesin thethermospheric composition Bulgaria.(e-mail:ikutiev•bgeam.acad.bg)
at middlelatitudes,
Planet. SpaceSci., $$, 807-811,1987.
Pr61ss,G.W., On explainingthe local time variationof ionospheric
stormeffects,Ann. Geophgts.,11, 1, 1993.
Pr61ss,G.W., and U. yon Zahn, Seasonalvariationin the latitudestruc-
tureof atmospheric
disturbances,
J. Geophgts.Res., 8œ,5629- (ReceivedMay 10, 1996;revisedNovember22, 1996;
5632, 1977. acceptedFebruary12, 1997.)

You might also like