Física de la Tierra ISSN: 0214-45572000, 12, 61-103

ImprovedQuality ofServicein IonosphericTelecommunicationSystemsPlanning

andOperation:COST251Major Achievements

R. HANBABAFranceTelecom-CNET,DMiR/TSI, 22307Lannion Cedex,France

B.ZoLESIIslitulo Nazionaledi Geofisica,

Via di Vigna Murala,605,00143Roma,Italy

ABSTRACT

COSTAcrion 251 <ImprovedQuality of Servicein lonosphericTelecommunicationSystemsPlanningandOperation)hasbeenadynamicgroupof 122 paruicipanísfrom46organisationsin 20 Nations working as a íechnical connnunity towardsa commongoal. Themainobjectiveswerelo collecíadditionalquantitiesandtypesof ionosphericdala, Lo generaleproceduresfor predictionof ionosphericmodeisoverEuropeand topromotetbeir use, lo extendubeexistingmodeislo give systemperformancestatisticsandlo developa melhodologyfor channelsimulalion.Themajorachievemenísincludeespeciallyte productionof databasesof ionosphericrneasuremenls,ube developmentofproceduresfor ionosphericandplasmasphericmodeUing,mappingand forecastingandmodeisfor usersfor overalíperformancepredictionof systemsin Europe.

Theresulísobtninedja te frame of COSTAction 251 during te period1995-1999aresummarisedwith particularreferencelo their applicalionfor terrestrialandEarth-spaceionospherictelecommunicationsystemsplanningandoperation.Reconunenda-tionsfor futureactivitiesarepresenled.

Kcy words:Ionosphere,Modelling, Mapping,Prediction,Forecasting,RadioWa-ves,RadioCommunications.

1. INTRODUCTION

Co-operationresearchon effects of Ihe upperatmosphereon terrestrialandEarth-spacecommunicationsal Ihe Europeanlevel is absolutelynecessarybecauseaccuratepropagationinformation is essentialLo supportte design,im-plementationandoperationof mostmodemterrestrialandsatellitecommuni-calion systems.Nowadays,communicationsthrough Ihe upper atmosplieresbouldmeetmore andmorerequirenients,evensothey seemLo beexcessive,unrealistic,or usefulness.Advancein perfonnancepredictionandforecastingshouldbeachievedby bringing togetbergroupsworking in two arcas,namelygroupswith objectivesradio engineeringapplicationsorientedandgroupsthaI

61 Físicadela Tierra2000,12,61-103

R.HanbabaandS.Zolesí improvedQuality ofSeri’ice in Ionospheric...

give moreemphasison radio sejence.Currently, thereis no structuredco-or-dinationeffort in Europein this domainoutsideCOST(EuropeanCooperationin tbe ficid of Scientific andTechnicalReseareh).

In this frame, work startedwith COST Action 238 (Predictionand Re-trospectiveIonosphericModelling over Europe - PRIME), which operatedfrom 1991 to 1995 with objectivesto developimproved modeisfor the Euro-peanareabetweenlatitudes35-550N and longitudes 100W-300Eusing jo-nosphericinformationtakenfrom existingmeasuringequipment.Studieswereconductedto the specificationof the monthly medianheightprofile of electrondensityup to a height of 1000 km andthe mappingas afunetion of time-of-day,seasonandsolaractivity of te monthlymedianvaluesof the Total Elee-tron Content(TEC) and standardionospheriecharacteristics.Proceduresofevaluationof thesequantitieson an individual daywere developed.

COSI251 (ImprovedQuality of Servicein IonosphericTelecommunica-tion SystemsPlanningandOperation)wasa naturalsuccessionto COSI 238w¡thout which resultswould notbe exploited. It wasbuilt on existing COSI238teamsandappropriatenewgroupsfor the additionalactivities. The parti-cipantsin COST 251 camemainly frompublie researchinstitutesandfrom uni-versities.Thework was organisedasa seriesof Work Packages,eachwith ownsub-objective,structuredwithin five Working Oroups.Al] Work Packageshadparticipantsfrom morethantwo countriesandhaveelectedtheirown lea-ders.A ManagementCommitteesupervisedprogress.

2. GENERAL DESCRIPTIONOF COST 251 AND ITS ORJECTIVES

Ihe mainobjectiveswereto collect additionalquantitiesandtypes of io-nosphericdata, to generateproceduresfor prediction of ionosphericmodelsoverEuropeand to promotetheiruse,to extendthe existingmodelslo give sys-temperformancestatisticsandto developa methodologyfor channelsimula-tion for HF systemsin Europe.me work hasbeensplit into five main areas,allocatedto five working groups: (1) modelsfor terrestrialsystems,(2) modelsfor Earth-spacesystems,(3) ionosphericmodelling, (4) variabiiity and fore-castingand(5) systemperformanceandspeetrummanagement.

Aceordingto the objectives,the principal activities werete following:

• Assemblyof a databankof vertical-ineidenceandoblique-pathionosp-hericmeasuremenísandTotal ElectronContent(TEC) for use in modelgenerationandtesting;

• Investigationof ionosphericpropagationeffeetson radio systemsusedfornavigation,geodesyandradio-astronomy;

• Studiesfurtherthe developmentof new modelsfor the monthly medianionosphericeharacteristiesfoF2, M(3000)F2andTEC;

• Improvementof existing techniquesfor predictionof ionosphericva-riability parametersof interestto HF systemplanningandoperationand

Física de la Tierra2~O, 12, 61-103 62

R.HanbabaandB. Zolesí ImprovedQualityof Servicein lonospheric...

otherionospherictelecommunicationssystemsparticularlyin higli la-títude;

• Investigationand improvementof ionosphericforecasífor 1ff systemoperation;

• Studiesof UF channelmodels,signalvariability andmodeisfor overallperformanceprediction;

• Productionof mapping functionsandassociatedcoefficients for themonthlymedianionospbericcharacteristicsfoF2,M(3000)F2andmC;

• Productionof spatial-interpolationalgorithmsfor instantaneousvaluesoffoF2,M(3000F2)andmC frona rneasureddatabr the sarneepoch;

• Elaborationof forecastingalgorithmsfor estimatingvaluesof foF2,M(3000F2)andTEC up to 24 hoursaheadof thepresent;

• Generationof expressionsfor te electron-densityheightprofile form intemisof valuesof foF2, M(3000F2),mcandotheravailableionosphe-ile andsolar-geophysicalparametersas necessary;

• Productionof amethodologyfor channelsimulationfor 1ff systems;• Interactionwith radio usersand comparisonof their performancedata

with modelresults;• Productionof computerprogramsin accordancewith therecommended

procedures.

Ihe adoptedCOSI 251 areais rectangularin geographiccoordinatesforoperationalconvenienceandeaseto interfaceto global models(Fig. 1).

lii total 122selentisísandengineersfrom 46 Organisationsin 20 Countrieshavebeenassociatedin the four-yearof the lifetime of the Action (1995-1999).SuccessfulAction outcomewas achievedby volunteercontributionses-timatedby peerexaminationto give generalagreement.

70z‘eo,0a‘1> 60-o4,

=

~ 50oro-‘uo> 40,o45o

30-

Figure 1. Map showing te COST251 areaof Europebetweenlatitudes35-7O~N ami longitudes10

0W-600Eandthelocationsof te 50 vertical-incidencelonosondescontributingto te databank.

-20 -10 0 10 20 30 40 50 60 70

Geographic icogifude (Degrees E)

63 Física de la Tierra2000,12,61-103

R.Hanhabaant.!B. Zolesi lmprovedQualityofServicein Ionospheric...

3. MEASUREMENTS AND DATABASES

Ihe studiesundertakenwithin COSI 251 werefor the mostparttheoreticalaudstatisticalinvestigationsmakinguseof ionosphericmeasurementdatasetsavailablefrom databasesestablishedunderAction auspices.Many pastmea-surements,especiallythoseof vertical-soundingsalreadyexistedand wereavailable,though sorneof tbem neededfurther analysisand assemblyintoconvenientformatsfor an easyuse.Speeialeffortshavebeenmadeto checkIhequality of the dataandrnanyparticipantsin tliis Action havecontrolledmostofthe storeddata.In this Sectionvertical- andoblique-incidencesoundings,elec-tron-densityheight profiles, TEC and UF spectraloccupancymeasuremenísthat werecarriedout underAction auspicesarereponedandthe databaseses-tablishedare described.

3.1. Measureinentscarriedoid under Action auspices

3.1.1. Ionosphericsoundings

Within and close the COSI 251 areaof Europe(Fig.1> Ihere havebeenabout50 ionosphericstationsproviding long-seriesof dala, mosíof them arestitl in operationandmaintainedby Action participants.In many stationsme-asurernentsperformedover4 solarcyclesarenow síoredin the databasesandmadeavailablein appropriatedigitisedform. Long-seriesdatasets are very im-portantto certain studies,particularly thoseundertakenwhich haveaddressedwhethertherehavebeenany detectableionospherictrendswithin the COSI251 areabeyondthoseassociaíedwith solar-cyclechanges,andfor Ihegene-ration of long-termmappingdalasets.

Regularvertical-soundingsare carriedoutsystematicallyonceeachhour inmost of the síationsaudonly al a few stationsevery 15-minute intervals,butalsoin Ihesecasestheregularreductionof Ihe datais madeonly atone-hourin-tervals.In thisregard theyrepreseníunder-sampledeffectivesnapshotsof jo-nosphericstate.Within-an-hourvariability canbelarge,especiallyat dawnanddusk. Regularsoundingsare also deficiení for the detailedinvestigationofthe day-to-dayvariability andfrom hour-to-hour.Coinprehensiveinvestigationof ionospheriedisturbancesindicative of ionosphericresponseto solar andmagneticdisturbancesrequiresmorerapid soundings.Morerapid soundingsarealsoneededfor spectralanalyses,which involve studyof temporalchangesasan aid to forecasting,and to spatial mapping on individual occasions.Iheneedis thereforefor coordinatedrapid-soundingmeasurementcampaigns.Sin-cedisturbancescannotyet be forecastit was necessaryto agreein advancetheepoehsof the campaigns.Campaignsof measurementsat 5-minute intervaistook placeduringdifferentsolarepochs.Difficuities havearisenin finding theneeessaryresourcesto analysethe largenumberof ionogramsfor theserelati-

Física de la Tierra2000.12.61-103 64

R. HanbabaanáB. Zolesí ImprovedQuatity ofServicein Ionospheric...

vely lengthy dataseIs. However, sorneof them measuredin Rome (Italy),TortosaandEl Arenosillo (Spain)havebeenanalysedandhavebeenusedinstatisticalstudiesduring the Action.

Obliquesoundersoffer thepotentialfor dalainversionto giveinformationon rnidpathionosphericconditions.The bestgeographicalimplementationof anetwork to monitordxc ionospherewouldbe lo haveregularlyspacedionosp-heriesounderscornbiningvertical andoblique transmissionsbetweenthern.Unfortunatelythis is not possible.However,a networkof suchsoundersexistsover Europeand a number of specialcarnpaignsof oblique-soundingwereperformedin Belgium, Italy, Spain,SwedenandUnited Kingdom during theAction.

3.1.2. Electron-densityhe¡ghtprofiles

Bottomsideelectrondensityprofiles havebeenobtainedby inversionofvertical-incidenceionograrns,usingeitherPOLAN (Titheridge,1985)or AR-TIST (Reinischetal., 1988).Topsideelectrondensityprofiles havebeenob-tainedfrom the COSMOS,Isis-AlouetteandIK-90 satellite sounders.memainapplicationis the validationof predictedelectron-densityheightprofiles.In thisaspectit was interestingto analysevertical-audoblique-soundingdataon the samepathsimultaneouslyin order to investigateIheir mutual depen-dencies.

3.1.3. Total ElectronContent

Total ElectronContent(ThC) datahavebeenderivedin DLR/DFD Fer-nerkundungsstationNeustrelitz(Gennany)from OPSmeasurernentscan’iedoutat stationsof the EuropeanOPS TrackingNetwork of the InternationalOPSService(Jakowskiet al., 1998a).Assumingasinglelayerapproximationfor theionosphere,slantThCdalawerethenrnappedto averticalatthepiercepointofthe ray pathwith the «ionosphericshell» fixed at 400 km height. Within dxcCOSI Action OPS measurementsover 4 years(1995-1998)werecollected,analysedandtrausformedto the COST251 dalaformat.

3.1.4. HF spectraloccupancy

1fF spectraloccupancywasmeasuredwith four autornaticsystemsaspartof ajoint UnitedKingdoni/Swedishexperiment(Gott et al., 1996).Ihesesys-tems werelocated at Baldock (United Kingdom),al Linképing audKiruna(Sweden)and al Munich (Gennany)and provide effective measurementofspectraloccupancyoverNorthernEurope.Also, Iherewasa fifth measurement

65 Física de la Tierra2000,12,61-103

R.HanbabaandB.ZoIesi hnprovedQualily of Service in Ionospheric...

systemat CobbettHill, nearFarnborough(United Kingdom).Ihe measureofoccupancyusedis definedas congestion,whieh is theprobabilitythatIhe out-put signalof abandpassfilter of given bandwidth,randomlyplacedwithin agi-venITU frequencyallocation,exeeedsa giventhresholdlevel,

3.2. Databasesestablishedunder Action auspices

3.2.1. fue COST251 databank

The COST 251 databankbasedatte AbdusSalamInternationalCentreforIheoreticalPhysics,Trieste (Italy) is operationalsince April 1997 frorn Webhomepage:http://www.eost251.ictp.trieste.itl. meCOSI251 databaseprovidesmontbuy-medianauddaily hourly valuesof up to 14 ionosphericcharacteristiesmeasuredat about50 stationslocatedin the FuropeanCOSI251 region andtheimmediatesurroundinggeographicarea, for the years 1944-1997.Ibis dalabankcontainsalsoanumberof electron-densityprofiles, as well asTEC andio-nosphericabsorptiondata. Participantsin COSI 251 havemadetestson tequality of te data, and a numberof errors in the original databavebeenco-rrected.Recognisingtbatthe databasehasapplicability for otherscientifxcin-vestigationsbeyondte lifetime of COSI251 it was decidedto includeup to 14ionosphericcharacteristiesalthoughnot alí of tesehavebeenusedto date.

Ihe databaseof vertical-incidenceionosphericsoundingshasbeenmadeavailableon CD-ROM by the RutherfordAppletonLaboratory(UnitedKing-dom),includingamodemWindows-basedinterface,the IIWO DAlA packagedevelopedby the OeophysicalInstitute of Sofia(Bulgaria).

3.2.2. TheIonosphericDespatchCentrein Europe

TheIonosphericDespatchCentrein Europe(IDCE) hasbeenestablishedasaninitiative of COSI Action251 (Stanislawskaet aL, 1999a).It is operatedsin-ce January1997 in HeliogeophysicalPredictionServiceLaboratoryof theSpaceResearchCentreof PAS (RegionalWarningCentreof the InternationalSpaceEnvironmentService,IRLES) in Warsaw(Poland).

lUCE allowsconvenientaccessto sornerecentionosphericdatafrornver-tical-incideneesounderslocatedmainly in Europe.Datafor the currenímontharepresentedfrom the beginningof the rnonth up to last rneasurementsavai-lable. IDCE offers the cataloguesof disturbedand quiet days,the list of dis-turbedperiodsof few hoursduration,dataon suddenionosphericdisturbancesas well as solar,magneticandotherionosphericinformation,predictionsandforecasting.IOCE providesalsovaluesof te MF2 indices(Mikhailov, 1999).Onte basisof thernost recentdalaapresentationof te instaníaneousmapsoffoF2 for Europefor is madeavailable.Data are availablefrom Web page:http:\\www.cbk.waw.plV-wc\idce.htmlandon ftp.cbk.waw.pl.

Física de la Tierra2000, 12, 6 1-103 66

R. Hanbabaant! B. Zolesi ImprovedQuality ofServicein Ionospheric...

4. IONOSPHERTC AND PLASMASPHERIC PREDICTION

4.1. The COST 251 model for the montbly median ionospheric

characteristiesfoF2, M(3000)F2and TEC

4.1.1. Modelfortite monthlymedianionosphericcharacteristicfoF2

The COSI251 modelfor the monthlymedianfoF2, knownasMQMF2R,is basedon SingleStation Models (SSM) and tbe multiquadriemethodforspatialapproximation.

