Name Designation Affiliation Date Signature Submitted by: Accepted by: Approved by: TROPOSPHERIC STABILITY AT CANDIDATE SKA SITES AUSTRALIA EDITION Document number .................................................................. WP3‐040.020.001‐TR‐003 Revision ........................................................................................................................ A2 Author ........................................................................................................ R.P. Millenaar Date.................................................................................................................. 2‐11‐2011 Status ................................................................................................... Final Confidential
24
Embed
Tropospheric stability at candidate SKA sites - AUS 1 · PDF fileTROPOSPHERIC STABILITY AT CANDIDATE SKA ... reports on the measurement campaign, the instrumentation, deployment ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
The tropospheric stability over the proposed core locations in the two candidate host countries,AustraliaandSouthAfrica,hasbeeninvestigated.Thisworkwasorganisedandpartlycarriedoutbythe SPDO,within PrepSKAWorkPackage3.4,which aimed to “Carry out studies of the effects oftroposphericturbulenceonhighfrequencyobservations.Studythehigh‐frequency limitsofphase‐referencingandself‐calibration,anddeterminethe implicationsfortheSKAdesign” [1].Onadvisefrom the IEAC and SSEC thework has been carried out as in‐situmeasurements of troposphericstability. The current document reports on the measurement campaign, the instrumentation,deploymentandmeasurementresultsuptoapointintimeaslateaspracticallypossible.Thelatterremark refers to the limited time that the measurement systems have been producing data, inparticulartheoneinAustralia.Thisreportthereforewillbeissuedasclosetothesetdeliverytimeaspossible,andwillberevisedwhennewdatahasbeenanalysedatlaterdates.Thedocumenthistorytablewillaccountfortheseupdates.
2 Scope
SiteTestInterferometers(STI)havebeeninstalledatthecorelocationsofthecandidateSKAhostsandmeasurementsweretaken.Thepurposeistocharacterisethetroposphericstabilityconditionsoverthesesitesforaperiodlongenoughsuchthatrepresentativeinformationisobtained,samplingall seasons.Becauseofslowdeployment themeasurementperiodhasbeen limited, to theextentthatsummermonthshavenotbeenobservedwithintheavailabletime.Thisreportprovidesgeneralinformationonthesystemsanddetails thepropertiesof thetwosites.Measuredrmsdelaysoverthebaselineofnominally200marepresentedanddiscussed.Plotsofmonthlyrmsdelayovertimeareincludedandmonthlystatisticsfordaytimeandnight‐timecumulativedistributionsofrmsdelayarepresented.Anoverviewofusabledataoverthereportingperiodisincluded.
3 Organisationofwork
3.1 Partnersandresponsibilities
Theprojecttocollecttroposphericstabilitydatahasbeenorganisedaroundtheseparties:SPDO–Hasorganisedtheproject,participatedindeploymentandcommissioning,carriedoutdatainspections,remotedataretrieves,dataprocessingandintermediateandfinalreporting.JPL – Has designed, produced and tested two identical STI systems, described fully in [2]. Eachconsistsoftwo0.84mreflectorswithmodifiedlnb’s.Themodificationallowsfeedingthemixerwithacommonlocaloscillatorsignal.Aweatherproofboxneartheantennacontainsantennaelectronics,consisting of a bandpass filter, amplifier and optical transducer for the IF path and transducer,amplifier, tripler, filter, amplifier anddoubler for the LO signal path.A central rack contains a LOgenerator,IFmoduleswithtransducer,filtersandamplification,feedingtheiroutputsintoIQmixersthat function as analog correlators. LO and IF connections between central rack and antennaelectronics boxes is through RF over fibre. A computer system takes care of data acquisition,monitoring,storageandethernetcommunications.Sites‐ Have participated in deployment, provided foundations for antennas, antenna‐boxes and acontrolled environment for the central electronics. Have taken care of subsurface routing of thefibres from central rack to both antenna electronics boxes. Have supplied personnel to carry outbasicmaintenanceandlocalsupport,includingapowersupplyprovisionandinternetconnectivity.
