AampA 556 A1 (2013)DOI 1010510004-636120079306ccopy ESO 2013

Astronomyamp

Astrophysics

The 23 GHz continuum survey of the GEM project

C Tello1 T Villela1 S Torres2 M Bersanelli3 G F Smoot45 I S Ferreira6 A Cingoz5 J Lamb7 D Barbosa8D Perez-Becker59 S Ricciardi410 J A Currivan5 P Platania11 and D Maino3

1 Divisatildeo de Astrofiacutesica Instituto Nacional de Pesquisas Espaciais (INPE) CP 515 12201-970 Satildeo Joseacute dos Campos SP Brazil2 Centro Internacional de Fiacutesica Bogotaacute Colombia3 Dipartimento di Fisica Universitagrave degli Studi di Milano via Celoria 16 20133 Milano Italy4 Lawrence Berkeley National Laboratory University of California 1 Cyclotron Road Bldg 50 MS 205 Berkeley CA 94720 USA5 Physics Department University of California Berkeley CA 94720 USA6 Instituto de Fiacutesica Universidade de Brasiacutelia Campus Universitaacuterio Darcy Ribeiro minus Asa Norte 70919-970 Brasiacutelia DF Brazil7 Physics Department University of California Santa Barbara CA 93106 USA8 Grupo de RadioAstronomia Instituto de Telecomunicaccedilotildees Campus Universitaacuterio de Aveiro Aveiro Portugal9 Departamento de Fiacutesica Facultad de Ciencias Universidad Nacional Autoacutenoma de Meacutexico 04510 DF Mexico

10 INAF Osservatorio Astronomico di Padova Vicolo dellrsquoOsservatorio 5 35122 Padova Italy11 Istituto di Fisica del Plasma CNR-ENEA-EURATOM Association via R Cozzi 53 20125 Milano Italy

Received 1 March 2013 Accepted 19 April 2013

ABSTRACT

Context Determining the spectral and spatial characteristics of the radio continuum of our Galaxy is an experimentally challeng-ing endeavour for improving our understanding of the astrophysics of the interstellar medium This knowledge has also become ofparamount significance for cosmology since Galactic emission is the main source of astrophysical contamination in measurements ofthe cosmic microwave background (CMB) radiation on large angular scalesAims We present a partial-sky survey of the radio continuum at 23 GHz within the scope of the Galactic Emission Mapping (GEM)project an observational program conceived and developed to reveal the large-scale properties of Galactic synchrotron radiationthrough a set of self-consistent surveys of the radio continuum between 408 MHz and 10 GHzMethods The GEM experiment uses a portable and double-shielded 55-m radiotelescope in altazimuthal configuration to map60-degree-wide declination bands from different observational sites by circularly scanning the sky at zenithal angles of 30 from aconstantly rotating platform The observations were accomplished with a total power receiver whose front-end high electron mobilitytransistor (HEMT) amplifier was matched directly to a cylindrical horn at the prime focus of the parabolic reflector The Moon wasused to calibrate the antenna temperature scale and the preparation of the map required direct subtraction and destriping algorithmsto remove ground contamination as the most significant source of systematic errorResults We used 484 h of total intensity observations from two locations in Colombia and Brazil to yield 66 sky coverage fromδ = minus5173 to δ = +3478 The observations in Colombia were obtained with a horizontal HPBW of 230 plusmn 013 and a verticalHPBW of 192 plusmn 018 The pointing accuracy was 6prime84 and the RMS sensitivity was 1142 mK The observations in Brazil wereobtained with a horizontal HPBW of 231 plusmn 003 and a vertical HPBW of 182 plusmn 012 The pointing accuracy was 5prime26 and the RMSsensitivity was 824 mK The zero-level uncertainty of the combined survey is 103 mK with a temperature scale error of 5 afterdirect correlation with the RhodesHartRAO survey at 2326 MHz on a T -T plotConclusions The sky brightness distribution into regions of low and high emission in the GEM survey is consistent with the ap-pearance of a transition region as seen in the Haslam 408 MHz and WMAP K-band surveys Preliminary results also show that thetemperature spectral index between 408 MHz and the 23 GHz band of the GEM survey has a weak spatial correlation with theseregions but it steepens significantly from high to low emission regions with respect to the WMAP K-band survey

Key words surveys ndash Galaxy structure ndash radio continuum general ndash radio continuum ISM

1 Introduction

The large-scale distribution of non-thermal radiation from ourGalaxy has been observed since the birth of radioastronomyat very low frequencies of a few tens of MHz (Jansky 19321933 1935) until the advent of the present era of precision cos-mology at frequencies above 20 GHz (eg Mather et al 1990Smoot et al 1992 de Bernardis et al 2000 Bennett et al 2003Meinhold et al 2005) By the time its synchrotron nature was

The survey is only available at the CDS via anonymous ftp tocdsarcu-strasbgfr (130791285) or viahttpcdsarcu-strasbgfrviz-binqcatJA+A556A1

recognized to be due to relativistic electrons spiralling along in-terstellar magnetic field lines (Kiepenheuer 1950 Mayer et al1957 Mills 1959 Westerhout 1962 Wielebinski et al 1962)free-free radiation from HII regions had already been identifiedas another Galactic emission component through the pioneer-ing work of Reber (1940) who sketched the first radio mapsof the Galaxy at 160 MHz and 480 MHz (Reber 1944 1948)Excluding atmospheric emission and the 3 K due to the cosmicmicrowave background (CMB) the sky signal on large angularscales is dominated by synchrotron radiation up to sim20 GHz andby free-free and anomalous microwave emission (AME) up tosim70 GHz where the rising spectrum of thermal dust takes over

Article published by EDP Sciences A1 page 1 of 20

AampA 556 A1 (2013)

the characterization of Galactic emission all the way into the in-frared including rotational line emission of CO molecules above100 GHz (Planck Collaboration 2013cd) Given this associationwith distinct astrophysical sources each Galactic emission com-ponent traces a unique spatial template In the case of the non-thermal continuum radioastronomers have always been activelypursuing the frequency dependance of the synchrotron templateto understand the structure and composition of the interstellarmedium In cosmology the nature of temperature anisotropiesin the CMB and its polarization cannot be fully understood un-less the contaminating role played by diffuse Galactic emissionis accurately detailed