Mostof te ionosondesin the COSI251 areahavemorethanonesolarcycleperiodof observations,enablingthecreationof fofl monthly medianmodels,ca-lledSSM’s, for tite locationof theseionosondes.meionosphericmdcxMF2 wasusedto derivete foF2regressionson solaractivity for eachionosondestation.Aset of coefficientsfor 12 monthsand24 UI momentsfor eaehjonosondeweregeneratedwith anon-linear(cubiepolynomial)dependenceon MF2 index.

lite choiceof ametitodfor spatialinterpolationwasa crucialpoint. Themultiquadriemetitod(Hardy, 1971)usedfor mappingin topographymaybesuccessfullyappliedto global as well as to regionalionosphericmappingaswasshownby TeryokhinandMikhailov (1992).Titis metitoddrawsa surfacestrictly over given setof pointsin distinetionfrom otiter methodsadoptedforionosphericmapping.AlI pointsinsidete COSI 251 areaas well as iii thebuf-ferzonewereusedfor drawingte surface.Thus thisapproachalwaysprovidesasmoothinterfacingto te IIU-R (1997) modeloutsidetite COST 251 area.Ihe metitod is numericallystableand doesnot requireadditional «soreenpoints» for numericalstability in data-sparseregions.A buffer zoneof 45pointsaroundtite COSI 251 areawasusedto interfacete COSI 251 mapstote global ITU-R maps(Fig. 2). It shouldbenotedtbatthe modelMQMF2Risbasedon tite MF2 indextat is availableonly for tite past.A conversionpro-cedurefrom R12to MF2 is adoptedin the modelMQMF2R.

4.1.2. Modelfortite monthlymedianionosphericcharacteristicM(3000)F2

lite COSI 251 modelfor te monthlymedianM(3000)F2,knownas UN-DIV, usesSSM’s,making no distinetionbetweensolarcycles.no distinetionbetweenrising andfalling partsof thecyclesandno accountingfor hysteresiseffects.

Since for constantlocal time no appreciablesystematicdependenceofM(3000)F2on geograpitielongitudewas foundover te COST238 areateoriginal UNDIV approacitwas to uselinear regressionswit geograpiticlati-tude.Testsdemonstratedtat titis approacitwasnot goodenoughfor te mucitwider COST251 area(Leitinger, 1998a).Tite useof magneticlatitude(dip la-titude andmodifieddip) insteadof geograpiticlatitude brougittslight impro-

Física de la Tierra2000,12,61-10367

R.Hanbabaant! fi. Zolesi improvedQuality of Servicein ionospheric..

March 1980 <05 UTI-40 -30 -20 -10 0 10 20 30 40 60 60 70 80 90

90 90

80 80

0> 70

o, 70

6060

o,

0 50o, 500503~ 40 40

‘~ 30 30

-J

20 20

10

Figure2. FxampleofCOST2SIÉoITU-RfoF2mapsinterfacing.The COST251 urea gronod-basedlonosondes(*) asidbuffer zonepoínOs(•) areshown.

vementsonly. Thereforegeograpitielatitude was maintainedand a linearde-pendenceon geograpiticlongitudewas added.Porconstantlocal time (LT) audmonth of the yearbi-linearregressionsin latitude andlongitudewereadopted.

Ihe monthlymediansof M(3000)F2were takenfrom the COST 251 data-baseandpreparedin tite following way: (i) for eachCOSI 251 stationnot re-servedfor testing,for eacitmonth andfor eachitourUI linearregressionsforthe R12 dependenceof M(3000)F2 were constructed,(u) level valuesofmonthly mediansfrom tite regressionswere calculatedfor two R12 levels(R1~=35 andR12=135),(iii) by meansof Fourier interpolationtite diurna! va-riation of tite level valueswere trausformedfrom UI dependenceto LI de-pendenceand(iv) for severa]different se]ectionsof stationsbilinearregressionswerecalculatedfor eachR12 level, for eacitmonth andfor eacithourLI.

lite buifer zoneprocedureadoptedin COSI 238 was used.Titis bufferzonehasa width of 5 degreesin latitude andlongitude,half of tite buffer zonebeinginside the COSI 251 area,half of it outside.Titis meansthat with titebufferzoneinterpolationtite «unadulterated»COSI 251 areais restrictedtobetweenlatitudes 37.5-67.5

0Nand longitudes7.50W- 57.50E. It should bementionedthat the bufferzonesolutionis an analyticaloneaudtitereforecon-tinuousin tite mapvalueandin alí its derivatives.

Tite M(3000)F2mapsfrom UNDIV slrnw asmoothbeitaviourwithoutsmall scalestructureswitich would be unrealisticfor a montitly medianrnap.Figure3 displaysexamplesas contourlinesover tite entireCOSI 251area.

lo

-40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90

[ongitude geodetic, deg

Física de la Tierra2000, 12. 61-103 68

R.Hanhabaant! fi. Zolesí ImprovedQuality ofServiceluzIonospherie...

4.1.3. Modelfor TEC

Tite COST 251 model for TEL, known as COSTTEC,is basedonmontitly andhourly mediansof electroncontentderivedfrorn tite Differen-tial Dopplereffect on tite signalsof tite polarorbiting NNSS satellitesfortlneesolaractivity interval. Tite mediansweregainedfor tite latitudes45,50, 55 and600N from latitudinal profiles of electroncontent.For eachoftiteselatitudesa setof 1212 Fouriercoefficientswasproducedby transfor-ming thebi-itourly andmonthlymedians.Tite Fouriersetswerecut down to55 coefficients(time independentvalue, diumal, semi-diumal,annual,semi-annualcomponentsandcombinations).Sincetite latitudedependence

COSIza

65

60

SS

50

45

-10 0 ¡0 20 30 40 50 60

Á4J0000’•2j. 0n6b.,~ 13 ~fl 2070

os

60

SS

SO

45

40

351S~í4Q3tJ no

70

65

60

SS

So

45

40

3560

50000,r2>~hlW ¡SC/It 612—2070

65

60

3— 55

50

4-5

40

35-‘0 0 ¡0 20 30 40 50 0<-

70

OS

60

SS

a50

45

40

35—10 Q¡02<J-5U00J¿¡ 3<

za

65

60

ss-ta

50

45

40

55-10 0 50 6<

Figure3. Regionalmapof tlse ionosphericcharacteñsticM(3000)F2for 1800h UT, asobtainedfromte COST251 recommendedmappingprocedure.The shadingindicatesmaxirea(lightestgrey) andmi-nima(darkestgrey).Seasonandsolaracíivity fromleft to right asidfrom top lo bottom:low solaractí-vity (R>,= 20) for January.July asidOctoberandhigh solaractivity (R,

2 = 150) for January.iuly andOctober.

A1A000lF;5. YjN~YARr .13 UIt 612— 20

‘0 30 4’,Cango uSe <5

10 20 30 40¡o,.gttnde -‘ -27

69 Física dela Tierra2006, 12. 61-103

R.¡-1anhabaant! fi. Zolesí ImprovedQualiíy of Servicein Ionospheric...

tumedout to benearlylinear,regressionuneswerefitted to the Fourierca-efficients leading to twa setsof 55 Fourier terms(valuesfor tite latitude52.50N andvaluesfor the latitudinal gradients)for eachof tite two leveisofsolaractivity.

Becauseno electroncontentdataexist (or were¡nadeavailable)fram titelangituderegionEastof 200E it was necessaryto transferexperiencewith thelongitudedependenceof peakelectrondensity ta electrancontent.This wasachievedby introducingtite equivalentslabthiekness,tite ratio electroncantentdivided by peakdensity,(Leitinger and Hochegger,1999). Iecitnically titelongitudedependencewas derivedfrom tite UTNDIV 251 faF2 map and for-mulatedwitit twamoresetsof Fouriercoefficients.

Figure4 displaysexamplesof the annualauddiumal variatianof IEC fortwa lacationsandtwo levelof solaraetivity.

e

Egure4. Annualasid diurnalvariationoboalaedfromCOSrrECfm latitude 400N; top: low solaracti-vity(R

12=20); bottom:high solaractivity (R>,=150);left: Iongitudeó00E;right: longitude io0w.

~520

3150

r—4. ~

Física de la Tierra2000,12,61-103 70

R.Hanhabaant! fi. Zolesi ImprovedQuality ofServicein Ionospheric...

4.2. The COST 251 model¡br the electron-densityhe¡ghtprofile

Tite COST 251 recommendedmodel COSTPROF for tite eleetron-den-sity iteigitt profile consistsof twaparts:

1. A bottomsidemodelfor tite heigitt regionbelowthe F2-layerpeakbasedun tite ionosphericeitaracteristicstoE, foFí, foF2 andM(3000)F2andon rocket soundings.

2. A topsidemodelbar tite heigitt region aboyetite F2-layerpeakbasedunO~-H~diffusive equilibrumwitit built-in mapsbar titreeparameters:titeoxygensealeheigitt atte F2-layerpeak,its heigitt gradientandtite 0~-H~ transitioniteight.

The model is continilausin alí spatialfirst derivatives,a necessityin ap-plicationse.g.ray tracingandlocationfinding.

4.2.1. Bottomsidemodelforthe heightregion belowtite £2-¡ayerpeak

Tite modelfor te E, Fi andF2regionsis a modified«Di Qiovanni-Radi-celia»(DOR) model(RadicellaandZhang, 1995).Regionalor global mapormeasuredvaluesare adoptedas input for tite ionosphericcitaracteristicsfoE,foFí, foF2 audM(3000)F2.lite modelusesfive semi-Epsteinlayers (Bossy,1987).Iwo semi-Epsteinareusedfor tite E-layer (topandbottom),twafor titeF 1 -layet(alsatop andbottom)audonefor te bottomof tite F2-layer.Tite ma-ximum of tite E-layer is fixed at 120 1cmaltitude,tite heightof tite Fl-layer ismodelledandtite heigitt of tite F2-layerpeakis derivedfrom M(3000)F2andtite ratio foF2/foE.

The D regionmodelcoveringtite 55-85 km iteigitt rangewas producedbyte University of Rostock,Germany,(Singeret al., 1995)anditas beenínter-facedwitit tite DOR model. lite formulationis basedun rocketsoundingdataandcomparisonwitit radio wavepropagationdala andusesdalafiles for at-mosphericpressurederivedfromtite CIRA-86model.

To combinetite D-regionmodelwitit tite DGR modelof tite E-Fregion,titemodelswerelinkedby aconnectingregionbetween85 and95 km to secureatransitionwititout-discontinuitiesin electrondensityandits first derivative.

4.2.2. Topsidemodelfor tite iteight region aboyetheF2-layerpeak

The topsidelonospiterefonnulation,for tite iteigitt regionaboyetite F2-la-yer peakup to 2000 km, usestitree pitysical parameters,namelyte oxygensealeheigitt atte F2-layerpeak,its heigitt gradient,andtite O~ — H~ transitioniteigitt. mesetitree parametersaremodelledaccordingto solaractivity, season,local timeandmodifieddip latitude (Leitingeretal., 1995). -

71 Físicadela Tierra2000,12, 61-103

R.Hanbabaandfi.Zolesí ImprovedQualiiy of Servicein Ionospheric...

Diffusive equilibrium doesnat give a maximumbut a decreaseof electrondensitywith increasingiteigbt. A Chapman’slayerexpressianis adoptedto forcea peakof tite electrandensityprofile. At tite peakelectrondensityit is assumedtitat tite electrondensityscaleiteight is approximatelytheplasmascalebeightbaratomicoxygenions. This scaleiteigitt increaseswithheigitt togetiterwith tite tran-sitianin ion compositionfrom atomicaxygento atamicitydragenians.Tite tran-sitianheigitt is tite iteightwitere tite densitiesof tite two ions areequal.

lite oxygensealeiteigitt wasfound by fitting CitapmanIayersta titeareticalmodelsin tite surroundingsof tite electrondensitypeak(Iitheridge, 1993).Thetemperatureof neutralgaswas takenfrom M51586 (Ivledin, 1987), tite tempe-raturesof ionsaudelectronswere calculatedfram simplifiediteatcanductionequatians(Titheridge, 1993), calibratedwitit abserveddata (Tititeridge,1976a,b)audby comparisonwitit tite resultsof moreaccuratebut mucit morecomplicatedcalculatians.

Tite iteigitt gradientof tite oxygenscaleiteigitt was derivedfrom ananaly-sis of observeddataaudof tite iteatflow equations(Leitinger 1998b).COSI-PROF usesa scaleiteight, witicit increaseslinearly witit iteigitt from tite F2-la-yer peakto aceiling iteight h~. Aboye a constantscaleiteight is takenfarelectrandensity.

Tite O~ — H~ transitionheigitt is abtainedin COSTPROFfram a diffusiveequilibrium formulation.

4.3. The COST251 metitod¡br instantaneousvaluesof foF2, M(3000F2)andTEL

Almost aH approacitesfor obtaining instantaneousmaps of faF2,M(3000F2)andTEC maygive significanterrorsespeciallyin data-sparsere-giofis, or evenon someoccasionscan leadto numericalinstability of tite map-ping procedure.lite accuracyof an ionosphericmapdependson its tite abilityto describetite differentpitenomenain tite ionospitere.Unfartunately,tite num-ber of measurementlocations is generaUyinsufficient for tite productionoffully accuratemaps,evenayerarestrictedgeograpiticalregion,andusemustbemadeof artificial screen-pointvalues to constraintite mappingcontoursin re-mote areas.Furtitermoreitigiter latitudesare signif¡cantly influencedby themíd-latitudetraughpitenomenaduringnighttime(seeSection4.5.12).

4.3.1. MethodforinstantaneousvaluesoffoF2,M(3000F2)

Tite COST251 metitadfar instantaneousvaluesof foF2 andM(3000)F2,knownas PLES(PolandPL, SpainES),combinesmontitly medianmapsof io-nospitericcitaracteristiesandasetof sereenpoints-measurementsusingtwo in-terpolationmethodsmadifiedtitis purpase:Kriging and«fitting».

Física de la Tierra2000, ¡2, 61-103 72

R. Hanhabaant! B. Zoles! ImprovedQualiiy of Serviceluz !onospheric...

Titebasisof Kriging (Matheron, 1971)is ta apply weightedinterpola-tion amongadjacentdatapointsusingvariogram,i.e. functionwitich illus-tratestite differentiationof tite valueof aparameterdependingon thedis-tancebetweentite differentmeasurements.Weigitting factorsaregiven totite measurementslii orderto assuretite mostaccurateestimationof tite unk-nown parameter.A scalingfactor is introducedin tite Kriging interpolationtecitnique.

Fitting (Rusit audEdwards,1976)usesa setof weigitting functions,whicitarea measureof tite staristicaldependenceof foF2 betweenpointsaix a spitere.

Tite janospiteriecitaracteristicsattwo clasepointscitangein time in ax’eiy similar way. Titecorrelationdecreaseswith tite distancebetweentheobservationpointsaudtite variagramis foundto be anisotropici.e. tite va-riogram18 different in distinct azimutits.mecorrelationradius,witicit is titedistancebetweentwo points for witicit the correlationof tite ionospitericcitaracteristiccitangesby 0.5, is adoptedto parameterisetitis propriety. Insornemetitodstite fulí correlatianfunction, calledweigitting function, isused(RusitandEdward,1976). Oneof thedisadvantagesof tite metitadistite fact titat tite resultantfunctiondoesnot passthrougittite input point va-lues. For COST 251 applications,JuchnikowskiandZbyszynski(1992)modifiedtite Rush-EdwardsinterpolationmetitodLo ohtainmoredesirableresults.

It is well knawntitat tite weakpointof everymappingprocedureis gene-rally the numberaud/or tite distribution of measurementlocations,witicitmaybeinsufficientfor tite productionof fully accuratemaps,evenoverares-trictedgeograpiticalregion.It maynotallow describinggradientsproperly.Toconstrainresultsto pitysicallyrealisticfigures,deviationsfram monthlyme-dianpredictedvaluesaremodelled.It itas beendemonstratedtitat titis tecit-nique smootitesout many discontinuities(Stanislawskaand Juclmikowski,1997). Titree or faur data pointsallow constructinga mapwith acceptableprecision.

Comparisonswitit measurementsIxave showntitat tite accuracyachievedit>’botit metitods is almost tite same.Nevertbelesstite modified Kriging algo-rithm is betterfor foF2(metitodPLES2),witile fitting is betterbarM(3000)F2(PLESS).So that both metitods¡¡ave beencombined(PLES2 for foF2 audPLESSfor M(3000)F2)to producetite COST251 instantaneoúsmappingmo-del for ionospitericcitaracteristics.