WP3‐040.020.001‐TR‐003 Revision:A
Page6of24
3.2 Installationandcommissioning
DesignandconstructionofthetwoSTIsystemswasdoneduring2010.Mid2010thetwositeswereadvisedtomakepreparatorymeasuresattheirsitestoallowrapidinstallationandcommencementof themeasurement campaign. This included the request tomake room available for the centralrackinanRFIshieldedenvironment.InNovembertheauthorvisitedJPLtoinspectanddiscussthesystemsthenbeingcompletedandtested.ThesewerereadyforshipmenttothetwositesbymidDecember2010.Itwashopedthatthesewouldarriveattheirdestinationsintimebeforethelocalsummerholidays.Unfortunately this turnedoutnot tobe thecase.Moreover itwas learned thatsitepreparationshadnotevenstartedinAustraliabythetimetheholidayswereover.Sitepreparation,andthereforealsoinstallationinAustraliahasbeendelayedsubstantially.Someofthedelaywascausedbythefactthatthesitecouldnotbereachedbecauseofheavyrainfallinthearea.Alsodelayswere incurredbecause theRFI campaign inAustraliawas in competition for thesamepersonnel.Furthermore,itturnedoutthattherewasnoshieldedenvironmentavailableatthesite, in which the central rack could be housed. That caused further delays because the rackelectronics needed to be repacked into a shielded rack that was acquired, and which alsonecessitatedEMItestingofthenewrack.ItwasnotuntilthesecondweekofMaythatitmadesensefortheSPDOteamtotraveltothesite.Onarrivaltheinstallationwasfoundnottobecompleteyet.During this visit a successful commissioningwas not possible because of this and other technicaldifficulties experienced at the site. The two antennas could be pointed correctly, but a completeworkingSTIcouldnotbeachieved.TheAustralianteamwouldcompletetheinstallationandtestingbut this was not completed before the second week of June, again because of technical andpersonnel difficulties. Usable data was being acquired starting 12‐6‐2011. Initially the routinemeasurementswerestartedwithasystemthatdidnothavethecabledelaysfullyadjusted.Thiswaseventuallycompletedon23‐9‐2011.Thishashadaneffectonthecorrelatedamplitudeofthesignal,but there is no sign that this has negatively affected the delay data. See also the discussion onamplitudeeffectsonthedelaydatainsection6.2.1.
4 Processingprinciples
Thesystemdocumentationprovidesgeneralprinciplesofoperationofthiskindofinstrumentationin addition to specific details for the equipment used, see [2] and [3]. Here the data processingprinciplesaresummarised.The STI correlator produces I,Q, and reducedphaseoutput data streams. Correlated amplitude iscalculatedfrom:
a = I 2 +Q2 Eq.1
Phase(raw)iscalculatedfrom:
p = tan−1(QI) Eq.2
The phase is unwrapped before being used. The reduced phase P is the raw phase p, with slowvariationsremovedbyanalgorithmexplainedinappendixA1of[3].The results are to be expressed in delay values, and in order to compare between sites, for astandardisedbaselinelengthandindependentofpathlengthdifferencesduetotheelevationofthesatellite. It is noted that there is no need to calculate projected baseline length as function ofsatelliteelevationandazimuthbecausethephasedisturbancesoriginateinathinatmosphericlayerclose to the antennas forwhich the distance for the piercing points through the layer effectively
WP3‐040.020.001‐TR‐003 Revision:A
Page7of24
equals to the baseline length. The algorithm to arrive at a standardised zenith delay thereforebecomes:
d =P
360 fsin(e) b0
b⎛⎝⎜
⎞⎠⎟β
, Eq.3
where f is the observing frequency, e the satellite elevation, b and b0 the actual and referencebaselines(forwhichwetake200m)andβ=5/6.ThisisascalingexponentialthatmustbeappliedtophasefluctuationsinducedbytheatmosphereusingKolmogorovmodelling,see[4].Thermsofthedelayover3000samples is calculated.Thesample ratebeing10Hz thisequates toa streamof5‐minuteintervalrmszenithdelayvalues.