In this article we present the first survey of an observationalprogram the Galactic Emission Mapping (GEM) project aimedat supplementing the frequency gap in radio-continuum surveysthat map its large-scale distribution at centimeter wavelengthsand the all-sky maps from the COBE WMAP and Planck satel-lite missions in the millimetric and submillimetric bands Weexplore this deficiency in Sect 2 as we review the present statusof the GEM project toward the preparation of spatial templatesof the synchrotron component between 408 MHz and 10 GHzIn Sect 3 we describe our portable single-dish scanning experi-ment at 23 GHz during three observing seasons in Colombia andBrazil In Sect 4 we outline the data reduction and preparation ofa combined map with 66 of sky coverage in total intensity be-tween δ = minus5173 and δ = +3478 Its baseline calibration anddestriping techniques are treated in Sect 5 together with the ab-solute temperature scale of the final map We discuss our resultsin the light of the RhodesHartRAO survey at 2326 MHz (Jonaset al 1998) in Sect 6 and finally we state our conclusions andfuture prospects in Sect 7

2 Large-scale Galactic emission and the GEMproject

Unparalleled efforts have always been dedicated to delineatethe complex morphology of the synchrotron-dominated Galacticemission at meter and centimeter wavelengths which reveals therich environment of supernova remnants with spurs and loopsemanating from the Galactic plane and in some cases engulfingour location near the inner Perseus spiral arm of the Galaxy Thework of Jansky and Reber led naturally to the completion of thefirst all-sky radio surveys The earliest of these was the 200 MHzsurvey of Droumlge amp Priester (1956) which combined the obser-vations of Allen amp Gum (1950) in the southern hemisphereThe advent of the worldrsquos largest radiotelescopes together withan improvement in the performance of the receivers promptedthe realization of several partial surveys of better sensitivityand resolution at 178 MHz (Turtle amp Baldwin 1962) 30 MHz(Mathewson et al 1965) 38 MHz (Milogradov-Turin amp Smith1973) 85 MHz and 150 MHz (Yates et al 1967 Landecker ampWielebinski 1970) which were used to synthesize three addi-tional low-frequency all-sky surveys the 30 MHz map of Cane(1978) the 85 MHz map of Yates (1968) and the 150 MHzmap of Landecker amp Wielebinski (1970) The integrity of thesemaps is however compromised by the need to assume spec-tral indices when combining different frequencies and by sig-nificant sidelobe level corrections due to differential ground andsky pick-up let alone the differences in the mapping strategy ofeach radiotelescope

A major improvement in the preparation of all-sky surveyswas achieved by Haslam et al (1981 1982) whose 408 MHzmap of the whole sky reflects the choice of a single observing

frequency and mapping strategy across four surveys to obtaina resolution under 1 with negligible sidelobe contaminationusing single-dish observations More recently another all-skymedium-resolution (sim05) survey was completed at 1420 MHzthrough the observations of Reich amp Reich (1982 1986) in thenorthern hemisphere and those of Testori et al (2001) in thesouthern hemisphere (Reich et al 2001) Single-dish surveys ofthe radio continuum for mapping the large-scale Galactic emis-sion have not been pursued at higher frequencies with the ex-ception of the RhodesHartRAO survey at 2326 MHz (Jonaset al 1998) over 67 of the sky with a HPBW = 0333 Higherresolution surveys in this frequency range have been obtainedwith single-dish antennas at 14 MHz (Reich et al 1990a 1997)24 MHz (Duncan et al 1995) and 27 MHz (Reich et al 19841990b Fuumlrst et al 1990) but their sky coverage has been limitedto within a few degrees of the Galactic plane At lower frequen-cies low-resolution antenna arrays have been used to concludethree other major surveys with 69 and 95 of total sky cov-erage respectively at 22 MHz (Roger et al 1999) and 45 MHz(Alvarez et al 1997 Maeda et al 1999)

In astrophysics medium- and low-resolution surveys com-plement high-resolution surveys in probing the full range of spa-tial scales that are necessary to explore the complexity of theISM In particular toward the Galactic disk where the majorityof the Galactic radio sources and the 21-cm continuum emis-sion of atomic hydrogen are concentrated Combining dedicatedsingle-dish observations and their sensitivity to diffuse emissionwith interferometric observations down to the resolution limit inthe Galactic disk has resulted in the International Galactic PlaneSurvey of HI which covers 90 of the Galactic disk extend-ing across three surveys the Canadian Galactic Plane Survey(CGPS Taylor et al 2003) the Southern Galactic Plane Survey(SGPS McClure-Griffiths et al 2005) and the VLA GalacticPlane Survey (VGPS Stil et al 2006)

21 CMB contamination

The detection of temperature anisotropies in the CMB radiationby the DMR experiment onboard the COBE satellite (Smootet al 1992) initiated a scientific breakthrough for cosmologybut it also showed from the analysis of the COBE all-skymaps at 30 GHz 53 GHz and 90 GHz (Bennett et al 1992)the need to extend our knowledge of diffuse Galactic emis-sion to higher frequencies Experiments to measure the CMBwith higher precision and resolution have since then been de-veloped or are currently under way These new generations ofexperiments are conducted from the ground (eg BEAST ndashChilders et al 2005 Meinhold et al 2005 OrsquoDwyer et al 2005QUIET ndash QUIET Collaboration 2011 2012 SPT ndash Keisler atal 2011 Reichardt et al 2012 ACT ndash Das et al 2011 Duumlnneret al 2013) from balloon platforms (eg BOOMERanG ndashde Bernardis et al 2000 EBEX ndash Reichborn-Kjennerud et al2010) or onboard satellites (WMAP ndash Bennett et al 2003 Goldet al 2011 Planck ndash Tauber et al 2010 Planck Collaboration2011a) All these efforts have established polarization measure-ments as one of the hottest topics in observational cosmologyThe success of high-precision CMB observations imposes aneven more reliable and accurate separation of the cosmologi-cal signal from the foreground emission of our own Galaxywhich cannot be removed by any improvement in the detectorswhether on the ground or in space Several techniques have be-come available from simple Galaxy masking of heavily con-taminated regions and point sources (eg Mejiacutea et al 2005) to

A1 page 2 of 20

C Tello et al The GEM 23 GHz survey

the novel approaches of component separation and foregroundremoval (Hobson et al 1998 Maino et al 2002 Delabrouilleet al 2003 2009 Eriksen et al 2004 2006 2008 Bonaldi et al2006 Cardoso et al 2008 Leach et al 2008 Stompor et al2009 Ricciardi et al 2010 Fernaacutendez-Cobos et al 2012 PlanckCollaboration 2013c) which exploit the spectral and spatial di-versity of the different astrophysical and cosmological compo-nents The multi-frequency design of the WMAP mission had al-ready revealed an impressive picture of unprecedented precisionof the CMB at 23 GHz 33 GHz 41 GHz 61 GHz and 94 GHz(Hinshaw et al 2007 2012 Bennett et al 2012) whereasthe Planck mission with its higher angular resolution and su-perb sensitivity at 30 GHz 44 GHz 70 GHz 100 GHz 143 GHz217 GHz 353 GHz 545 GHz and 857 GHz has at the time ofwriting made available an overwhelming set of results that pushthe standards of cosmological and astrophysical research to astate-of-the-art level (Planck Collaboration 2013b)