4.3.2. MethodforinstantaneousvaluesofTEC

lnstantaneousvaluesof IEC mapsaregained by using tite COSI 251electron-densityiteigitt profile in its «instantaneous»modeandnumericallix-tegrationalongverticalor slantrays.

73 Físicadela Tierra2000, ¡2,61-103

R.h’anbabaandB.Zolesi ImprovedQualiry of Servicein Ionospheric...

4.4. Thc COST 251 forecast¡ngprocedure

Tite COSI 251 forecastingprocedureis basedon tite auto-correlatianpro-cedureCORLPREVdevelopedby Muittarov audKutiev (1999) in tite frame-work of tite Actian. A networkof 23 vertical-incidenceionosandesprovidestitebasicinputs. Mappingis possiblefor tite areaextendingfram 100W to 900Eandfrom 300 to 700N, which includestite COST 251 region.Dalaarecurrentlyupdatedevery24 itours on weekdays.

Tite difficulty is thatgoadstatisticalestimatesof tite autocorrelationfunc-tion require very long observationperiods.Titis is impractical since tite un-derlyingpitysicatprocessesarenot stationaryayerlong periodsof time. lix or-derto avoid titis, a predeterminedsitapewitit a finite numberof parametersisimposedto tite autocorrelationfunctian.Theseparametersarecitasenin suchaway titat diumal variationsaudmedium-termdecorrelatianof tite ionospitericcitaracteristicof interestarewell represented.A time period of 20 daysis usedfor tite calculationof tite empiricalautocarrelationfunction.Iestsata varietyoflocationsitave demonstratedtitat tite autocorrelationmetitod perfarmssubs-tantially bettertitan usingregularlyupdatedmantitly medianvalues.It is cleartitat titis metitodwill breakdownat tite onsetof ionospitericstorms,as it cannotitaveprior warningof disturbances.Howeverit itas temedout to be surprisinglyrobust in disturbedconditions.Tite autocorretatianmetitodis applied to pro-duceforecastvaluesof foF2andtite F2-layerbasicMUF far a3000 km range(MUF(3000)F2)at integeritours UT up to 4 days aiteadateacitvertical-mci-dencestationwiten sufficientmeasurementsareavailable.This is necessarytoensurea forecastfor up to 24 hoursaiteadis alwaysavailable,despitewee-kend’sdifficulties. Valuesfrom tite past60 daysareusedtaconstmctan auto-regressivefilter.

foF2(MI-~) 12MAY1O9S 12OXIUT

2

- 6.50V

40 50 60 70 80

Gecg~MtIc~ÉtttÉ(E~es E)

Figure5. Map derivedfrom 24 hoursaheadforecasoof foF2 for 12 May 1998 80 1200 h UT.

Física de la Tierra2000,12,61-103 74

R.Hanbabaant! fi. Zotes! ImprovedQualiiy ofServicein Ionospheric...

Forecastsmapsof foF2 andMUF(3000)F2are drawnusinga conxmercialpackagewith aKriging option, witicit is particularly suitablefor contouringsparsedala(Fig 5). An operationalforecastingtool is availableon te WorldWide Webbar interactiveuse(itttp://www.rcru.rl.ac.uk/iono/maps.ittm).CORL-PREV softwareis available aL SpaceResearchCentre,Warsaw(Poland),1mwanonymousftp://itaydn.cbk.waw.pl(seeSection8).

4.5. Other stud¡esand models

Wititin titis line of activities studieswereconductedfirst of al on tite de-velopmentof modelsrelevantto ionospiterictele~ommunicationapplicationslar tite COSI25] area.In Ihis Sectiontesemodeisarebriefly describedandtite mainachievementsof differentstudiesarereported.

4.5.1. Monthlymedianionosphericcharacteristicsfoff2¿mdM(3000)F2

4.5.1.1. IonosptxericSingleStationModeis

As it is generallyaccepted,ionosphericSingle StationModels (SSM’s) givetite mostaccuraterepresentationof tite modelledionospitericparametersun teinimediatevicinity of aix ionaspitericstation.For titis reasonSSM’scanbeusedin severa]applications,as,br instance,fiuing datagaps,either retrospectivedalagapsof activestationswitit long enougitseriesof rneasurements,or to ex-tendtite seriesof valuesof clasedstations,while asecularvariation is not fo-reseen.Anotiter importantuse is Lo producescreenpoints for mappingio-nospitericcharacteristics.‘me work itasbeendoneasfollows: i) to producenewSSM’s, u) to comparetitem with tite previouslyproducedmodeisduring titeCOST238projectandiii) to usethemin ionaspbericmapping.

TwanewfoP2modelswereproduced.Xenoset al. (1996)useaquadraticre-gressionbetweenmontitly itourly medianvaluesof foF2 andtite AristotelianUni-versity lonospiterieS-index.Solé(1998)modelintroducesageomagneticdepen-dence.litis madelis basalon astatisticalseconddegreemultiregressionbetweenfoF2, te monthiy ionosphericTmdcx ami tite monthlygeomagneticindexAp.

‘mesemodeisandtite IIU-R modelitave beencomparedwith te modeisproducedin te COST238 projectand testedwitit datafrom 22 ionosphericstationstitat coverte wboleCOSI 251 area(AlbercaeL al., 1998).As expee-ted, te SSM’sgivebetterresultstitan globalmodels.‘me bestresultsarepro-ducedby Solé‘s modelandte modelof Mikhailov andMikitailov producedduringCOSI 238 (Bradley, 1995). ‘me resultsof botit modelsarevery clase,but titoseof Salé’smodelare tite best.A predictionof te Ap magneticindexma>’ be neededto makeuseof titis model.lo acitievetitis, aneuralnetworkprocedurewas introducedin tite model, obtainingte index witit aix error of

75FísicadeJaTierra

2000,12,61-103

1?. Hanbabaant! fi. Zoles! lrnprovedQua!isyofSen~iceiii Iouzospheric..

25%.Furthetmorc,Soléel al. (1999)bavedrawnatientionjo tite fact thai titedependenceof tite Ap índexon Ihe resultsof tite modelisnoí high. II ~sfornidthat a 30%error lo tite Ap indexaffectstite lina! resultof foF2 in oniy 0.1 MHzlii average.Wilh tusAp índexpredictionprogram,Solé’smodel canbeadop-lcd lo predictfoF2, Consequentíy,tite modelof Soléis tite SSM of fof’2 re-commendedby COSIAdían251.

Desudestite foF2 SSM’s~tinte M(30005F2 SSM’s itave beenproduced.Míkhadovet al. (1996) modeluse a seeonddegreerela.tionshipbeíwcenM(3000~F2and1< Solé(1998) utilisestite saniea]gorithmlot M(3000)F2thaitite oneLar fofl. 3%.enosami Alberca(1999)usenew-al network tcchniquclarM(3000)F2,Mosíof tite valuesarepredicíedwitit lesstitan 5% error.

4.5.1.2. ISIRM (ImprovedSimplitied lonospiterieRegionalModel)

¡51kM is a regional inodel of tite ionospitericcitaracteristiesfoF2 andM(3000)F2,evolvedby SIRM (Sirnplified lonospiterieRegionalModel) (Za-lesí el aL, 1996)amI appliedto a moreextendedareataldng lino accounítiteconsequencesofitigh latituderegion.lite ¡nade]is basedon tite Fourieranaly-sísof tite monthlymedianvaluesof tite tonospiterieparamekrsmeasuredat11wsíatíonsni tite Luropeanami nearFastre,giorísant] colleetedundertite COSI25! projecí.Tite bit-st stepof tite procedureis tite linearregressionanalysisoffuemontitly medianvaluesof a givenionospitericcitaracteristieagainsítheso-lar actívííy mdcxR

11. Tite seceodstepis aFourieranalysísof tite dataLar twalixed valuesof R, ~0andR1., z 1Q~~ In orderto benerreproducetite moníhlybeitaviourtite Fourier anoiys~werepeifonnedmontl-i by month insteadofalongtite twa virtual yearsoftite two solarepocits.

Reimproved811kM muodel is still a very simpleprocedure,001 only lot liscas>’maúxematicalformulationant] Lot tite reducednumberof numericalcoef-ticients,buí aboyealí foi- tite sitort softwareprogramIhaícan beeasilyusedant]liniced wlih otersoftwareprocedures.

4.5.1..3. ILCNN (JonospitereLocal Copy by NeuralNeíwork>

ILCNN is a ocotal networkproceduredesignedto ruodel ant] predicí titemoníitly medianfoF2and M(3000>F2over tite COSI 251 area(Larnming amiCander.1999).A mulíi-layerpercepíronwainedwiíh back-pronagaíionis adop-ted. ‘tite numberof híddenunits ls 16 ant] 9 on lirst ant] secondliUden layers,respectíveíy.U is welI knownthai titeknowledgeof aneural network is coníainedin its weight.Titeseweightsareami cae-it inputof al] itiddeux unitsandalt out—put unlís. lite aim of tite back—propagaríonalgariíhmis Lo fínd tite weigbts.loimplementthe ILCNN model,only tite architectureof tite new-alnetwork, titeweigbteoefflcients,sígmoidfuncrionami nonnalisedfunctíonsareneeded,

Física de lo Tierra2000, 12,Ñ-¡03 76

R.Hanbabaant! fi. ZoIesi ImprovedQualily of Servicein Ionospheric...

4.5.1.4. MQMF2

‘me MQMF2 metitodbar te montitly medianfoF2andM(3000)F2usestitesameSSM andinterpalatianprocedureas un tite metitodMQMF2R describedin Sectian4.1.1 but tite solaractivity index in MQMF2 metitodis MF2 insteadof R12.

4.5.1.5. PLESíaudPLES2(PolandPLSpainES)

PLES1 andPLES2 are tite implementationfor montitly medianmade-lling of tite appraacitdescribedin Section4.3.1.Titesemetitodsinvolve titegenerationof SSM acitievedby Solé’sprocedure(Solé,1998). ‘me modelre-quirestite montitly averagedAp and1 indices as input for SSM’scalcula-tions. Tite valuescreatedby SSM’sareinterpolatedfor tite witole COSI 251areawith tite modifiedKriging technique(PLES2)or witit fitting (PLES1).Different responsesta geomagneticdisturbancescanbe takenunto accountwitit tite inclusionof tite Ap mdcx. Titis index canbe successfullyusedbarretrospectivemodellingwiten Ap index is knawn.However far long-termprediction tite difficult>’ of sufficient accuracyin its predictability nowa-dayssitouldbe noted.

4.5.1.6. SWILM (Space-WeigittedIanospitericLocal Model)

SWIiLM is a tecitnique for modelling tite foF2 montitly median,witicituses data andtite R mdcx. Twa setsof coefficientsarecalculatedpast series 12by linearrelationsbetweenfoF2 audR bar 20 stationsin tite COSI 251 area,

12for eacitmontit audbar eacititaur. Framte twa setof coefficientsant] from gi-venvaluesof tite inputparameterR12, tite faF2montitly mediansarecalculatedfor tite penad1957-1996,far eacitslationof tite database,for eacitmantit andbareacititaur.‘me 20 slationsaretiten assembledin dibferentsectorsdependingaix tite latitude distancefrom tite locationwherepredictedvaluesof faF2itaveto be calculated.Eacit sectoritas a widtit of 5 degreesin latitude.‘me predictedfaF2 valuesare finally basedon tite values of foF2 coming fram differentsectors,opportunelyweigittedaix tite spatialscale.

4.5.1.7. UNDIV

‘me UNDIV metitodbarmodellingte foF2 montitly medianusete sameprocedureas describedin Section4.1.2 barM(3000)F2.A parabolicregressionfor te R12 dependenceof foF2 is adapted.Titree levelvaluesof montblyme-dians were calculatedfrom tite for R levels (35, 85 andregressions titree 12

77Física de la Tierra

2000,12,61-103

1?. Hanbabaant! fi. Zolesi Iniprovet!Qualityof Servicein lonosplierie.. -

135). Ihe resu¡tsare12 x 24 x 3 level valuesperstatianfar foF2. Bilinearre-gressionsfor eacitR>2 level, for eacitmontit audfor eachhaur(LI) arederived.

4.5.1.8. SAILT

SAILI is an empiricalmodelof foF2 for tite 35(400 latitude belt duringquietant] disturbedcanditians(Eliseyevant] Besprazvannaya,1998).Tite mo-del is basedan a slatisticalanalysisof multiyearobservationat tite vertical-un-cidencesoundingnetwork.Tite ¡evel of tite noanfoF2 is approximatedbyproductof twafunctians.‘me foF2for otiter haursarerelatedta tite noanvalueby a simpledependence:

foF2 = K. .foF2

Tite valuesareobtainedby interpalatianbetweentablesof coefficients.

4.S.2. ModeisforTotal ElectronContent

Tite regionalTEC nxadel NTCM (NeustrellizTEC Mode»,producedatDLRJ’DFD Fernerkundungsstatian Neustrelitz (Oermany)framOPSmeasure-ments,itas beenimpravedant] updated(Jakawski,1998, 1999).Twaversionsof tite madelweredevelapedant] appliedta mapcanstruction.

NTCM 1 Tite algarititmincludesfundamentaljanospiterievariatiansandsolaractivity dependence.‘me coefficientsof tite modelarebasedaix IEC-Fa-radayratationdalaoblainedat tite EuropeanstationsNeustrelitz(Oermany),Graz(Austria)andFlarence(ltaly) during tite years1976-82.

NTCM 2 ‘mis versioncomparedwitit NTCM 1 includesadditianallyageamagneticlatitudedependenceterm.‘me coefficientsaretakenfrom GPSderivedTEC mapsayerEuropeobtainedfram February1995 until January1996.

An evaluationitas beenmadeof tite globalrms error,given by:

a=(ji~(P, ~mí)2j}

witereN is tite total numberof samples,istite valuepredictedby tite modelforsample¡ ant] is tite measuredvaluefar sample1. ‘me rmserrarof NTCM 2 waslesstitan 1 xlO’

6nr2 for 1995(Jakowskietal., 1998b).It is cleartitat titis modelprovideshighaccurate‘WC dataayertite COST251 areabar law solaractivitycondition.

Física de la Tierra2000, 12,61-103 78

R.Hanbabaant! fi. Zolesí ImprovedQuatity ofServiceluz Ionospheric...

4.5.3. MetitodsforinstantaneousvaluesoffoF2, M(3000)F2¿mdTEC

4.5.3.1. MQMF2-IM

‘me metitodMQMF2-IM for foF2 audM(3000)F2instantaneousmappingusestite following: (i) SSM’sbar faF2 audM(3000)F2,(u) screenpaintsrnsi-detite area, (iii) effectiveitourly MF2Cff andR12ff indexes,(iv) bubferzane,(y)

main ionospherictrougit model and (vi) multiquadricmetitodfor spatialap-praximation.

‘me ionaspitericindexMF2 is usedto find tite foF2relatiansitipwitit salaractivity. HoweverformadellingM(3000)F2tite index R

12 is adaptedas it pro-videsbetterregressionaccuracy.

SSM madelma>’ beusedto producefoF2 aud/arM(3000)F2screenpointvalueif tite nearestavailablecurrentobservationis locatedfurtiter titan 1O( inlongitudeand 5( un latitude. In tite areaswitit sparseionosondesadditianalscreenpointsare intraducedwitit tite ITU-R model.

A bufferzaneof 45 pointsis definedas aconstraintto avoidunrealistiebe-itaviour of foF2andM(3060)F2alangtite baundar>’of tite area.Local MF2~ lii-dicesarederivedfor eacitionosondelocationfram tite observedfoF2valuestodrawaMF2 surfaceayertite areawitit tite multiquadricmetitadbar spatialap-proximation.MF

26~~valuesbar tite buffer zanepaints arededucedfram tite re-

gressianabMF212versusR>2. ‘mis is possiblewitit aminimumof 3 abservedva-lues of faF2. If titis canditianis nat met, tite MF

2Cff far previausitaurs (ormontitly medianvalue)maybeused.Iii tite sameway, aneffectiveint]ex R isadoptedastite input to tite M(3000)F2SSM’saudbar tite buffet zaneta c’álgu-latefoF2 ant] M(3000)F2usingtite ITU-R model. ‘mis index R

1 eff 18 deducedframtite mininiisationof tite standarddeviationbar tite calculatedIaF2with res-pectto tite abservedonesayertitewitole areain questionbarall integersUT. Ef-fectiveR1~ff mdcxis usedasinput to tite IIU-R modelun tite bufferzane.