WP3‐040.020.001‐TR‐003 Revision:A
Page8of24
5 Sitedescriptions
5.1 Boolardy,Australia
Theantennashavebeenmountedtolarge,heavyconcreteblocksthatrestuponthesurface,seetheillustration inFigure1.Underground fibre routing leads tooneof the temporaryhuts thathousesthecentralelectronicsinitsshieldedcabinet,about200metresawayfromtheantennas.Atthetimeofinstallationandalsoduringthecourseofthemeasurementsconstructionactivitiesintheareaaretaking place, for building antenna foundations, excavation works, road building and telescopeconstruction.Thishasmade theenvironmentaverydustyplace,whichposedchallenges tocleanfibrehandling.
Insection3.2theperiodinwhichmeasurementsweredonewasdiscussedinbroadterms.Heretheactualperiodsarespecifiedinwhichdatawasacquiredthatwereultimatelyusedforanalysis.Thisisdone intheformofpresentingmonthlyplots insection6.1.1.Thesearestandardplotsthatcomeout of the data processing package using all acquired data, regardless of validity due to systemstatus.Completemonthsareshown.Foreachof theplotsabriefexplanationofdataanomalies isgiven,togetherwiththenetperiodsofusabledatapermonth.Eachoftheplotsinthischapterontheoverviewofavailabledata,andalsointhechaptersondataanalysis(6.2),containfourpanels,fromtoptobottom:
1. Raw unwrapped phase: For correct operation sinewave‐like variations should be seen ascausedbythediurnalmovementofthesatelliteinitsgeostationaryorbitcube.Becausenewfilesarestartedat0UTthephaseisresettozeroatthattime,whichgivesrisetooccasionalsmall steps, or larger steps at that time when during the previous day the unwrappingalgorithmwasdisturbedorthefilewasn’tstartedat0UTforsomereason.Itisquiteeasytoseeinthissubplotwhethervaliddataispresentornot.
2. Correlated amplitude: Nominal levels ~200 to 1100 units in Australia indicate correctoperation.Thevariationthatisseenislikelyduetotransponderchannelsbeingswitchedonandoffwithinthe200MHzofavailablebandwidth.Incorrectoperationisindicatedbyverylowamplitudeorwildlyvaryingvalues.
Even though the first two panels are sufficient for judging whether valid data is present, thefollowingtwoareincludedtoshowtheimpactonthesederivedparameters,andalsotoprovideacompleteoverviewthatincludesmallscaleanomaliesorglitchesthatneedattentionfortheanalysisinsection6.2.3:
3. Filteredphase:Asinusoid is fittedand removed fromthe rawphaseover theprevious10minutes.Usuallyquitecleardiurnaltroposphericvariationbecomesvisible.
4. Rmsdelay:Theseare corrected zenithdelayvalues,asdescribed in section4.Thediurnalvariationsshouldbeobvious.
September 2011: Usable data from 1/9 through 6/9. Power was interrupted early on 7/9. Afterbeingrestoredon9/9theLOdidnot lock,whichwentundetectedat that time.WhentwoweekslaterateamworkedonthesystemtoreplacetheLOsynthesizerthatturnedouttobedefectivethedelayadjustmentswerefinallymade,whichmadeagreatimprovementinthecorrelatedamplitude,aftert≅513(earlyon22/9).Effectivelythisresultsin14fulldaysofusabledata.Figure6:OverviewAustraliaOctober2011October2011:Usabledataisacquiredfrom1/10through9/10.On10/10theLOhaslostlock,whichis resetonthe11th.On12/10 through20/10data isagainacquired,butthequalityon18/10and19/10appearspoor (seedataanalysis section). On21/10again theLOhas lost lock,until 27/10.Usabledatafrom28/10through31/10.Atotalof22fulldaysofusabledata.