Yet there remains a considerable gap in our knowledgeof diffuse Galactic emission between the upper-frequency endof the large ground-based radio-telescopes and the lower endachievable from space or balloons which confronts radioas-tronomers and cosmologists alike In this frequency range theincreased resolution of single-dish observations imply that thetime allocation at radio-astronomical facilities capable of sur-veying a sizable area of the sky meaningful enough to ad-dress the large-scale properties of this diffuse component be-comes practically prohibitive The few examples we find inthe literature where high-resolution surveys were undertaken inthis frequency gap default to efforts devoted to construct cata-logues of radio sources against the high-intensity emission of theGalactic disk Among them we find arcminute-resolution sur-veys at 49 GHz (Altenhoff et al 1979) 5 GHz (Haynes et al1978) and 10 GHz (Handa et al 1987) Without a dedicatedinstrument free of time constraints in its observational sched-ule diffuse Galactic emission will continue to lack the cover-age it needs to map the sky away from the Galactic plane inthis frequency gap This is a crucial step in our understand-ing of the astrophysics of the ISM with a direct impact forthe cosmological problem of foreground contamination First ofall without a detailed morphological description of the varia-tion of the synchrotron spectral index across the sky the use-fulness of a synchrotron spatial template looses its appeal inthe study of CMB anisotropies Second the roll-off in the en-ergy spectrum of relativistic electrons as they leave their con-finement volumes in the Galactic disk is expected to intro-duce breaks in the spectrum of the non-thermal component upto a few GHz with significant implications regarding (a) theelusive nature of anomalous microwave emission (AME) fromspinning dust grains (Draine amp Lazarian 1998 1999 Lagache2003 de Oliveira-Costa et al 2004 Watson et al 2005 Davieset al 2006 Dobler amp Finkbeiner 2008 Dickinson et al 2009Ysard et al 2010 Planck Collaboration 2011b) (b) the iden-tification of spectral distortions in the black body curve of theCMB signal (Burigana et al 1991 Bartlett amp Stebbins 1991)and more recently (c) the residual radio signal in the ARCADEdata (Seiffert et al 2011 Kogut et al 2011 Fixsen et al 2011)and (d) the existence of the Galactic haze (Finkbeiner 2004Dobler 2012 Planck Collaboration 2013a)

The lack of surveys in this frequency gap is even more crit-ical for polarization surveys Synchrotron radiation is intrinsi-cally polarized and its contaminating role poses higher risksfor polarization than for total intensity In addition the selec-tion effect of Faraday depolarization is partial only toward thelower end of the gap The first all-sky absolutely-calibrated

polarization survey (Wolleben et al 2006 Testori et al 2004)was a rather recent achievement at 14 GHz Still the gap ex-tends up to the lowest of the WMAP frequency bands (Page et al2007 Hinshaw et al 2009) From the point of view of cosmo-logical implications it becomes more urgent to bridge radio andmicrowave polarization surveys whereas from the astrophysicalpoint of view the precise elucidation of the Galaxy as a Faradayscreen is of key importance to the study of magnetic fields inthe Galaxy (Beck 2007) via the synchrotron mechanism Clearlypolarization surveys at frequencies higher than a few GHz (forexample Leonardi et al 2006) are needed to address the con-tamination of the CMB at higher frequencies where the currentPlanck mission with a wider frequency coverage and a bettersensitivity will undoubtedly deepen our cosmological under-standing (Fauvet et al 2012) Despite their lack of sensitivity todiffuse Galactic emission Galactic plane surveys also producea wealth of information regarding the magneto-ionic mediumwhich help us understand the complexity of the interplay be-tween electron densities and magnetic field strengths in gener-ating the pervading synchrotron radiation of the low-frequencysky (Gao et al 2010) Below we outline the scope of a novelobservational program aimed to bridge the mapping of diffuseGalactic emission in this frequency gap

22 The Galactic Emission Mapping project

The concept of the GEM project dates back to 1992 when thefirst detections of CMB anisotropies called for an improved un-derstanding of diffuse Galactic emission (Smoot et al 1992Bennett et al 1992) to eliminate foreground contamination fromgenuine cosmic signals The existing radio-continuum surveyssuffered from two major shortcomings First they lacked mutualconsistency in their baseline levels (Lawson et al 1987 Bandayamp Wolfendale 1991) to be able to reliably predict the spectrumof Galactic synchrotron emission in the cosmologically interest-ing windows Second the spatial templates traced by the non-thermal component in the 408 MHz and 1420 MHz surveys werecompromised by striping along the scanning direction of the ra-diotelescopes (Davies et al 1996) Thus given the frequencygap mentioned above the initial goal of the GEM project be-came the production of an atlas of total sky brightness mapsat 408 MHz 1465 MHz 23 GHz 5 GHz and 10 GHz whichwould overcome these undesired radioastronomical shortcom-ings (De Amici et al 1994 Smoot 1999) using a full-time ded-icated experiment from different observational sites Improvedmapping of the spectral and large-scale properties of Galacticsynchrotron radiation would result from an observational strat-egy aimed to minimize ground contamination and long-termsystematics to guarantee baseline uniformity In addition theproduction of a self-consistent set of sky maps would allow there-examination of the baseline and destriping in previous surveysto take advantage of their superior resolutions

At the core of the GEM project stood the development of aportable 55-m double-shielded radiotelescope which could bedeployed at different geographic latitudes to maximize homoge-neous sky coverage Since ground contamination is one of themost common sources of systematic errors in this type of sur-vey experiments the two shields were built to minimize side-lobe pick-up of stray radiation and to level out the horizon pro-file at the site location These shields are schematically drawnin Fig 1 and consisted respectively of (a) a 21-m-long rim-halo of aluminum panels extending tangentially from each ofthe 24 dish petals and (b) a concentric network of inclined wire-mesh ground screens which levels the horizon with a 10-high

A1 page 3 of 20

AampA 556 A1 (2013)

Fig 1 Ray-tracing diagram (dotted lines) of the illumination propertiesof the double-shielded GEM dish

azimuth profile Under full illumination the dish subtends anangle of 158 at the prime focus Tello et al (1999 2000) mod-eled the diffraction and attenuation properties of this double-shielded radiotelescope to estimate the spillover and diffractionsidelobe contamination seen in elevation scans with back-fire he-lical feeds at 408 MHz and 1465 MHz

Diurnal and seasonal variations in the temperature ofthe ground also add low-frequency noise to survey measure-ments carried out continuously over extended periods of timeSimilarly the variability of atmospheric emission introduces inaddition to low-frequency noise short-term contributions on thescale of minutes To make individual scans insensitive to atmo-spheric fluctuations while still being able to map large areas ofthe sky we adopted azimuthal scans of sufficiently high constantspeed (1 rpm for Z = 30 scans at 23 GHz) Coupled to the ro-tation of the Earth these zenith-centered circular scans achievecoverage of full declination bands with around-the-clock obser-vations as shown schematically in Fig 2 where sample circu-lar scans spaced at one-hour intervals illustrate the coverage indeclination for a site in the southern hemisphere