‘me main ionaspiterictrougit model basedaix COSMOS-900ant] ínter-COSMOS-19satelliteobservationsdevelopedun IZMIRAN, Russia(Annaku-liev et al., 1997)is introducedlix tite madeloptionally (Mikitailav etal., 1998).

4.5.3.2. PLES3audPLES6(PalandPL SpainES)

PLES3audPLES6aretwa otiter versiansof tite recaminendedCOST251instantancausmappingmediaddescribedin Sectian4.3.1.lix titasemethadstitemid-latitudetraugitmadelis intraduced,as well as SSM’s,develapedby Salé(1998), witicit supplementedtite monthly medianmodel. ‘me traugh madelsuppliesadditionalscreen-paintsbarmapcanstruction.‘me madelof tite mit]-latitudetraugit,describedlii Sectian4.5.12,itasbeenincludedin PLES3 only,witere tite interpalatianpracedureis tite modifledKriging. PLES6makeuseofSSM’sandfitting.

79Física dela Tierra

2000,12,61-103

R.Hanhabaant! B. Zolesi ImprovedQualiiy of Se¡-vicein ¡onospheric...

4.5.3.3.SAIM

Tite objectivewas to developa metitodwiticit canprovidesmapsof foF2evenin situationswiten foF2 abservationsare availableanlyfrom 1-3 iona-sondes(EliseyevandBesprozvannaya,1998).An elfectivelevelof solaracti-vity 18 evaluatedby tite analysisof tite noandatafrom tite previausdays.Realtimeionaspitericdataare titen introduced.As areplacementof tite Kp-index, aneffectiveKp valueis estimatedfrom tite experimentaldataun ardertitat tite pre-dictedfoF2 value is equalto tite observedoneat tite ianospitericstation.Titetraugitpasitianaudtite effectiveKp aregiven by monitoringtite equator-warddiffuse auroralbaundary.

4.S.4.Instantanecusmodelforelectron-densityheightproflle

A neuralnetwork scitemewas proposedby Stanislawskaet al. (1999b)for tite generationof an instantaneousmodelfor electron-densityiteígitt profi-le. Insteadof adirect predictianof Uds profile it was propasedtapredicttite jo-nogramwiticit is unambiguouslyconnectedwitit tite profile.

lite model derivesfram a feed-forward multi-layer structure. A predetermi-nedprofile formulais establishedto describepastprofilesabtainedattite locationof interest. lite training processuseselevenpointson tite trace of a set of pre-viously recordedtypicalionograms.‘me final profile formula is developedstep-by-stepby aix iterativepracess.Tite neuralnetwarkis usedas auniversalappro-ximate connectingtite profiles witit five citaracteristicsparametersof titeionagram.Tite sameparametersare titenusedas inputswhenthe trainingprocessendsup witit aprofile formulathatis bestf¡ttet] ta tite set of training profiles.

4.S.5. Forecastingprocedures

Wititin tite framewarkof titis COSI 251 activity dierearea numberof fo-recastingproceduresrelevantto ionospherictelecommunicationapplicatians.Amajartitrustof tite workitas beentite applicationanddevelapmentof non-li-neartechniquesLa improveianospitericforecastingcapabilities.

Tite DERA (UnitedKingdom) IonospitericForecastingServiceneuralnet-work modelcanpravidepredictionsof tite hourly variationof foF2 from 1 to24 itours ahead.However, tite operationalpackageof tite modelhasbeende-signedun sucita way titat it wouldbe a simplemattertaincorparatepredictivemodelsfor additianalgeaphysicalparametersunto tite sameframework.

A neuralnetworkbasedstand-alonemodelwitit intrinsic inputsitas beendevelopedby Tuluna>’etal. (1999)to forecastfoF2oneitaur in advance.lix titisresearcit,tite feedforwardmultilayerperceptronneuralnetworkwas preferredas arcititecture.Standardback-propagationalgorititm was adopted.

Física de la Tierra2000, 12, 61-103 80

R.Hanbabaauzdfi.Zolesí ImprovedQualiiy of Servicein Jonosplierie...

A mediadfor foF2predictianfrom 1 ta 24 itours aheaditas beenproducedattite AtmosphericSoundingSlation «El Arenasilla»(Spain)byMilcitailov etal. (1999). ‘mis metitadis basedaix amulti-regressianof deviatianof itaurlyfoF2 fram runningmedianwith tite previausabservationsant] Ap índex.Spe-cial proceduresitavebeendevelopedby Mannet al. (1999)to predictfoF2 t]u-ring severestarm periodswitit leadtimes larger titan 3 hours. ‘me metitodcomprisestite basicversionof tite mediadtabe usedduring quiet time andma-

- t]eratelydisturbedconditianswitit switcliing taaspecialmodeto predictduringstronglydisturbedperiods.

Finally, tite COSI 251 forecastingprocedurefor estimatingvaluesof foF2audM(3000)F2itas beenextendedto includefarecastsof TEC ayerEurope(Canderetal., 1999).

4.5.6. Merit of íonosphericindices

Researcitín thisfleid was carriedautwitit tite daubleami of citaracterísingbotit tite relationsitipsbetweenindices and also seasonalcitanges.SeparatecubicpolynaniialrelationsitipsitavebeenderivedbetweenR,2audMF2 indicesfor eacitmantit of tite yearandfor tite up-gainganddawn-comingitalf salarcycles,but adoptedfiguresaretitasefar both italf cyclescombined.A simpli-fied linearrelatiansitipín termsof hasbeenproposedfor usewidi diesepastMF2 for input to mappingcodeswitich arespecifiedin termsof R12, since insamecasestite alternativeuseof die cubierelationswauldresultun inconsis-tenciesandwouldnot bepassible.Small systematicseasonalchangesin MF2itavebeendetectedbut sucit featuresareabsentin tite casesof aud1 indi-ces.‘me reasansfar titeseeffects arecurrentlynot clear,but arebelievedto berelatedto tite fact titat MF2 is givenentirel>’ from Nartitern hemispiteremea-surements,witereasT is basedaix measurementdata for both itemispiteres.litis aspectis subject to furtiter review, but it is suggestedtheseseasonalcitangesarenat impartantsincetitey arecampensatedin tite way tite faF2re-gressionrelatiansitipsaredetermined.

As far as indicesbar usein sitort-termionospitenicpredictianswerecan-cerned,titree separateinvestigatiansconcerningcanbereported:(1) PerraneandDeFrancesciti (1998)itave reviewedtite mast cammon solar, ionaspiteric ant]geamagneticindices witit particularreferenceta titeir application for radio-communicationpredictianspurpases,(u) Benczeetal. (1998) itave undertakenstudiesta deriveadaily MF2 index witicit is perturbedfrom tite mandilyvalueby anexponentialfunctionwiticit dependsaix neutraldensitycitangedueto ge-omagnetieactivity as citaracterisedby die Kp índexaud(iii) Muittarov ant] Ku-tiev (1998a) itave developeda similar typeof índex as tite basisof anewsitart-tennpredictianmediadun witicit regularmanthlymedianvariatiansof agiven íanaspitericcitaracteristicarecarrectedby afactorwhich dependsUne-arly aix tite assaciatedauto-correlatianfunctionexpressedun termsof Kp.

81 Física de la Tierra2000,12,61-103

R. Hanbaba ísndfl. Zotes! ImprovedQuatítyofScrvicebr lonospheric..

4.5.7.Long-termtrendsof¡onospiteriechange

Síndiesiii tite areadealtwitit long-termtrendsobservationsof t]ifferent io-nospiterieparanxeters(Benezeeta]., 1998; Bremer, 1998, 1999a,b;Danilov ant]Mikhailov, 1998; Lastovicka, ¡997). Here mainly ionasondedalaitave bernanalysedconsideringtite following twa main problesns:a) Are Itere markedtrent]s iii tite jonospiterewliicit canidbe importantlar tite ionospbericHP pro-pagatiananil its prediction?it) Are titere trendswitich are cannectedwitbanthropogenicaflycansedatmosphericinfluences(e.g.greenhauseeffeet)?Intite frameof tite COSI251 projeettite first problernis moreimportantwhere-asthe secondproblemis ageneralsejentifieandalsopolitical questian.Portiteinvestigationof tite secandproblernalsosorneadditional ianospitericamI al-maspiterieparametersita-ve beenanalysed.

It itas beensitown that tite detectedionaspiterictrendsare relatively srnallcomparedwidi tite solarant] geomagnetieinfluences.literefore,duringtite nextyearsit is net neeessaryto take mío acceunítheir influencesente ionospitericHP radio propagation.Neve-rthelessdiis effect itas Lo becarefully monitorediiitite future. Especiallyit sitouldbe noticedthattite scientific problemof apos-sible increasingalmaspiteriegreenitonseeffectrequiresfurtiter investigadomis.

ano] lii tite F2 regiantite resultsof tite treno]analysesarepartlycontroversia]eannotbe explainedby tite greenhauseeffect.

4.5.8. Assessrnen¿tofrrans-ionospher¡cpropagation

Researcitiri Ibis areawas partof tite issuescancerninginvestígationof io-nospiteriepropagationeffecísami radio systemsusedbr navigation,geodesyant]rat]io-astronomy.Propagafloneffecís due to baekgroundjonization (retarda-don, dispersion,plane-af-polarisationrotationant] absorption)ant] dat tairte-gularíties(seintillation>wereanalyset]ant] predictionteefiniquesproduced.

Attention primarily focusseden day-to-dayvariability. Due lo eonxplexsolar-terrestnalrelatiansitipsionosphericparameíerssucit as tite peakcloe-tran densityant] ‘EEC sitow a ratiter higit day-to-dayvaríability duela astrongcompetitionalvarlausprocesses.Variability studieslar ‘EEC itavebeencarnet]out, especiallyami tite basisof tite higit resalutionFaradaydalathatweregainedbetween1978 ano] 1989 at Florence(líaN). It tumedaul thai equivalentslabtbicknesscanbe usedtetransíertite variability infarmationlot foF2 in tite fo-llowing way: a) calculatepeakelectrondensityfrom foF2 ano] b) multiplypeakeleetrondensiíywith madelvaluesof slabtiticknessano]usetite resulítoassessvariability of verticalelectroncanlcnt.‘mis sehemeis proposedlot titefuture becauseit is not possíbleta replacetite higit resaludanFaradaydalabyOPS-IEC.

lite secano]topíc aix whichattentienwasfocusedwas tite studyof stonns.During stormsTEC JataindicatelargedeviatiansIt-orn tite meanbeitaviourup

Efsicii ¿ría Tierra2000, I2,6t-l03 82

R.Hanbabaant! fi. Zolesí ImprovedQuality ofServicein Ionospheric...

to200%(Jakowskiant] Schlueter,1999).Tite siatisticalstut]y hasrevealedsig-nificant differencesin tite temporaldevelopmentbetweenSummerandWinterstorms.Ta stut]yte ionospitericvariability un termsof ThC, armsanalysishasbeencaniedout referringta a meandiurnalTEC variation.‘me absalutermsvaluesrangefromabaut1- 4x10’6m-2,witereastite itígbestvaluesappearatlo-wer latitudesprobab¡ydueta tite itigiter absalutelevel of TEC at day-times.‘me percentagedeviationsrangefrom about20% atlower latitudesup to 60%un aurorallatitudes.No signifzcantlongitudinaleffectappears.Usingglobal orregional permanentoperatingOPS ground station networks(e.g. diat fromIGS ) it is suggesteddiattederived‘WC mapsmaybeusadto studylargesca-le structuresun electrandensityano] titeir mavementduring ionospitericper-turbations,‘mus,bat individual storm citaracteristicsof tite geomagnetic/ia-nospiteriestormsas well as commonfeaturesderivedfram statisticalstudiescanbe analysed.

Tite generationof ‘WC mapsallows alsotite computationof ring mapsayerte Europeanarea.Tite applicationof sucitmapstafmt] aut areasof particularvariability was cansidered.GPS basedIEC manitoring can effectively beusedta studylargescaleianospitericperturbatians.

4.5.9. Predictionoftite FI layer occurrence¿md«L condition»

Tite critical frequencyfaFí predictedby tite Du Citarme formula (DuCitarmeet al., 1973)assumeslimits bar tite presenceof die layerasafunetionof te solarzenitit angleandof die solaractivity given by tite index. un testudyundertakenin die frameof COST251, a probabilityfunctiontaevaluatetite occurrenceof die Fi layerano] «L condition»(caseswitere electrandensityprofiles ante ianagramstracessitow a ledgeratiterdianaremarkablecusp,sono critical frequenc>’canbe assignedto tite layer)wasproposedtereplacetitelimits mentionedaboye(Scotto,1999).

4.5.10. fonosplierievariability

A lot of effortsitas beendevotedto te establisitnientof statisticsof te re-latíve deviatíonsof foF2 ant] M(3000)F2Irom a referenceleve] ayertite Ru-rapeanregion(Kaurisetal., 1998; Muhtaravetal., 1999). ‘me relativecontú-butions abquasi-periadicoscillatiansfrom 2 ta 35 days to te variability offoF2 at middle Nortitem latitudesun Europeitave beeninvestigated(Altadilí,1996;Lastovickaano]Mlch, 1996; Altadill andLastovicka,1996;Altadilí a al.,1998;Apostalavetal., 1998; Soléetal., 1999). ‘me resultssuggestdiatdiereisaclearseasonalvariationwit maximumun Suminersolsticeano]mínimuminWinter ene. lite contributiensaremodulatedit>’ tite solarcycle ant] simúlta-neauslyinfluencedby te long-termgeamagneticactivity variations.Tite exis-

83 Física de la Tierra2000. 12,61-103

R. Hanbabaant! fi. Zolesi Improved Quahly of Servia’ in Ionospheric...

tenceant] developmentof tite quasi-2-dayoscillationsin tite plasmafrequencyvariationsof tite F-regionat Nortitern middle latitudesitas beenconfirmedusinga methadologytitat allows to do suchstudyatfixed heigitts. lanospitericquasi-periodicvariability causedby gravity ant] planetarywaveshas beenanalysed(Lastovicka,1999). Effectson tite F-regianof titree typesof waves,witicit comefrom belowant] affect te ionospitere,werecansidered.litesewa-vesareplanetarywaves(periodsof days),tides(periodsof 24, 12, sometimes8hours)andgravity waves(a few minutesto a few hours— partlycoming fromauroralionaspitere/titermospitere,partly coming from below).

Tite annualant] latitudinal variationsof spreadF itave beenstut]ied atmit]-latitudes(Benczeano]Márez,1998, 1999; Benczeano]Poar,1999). ‘me spreadF occurrencefrequencyhasbeen determined.litis occurrenceindicatesanannualvariationwith a maximumin the Wintermontits ano] aminimumin theSummermonthslike the annualvariation of the occurrenceat high latitudes.1-lowever,tite amplitudeof tite annualvariationof tite occurrenceis smallerinyearsof maximum salaractivity titan in yearsof minimum salar activity. It wasshawntitat tite occurrencefrequencyof spreadF decreaseswith increasingge-omagneticactivity,witich is appositeto tite situationat itigh latitudes.Tite ob-servedlatitudinalextensianof tite irregularitiesvanesbetween150 km and.1 lOO km.

In additioncharacteristicschangeshavebeenestablisitedof electrondensityprofile in responseta gravity waveupwardpropagation.Gravity wavespropa-gatingthrougit tite ionospiterecausea cyclic variatianof the verticalelectront]ensity profile. litis signaturehasbeenobservedin a numberof single site ia-nogramsequencestakenat 5 minute ant] 15 minuteintervalsby MoariteadandRadicella(1998). ‘me wavelengthof tite gravity waveswas 600 to 2000 km:lite temporalvariationobservedat the saunderlocationwas transiatedinto aspatialmodelof tite gravity wave.Someparametersof tite channelscatteringfunctionhavebeendeduced.