Figure5:OverviewAustraliaSeptember2011
WP3‐040.020.001‐TR‐003 Revision:A
Page13of24
6.2 Troposphericdelaydata
Thisreportcontainsanalyseddataintwoforms:timeseriesplotsoffourparametersandcumulativedistributionplotsofmeasuredzenithrmsdelay.6.2.1 DelaytimeseriesTheprocedurethatwasfollowedwastoinspecttheavailabilityofusabledata,whichresultedintheassessment in the previous sections. Next only full days were processed, resulting in a series ofmonthlyplotsthatarepresentedinthesectionsthatfollow.Theexplanationonwhatisshowninthepanelsineachoftheplotsisrepeatedhere(fromsection6.1),fromtoptobottom:
1. Raw unwrapped phase: For correct operation sinewave‐like variations should be seen ascausedbythediurnalmovementofthesatelliteinitsgeostationaryorbitcube.Becausenewfilesarestartedat0UTthephaseisresettozeroatthattime,whichgivesrisetooccasionalsmall steps, or larger steps at that time when during the previous day the unwrappingalgorithmwasdisturbedorthefilewasn’tstartedat0UTforsomereason.
2. Correlated amplitude: Nominal levels of ~200 to 1100 units in Australia indicate correctoperation.Thevariationthatisseenislikelyduetotransponderchannelsbeingswitchedonand off within the 200 MHz of available bandwidth, see discussion at the end of thisparagraph.
3. Filteredphase:Asinusoid is fittedand removed fromthe rawphaseover theprevious10minutes.Usuallyquitecleardiurnaltroposphericvariationbecomesvisible.
4. Rmsdelay:Theseare corrected zenithdelayvalues,asdescribed in section4.Thediurnalvariationsshouldbeobvious.
Notethattheseplotsdonothavethesametime‐scale,dependingonavailabilityofusabledata:theyareshownattheirgreatestresolutionhere.Thisisalsotruefortheverticalscalesinthebottomtwopanels:thescaleusedallowsbestvisibilityofdetails.Intheseplotsitwillbeapparentthatthecorrelatedamplitudetracesshowvariabilitythatonemightnotexpect.Someof thismaybecausedby inexactpointingof (oneof thedishes),asmaybe thecase inAustraliawhereadiurnalpattern is seenduringmuchof the time. InSouthAfricaweseeevidence of changes in received signal within the 200 MHz bandwidth. Turning on and offtransponderchannelsbythesatelliteoperatormaybecausingthevariations.This,however,cannotbebackedupbyspectrumanalyserevidence.Timeconstraintshavenotallowed investigatingthiswithin the reporting period.Onemight suspect that the changes in signal to noisewould have anoticeable effect on the phase data. This, however, has not been observed in the data. Itwouldappear that the signal to noise levels have been sufficient for not detecting a clear effect in thefiltered phase data, and that the processing to arrive at the rms delays over 3000 seconds hasremovedanyremainingeffect.ThisconclusionissupportedbythedemonstrationintheAustraliansystemwhereadjusting the cabledelays causedadoublingof the correlatedamplitude (see3.2),withoutseeingdifferencesinthefloorvaluesinthermsdelaydata,beforeversusafterthechangetothehardware.6.2.2 DelaycumulativedistributionFurther inspection of the time series plots reveals data that was affected by non‐troposphericalcauses.Itwouldbeincorrecttoincludethatdatainthecumulativedistributionplotsofzenithrms
WP3‐040.020.001‐TR‐003 Revision:A
Page14of24
delayvalues.Thatdata is removedbeforemaking thesedistributionplots.Occurrenceof thiswasindicatedinthedescriptionsofthetimeseriesplots.Theplotsshowthreetraces:
1. Night‐time cumulative distribution (dotted trace): An 8 hour period, centred on localmidnight.
The markers indicate glitches that should be disregarded as invalid delay points that have beenremovedfromdatabeforemakingthecumulativedistributionplots.Thebriefinterruptionatt~246willresultonlyinainglezerodelaysampleinthecumulativedistributionandisnotremovedfromthedata.
Figure8:TimeseriesAustraliaJuly2011
WP3‐040.020.001‐TR‐003 Revision:A
Page17of24
The markers indicate glitches that should be disregarded as invalid delay points that have beenremovedfromthedatabeforemakingthecumulativedistributionplots.
The plots show the night‐time distribution of zenith rms delay in the dotted trace, the day‐timedistributioninthedashedtraceandtheoveralldistributioninthesolidline,forAustraliaconsistentlyinblue.Theplotsarealwaysscaledsuchthatthelargestdelayintheseriesmatchestheright‐handsideoftheplot.