Table 1 lists a summary of the main GEM observational cam-paigns since its first deployment in the Owens Valley desertnear Bishop CA in 1993 In this period the ground- andrim-halo shields have been continuously refurbished and up-graded to achieve satisfactory levels of ground rejection up to10 GHz A new rim-halo was re-designed and installed in 2001whereas the ground screen has gone through two major upgradesto incorporate a finer mesh along with a more robust supportstructure (two ground shields have been lost due to unusuallygusty and strong winds) Another great impediment for radioastronomical surveys is the ever increasing usage of the spec-trum by human-related activities which plagues the radio en-vironment with non-negligible levels of radio-frequency inter-ference (RFI) 100 duty-cycle interferers have aborted GEMoperations several times in the past (the 408 MHz campaign in1994 the 1465 MHz runs in 1995 and follow-up observationsat 23 GHz in 2004)

Not included in Table 1 are the polarization measurementswe have been conducting at 5 GHz since 2006 which havealready been used to prepare a preliminary version of ourfirst polarization survey (Ferreira et al 2008) from the south-ern hemisphere The recent inclusion of Portugal in the GEM

Fig 2 Schematic representation of the mapping strategy with the GEMexperiment

collaboration promises to extend our 5 GHz and 10 GHz polar-ization templates of Stokes I Q and U parameters to the north-ern hemisphere which can provide substantial ancilliary data forthe Planck satellite mission and other future CMB polarizationexperiments The GEM-P program will certainly benefit from acarefully chosen site (Fonseca et al 2006)

3 The experiment at 23 GHz

The portability of the GEM experiment reduces its operation tothree main modules (a) a dish pedestal with an altazimuthallyrotating platform (b) a radiometer and (c) a control unit Theseare graphically displayed in Fig 3 with a slip-rings assemblyaligned with the axis of rotation along the base column of thedish pedestal Their task is to distribute power to the scanningdish and to establish a communication protocol with the ra-diometer Two interface boxes one inside a ground enclosureand another attached to the co-rotating base column providethe connecting ports for the slip-rings A 1-hp AC-motor con-trolled by frequency inverters and coupled to the main gear ofthe pedestal by means of a speed reducer provided the motion inazimuth An optical shaft encoder was fixed to the bottom of theslip-rings assembly to measure the angular variations of the ro-tating platform Similarly an elevation encoder was installed onthe dish horizontal axis The analog signals of both encoders arealso sent to connecting ports in their respective interface boxesfrom where they are routed to the data acquisition box located onthe elevation support arm beneath the dish The data acquisitionsystem (DAS) assembles the data frame by digitizing the analoginputs from the encoders the radiometer and a noise source be-fore sending it seamlessly through the slip-rings up to the controlpanel for storage on a PC

31 Data acquisition system

The data acquisition system (DAS) uses three single-width NIMmodules to interface with the experiment and a data-loggingPC The control module has an EPROM with a 16-channeldata frame burnt-in and an output connector for a serial output

A1 page 4 of 20

C Tello et al The GEM 23 GHz survey

the novel approaches of component separation and foregroundremoval (Hobson et al 1998 Maino et al 2002 Delabrouilleet al 2003 2009 Eriksen et al 2004 2006 2008 Bonaldi et al2006 Cardoso et al 2008 Leach et al 2008 Stompor et al2009 Ricciardi et al 2010 Fernaacutendez-Cobos et al 2012 PlanckCollaboration 2013c) which exploit the spectral and spatial di-versity of the different astrophysical and cosmological compo-nents The multi-frequency design of the WMAP mission had al-ready revealed an impressive picture of unprecedented precisionof the CMB at 23 GHz 33 GHz 41 GHz 61 GHz and 94 GHz(Hinshaw et al 2007 2012 Bennett et al 2012) whereasthe Planck mission with its higher angular resolution and su-perb sensitivity at 30 GHz 44 GHz 70 GHz 100 GHz 143 GHz217 GHz 353 GHz 545 GHz and 857 GHz has at the time ofwriting made available an overwhelming set of results that pushthe standards of cosmological and astrophysical research to astate-of-the-art level (Planck Collaboration 2013b)

Yet there remains a considerable gap in our knowledgeof diffuse Galactic emission between the upper-frequency endof the large ground-based radio-telescopes and the lower endachievable from space or balloons which confronts radioas-tronomers and cosmologists alike In this frequency range theincreased resolution of single-dish observations imply that thetime allocation at radio-astronomical facilities capable of sur-veying a sizable area of the sky meaningful enough to ad-dress the large-scale properties of this diffuse component be-comes practically prohibitive The few examples we find inthe literature where high-resolution surveys were undertaken inthis frequency gap default to efforts devoted to construct cata-logues of radio sources against the high-intensity emission of theGalactic disk Among them we find arcminute-resolution sur-veys at 49 GHz (Altenhoff et al 1979) 5 GHz (Haynes et al1978) and 10 GHz (Handa et al 1987) Without a dedicatedinstrument free of time constraints in its observational sched-ule diffuse Galactic emission will continue to lack the cover-age it needs to map the sky away from the Galactic plane inthis frequency gap This is a crucial step in our understand-ing of the astrophysics of the ISM with a direct impact forthe cosmological problem of foreground contamination First ofall without a detailed morphological description of the varia-tion of the synchrotron spectral index across the sky the use-fulness of a synchrotron spatial template looses its appeal inthe study of CMB anisotropies Second the roll-off in the en-ergy spectrum of relativistic electrons as they leave their con-finement volumes in the Galactic disk is expected to intro-duce breaks in the spectrum of the non-thermal component upto a few GHz with significant implications regarding (a) theelusive nature of anomalous microwave emission (AME) fromspinning dust grains (Draine amp Lazarian 1998 1999 Lagache2003 de Oliveira-Costa et al 2004 Watson et al 2005 Davieset al 2006 Dobler amp Finkbeiner 2008 Dickinson et al 2009Ysard et al 2010 Planck Collaboration 2011b) (b) the iden-tification of spectral distortions in the black body curve of theCMB signal (Burigana et al 1991 Bartlett amp Stebbins 1991)and more recently (c) the residual radio signal in the ARCADEdata (Seiffert et al 2011 Kogut et al 2011 Fixsen et al 2011)and (d) the existence of the Galactic haze (Finkbeiner 2004Dobler 2012 Planck Collaboration 2013a)