4.5.11. Disturbanceeffects

lite activitiescarnet]with titis areareferto tite investigationof disturban-ceeffects ami tite behaviourof ionosphericparameters(Bradleyet al., 1997;Bremer, 1996; Bremeret al., 1996; Davis et al., 1997; Fuller-Rowell et al.,1998; Gordienkoant] Kaliev, 1998; Lastovicka,1996; Iulunay, 1996).Tite re-actionof tite middle latitudeionospiteretageamagneticstormsitas beenmo-delledempirically by Fourierdecomposition(Kutiev et al., 1998; Muhtarovant] Kutiev, 1998b)ant] an algorititmbasedon linearpredictionfilters hasbeendevelopedta forecastfoF2 citangesrelatedto stronggeomagneticstonns(Va-silievie ant] Cander,1997). lite poststorm effectsin the D reglanionisationtitat influencesionospitericpropagatedradio waves hasbeen stut]ied withpromisingresults.

Física de la Tierra2060,12,61-103 84

R.Hanbabaant! fi. Zotes! ImprovedQuality ofServicebr Ionospheric...

Specialconsiderationitas beengiven to itigit latitudedisturbanceefbects.Directcorrelationstudiesbetweengeomagnetieindicesant] ionospitericpara-metersgayeun te pastpoarresultstul to tite intraductionby Wrenn(1987)ofasimpleway of treatingaselectedmagneticindex by meansof a time seriesaccumulationob tite index itself. lix tite frameof COSI Action 251, a similaranalysishasbeencarriedout but relatedto te itigit latitude stationsEurope(Perrene¿mdDe Franceschi,1999).Metbodswereexploredteevaluatetite ca-rrelatiandegreeseparatel>’foreachmondi ob te yearano]bar teup-goingandt]own-comingitalf solarcycle. It was demonstratedespecialí>’titat tite corre-lation degreesitows a clear seasanaldependence,being maximumduringSummer.A bettercorrelationwas foundduring tite rising phaseof tite solarcycle.

‘me effects of strongnietearologicalevents on te F-regionionospitereitavebeenstudied.It seemsto bedevelopedbestatiteigitts near200km but af-fectstite witole F2region up taits maximum.it is believedto be transferredup-wardsby gravity wavesof troposphericorigin. Ibis gravity waveactivity wasfauno] to increaseremarkablyafter te Mt. Pinatubovolcaniceruptionun titelong-periodrange(1.5-3 itours) ano] taincreasewit increasingsolaractivity inte 1 1-yearsolarcycle (Lastovickaetal., 1998).

Lastly mentíansitouldbe ¡nadeof investigationsof geomagneticstorniefbectsaix tite Fi -layer.

4.S.12. Troughmodetling

‘me main or mid-latitudetrougit is tite main ianospitericstructureat mio]-andsub-aurorallatitudesduringnigitttime. Thereitavebeena variety of tech-niquesto observetite trougit, including ionosandes,incoiterentscatterant]‘EEC, as well as satellite-bomeexperiments.Unfartunatel>’tite ionosandenet-work is not denseenougitaudnigitttimeionogramsfrom traugb-regionlatitudesobtensuffer fram disturbances.Topsidesoundersare a possiblesourceof in-formationabouttrougit sitape,but tite latitudinalresolutionis insufficient to de-fine tite troughmorpitalogy.Direct useof the latitude dependenceof verticalTECitas pravidedcasestudiesbar tite steepnessof tite walls of te traugit, itswidtit ano] statisticaldatafar teseparameters,but witit camparativel>’largeerrormargins.Relationsitipsbetweente latitudinalpasitionof te trougit ano]local time ant] geomagneticactivity indices havebeengenerated.Howeverdespitetite long itistory of titeoretical work, aur knawledge18 noÉ far-rea-citing.Neverthelessit is ver>’ importanttaitave reliableinfonnatianaix tite sita-peof tite trougit un electrondensityat tite F-layerpeakforterrestrialant] tran-sionospiterictelecammunicatiaixsapplications.

A new approacitto madellingte trougit itas beendemonstrated.‘mis met-ited is basalen a temagraphicimageof tite ionospitereover UnitedKingdomthatis extrapolatedteodier longitudesacrosstite Europeaixsector(Fig. 6). Ini-

85 Física de la Tierra2000,12,61-103

R. Hanbabaant! fi. Zolesí ImpnovedQualiiy of Service¡a Ionospherk...

tial resultsfrom the mappingwereshownto comparewell with observationsfrom jonosondeslocatedbetween12 to 30(Eastof tite tomographyreceiverchain (Mitchell et al., 1997). lix a separatestudya seasonalvariation hasbeenrevealedin the latitudinal position of tite trougit, showing te trougit to befurtiter southduring tite winter titan tite summer(Mitcitell et al., 1999a,b).

Tomograph¡c Iniage: 15/10/93 21:36 UT

Electron density <xl O’ ni’)

700— —700

500— ¡—500

1-rCD ~ - -,

300300—- - -

O.OO3vfl

100-

64 68 70 74

LAIIIUDE (degrees>

Figure 6. Modelof electrondensityover Europeusingthetoniographicimagefroin asateflite pass.

‘me basisfor tite COST251 traugit model is acombinationof a small-sca-

le trougit modelwitit a large-scalebackgroundmadelfor electrondensity:

Ne(h~ 9,2) =ML(h,~¾A+T(h, q~,2)

witere : electrondensity;ML : ¡arge-scalemadel; T : troughmodel; hiteigitt; 4) latitude;2 : longitude.A Gaussiandepressioncould serveas oneof tite simplestpossibilitiesfor T. Aix investigationwascarnet]out un ardertoevaluatethe positionof tite traugitminimumin invariaixt geomagneticcoar-dinateson te basisof alargedatabasefrom the DE2 satellite(Stanislawskaetal., 1 999c). lite sitapeof tite trougit is definedby te simplification of thcmid-latitudeionospiterictrougit modelbasedaix EuropeanfoF2 values den -

vedfrom analysisof Magion-3 HF radio spectrometerdata (Bradleyetal.,1998). Aix exampleof instantaneousmapof foF2 generatedby the PLES3procedurewitit titis troughmodel during an intensemagneticstorm is pre—sentedun Fig. 7. Following companisonwitit otiter modeis,it was demonstra-

66 72

Física de la Tierra2000,12,61-103 86

R.Hauzbabaant! B. Zolesi ImprovedQuatityof Servicein hauzosplierie...

ted titat titis modelis well suitedfar telecommunicationsapplications(Tulu-nayetal.,1998).

70

6sa’0,

~‘- 60

-n4~I

SG4-,nl

O 50-‘--fi

0-<si~,

aa,40

60

Figure 7. Exampleofinstantaneousmapof foF2 (xlO/MHz) for 10 Muy 1992st 2200 h UT generatedby te PLES3procedurewith thenew mid-latitudetroughmodel asidfive measurements.

4.6. Comparisonproceduresandresults

‘me overalí accuracyof differentfaF2, M(3000)P2ant] electrondensityprofile modeisdevelopedt]uring tite projectneedstebe assessed.lii arderteobtaincomparableresults,tite samedatasetandtite sameprocedurewasap-plied witenmakingcamparisansbetweenmeasurementsant] te resultsof titemodels.Testswerecarneo]ant independentlyun Bulgaria,Italy ano] the UnitedKingdam, witit identicalresults.

‘mineen vertical-incidenceianosandestationswere selectedbar testing,giving goadspatialsamplingof tite COSI 251 area.Por long-termmapping,montly met]ianswereused,witile bar instantaneousmappings,itourly datafram te 5»~ da>’ of eacitmontit were used.Quality cbecksaix tite testingda-tabasesresultedin tite eliinination of mostof teunreliabledala.Farlong-termmapping,mondiswidi gapsof total lengthgreatertan 15 days,or singlegapoflengt greatertitan 10 days,ar extremevaluesexceet]ing20% of tite totalnumberof vaho]valuesfor tat mondiwereelinúnated.Honrswit numberofvalid values lesstban5 werealsoeliminated.Porinstantanenusmapping,ex-tremevalueswereeliminated.A titresitolo]probability of 0.99 was usedto de-fine te extremevaluesbar eachstatian.

Tite testing databasefor electron-densityheiÉitt profile includedtapsideprofiles obtainedfrom COSMOSant] ISIS-Alouettetopsidesaunt]ersano]bat-tomsit]eprofilesframfive statiansin Ita]>’, Polaud,Spainant] UnitedKingdanx,obtainedby inversianof vertical-incidenceianagramsusingPOLAN (Titite-

0 10 20 SO 40 50Geographic 1cngit~uc1e /deg

87 Física de la Tierra2000,12, 61-103

1?. Hanbaba ant! fi. Zolesí improvedQuahlyofServicein Ionospheric.-.

ridge, 1985) ant] ARIISI (Reinischet al., 1988). An overalí rms error wascamputedfor eacitprofile, andtiteseweretitenaveragedayerte wholedata-base.For bottomsit]eprofiles,it wasdecidedtofocusami tite F2-layerin view ofuncertaintiesin tite invertedprofilesun tite D- ano] F1 -layer.

4.6.]. Long-termmappingofmonthlymedianfoF2 ant! M(3000)F2

The rnodelswere rankedaccordingto the global rmserror. In addition,au-xiliary testswere carriedout to ensurea goadmergewith tite ITU-R globalmappings.Tite madellerswereasket] to supply24 maps,un te fornv¡ of a lati-tude-longitudegrid of results (resolution2.5( far latitude, 5( far longitude)for tite COSI 251 area,for aselectionof montits ano] itaurs. Mapswererequi-red for 0000,0600, 1200ant] 1800 it UI for tite selectedmanths.A grat]ientciteckwas carriedout at tite bardersof tite maps.

Mostmodelsimproveon tite currentITU-R model.The gradientcheckswere acrucialpart of tite restingprocedure.A numberof modeisdio] natpassthat test,somewitit veryseriousproblems,somewitit minarproblemsonly. Far foF2, in retrospective¡nade, MQMF2 is tite only modeltitat sa-tisfies tite gradientchecksano]hasrms error lesstitan0.5 MHz. Tite nextbestmodel witicit satisfiesthe gradientciteck is PLES1. Of tite madelswitit R12 inputoní>’ ano] satisfyingtite gradientcitecks, MQMF2Ritas thebest rms error. For M(3000)F2, UNDIV is tite model with tite bestrmserrar. It alsa satisfies the gradientchecksant] R12 input requirements.MQMF2 is aclasesecano].

4.6.2. JnstantaneousmappingoffoF2 andM(3000)F2

Againte modelswererankedaccort]ingto mserror. For foF2 mapping,tite Kriging algorithmsin tite methodsPLES2 ano] PLES3 pei-form bettertitan tite fitting algorititms un the metitot]s PLESSant] PLES6ano] give exce-llent resultswitile for M(3000)F2mappingfitting algorititnis un tite metitodPLES5providetite bestaccuracy.Forboth citaracteristicstite performanceofMQMF2 is excellenttaoano]hascomparableaccuracy.Tite useof tite troughmodel in PLES3doesbring a smallimprovement,but requiresat]ditionalin-put data.

4.6.3. Electron-densiíyiteightprofile

Comparisonshavebeenmadebetweentite profilesobtainedfrom COSI-PROFant] IRI modelsant] measuredtopsideant] battamsideelectran-densityiteigitt profiles.

Física de !~ Tierra2000,12,61-103 88

R.Hauzbabaant! fi. Zotes! hnprnvedQuality ofServicein Ionospheric...

‘me COSIPROFmodelgivesa substantialimpravementayertite Rl mo-delfor tapsideprofiles aix tite testingdatabase.‘me performanceof te COST-PROFant] Rl modelis similaron tebattomsideprafilestesteditere,widi botitrnot]elsgiving aver>’ small rms errar.Haweverit sbouldbenatedtitat tite IRIitas sorneserioussitortcomingsin its formulationfor te upperF region.Oneoftiteproblemsappearsun itigiter latitudesatitigit solaractivity. Tite profilesteno]to acquireunrealisticallylargeprofile scaleiteigbts.Evenixegativesealeiteightsappear(exponentialincreaseob electrondensitywidi heigitt aboyetite F2-layerpeak).COSTPROFremainsrealistieunder alí conditions,botit witit global(ITU-R) ano]regional(COSI251) inputparameters.

6. METHODOLOGY FOR CHANNEL SIMULATIONFOR HE SYSTEMS

‘me primar>’ objectivewas tadefmea bunctianalcitannelsimulatorarciti-tectureTer HP systemsin Europe.‘mereareclearshortcomings¡ix tite Watter-sonmadel(Wattersonet al., 1970)witicit is widely usedas abasisof existingsimulators,not leastbeingte limited dataaix witicit te modelwas originallybased.Also, tite Wattersonmodel inhereixtly describesnarrowbano](appraxi-mately3 kHz bandwidtit)citanneis,witereasfuture simulatorssitaulo]becom-patiblewith arangeof systembandwidthsnp teatleast1 MHz, correspondingta atime resolutionof 1 micrasecono].

A new basicstructurefor tite realisationof a channelsimulatoraix ge-neralpurposeDSP-boardsitas beenproduced(Van derPerre,1997; Van derPeneet al., 1997).Tite functional elementsof tite simulatorarespecifiedunsuch a wa>’ tbat a completerange of citannel types,appropriatefor HPtransmissionsystemtestingup to bandwidtitsof at least1 MHz, canbe im-plemented.lix additian,a Wattersanmodelma>’ be simulated,if required.Consideratianwasgiven to tite applicability of a replay-typesimulator unte COST251 context.In general,it was felt tat tus dio] not providean ap-propriatebasisof flexibility for dxcrangeof applicationsenvisaged.Ha-wever,barunpredictableeventssucit as itigit latitudedisturbances,it wouldbe passibleta incorparatereplay segmentsunto tite overalí timeprofile oftite simulatortestprogram.

7. MODELS FOR USERSFOR OVERALL PERFORMANCE

PREDICTION

7.1. General propagation modell¡ng

Researcitactivity conceminggeneral propagation modelling was orientedaboyealí towards producing simulatian tools bar overalí perfonnancepret]ic-

89 Física de la Tierra2000. 12,61-103

R.Hanbabaauzd fi. Zolesi improved Quality of Service in Ionospheric. . -

tion. Numericalmethoo]sfor tite modellingof 1fF radio wavepropagationin titereal ianospherehasbeendeveloped(Gitermet al., 1998a).Tite term «realio-nospitere»is understoat]to describeamediumwitit threedimensionalslowly-varying electromagnetiepropertiesof tite backgrount]andtime dependentin-itamogeneitiesas well as deterministicandrandominhomogeneities,inclut]ingextrabent]ingof rays,multi-rayeffectsano]diffraction effects.‘mesemethodsarebasedon te camplexpitasemethodfar wavepropagationin a slowly var-ying mediawith local initomogeneitiesembedded.

Specialattentianitas beengivento tite problemof the HP fluctuatingcitan-nelcitaracterisationun tite caseof frequenciesclaseano] aboyetite MUF. Ihispanicularsituationof askip field is describedby a specifxcally developedwave fiel] integral representationin termsof t]iffracting camponentwaves(Oitermetal., 1998b, 1999). In at]dition Éhetecitniqueitasbeendevelopedbarmodellinglarge-scaleionosphericstructureswit tite itorizontalgradientsof titeionaspitericelectrandensity.Researcitactivitiesaix propagatianmot]elling in-volvedalsotite constructionof a computercodeto take into accaunttite ma-vementof tite main ionaspiterictrougit.

7.2. F2-Iayer basicMUF variabllity

lix tite higitly cangestedHP bano] thesignal quality generallyincreasesastite frequencyapproacitestite MUF. HF propagationpredictiansproceduressucit as titat of RecammendationITU-R P.533(ITU-R, 1995)canprovideguidanceaix tite expectedperformanceof eircuits. liteseproceduresareba-seo] aix averagebeitaviourof tite ianospitereano] sitaulo] not be expectedtagive informatianbeyandtiteir statistical capabilities.It sitaulo] be notedthat generalí>’ operatorsano] frequencymanagersteno] to select frequen-cies,whicit areanywiterefrom 65 to 90%of tite predictedMUF. If madeissitaulo] pravidea betterpredictionof tite distribution of tite daily MUF’saraun]tite mandil>’ medianMUF, thenit might bepossiblesuccessfulope-rationat frequenciescloserto tite predictedMUF. Titis woult] be very ad-vantageousfar spectrumoccupancysince it wauld widen tite availablebandsfar use,especialí>’t]uring periodsof low solaractivity. Furthermoretite evaluatianof the MUF variability ma>’ facilitate improvedfrequencysitaríngun HF terrestrialsystems.

lite variabi!ity of tite basicMUF (i.e. tite itigitest frequencyby witicit ara-dio wavecanpropagatebetweengiven terminals,an a specifiedoccasion,byionaspitericrefractionanly) with patit lengtit itas beeninvestigatedby Bradleyetal. (1998).It itasbeenfauno]titat tite upperdeciledeviationfrom tite medianusuallyincreasessligittly with distancewitilst tite lower deciledeviationfalís,but citangesarerelativel>’ small. Oneof tite key aspectstreatedwas tite esta-blisitmentof boundsfor tite relativemagnitudesof tite day-to-dayvariationsinfoF2 ano] M(3000)F2on quiet ano]disturbeddaysano] itow titesecitangewitit

Física de la Tierra2000,12,61-103 90

R. Hanbabaauzd fi. Zotes! Improved QuaIiiy of Service in Ionospheric...

season,locationant] solaractivity (Fotiadisano] Kouris, 1998).Tite day-ta-dayvariation ab tite basicMUF itave beencomparedwitit titase of foF2 ant]M(3000)F2(Kouris etal., 1999).Tablesof valuesfor botit tite ratio of te up-perant] lower decilevaluesto te medianbasicF2-MUIF itavebeenprovided.lite F2-layerbasicMUF variability ma>’ beevaluatedbar tite COSI 251 areafrom tesenew referencefigures,whicit aresligittly t]ifferent from tosegivenun CCIR (1980).