The lack of surveys in this frequency gap is even more crit-ical for polarization surveys Synchrotron radiation is intrinsi-cally polarized and its contaminating role poses higher risksfor polarization than for total intensity In addition the selec-tion effect of Faraday depolarization is partial only toward thelower end of the gap The first all-sky absolutely-calibrated

polarization survey (Wolleben et al 2006 Testori et al 2004)was a rather recent achievement at 14 GHz Still the gap ex-tends up to the lowest of the WMAP frequency bands (Page et al2007 Hinshaw et al 2009) From the point of view of cosmo-logical implications it becomes more urgent to bridge radio andmicrowave polarization surveys whereas from the astrophysicalpoint of view the precise elucidation of the Galaxy as a Faradayscreen is of key importance to the study of magnetic fields inthe Galaxy (Beck 2007) via the synchrotron mechanism Clearlypolarization surveys at frequencies higher than a few GHz (forexample Leonardi et al 2006) are needed to address the con-tamination of the CMB at higher frequencies where the currentPlanck mission with a wider frequency coverage and a bettersensitivity will undoubtedly deepen our cosmological under-standing (Fauvet et al 2012) Despite their lack of sensitivity todiffuse Galactic emission Galactic plane surveys also producea wealth of information regarding the magneto-ionic mediumwhich help us understand the complexity of the interplay be-tween electron densities and magnetic field strengths in gener-ating the pervading synchrotron radiation of the low-frequencysky (Gao et al 2010) Below we outline the scope of a novelobservational program aimed to bridge the mapping of diffuseGalactic emission in this frequency gap

22 The Galactic Emission Mapping project

The concept of the GEM project dates back to 1992 when thefirst detections of CMB anisotropies called for an improved un-derstanding of diffuse Galactic emission (Smoot et al 1992Bennett et al 1992) to eliminate foreground contamination fromgenuine cosmic signals The existing radio-continuum surveyssuffered from two major shortcomings First they lacked mutualconsistency in their baseline levels (Lawson et al 1987 Bandayamp Wolfendale 1991) to be able to reliably predict the spectrumof Galactic synchrotron emission in the cosmologically interest-ing windows Second the spatial templates traced by the non-thermal component in the 408 MHz and 1420 MHz surveys werecompromised by striping along the scanning direction of the ra-diotelescopes (Davies et al 1996) Thus given the frequencygap mentioned above the initial goal of the GEM project be-came the production of an atlas of total sky brightness mapsat 408 MHz 1465 MHz 23 GHz 5 GHz and 10 GHz whichwould overcome these undesired radioastronomical shortcom-ings (De Amici et al 1994 Smoot 1999) using a full-time ded-icated experiment from different observational sites Improvedmapping of the spectral and large-scale properties of Galacticsynchrotron radiation would result from an observational strat-egy aimed to minimize ground contamination and long-termsystematics to guarantee baseline uniformity In addition theproduction of a self-consistent set of sky maps would allow there-examination of the baseline and destriping in previous surveysto take advantage of their superior resolutions

At the core of the GEM project stood the development of aportable 55-m double-shielded radiotelescope which could bedeployed at different geographic latitudes to maximize homoge-neous sky coverage Since ground contamination is one of themost common sources of systematic errors in this type of sur-vey experiments the two shields were built to minimize side-lobe pick-up of stray radiation and to level out the horizon pro-file at the site location These shields are schematically drawnin Fig 1 and consisted respectively of (a) a 21-m-long rim-halo of aluminum panels extending tangentially from each ofthe 24 dish petals and (b) a concentric network of inclined wire-mesh ground screens which levels the horizon with a 10-high

A1 page 3 of 20

AampA 556 A1 (2013)

Fig 1 Ray-tracing diagram (dotted lines) of the illumination propertiesof the double-shielded GEM dish

azimuth profile Under full illumination the dish subtends anangle of 158 at the prime focus Tello et al (1999 2000) mod-eled the diffraction and attenuation properties of this double-shielded radiotelescope to estimate the spillover and diffractionsidelobe contamination seen in elevation scans with back-fire he-lical feeds at 408 MHz and 1465 MHz

Diurnal and seasonal variations in the temperature ofthe ground also add low-frequency noise to survey measure-ments carried out continuously over extended periods of timeSimilarly the variability of atmospheric emission introduces inaddition to low-frequency noise short-term contributions on thescale of minutes To make individual scans insensitive to atmo-spheric fluctuations while still being able to map large areas ofthe sky we adopted azimuthal scans of sufficiently high constantspeed (1 rpm for Z = 30 scans at 23 GHz) Coupled to the ro-tation of the Earth these zenith-centered circular scans achievecoverage of full declination bands with around-the-clock obser-vations as shown schematically in Fig 2 where sample circu-lar scans spaced at one-hour intervals illustrate the coverage indeclination for a site in the southern hemisphere

Table 1 lists a summary of the main GEM observational cam-paigns since its first deployment in the Owens Valley desertnear Bishop CA in 1993 In this period the ground- andrim-halo shields have been continuously refurbished and up-graded to achieve satisfactory levels of ground rejection up to10 GHz A new rim-halo was re-designed and installed in 2001whereas the ground screen has gone through two major upgradesto incorporate a finer mesh along with a more robust supportstructure (two ground shields have been lost due to unusuallygusty and strong winds) Another great impediment for radioastronomical surveys is the ever increasing usage of the spec-trum by human-related activities which plagues the radio en-vironment with non-negligible levels of radio-frequency inter-ference (RFI) 100 duty-cycle interferers have aborted GEMoperations several times in the past (the 408 MHz campaign in1994 the 1465 MHz runs in 1995 and follow-up observationsat 23 GHz in 2004)

Not included in Table 1 are the polarization measurementswe have been conducting at 5 GHz since 2006 which havealready been used to prepare a preliminary version of ourfirst polarization survey (Ferreira et al 2008) from the south-ern hemisphere The recent inclusion of Portugal in the GEM

Fig 2 Schematic representation of the mapping strategy with the GEMexperiment

collaboration promises to extend our 5 GHz and 10 GHz polar-ization templates of Stokes I Q and U parameters to the north-ern hemisphere which can provide substantial ancilliary data forthe Planck satellite mission and other future CMB polarizationexperiments The GEM-P program will certainly benefit from acarefully chosen site (Fonseca et al 2006)

3 The experiment at 23 GHz

The portability of the GEM experiment reduces its operation tothree main modules (a) a dish pedestal with an altazimuthallyrotating platform (b) a radiometer and (c) a control unit Theseare graphically displayed in Fig 3 with a slip-rings assemblyaligned with the axis of rotation along the base column of thedish pedestal Their task is to distribute power to the scanningdish and to establish a communication protocol with the ra-diometer Two interface boxes one inside a ground enclosureand another attached to the co-rotating base column providethe connecting ports for the slip-rings A 1-hp AC-motor con-trolled by frequency inverters and coupled to the main gear ofthe pedestal by means of a speed reducer provided the motion inazimuth An optical shaft encoder was fixed to the bottom of theslip-rings assembly to measure the angular variations of the ro-tating platform Similarly an elevation encoder was installed onthe dish horizontal axis The analog signals of both encoders arealso sent to connecting ports in their respective interface boxesfrom where they are routed to the data acquisition box located onthe elevation support arm beneath the dish The data acquisitionsystem (DAS) assembles the data frame by digitizing the analoginputs from the encoders the radiometer and a noise source be-fore sending it seamlessly through the slip-rings up to the controlpanel for storage on a PC

31 Data acquisition system

The data acquisition system (DAS) uses three single-width NIMmodules to interface with the experiment and a data-loggingPC The control module has an EPROM with a 16-channeldata frame burnt-in and an output connector for a serial output

A1 page 4 of 20

AampA 556 A1 (2013)

Fig 1 Ray-tracing diagram (dotted lines) of the illumination propertiesof the double-shielded GEM dish

azimuth profile Under full illumination the dish subtends anangle of 158 at the prime focus Tello et al (1999 2000) mod-eled the diffraction and attenuation properties of this double-shielded radiotelescope to estimate the spillover and diffractionsidelobe contamination seen in elevation scans with back-fire he-lical feeds at 408 MHz and 1465 MHz

Diurnal and seasonal variations in the temperature ofthe ground also add low-frequency noise to survey measure-ments carried out continuously over extended periods of timeSimilarly the variability of atmospheric emission introduces inaddition to low-frequency noise short-term contributions on thescale of minutes To make individual scans insensitive to atmo-spheric fluctuations while still being able to map large areas ofthe sky we adopted azimuthal scans of sufficiently high constantspeed (1 rpm for Z = 30 scans at 23 GHz) Coupled to the ro-tation of the Earth these zenith-centered circular scans achievecoverage of full declination bands with around-the-clock obser-vations as shown schematically in Fig 2 where sample circu-lar scans spaced at one-hour intervals illustrate the coverage indeclination for a site in the southern hemisphere

Table 1 lists a summary of the main GEM observational cam-paigns since its first deployment in the Owens Valley desertnear Bishop CA in 1993 In this period the ground- andrim-halo shields have been continuously refurbished and up-graded to achieve satisfactory levels of ground rejection up to10 GHz A new rim-halo was re-designed and installed in 2001whereas the ground screen has gone through two major upgradesto incorporate a finer mesh along with a more robust supportstructure (two ground shields have been lost due to unusuallygusty and strong winds) Another great impediment for radioastronomical surveys is the ever increasing usage of the spec-trum by human-related activities which plagues the radio en-vironment with non-negligible levels of radio-frequency inter-ference (RFI) 100 duty-cycle interferers have aborted GEMoperations several times in the past (the 408 MHz campaign in1994 the 1465 MHz runs in 1995 and follow-up observationsat 23 GHz in 2004)

Not included in Table 1 are the polarization measurementswe have been conducting at 5 GHz since 2006 which havealready been used to prepare a preliminary version of ourfirst polarization survey (Ferreira et al 2008) from the south-ern hemisphere The recent inclusion of Portugal in the GEM

Fig 2 Schematic representation of the mapping strategy with the GEMexperiment

collaboration promises to extend our 5 GHz and 10 GHz polar-ization templates of Stokes I Q and U parameters to the north-ern hemisphere which can provide substantial ancilliary data forthe Planck satellite mission and other future CMB polarizationexperiments The GEM-P program will certainly benefit from acarefully chosen site (Fonseca et al 2006)

3 The experiment at 23 GHz

The portability of the GEM experiment reduces its operation tothree main modules (a) a dish pedestal with an altazimuthallyrotating platform (b) a radiometer and (c) a control unit Theseare graphically displayed in Fig 3 with a slip-rings assemblyaligned with the axis of rotation along the base column of thedish pedestal Their task is to distribute power to the scanningdish and to establish a communication protocol with the ra-diometer Two interface boxes one inside a ground enclosureand another attached to the co-rotating base column providethe connecting ports for the slip-rings A 1-hp AC-motor con-trolled by frequency inverters and coupled to the main gear ofthe pedestal by means of a speed reducer provided the motion inazimuth An optical shaft encoder was fixed to the bottom of theslip-rings assembly to measure the angular variations of the ro-tating platform Similarly an elevation encoder was installed onthe dish horizontal axis The analog signals of both encoders arealso sent to connecting ports in their respective interface boxesfrom where they are routed to the data acquisition box located onthe elevation support arm beneath the dish The data acquisitionsystem (DAS) assembles the data frame by digitizing the analoginputs from the encoders the radiometer and a noise source be-fore sending it seamlessly through the slip-rings up to the controlpanel for storage on a PC

31 Data acquisition system

The data acquisition system (DAS) uses three single-width NIMmodules to interface with the experiment and a data-loggingPC The control module has an EPROM with a 16-channeldata frame burnt-in and an output connector for a serial output

A1 page 4 of 20

C Tello et al The GEM 23 GHz survey

Table 1 Total intensity observations with the GEM experiment

Frequency HPBW Sky coverage Duration Observational campaign(MHz) (degrees) (degrees) (hours)

408 105 +072 lt δ lt +675 40 Bishop CA USA 1994 (1)1465 42 +072 lt δ lt +675 70 Bishop CA USA 1994 (2)408 113 minus244 lt δ lt +356 447 Villa de Leyva Colombia 1995 (3)

2300 37 minus244 lt δ lt +356 231 Villa de Leyva Colombia 1995 (4)1465 42 minus521 lt δ lt +067 709 Cachoeira Paulista SP Brazil 1999 (5)2300 230 times 185 minus5173 lt δ lt +0636 532 Cachoeira Paulista SP Brazil 1999 (6)1465 42 minus521 lt δ lt +067 510 Cachoeira Paulista SP Brazil 2004ndash2005 (7)1465 42 minus521 lt δ lt +067 417 Cachoeira Paulista SP Brazil 2009 (7)

References (1) Rodrigues (2000) (2) Tello (1997) (3) Torres et al (1996) (4) Torres et al (internal report) (5) Tello et al (2005) (6) Tello et al(this article) (7) Tello et al (in prep)