7.3. Stat¡st¡calcharacterisationof HFchannel

‘me statisticalcitaracterisatianof tebeitaviourof te transniissioixcitannelwas animpartanttopic of researcit.Sincejonospiterevaneswitit time, spaceant] frequencyin man>’ scales,tustaskwasnatatrivial activity. Difberentap-praacitesto describete scatteringfunetionwerestudiet] (OitermandZernov,1998;Zemavetal., 1999)ano] metitodsto statisticallyanalyseano]citaracterisete variationsof the HE citannelresponsehavebeendeveloped(Arikan ano]Eral, 1998; Arikan, 1999).

Sincemostof te widely availabledatasetsconsistof vertical-arablique-incidencesoundingdata,citaractenisationparameterscanbe extractedfromsucit data sets utilising oní>’ tite amplitude infarmationob receivedsingletonesignals.By te propasedmetitodalogy,robustestimatesta time var-ying meanano] varianceof te citannelresponseis obtaiixeo].For titis purpose,slidingwindaw statisticsof te availabledatais used.un applyingte slidingwindow statistics,tite window size is citasenas longas possibleto pravidebetterestimatesano] still sitortenougit to capturete underlyingtimevaniationob tite citannelresponse.Basedan tite estimatedvarianceof obtainedre-sults,detaileo]justificationante properwindaw size is given. Iii ardertaob-tain morereliable estimates,tite data is medianfiltered prior to statisticalanalysis.‘mis new statisticalanalysisapproacitis appliedto availabledataob-tainedfrom measurementcampaigns.Tite resultsob tite statisticalanalysisconfirmedtite expectatianson tite pit>’sical beitaviourof tite ianospitericcitannel.It was fauno]tat tite midíatitudesinglet’requencycitannelis slowlytime varying ano] localí>’ stationaryun a sliding window of 22 secands.Itwascont’irmeddiat during earl>’ morningitours ant] mghtitours,te citanneliscansiderablymorestablebarcamniunicationpurpasescamparedto dayano]earlyeveniixgitaurs.

Witen consideringte establisitmentof ebficientano]reliablecommunica-tion links thraugitte ianospitericcitannel,scatteringfunction ano]oter citan-nel coherencebunctionsare usualí>’ computedano] examined.Twamajaralter-nativeswerecansideredbar tite numenicalcamputationof diesefunctions.Intite first metitod,tis is acitievedtitrougit tite estimationof tite autocarrelationof citaxinel impulseutilising te transmitteo]ant] receivedsignalsfrom contra-íleo] experimentresponse(Arikan ant] Arikan, 1997). lix te secano]suggested

91Física de la Tierra

2006, 12,61-103

1?. Hanhaba¿mdfi. Zol¿si lmprov¡sdQuah¡y ofSen’kc iii Ionospheric...

metitod,estimationof te HE citannelimpulseresponseis tite fu-st stepant] thescatteringfunctjan is computedutilising tite channelresponse(Aúkan, 1998,1999).

Afta assessingtite elfieieney,performance-,coniputationaleomplexityano] casÉof thesemetitot]s la estímatitigte HP channelresponse,it is con-cluded that ¡CalmanFilía estimatorwitit properadaptationprovidestite-besÉesrimate.‘Ehusa meíhodalogywas developedtaestimaretimevarying,fading, ¡nultipath HP channelresponsenumerically titrough aix adaptiveKalman Fílter basedchannelestimatorutílisíng tite data froni cantrolledexperiments(Arikan, 1998, 1999). Majar advantageof adaptiveKaInianFilter basedchamad esúmatoris thatit can efficiently trackthe variationsiiitite channelresponse.Titas tite physiealmoo]eiof thc channelcanbe incor-poratedloto te estimatar.Titis capabili¡ydaesnoÉ exist un ata ¡nethods.Adaptationof te ¡CalmanFilter tu rite variarionsiii te ehannelresponsehasbeenimplementeo]atvarjoasleveisant] sealesespeciallyay tite initialisationano]parameteradaptation.Basedon simulationsof HP channeisundergoad,moderateant] poarionosphericstatesant] lar Signalta NaiseRatias(SNR)from 10 to 40 dB, it hasbeenobservedthar adapúve¡CalmanFilter conver-ges fasta for lowesterror not only for goadano] moderateconditionsant]high SNRsbut alsakw poarknospitericstate-ano] low SNRs.Titas tite sug-gesteo]adaptatiansignificandy improvesthe perfornianceof te standard¡CalmanFilter estimator.

7.4. Modellingof HP spectral occupancy

Measurementsof HP spectraloccupane>’allowedtite developmentof (i)low- ant] higit-angleoccupancyovei~NarthemEuropeano] (u) variationof oc-cupancywith azinxuth(EeonomoucÉ al., 1998; Panújarosetal,, 1998).

Tite logit ¡nade],witit the appropriatemodel index funetiony~, forms titeLaycockXiottmadelfor esrimatedcongestionQk for ¡TU frequencyallocationk (Pantjiarosetal,. 1997).Tite mdcxfunetiony~ of tite paranreterson whicit oc-cupancydependsis detenninedby statisticalanalysisof tite experimentaldata,Por cae-it receivingsite-, estimatesof Yk aredeterminedfor eachof tite 95 ITIJfrequencyallocations,ano] tite variatianof diese95 index functionswith fre~quencyistitenmadelled,ta giveasingle modelíndexfunetionapplyingto titecatireHP spectrutn.Uy dcvelopingtite samefarmof moJellar cadi ¡ite~ ant]comparingtite coeffieientsof tite parametcrsof titesemodeis,a singlemodelkw loar sitesis derivedby attributingdiffere-ncesin me-asure-Ucongestionvaluesto differencesin te locatiomisof tite reeeivingsites.Tite titting pracedureay-tenxptsto modeltite systematiecamponentof tite dala,witilsí allowing Lot titerandomconiponent.‘mis givesa meidelwhich is betterable to pre-dict futurevaluesof accupaney,ant] witicit maybe develapeclreadiiyby tite- inclusionofadt]itional measureme-nts.

tisicede Mg Tierra2000, 12,61-lOS 92

R.Hanbabaauzdfl.Zotes! ImprovedQudhtyof Servíceiii Ionospheric...

7.5. RealTimeFrequencyManage¡nent

Develapinentsun tite availability of flexible ant] rapit]ly-contrallableequipment,un broad bano] antennasano] un tite regulataryenvironmentnowpermit tite economicpossibilityof adaptiveEF systemsto provideitigit per-bormanceusingrealtime frequencymanagement.Studieswititin this COSIAction itavebeenconcernedwith aspectsab tite control algoritms,modelsano] diagnosticmeasurementmetitodsto enitancetite frequencymanagementfunetion.

‘mere is now aix increasingneedLo provide higit dataratecaminunicatianano]die averalíutilisation of tite spectrumwill beenitancedby considerationofat]aptivity in tite modulationwavefornxsant] dataratesin responsetatite cha-racteristicsof tite citanneltransberbunction.

‘me tapiescansideredo]uring tite COSI Action relevantta titis subjectweretite following.

1. The effectiveuseof ionospitericsoundingwitit very low pawers itasbe-enintroduce-o],titusminimising tite interferencecauseo]by te neces-sarydiagnosticmeasurements

2. ‘me probabilisticmodellingof interferencewas undertaken.‘mis workma>’ be usedwitbin tite frequene>’selectionalgoritlxm.

3. Tecitnique-sfor tite measurementof tite scatteringfunetion,basedaixtite spreao]ingob signaldela>’ ant] Doppleror on afractal measurementabtite funetion,itave beendevelopedano]sitaulo] leadta statisticalmo-deis.

4. Sitort-term forecasting(itours aitead) ano] ver>’ sitort-termforecastíng(minutesahead)itas beenconsidered.Iii particular,metitodsof datafu-sianant] aneuronalapproachwereconsideredwhicit enableestimatesofinterferenceto bemade.

8. THE COST251 COMPUTERPROGRAMS

Sevencomputerprograms(Table 1) itavebeenpreparedto providepredic-tions in accordancewitit te recommendedpracedures(Stanislawskaet al.,1998). ‘me calculationrelatedeititer to resultsfor difberent times at a singlecitasenplace,or as mapsayeraspecifiedareaaL a single time. Locatiansano]mapsmaybe anywiterein tite world, witit interfaceas apprapriatetate ITU-R global inaps of ionospitericcitaracteristiesant] to tite ]RI electran-t]ensityiteigitt profile model. ‘me computerprogramsare availablefrom anonymausftp ://haydn.cbk.waw.pi.

93 Físicade la Tierra2000, 12,61-103

R. Hanbaba ant! fi. Zoles! improvedQualiíy of Servicein Jonospheric...

Table 1: Lisí ofprogramíant! outputparameteríavailab!efromanonymousftp :1/hayt!n.chk. waw.pl.

Programname

AuthorlOrganisation/Country

Jonospheriecharacteristie

Prediction

MQMP2R A. Mikitailov, Instituteof Apptiet] Geopitysics,Russia

foF2 Long-term mapping

UNDIV R. Leitinger, Universityof Graz,Austria

M(3000)E2 Long-termmapping

PLES2 1. Stanislawska,SpaceResearchCentre,Polant]

foF2 Instantaneousmapping

PLESS 1. Stanislawska,SpaeeResearchCentre,Poland

M<3000)F2 Instantancausmapping

COSTPROF R. Leitinger,Universityof Graz,Austria

electron-densityheigbtprofile

Long-termmappingant]instantancausmapping

COSTTEC R. Leitinger.Universityof Graz,Austria

TEC Long-term mapping

CORLPRED 1. Kutiev, GeophysicalInstitute,Bulgaria

foF2 Short-termforecasting

9. CONCLUDING REMARKS AND FUTURE ACTIVITIES

COSI 251 itas undertaken ano] completeda large variety of tasks. Tite pri-mary objectiveof tite prajectwas to recommendimproved proceduresto im-prove qualityof ionospitericteleconiniunicationsystemsplanningant] operationin Europe.Majaradvancesitavebeenabtainedin janospiterieant] plasmasp-iteric modellingant] systemperformancepredictionundernormalant] disturbedconditians.Tite resultsobtainedt]uxáng tite Action can be applieddirectly taoperatianaluse.Tite Final Repon,basedaix materialprovidedby 34 contribo-ting autitorswitin titis Action, is et]itedby Hanbaba(1999)ant] publisitedbytite SpaceResearcitCentre,Warsaw,Poland.It is anticipatedto annauncetitattite newpracedureswill meetusersrequirementsant] wiIl be adoptedfor a longtime.

Co-operationwititin COST itas pravidedefficiency ant] informality oftite co-aperationwith otiter scientistsbeyondtite political ano] administrativeconfinesof Europe.Participaiitsgayeempitasisaix spe-eifietemesfollowing

94Físicade la Tierra2000,12,61-103

R.Hanbabaant! fi. Zotes! ¡mprovet!QuahtyofServiceir> Ionospheric...

titeir own expertiseano] interest but witit an applicationarientedgoal. Itsitaulo] be notedtitat COST insuresa crassbertilisationof ideasano]warkingtogedierin a largergroupta providebasterprogresson agreaternumberabfrants tan anecountrycault] managealone. It was recagniseddiatte COSIframework is tite only frameworkto developtitis researcitindependentlyofecanomicalinterests.Only tite scientific interestano] tite potentialfuture ap-plicatianswerete key factars.‘me addedvaluein carryingautresearcitunderte COSI frameworkis f¡rst of aH tite establisitmentof new networks.‘me ca-llabarationestablisitedo]uring COSI 251 pennittedte establisitmentabviablecombinedteamswitit sufficient size ano] resaurcesano] enlargedtite callaba-ratianbetweeninstitutesin PuropeanCauntries,stimulatingresearcitin manyof titem.

Taking unto accountte new horizansof Eartit-spacesystemsdevelop-me-nt including titat interestsEuropeamore carefuldefinition of tite upperatmaspiterestructureano]variability particularlyin its topsideregianis ne-eded.Suchdefinitionis an importantbasisfor a COSTActianorientedta-wards tite evaluatianof tite upper atmospitereeffects on tite advancedEartit-spacecommunicationSsystemsincludingnavigationalsystemsano]LEO ano]MEO satellitesconstellatians.Cansistentwitit titis, afour-yearfo-llaw-onActian «Effectsof tite upper atmospitereaix terrestrialano]Eartit-spacecommunicatians»itasbeenpropase-o]witit tite fallowing objectives:(i) ta perfarmstudiesto influencetite tecitnicaldevelapmentano] tite im-plementationof new cammunicatianservices,(ji) ta developmetitodsano]algarititmsto pre-dict ano] ta minimize tite effectsof ionospitericperturba-tionsano] variationsaix cammunicatians,(iii) ta collectadditionalano]newianospitericano] plasmaspitericdatafar naw-castingano] forecastingpur-posesano] (iv) ta stimulatefurtiter ca-aperationun tite damainof ianosp-iteric ano] plasmaspitericpredictionano] forecastingbar tenestrialano]Eartit-spacecammunications,including interactiverepercussionsaix titecarrespandingstandardsun titis fiel], taking into accountusersprese-ntano] future need.

ACKNOWLEDGEMENTS

‘me work caniedout duringCOSI 251, describedin titis paper,itas madeit passibleby alargenumberof participantsof dibferentEurapeancauntriesinputting tagedierteir camplemeixtarycompetenceto appraacittecommonpro-blem ab ano] to propaseoperationalsolutions.A specialacknawleo]gementismadeto tite invaluablework abalí participantsin titis Actian. ‘manksta titeirdedicatedefbortsant] activeca-operationayerte four-yearperiod,COST 251was alargesuccess.Man>’ tanks un particularbarprovisionobmaterialinclu-dedun titis paper.Finally aur tanksgo to AngelaVernanbar iter entusiasmano]assistance.

95Física de la Tierra

2000. 12, 61-103

R.Hanhabaant! fi. Zotes! improved QuaIily of Service in Jonospheric...

REFERENCES

ALBERCA, L. F., JIJCHNJKOWSKI, G., KouRís, 5. 5., Mnca~n~ov, A. V., MIKHAILOV, V. V.,MIRO, O., DE LA MORENA, B. A., MIJKHTAROX’, P., PANCHEVA, O., SotÉ, J. O.,STANI5LAW5KA, 1., VILLANUEVA, L. amidXENOS,TH. (1998): «Comparisonof vanausfoF2Single StationModeisbar Europe-aixArea»,ActaGeophysicaPolonica,46,(1),77-91.

ALTADILL, D. (1996): «Onohe 18-dayquasi-periot]icoscillationin thc ionosphere»,,4nn.Geophysicae, 14, (7), 716-724.

ALTADILL, 13. ant] LASTOVICRA, 3. (1996):«Quasi-five-andten-dayoscillationsiii foF2ant] their possibleconnectionwith oscillationsat lower ionosphericheights»,Ann.Geofis., 39, 705-7 12.

ALTADILL, 13., SOLÉ,J. G. andAPOSTOLON’, E. M. (1998):«Pirstobservationsof quasi-2-day osdillations in ionosphericplasmafrequencyat fixed iteiglios», Ann. Ge-ophysicae, 16, 609-617.

ANNAKUIÁE’V, 5. K., AFosaN,V. V., DEMJNov, MG., asid KARPACHEV, A. T. (1997):«An empirical formulafor tite positionof rnaínionospherictroughduringtite peñadof geomagneticstorm»,GeomagnetismyAeronomy,37, 183-185, (in Russian).

APosroLov,E. M, ALTADILL, 13. andHANBABA, R. (1998):«Spectralenergycontribu-tiosis of quasi-periodicoscillatiosis (2-35 days)to tite variability of the f

0F2»,Ann.Geophysicae,16, 168-175.

ARIKAN, F. (1998):«Estimationof HP impulseresponseant] computationof scatteringfunction»,2nd COST 251 WorkshopProceedings,Side,Turkey, March 1998, (cd. A.Version,Rutherfort]AppletonLaborator>’,UnitedKingt]om), COST2S1113(98)005,273-280.

ARIKAN, E. (1999):«Comparisonof scatteringfusictioncomputations»,4íh COSI251WorkshopProceedings,Madeira,Portugal,March 1999, (cd. A. Vemon,Rutherfort]Apple-tanLaboratory,UnitedKingt]om), COST25lTD(99)008,25-36.

ARIKAN, E. mit] ARtKAN, 0. (1997): «Reviewof equalizationme-thodsfar HE channe-Is»,Jamo COSI251/IRI WorkshopandWorking Group SessionsProceedings,Ktih-lungsbom,Germasiy,May, ¡997, (cd. B. Lundborg,Defence-ResearchEstablish-mení,Swet]e-n),COST25lTD(97)006(PartII), 1-5.

ARIKAN, U. ant] ERaL, C. B. (1998) «Statisticaleharacte-rizationof time variability inmidíatitudesingle toneHF channelresponse»,RadioSc!.,33, (5), 1429-1443.

BENCZE, P. and MÁRCz, E. (1998): «Cantributionto the variability of ionosphere-

spreat]E at midíatitudes»,2,d COSI 251 WorkshapProcee-dings,Side, Turkey,March 1998,(cd. A. Vernon,RutherfordAppletonLaboratary,United Kingdom),COST25lTD(98)005, 202-210.

BENCZE, P. AND MÁRCZ, E. (1999): «Furtiter investigationsrelated00 tite mid-latitudespre-at]E pitenomenon».3~ COSI 251 WorkshopProceedisigs,El Arenosillo,Spain,October1998,(ed. R. Haubabaant] ItA. de laMorena),COST25ITD(99)003, 195-205.

BENCzE,P. AND POOR,A. (1999):«lnvestigationsreferring to Ohe forecastingof spreadE», 41h COSI251 WorkshopProceet]ings,Madeira,Portugal,March 1999,(cd. A.Ve-man,RutiterfordAppletonLaboratory,UnitedKingdom), COST25LTD(99)008,183-186.

BENCzE,P., SOLÉ, O., ALBERCA, L. E. asid PooR,A. (1998): «Long-term changesofhmF2:possiblelatitudinal asidregionalvariations»,2nd COSI251 WorkshopPro-

Físicade la Tierra2000, 12, 6 1-103 96

R.Hanbabaant! fi. Zolesi ImprovedQuahlyofServiceir> Ionospheric...

ceedings,Side, Turkey, March 1998,(cd.A. Vernan,RutiterbordAppletanLaba-rataiy,UnitedKingdam>,COST25ITD(98)005,107-113.

Bossv,L. (1987): «Tite determinationof LAY-para¡netersbar a givenprafile»,Adv.SpaceRes.,7, 35-37.

BRADLEY, P.A. (1995):«PRIME(Predictianano]ReLrospectiveIonasphericMadellingayerEurope),COST Action 238 Final Repon,AdvanceIssue’>, October 1995,Cammissionof tite EuropeanCommunities,Brussels.

BRAOLEY,P. A., CANDER,Ls. R.,KurrEv, 1. ano] I-IANBABA, R. (1997):«PRiME(COSI238)studiesof ionospitericstarmefbects»,Ádv.SpaceRes.,20, 1669-1678.

BRADLEY, P. A., JIJCHNIKOW5KI, G., ROTHKAEHL, H. andSTANI5LAW5KA, 1. (1998):«Instantancausmapsof tite Europeanmiddle ano]high-latitudeionospherefor HEprapagationasse-ssments»,Adv.SpaceRes.,22, 861-864.

BREMER,J. (1996): «Tite influenceof itigit speedplasmastreamson tite ionospliericplasma»,J.Atmos.Terr. Phys.,58, 845-854.

BREi~¶n~,1. (1998): «Trent]s lii tite ionosphericE ant]F regionsayerEurope»,Ann.Ge-ophysicae,16, (8), 986-996.

BREMER,1. (1999a):«Leng-termtrendslii ionosphericplasmaparasnetersayerEurope»,3~ COST 251 Worksitop Proceeo]ings,El Arenosillo, Spain,October1998 (cd. R.Hanbabaano]B.A. delaMorena),COST25íTD(99)003,289-296.

BREXffR, 1. (1999b):«Long-te-mitreno]sof ionasphericcitange»,4~COSI251 Works-hopProceedings,Madeira,Portugal,March 1999,(cd. A. Vernon,RutherbordAp-pletonLaborator>’,UnitedKingt]om), COST25lTD(99)008,138-141.

BREMER,1., LASTOVICKA, J. ano] TUIUNAY, Y. (1996):«Influenceobtite interplanetarymagneticile-Id aix tite variability of tite mid-latitudeF2-layer»,Ann. Geofis.,39,721-727.

CANDER,Ls. R.,Lnvy, M. F.,Dxx, M. 1., MuHmRov,P.,Ku’nEv, 1., RADICELLA, S.M.,NAVA, B. aud LEITINGER, R. (1999): «Ianosphericborecastmodel for faF2,M(3000>F2ant] TEC ayerEurope»,

4d, COSI251 WerksbopProceedings,Madei-ra, Portugal,March 1999,(cd. A. Vanan,Rutherfort]ApplctonLaboratory,UnitedKingdom),C05T25lTD(99)008,206-213.

CCIR (1980):«Secano]C.CIR. computer-basedinLerim meritad for estimatingsky-wavefiel] strengtitant] transmissionlassatfre-quenciesbetween2 ant] 30 MHz»,Supplementso Reporí252-2,InternationalTelecommunicationUnion, Geneva.

DaníIov AD. ant] Mildiai¡ov AV. (1998)«Long-termtrcndsof tite F2-¡ayerparame-te-rs: A newapproacit»,2nd COSI251 WorkshopProceedings,Side-,Turkey, Marcit1998, (cd. A. Vernan, Rutherfort] AppletonLaboratory,United Kingt]am),COST2SITD(98)005,114-121.

DAVIS, C. J., Wwo, M. N., LOCKWOOD, M. ant] TULUNAY, Y. (1997): «Ionospitericano] geamagncticresponsesta changesin IMF Hz :A SuperposedEpocit Study»,Anuz.Geophysicae,15,217-230.

Du CHARME, E. 13., PETRIE, L. E. ano]Eynua,R. (1973): «A metitodforpredictingtiteFi layercritical brequencybasedaix tite Zurieit smoothe-dsunspatnumber»,RadioSci., 8, 837-839.

EcoNoMou,L. y., GREEN,P.R.,Gorr,G. F., LAYCOCK, P.J., BaoMs,M. ano] BOBERa,5. (1998): «HFspectraloccupancyathigit latitudes»,ff98, NordieShartwaveCan-ference,Gotland,August1998,1.3.1-1.3.10.

ELI5EYEV, A. Y. ant] BESPROZVANNAYA,A. 5. (1998):«Adaptivemodelof baF2t]istri-bution for instantane-ausniappingano]long-tennpredictian»,2~’ COSI251 Works-

97 Físico de la Tierra2000. 12, 61-103

R.Hanbabaant! fi. Zolesí Improved Qualiíy of Service in Ionospheric. -.

hop Procee-dings,Side,Turke-y,March 1998,(cd. A. Vemon, Ruthe-rfordAppIe-tonLaboratory,UnitedKingdom),COST25IID(98)005,257-262.

FoTIADIs,O. N. andKouius,5. 5. (1998): «Estimationof sornecharacmeristicparametersof tite F-re-gionvariability»,

2nd COSI 251 WorkshopProcee-t]ings,Side-, Iurkey,March 1998, (ed.A. Vernon,Ruthe-rfordApple-tanLaboratory,lJnited Kingdom),COST251TD(98)005,196-201.

FULLER-ROWELL, 1.3.,CODRESCU,M. y., ARAUJO-PRADERE,E. ant] KUTJEV, 1. (1998):«Progressin developingastorm-timeiosiospite-riccorrectionmodel»,Adv. SpaceRes.,22, 821-827.

GHERM, V. E. andZERNOV, N. N. (¡998): «Scarteringfunctionof thefluctuating lo-nosphere-un tite HF bant]»,RadioSc!.,33, 1019-1033.

GHERM, V.E., ZERNOV, N. N. ant] ELI5EYEv, A. Yu. (1998a):«A compute-rprogran3fortite calculationof statisticalcharacteristicsof 1-IP fie-Id in the- ionosphericchannel»,2nd COSI251 WorkshopProcee-dings,Side,Turke-y,March 1998, (ed. A. Version,Ruthe-rfort] Apple-tasi Laboratory,Unired Kingdom), COST2SlTD(98)O05,291-293.

GHERM,y. E., KvACHAOZE, V. R. ant] ZERNOv, N. N. (1998b):«HFile-Id ne-artite « skipdistasice»subjecte-t]by the influence-of tite- ionosphericelectronde-nsity fluctua-tions»,2nd COST 251 WorkshopProceet]ings,Side, Turke-y, March 1998, (cd.A.Ve-mon,Rutherfort]Apple-tanLaboratory,UnitedKingdom), COST25¡ TD(98)005,266-272.

GHERM, V. E., KvACHADzE, V. R. ant] ZERNOV, N. N. (1999): «On UF propagationatfrequenciesaboye- MUF», 3n1 COSI 251 WorkshopProceedings,El Arenosillo,Spain,October1998, (ed. R. Hasibabaant] H.A. de la Morena),COST25ITD(99)003, 27-33.

GORDIENKO, G. 1. andKALIEXT, M. Z. (1998): «Tite- connectionbe-tweent]ay-to-dayva-riatiosisof tite ionosphe-reant] solaractivity citanges»,Re-p. UAG 105,Computerai-dedprocessingof ionogramsandionosonde-recort]s,World DataCenterA for So-lar-TerrestrialPhysics,National Ge-ophysicalCenter,Boult]er, Calo., USA,132-133.

Gorr,G. F., PANTJIARO5, C. A., WYLIE, J. A. andLAYCOCK, E. J. (1996):«HPspectraloccupancymeasure-me-ntsin Europe»,LEE Colloquium «Freque-ncySelectionandmasiage-me-ntTechixiquesfor HF communications»,London,UnitedKingdom,6/1-6/6.

HANBABA, R. (1999): «COSTAction251 FinalRepon»,cd. SpaceRese-archCesitre-,00-716 Warsaw,Polano].

HARDY, R. L. (1971): «Multiquat]ric e-quationsof topographyant] otiter irregularsur-faces»,J.Geophys.Res.,67, 1905-1911.

HEDIN, A. (1987):«MSIS-86chermosphe-ricmodel»J. Geophys.Res.92,4649-4662.ITU-R (1995) «HPpropagationpredictionmethod»,Recomment!ation ITU-R P.533,In-

ternationalTelecommunicationUnión, Geneva.ITU-R (1997): «flU-R reference-ionosphe-riccharacte-risticsant] methot]sfor basic

MUF, operationalMUP and ray-pathspredictions»,RecommendationJTU-RP.1239,International TelecommuninationUnión, Geneva.

JAKOWSKI, N. (1998): «Ge-ne-rationof TECmapsayertite COST251 areabase-donGPSme-asurements»,2M COST 251 WorksitopProce-e-dings,Side-,Turkey, March 1998,(ed. A. Vemon,RutherfordApple-tanLaboratory,UnitedKingdom), COST25lTD(98)065,51-57.

Físicade la Tierra2000,12.61-103 98

Inzprovet!Qualily of Servicein Ionospher!c...R. Hanbabaant! B. Zohes!

JAKOWsKI, N. (1999): «Capabilitiesob radiooccultatianmeasure-mentsaix boardLEOsatellitesbar ianospitericmonitoring ant] re-search»,4fl’ COSI 251 WorkshopPro-ceedings,Madeira,Portugal,March 1999,(ed. A. Ve-man,Rutiterfort] Apple-tanLa-boratory,UnitedKingdom,COST251ID(99)008,116-121.

JAKOWsKI, N. ant] SCHLIJETER, 5. (1999): «lonospite-rieStormsDe-tectedUy GPSBasedTEC Monitoring», 3~ COSI 251 Worksitap Proce-e-dings,El Arenosillo, Spain,October 1998, (cd. R. Hanbabaant] B.A. de- laMorena)COSI25ITD(99)003,103-107.

JAKOWSKI, N., SCHLUTER,5. ano] JUNOSTANO, A. (1998a): «Monitoringtite ionosphe-reayerEuropeant] re-lateo]ionaspitericstut]ies»Proceedingsof 1998IGS anaiysiscen-ter workshop(ed.J.M. Dow, J. Kouba,T. Springer),ESA/ESOCDarmstat]t,Fe--bruary 1998,265-271.

.IAKowSKI, N., SARDON, E. ant] ScHLIJ’rER, 5. (1998b):«GPS-base-dIEC observationsincomparisanwitit 1R195 ant] tite Europe-anIEC modelNICM2»,Mv. SpaceRes.,22,803-806.

JUcHNIKOwSKI, O. andZByszYNsKr,Z. (1992): «A Modification of foF2statisticalmo-del using vertical soundingdata»Proceedingsof III PRIME Warksitop,Rome,January1991,222-224.

KouRís,S. 5., BRADLEY, P. A. ant] DoMmncl, P. (1998):«Solar-cyclevariation of titet]aily foF2ant] M(3000)F2»,Ann.Geophys!cae,16, (8), 1039-1042.

KouRrs,5. 5., FoTL&Dís, 13. N., HANBABA, R. ant] IULUNAY, Y. (1999): «Sorne-consi-de-rationson tite beitayiourof MUF ayerEurope»,

3Ñ COSI251 WorksitopProce-et]ings,El Arenosillo, Spain,October1998,(cd.R. Haubabaant] B.A. de- laMore-na), COSI25ITD(99)003,157-167.

Kurinv, 1., MIJHTAROv, P. ant]BRADLEY, l1~. A. (1998): «Pe-netrationof ionaspite-ricdis-turbancesinto tite Europeanre-gionduringgeomagneticstorms»,Adv.SpaceRes.,22,865-867.

LAMMING, X. ant] CANDER, Lj. R. (1999): «Monthly median foF2 madelling ayerCOST 251 are-aby ne-ural ne-tworks»,PhysicsandChemistryoftheEarth,PanC:Solar-Terrestrialant!PlanetaryScience,24, (4), 349-354.

LAsTovIcKA, J. (1996):«Effectsof ge-omagneticstormsin tite- lower ionospitere-,mit]d-le- atmospitereandtrapospite-re-»,1. Aímos.Terr. Phys.,58, 831-843.

LAsTovIcKA, J. (1997): «Lang-te-rmtrendsin tite- uppermiddle atmosphe-reas de-tectedUy ionospitericmeasure-ments»,Mv. SpaceRes.,20, 2065-2073.

LAsTovIcKA, J. (1999):«COSI251 resultson tite ianospitericquasi-periodicvariabilitycauseo]Uy gravity ant]plane-tarywax’e-s — A revie-w»,4”’ COSI251 WorksitopPro-ceedings,Madeira,Portugal,March 1999, (cd.A. Vemon,RutiterfordAppletonLa-baratory,UnitedKingdam),COST25lTD(99)008,142-154.

LAsTovIcKA, J. ant] MLcH, P. (1996): «Solarcycle-e-ffe-ct on oscillationsun tite periodrange-of 2-20daysin tite- F-regianof tite- ionaspite-re-»,Ann. Geofis.,39, 783-790.

LAsTOvIcKA J., BURESOvA, 13. ano]BosKA, J. (1998) «DaesdicQBO ant] Mt. Pinatubovolcanic eruptionafbecttite- gravity wave activity in tite lawe-rionosphe-re?»Siudiageoph.er geod.,42, 170-182.