Fig 3 Graphical layout of the GEM experiment at 23 GHz

line The analog multiplexerADC module receives 16 differen-tial analog inputs of FSD plusmn10V with a common-mode rejectionof 86 dB The ADC limits hard at plusmn10 V with no bleed-overbetween channels and overloads of plusmn30 V In addition to theEOF signal the DAS also times the firing of a thermally sta-ble noise source every 80 frames The noise source is kept un-derneath the dish surface inside an insulated box and transmitsan asymp140 K reference signal from a small front-fire helix sittingat the junction of two of the dish petals Two analog channelsare dedicated to record the temperature and the voltage of thenoise source Finally a serial-to-RS232 module takes the seam-less serial data and synchronizes each frame with the hexadeci-mal EB90 word for decoding in the PC LabView routines wereused with Macintosh and Pentium PC to log the data Propertime-stamping of the data followed by adding UT from a WWVdigital receiver clock to the data frame for the observations inColombia whereas the WGS84 datum from a GPS receiver pro-vided the time reference for the observations in Brazil

32 Radiometric characterization

For the observations at 23 GHz we used the three-legged sup-port structure of the parabolic reflector to secure a direct-gaintotal-power radiometer at its focal plane by means of a latchingmechanism Its feed is built-in at the bottom of the radiometerenclosure as shown in Fig 4 and consists of a flanged cylin-drical horn with 14-λ chokes and a measured voltage standingwave ratio (VSWR) of 1023 at 23 GHz (return loss of minus39 dB)Its circular aperture provides suitable illumination of the primarywith a symmetric beam and a discrimination of minus16 dB at theedge of the reflector Since the Moon is a reasonable approxima-tion of a point source for our beam horizontal and vertical beampatterns were obtained directly from the survey data by mappingthe signal onto a Moon-centered coordinate system With our cir-cular scanning strategy the horizontal pattern is derived from theantenna response when the Moon reaches the same elevation asthe scanning circle while the vertical pattern is established from

A1 page 5 of 20

AampA 556 A1 (2013)

Fig 4 Focal configuration at 23 GHz with a direct-gain total powerradiometer

the antenna response when the dish azimuth coincides with thatof the Moon These beam pattern profiles are shown in Fig 5 forthe data collected at the Colombian and Brazilian sites UsingGaussian fits to these profiles we obtained weighted averagebeam widths of 231 plusmn 003 for the horizontal and 185 plusmn 010for the vertical planes of the pattern from the two experimentalset-ups Assuming that the secondary introduces no beam dis-tortion the 25 larger horizontal beam is an upper limit to thesmearing of the beam due to the rotation of the antenna Usinginternal errors only this smearing of the beam is significant atthe 45σ level

The power fed by the reflector is captured at the throat ofthe horn by a waveguide probe which sends it down a sec-tion of semi-rigid coax into a front-end high electron mobil-ity transistor (HEMT) amplifier As depicted in Fig 3 theRF chain of the receiver section attains a nominal total ampli-fication of sim70 dB The HEMT amplifier has a flat responseacross the entire 100 MHz width of the bandpass-defining filterand a noise temperature of asymp30 K at an operating temperatureof 300 K After the filter the signal is rectified by a square-lawdetector diode and its DC voltage is amplified by a factor of500 (low gain) or 1000 (high gain) The DC amplifier is housedwith a thermo-regulating circuit and a timing board inside a de-tached electronics box The timing board subdivides the pulsesfrom an end-of-data frame (EOF) signal generated in the dataacquisition box to sample the DC voltage with a time constantof τ = 056002s

For the receiver to achieve an optimal performance its gainvariations were kept to a minimum by thermally stabilizing theinterior of the RF-tight radiometer enclosure asymp10 C below thetemperature set-point of the aluminum RF plate To accomplishthis functionality we used a Peltier cooling unit on the side of theradiometer to continuously extract heat from its interior whilethe thermo-regulating circuit maintained the RF plate (physical)temperature at its set-point by means of a sensing diode on theplate to actively control a set of power resistors The rest of the

RF enclosure was coated internally and externally with insulat-ing styrofoam Four calibrated temperature sensors constantlymonitored the thermal state of the radiometer at the detectordiode the HEMT amplifier the horn flange and the electron-ics box

The radiometric system was calibrated against the ampli-tude of the Gaussian fits to the Moon-signal profiles shown inFig 5 which we compared with the expression of Krotikov ampPelyushenko (1987) for the brightness temperature at the centerof the lunar disk at 23 GHz according to

T MoonMB = 22163 + 314 cos (Ωt minus 45) (1)

where Ωt denotes the dependance on the optical phase of theMoon and sets the full Moon atΩt = 0 This expression is equiv-alent to the one obtained by Mangum (1993) except that the oneabove allows for the averaging action of the antenna beam pat-tern A summary of calibration constants and antenna parametersobtained with this method is included in Table 2

There are some disadvantages with this method that impactthe accuracy of our estimates but we have assessed the uncer-tainties in several ways to guarantee a reliable temperature scalefor our map Independent estimates of the total and main beamsolid angles are the main source of uncertainty since the ratioof the beam efficiency εM and the aperture distribution factorκp which scales the product of the horizontal and vertical beamwidths into the main beam solid angle enter our gain calcula-tion Assuming a ratio of εMκp = 075105 we have tested thevalidity of the Moon calibration method by factoring out the to-tal beam solid angle from its relation to the effective aperturesuch that

S Jy = πR2Moon

2kλ2

⎛⎜⎜⎜⎜⎜⎜⎝

TMB

S peakvolts minus S ref

volts

⎞⎟⎟⎟⎟⎟⎟⎠

Moon

times⎛⎜⎜⎜⎜⎜⎜⎝

S peakvolts

1 minus ηGΔT peakcal

minus S refvolts

1 minus ηGΔT refcal

⎞⎟⎟⎟⎟⎟⎟⎠

source

times 1026 (2)

where S Jy is the flux density of an observed known source injanskies RMoon is the angular radius of the lunar disk in radiansS volts is the signal output of the radiometer in volts ηG is the ther-mal gain susceptability in Table 2 and ΔTcal is the temperaturedrift of the RF plate from its mean temperature during the MoonS peak

volts measurements We chose Vir A and Cen A to test the cal-ibrating Moon amplitude in the Colombian and Brazilian datasets respectively Vir A is sufficiently smaller than our beamsize and we obtained 134 Jy using Eq (2) compared to 140 Jyusing the Baars et al (1977) semi-absolute spectrum between400 MHz and 25 GHz as given by

log S VirA[Jy] = 5023(plusmn0034)minus 0856(plusmn0010) logνMHz (3)

The Baars scale was updated by Ott et al (1994) to assess sourcevariability and they obtained new measurements at λ = 21 116 and 28 cm for Vir A which allowed them to fit a second-orderpolynomial between 1408 MHz and 1055 GHz such that

log S OttVirA[Jy] = 4484 minus 0603 log νMHz minus 0028 log2 νMHz (4)

However at our frequency of 23 GHz this revised scale pro-duces 138 Jy which does not differ significantly from the Baarsscale Cen A on the other hand is an extended source coveringsome 4times10 in celestial coordinates so we targeted its brightestcomponent the northeast inner lobe and obtained 94 Jy com-pared to 99 Jy using the Alvarez et al (2000) integrated fluxdensity relation between 80 MHz and 43 GHz as given by

log S CenA[Jy] = 435(plusmn008)minus 070(plusmn002) logνMHz (5)