LEITINGER, R. (1998a): «Cansequencesfrom tite- extensionin longitude- of tite COSIare-afor faF2 ant]TEC monthly medianmapping»,2nd COSI251 WorksitapPro-ceedings,Side,Iurkey, March 1998,(ed.A. Ve-man,Rutiterfaro]AppletonLaba-ratary, UnitedKingdom),COST251TD(98)005,125-l39.

LEITINOER, R. (1998b):«Ne-UaG-plas- An casyto useglobalelectrondensitymodelfor

99Física de la Tierra

2000, 12,61-103

/?. Hanhabaant! fi. Zoks¿ improvedQualñyof Serviceluz Icnospheric.

tite ionospliereant] tite plasmaspite-re»,2”<~ COSI 251 WarkshopPracee-dinga,Side,Turkey,March 1998,(ea. A. Ve-man,RutherfordApple-tenLabaratory,Uni-te-o] Kingdam),COSflSITD{98)0O5,140-150.

LEmNwR, R. ant] HOCHEOGER, 0.(¡999): «Mappingmonthlyquantilesof e<ectroncan-te-nt andfoP2»,Physicsant! Chemislryof¡he Eanhi,PanC: Solar-Terreszrialant!PlímerarySelence,24,359-363.

LErrINGsR,R.,TtTHEkIDGE,J. E., KnrcmsNoAsT,0., andRo¡HwrJER, W. (1995): «A«simple»globalempine-al¡nadelfon tite- E [ayerof rhe ionosphcre-»,Wiss.Serie/U1/1995,IMO, Unive-rsity of Oras,Austria.

MMrl, 0,, MIRo. <3. andMnu-iMuav,A. V. (1999): «A me-titadfor foEZ shon-termpre-diction», 4ll~ COSI 251 WarkshapProcee-dings,Madeira,Portugal,March 1999,(cd. A. Version, Ruthe-rfordApple-tun Labarato¿-y,United Kingdom),COST2SITD(99)008,214-222,

MAnIERON. <3. (1971): «Tite the-oryof re-gionalizedvariablesano] its application»,Lescahieí-sdii CentredeMorphologueMaihématiquesdeFonrainehleau5, Pat-ls.

MnalAILOv, A. \T~ (1999): «Ionosphericint]e-x MP2 losmornhly medianfoP2madellingant] long-te-rinpre-clictiosiovenBurepeanare-a».PJzysics<md Chemistryof dic flan)),ParÉ C: So)ar-Terrestrialant! Planean-yScience,24,(4), 329-332.

MwHáunv,A. V., MIKHAÁLOV, V. V. ano] SKOBLIN, M. <3. (1996): «Monthly medianfoP2ano]M(3000)F2lonospiterie-madelayerEurope»,¡tn¡i. Geofis.39, 791-805.

MIKHA!LOV, A.V., PULUNEIS, 5. A., MIICHAILOS’ VV., KARPACHEV, A. 1. ant] SHUBIN, V.N. (199$>: «An attemptlo me-lude-mid-latitude- trough to tite foP2 instantane-ausmappingmodelMQMP2-IM», 2~COSI251 WorkshopProce-edings,Side,Tuñwy,March 1998,<cd. A. Ve-man.Rutherfart]Apple-LoaLabonatory,UnitedKingdom),COST2SI’I’D<98)005,100-103.

MÍKI±swov, AV., DELA MORENA, B. A. ant] Mma,<3. <1999): «A Me-titadlbsfoP2 Ma-nitaning ayerSpainUsing the El Are-nosillo DigisandeCune-ntObservations»,3’~COSI 251 WarkshapPnoce-e-dings,El Areriosillo, Spain,October 1998, (cd. R.Hanitabaant] RA. de- laMorena),COST25ITD(99)003,185-194.

MVWHELL,C. N., KER5LEY, L. ant] PEysn,5. E. (1997): «Thee-ffe-ctsofre-ceivenlocationin twa-stationexperimentalianospiterictaniography»,J.Armas.Solar-Terr.Phvs.,59, 1411-1415.

MTICI-IELL, C. N., KERSLEY, L., PRYsn5. E., HEATON .1. A. ant]CANNON. P. 5. (199%):«A new appraachto rnode-lling of tite maisi iasiasphe-rictroughovenEurope- usingtemographicimages>’,3’~ COSI251 WorkshopPreceedings,El Are-nesillo,Spain,Octoher 1998, (ccl. R. Hanhabaano] RA. de- la Morena), COST25ITD(99)0013,75-86.

MTWHELL, C. It, KER5LEY, L, ami CANNON, P. 5. (199%): «me-latitudinalpasitianofthe mid-latitudetrough - seaseríalvariatians»,4” COSI 251 WerkshopProcee-dings,Madeira,Portugal.March 1999,(ed. A. Ve-man,RuthezfordApple-tanLa-boratory,UnitedKingdam).COST2SITD(99)008, 110-109.

MooRu~xti,M. 13. ano]RAOICELLA,S. M. <1998):«Using20 raytracingto modeltite ef-fe-cts of vertical ionasphericprofile variability on ohliquepathprapagationat HPfre-quencies»,2»~ COSI 251 WonkshapProe-eeclings,Sitie, Turkey, March >998,(cd. A. Ve-man, Ruthenfart]Apple-tanLaboratory, United Kingdom),COST2SITD(98)005,226-233.

MUHTp,Rov,1’. ant] Ku—nuv, 1. (1998a):«GeomagneLicallycorre-late-o]statisticalmedd(GCSM) for shart-te-rmpredictionof lonospiteñeparame-tcrs».Yd COSI 25 1

Física del« Tierra2000, 12, 61 -103 loo

R.Hanbabaant! B. Zoles! ¡rnprovedQuahityof Serviceir> Ionospheric...

WorksitapProce-eo]ings,Side, Turke-y, March 1998, (ecl. A. Ve-man,Rutiterforo]Apple-LanLaboraÉory,UniLed Kingdorn),COST2S¡113(98)005,246-251.

MUHTAROv, P. ano]Ku’nEv, 1. (1998b):«Empiricalmode-llingof ianosphericstormsatmidiatitudes»,Adv. SpaceRes.,22, 829-832.

MLIHTAROv, P. ano]Ku’rmv, 1. (1999):«Autocorrelationme-titadfar temporalinterpo-latianano]short-tennpredictionof ionaspite-ricdata»,RadioSci.,34,(2), 459-464.

MUHTAROv, P., Ku’r¡~v, 1. ano]BRADLEY, P. A. (1999):«Prababititystatisticsof tite- re-latiye de-viatiansob foF2 from a refe-rencele-yel ayertite Europe-anre-gion», 4”’COSI251WorksitapProe-eedings,Madeira,Portugal,March 1999,(ecl. A. Ve-man,RutiterfardApple-tan Labarator>’, United Kingdam), COST251113(99>008,128-137.

PAwrnAaos,C. A., LAYCOCK, 1’. J.,Garr,O. F. ano] CHAN, 5. K. (1997):«Develapmental Lite Laycock-GaLLoccupancymodel»,¡RE Proc. Coinmun.,144, (1), 33-39.

PANTJXAROs, C. A., Wmw, J. A., BROWN J., Gorr, O. F. andLAycocK, P. J. (1998):«Extendedmode-Isbar HF spe-ctraloccupancy»,¡EF Proc. Commun.,145, (3), 168-174,

PERRONE, L. ant] DE FRANCESCHI, 0. (1998):«Solar,ianosphericandge-amagne-ticin-dices»,Anuz.Geofis.,41,(5-6), 843-855.

PERRONE, L. ano] DE FRANCIESCHI, 0. (1999): «A carrelatianstudybetweentime-we-igit-tedmagneticindicesant] tite- higit latitude ionaspite-re-»,Physicsant!ChemistryofIheEarrh, ParÉ C: Solar-Terrestrialant!PlanetaryScience,24,(4), 389-392.

RADICELLA, 5. M. ant]ZUANO, M. L. (1995): «‘me- improvedflOR analyticalmadelofe-le-cuandensityheightprofile- ano] total electroncante-ntun tite ionosphere»,Ann.Geofis.,38,35-41.

REINI5CH, B. W., GAMACHE, R. R., HUANO, X. ant] Mc NAMARA, L. F. (1988): «Realtime electrande-nsityprofilesfrom ionograms»,Adv. SpaceRes.,8, (4), 63-72.

RusH,C. M. ant]EDWARDS Ir., W. R. (1976):«An audiomatedmappingteclmiqueforre-presentingte-hourlybehaviarof tite- ionospitere»,RadioSc!.,11,931-937.

Sca’rro, C. (1999): «meprababilityof occurrenceof Pl laye-randL conditionesti-mate-o] by formulasant]by tables»,3’~ COSI251 Worksitop Procee-dings,El Are--nosillo, Spain,October 1998, (ed. R. Hanbabaano] B.A. de la Morena),COST251113(99)003,349-353.

SINGER, W., KNYAZEv, A. K. ant] BREMER, 1. (1995): « Empirical13-re-gionelectronden-sity model for mid-latitudes»,Proce-e-o]ingsof tite- C05T238Workshopon «De-ve-]opmentant] testingof an electron-densityheigbtprofile ¡nade]barPRIME», ElArenasilla,September1994,311-323.

SoizÉ, J. G. (1998): «Relationsbetweenitourly manthlymedianyalue-sof boP2ano]sornegeopbysicalindices.Tite-ir applicationto aix ianasphericsingle stationmodel»,Acta GeophysicaPolonica,Vol XLVI, (1), 77-88.

SOLÉ, J. O., ALTADILL, 13., ano] APosToLav,E. M. (1999):«Variationsof tite relativeamplitudesof te foP 2-, 3-,5- ano]6.5-dayquasi-periodicascillationsayerEuro-pe»,3”’ COST251 WorksitopProceedings,El Arenasillo,Spain,Octaber1998(ecl.R. Hanbabaano]B.A. de-laMorena),COST25ITD(99)003,175-184.

STANI5LAWSKA, 1. ano] JtJOHNIKOW5KI, 0. (1997): <‘A noteanuse-obscre-enpointsin re--gianal ionaspite-ricmapping>*,ActaGeophysicaPolonica,Vol.XLV, (4), 355-362.

STANI5LAWSKA, 1., JUCHNfl<OWSXI, O., HANHABA, R., CANDEn, Ls. R., ¡<Los, Z. andZBx’szYNsKí, Z. (1998):«An automaticsystemforHP propagationcanditionspre-dictianant] fare-cast»,~ COSI 251WarkshopProce-edings,Side,Turkey, March

101 Físicadela Tierra2000, ¡2, 61-103

R. Hanbaba and B. Zolesi lmproved Quality of Service in lonospheric...

1998, (ed. A. Vernon, Rutherford App1eton Laboratory, United Kingdom),

COST251TD(98)OO5,294-297.STANISLAWSKA, l., GULYAEVA, T. L. and HANBABA, R. (1999a): «Ionospheric Des-

patch Centre in Europe», Physics and Chemistry ofthe Earth, Part C: Solar-Te-rrestrial and Planetary Science, 24, (4), 355-357.

STANISLAWSKA, l., ZBYSZYNSKI, Z. and JUCHNIKOWSKI, G. (1999b): «Comparison of dif-ferent instantaneous mode1s of electron concentration height profiles at single 10-cation», Physics and Chemistry ofthe Earth, Part C: Solar-Terrestrial and Plane-tary Science, 24, (4), 385-388.

STANISLAWSKA, l., JUCHNIKOWSKI, G., RornKAEln.., H., ZBYSZYNSKI, Z. and TuLUNAY, Y.(1999c): «Considering fue trough phenomena in COST 251 models», 3rd COST 251Workshop Proceedings, El Arenosillo, Spain, October 1998, (ed. R. Hanbaba andB.A. de la Morena), COST251TD(99)OO3, 65-73.

TERYOKIllN Yu. L. and MIKHAILov, A. V. (1992): «A new approach to the ionosphericmapping», Solar Terrestrial-Predictions Workshop Proceedings, Ottawa, Ganada,US Government Printing arrice, Washington, DC 20402, 558-567.

TUHERIDGE, J. E. (1976a): «Ion transition heights from topside electron density profi-les», Planet. Space Sci., 24, 229-245.

TITHERIDGE, J. E. (1976b): «Plasma temperatures from Alouette 1 electron densityprofiles», Planet. Space Sci., 24, 247-259.

TITHERIDGE, J. E. (1985): «Ionogram Analysis with fue Generalized Program PO-LAN», UAG-93, World Data Center A for STP, NOAA, Boulder, CO 80303, USA.

TUHERIDGE, J. E. (1993): «Atrnospheric winds calculated from diurnal changes in fuemid-latitude ionosphere», J. Atmos. Terr. Phys., 55,1637-1659.

TuLUNAY, Y. (1996): «lnterplanetary Magnetic Field (IMF) and its possible effects onfue rnid-latitude ionosphere 111», Ann. Geofis., 39, (4), 853-862.

TuLUNAY, Y., STANISLAWSKA, l. and ROnIKAEL, H. (1998): «A review on fue ARIELtrough work», 2nd COST 251 Workshop Proceedings, Sirle, Turkey, March 1998,(ed. A. Vernon, Rutherford Appleton Laboratory, United Kingdom),

COST251TD(98)OO5,71-94.TuLUNAY, Y., TuLUNAY, E., ÜZKAPfAN, C. and KAYA, A. (1999) «Neural network based fo-

recasting of ionospheric critical frequency foF2 during IMF B polarity changes», 4d1COST 251 Workshop Proceedings, Madeira, Portugal, March 1999, (ed. A. Vernon,Rutherford Appleton Laboratory, United Kingdom), COST251TD(99)OO8, 242-245.

VAN DER PERRE, L. (1997): «Propagation Modelling and Simulation of the HF channelwith the objective of testing intelligent narrowband and wideband modems», Ph.D.Thesis of fue Katholic University of Leuven, Belgium.

VAN DER PERRE, L., VAN DE CAPELLE, A., A DE M. and LAUWEREINS, R. (1997): «Im-plementation of an extended simulator for fue HF-channel on multi-DSP board»,lEE Conf Publ., 441, 373-377.

V ASILIEVIC, l. M. and CANDER, Lj. R. (1997): «Forecasting of ionospheric storms: a caseat rnid-latitudes», Joint COST 251/1R1 Workshop and Working Group Sessions Pro-ceedings, Kühlungsborn, Germany, May 1997, (ed. A. Vernon, Rutherford Ap-pleton Laboratory, United Kingdom), COST251TD(97)006 (Part 1),98-108.

W ATTERSON, C. C., JUROSHEK, J. R. and BENSEMA, W. D. (1970): «Experimental con-fmnation of an HF channel model», IEEE Trans. Commun., COM-I8, 792-803.

WRENN, G. L. (1987): «Time-Weighted accumulations ap(t) and Kp(t»> J. Geophys.

Res., 92, 10,125-10,129.

Ffsica de la Tierra 1022000,12,61-103

R.Hanbabaant! fi. Zotes! ImprovedQuahilyofServiceir> Ionospheric...

XENos, It. 13. ant] ALBERCA, L. E. (1999): «Applicationof Ne-uralNetwarks inM(3000)P2Single-StatianModelling», 3’~ COSI251 WorkshopPrace-e-dings,ElAre-nosillo, Spain, October1998, (cd. R. Haubabaano] 8. A. dc la Morena),COST25íTD(99)003,273-278.

XENOs, TH. 13.,Koums,5. S. ano] PAPANDONXOU, V. Ph. (1996):«Single-stationma-de-ls»,Praceet]ingsof tite PirstWorkshapof COSI251,Prague,Cze-citRe-public,27September1996, (ed.J. Lastovicka),C05T25110(96)017,15-31.

ZFRNov, N. N., GHERM, V. E. ant] LUNDBORG, B. (1999): «Ontite- scatteringfunctianbarte- widebandPIE ianosphe-rictransmissioncitannel»,

3Td COSI 251 WorkshopProce-e-t]ings,El Are-nasillo, Spain,Octaber1998,(e-o]. R. Haubabaano] 13. A. de- laMorena),C051251TD(99)003,35-38.

ZoLEsí,8., CANDER, Lj. R. ano] fin FRANCESCIHI,0. (1996): «Ontite- patentialapplica-bility of tite simplified ionaspite-ricregionalmodeltadifferentmid-latitude-arcas»,RadioSc!.,31, (3), 547-552.

103 Física de la Tierra2000, 12,61-103