A1 page 6 of 20

C Tello et al The GEM 23 GHz survey

Fig 5 Beam pattern characterization at 23 GHz from Moon measurements during routine survey scans away from the Galactic plane

The less than 5 discrepancy between our estimates of bothsources and the published values lends substantial support tothe use of the observed amplitude of the Moon signal to cali-brate our antenna temperature scale The ratio εMκp does notenter Eq (2) explicitly but it is implied in the calculation ofthe RF plate physical temperature drift Given the good agree-ment between observed and published fluxes the assumed ratiorestrains the beam efficiency to the range defined by the uncer-tainty quoted in Table 2 on behalf of the expected range for κpalone (100 le κp le 105)

An additional source of uncertainty with the Moon calibra-tion method is the need to estimate the temperature of the ref-erence background level above which the Moon signal is super-imposed This level consists of two isotropic components theCMB radiation and a small contribution of the diffuse back-ground of extragalactic emission but it also includes Galacticforeground radiation atmospheric emission and stray radiationfrom the ground through the sidelobes The CMB contributionhas been measured accurately to be 2725 plusmn 0002 K (Matheret al 1999) while we estimated the extragalactic backgroundat 0027 K (Lawson et al 1987) Likewise we estimated the at-mospheric emission at 164 plusmn 007 K for the Colombian site andat 229 plusmn 013 K for the Brazilian site according to the modelingprescription of Danese amp Partridge (1989) As for the Galacticforeground component the observations of the Moon took placein regions away from the Galactic plane so that an estimateof 08 plusmn 04 K was used in agreement with successive refine-ments of the calibration constants during the removal of thestray radiation component The assessment of ground contam-ination proved to be the single most challenging aspect in thepreparation of the GEM 23 GHz map as will be seen in the

next section and for this reason the system temperature estimategiven in Table 2 includes the contribution of the ground

4 Data description and processing

The data used in the preparation of the GEM 23 GHz map werecollected during three observational seasons at the Colombianand Brazilian sites listed in Table 1 A brief description of eachobservational site and their associated data sets is given in thefirst part of Table 2 We used Z = 30 scans for a total combinedsky coverage of asymp66 In Fig 6 we have summarized the majorsteps in the preparation of the combined map

41 Time stamping and pointing calibration

The Colombian data set had each frame tagged in seconds andwe adjusted a simple linear progression with a step size of τ sec-onds to time-stamp each frame For the Brazilian data we reliedon an iterative process to minimize the difference between thetiming of the frames in the timeline of 1 s updates of the GPS re-ceiver and the expected cadence of the seamless data stream Webroke the latter into equal time slots of 2h and 34m to facilitatedetection and correction of systematics during the data collec-tion Shorter time slots were inevitable whenever interruptionsof the scanning process were encountered in lining up the TODor when the data stream was found to be corrupted

Readings from the azimuth angle encoder were first scaledto match rotations of 360 at constant speed The ephemeris ofthe Sun was used to estimate the offset between the astronomi-cal azimuth of the Sun and the encoder readout of the radiometerpeak signal during scans with closest approach in the direction

A1 page 7 of 20

AampA 556 A1 (2013)

Table 2 GEM 23 GHz survey

Description Colombia Brazil

Observational site Villa de Leyva Cachoeira PaulistaLongitude (WGS84) minus7335prime053primeprime minus4459prime5434primeLatitude (WGS84) +537prime784primeprime minus2241prime074primeprimeAltitude (masl) 2173 572Observing runs 1995-Jun-1ndash18 1999-May-18ndashJun-17

1999-Oct-11ndash26Antenna mounting altazimuthal altazimuthalAzimuth scanning speed (rpm) 099632 plusmn 000036 100290 plusmn 000063Sky coverage () 463 468Pointing accuracy 6prime84 5prime26Center frequency (MHz) 2300 2300Pre-detection BW (MHz) 100 100Gain (K Vminus1) 54675 50928Gain susceptability (Kminus1) minus002381 plusmn 000028 minus002164 plusmn 000006System temperature (K) 85466 61644RMS sensitivity (mK) 1142 824RF plate Tcal (K) 310572 308031Horizontal HPBW () 230 plusmn 013 231 plusmn 003Vertical HPBW () 192 plusmn 018 182 plusmn 012Beam efficiency () 750 plusmn 35 750 plusmn 35Aperture efficiency () 380 399PSS (Jy Kminus1) 306 291

of the Sun To guarantee a consistent mechanical configurationfor all scans we used a calibrated steel bar to lock the dish intopre-defined elevation angles Fine-tuning of the actual azimuthand elevation angles was accomplished by re-projecting the po-lar coordinates θ and φ of the boresight onto a celestial gridcentered on the ephemeris-specified position of the Moon asshown in Fig 7 We estimated the pointing accuracy of the sur-veys obtained from the Colombian and Brazilian data using therms deviation of the centroid of the Gaussian fits in Fig 5 fromthe origin of the Moon-centered coordinate system We obtainedrms values of 6prime84 from the Colombian data and 5prime26 from theBrazilian data

42 Total power systematics

Once the time-ordered data were properly synchronized we ex-tracted the raw signal of the Moon and calibrated the radiometerconstants as described in Sect 32 Of these constants the ther-mal gain susceptability ηG required extraction of the raw sig-nal observed in pre-defined regions of low Galactic emission orcold sky regions as shown in the 408 MHz survey of Haslamet al (1982) to monitor the linear response of the radiometersignal output as a function of the RF plate physical tempera-ture Fortunately these low-emission regions were scanned dur-ing routine day-time hours when the drift in ΔTcal displayed itslargest excursions and thus enabled accurate linear fits to thesignal-RF plate temperature correlation A set of three regionsof 5 times 5 were chosen for each site as listed in Table 3 andthe linear correlation coefficients of each set were averaged andscaled by the corresponding ratio of gain to system temperatureto obtain ηG

Total power systematics are both linear and non-linear in na-ture Thermal gain susceptability corrections address some ofthe systematics of the linear type Non-linear gain variationson the other hand cannot be corrected for without additionaltime-series analysis of the data stream Initially we had intendedto use the periodical firings of the noise source to correct gainvariations in general However after some careful analysis of

Table 3 Low Galactic emission regions

Region Celestial coordinates ndash Epoch 2000(degrees)

Colombia 1 1410 le RA le 1460 300 le Dec le 350Colombia 2 1345 le RA le 1395 285 le Dec le 335Colombia 3 1360 le RA le 1410 15 le Dec le 65Brazil 1 450 le RA le 500 minus330 le Dec le minus280Brazil 2 540 le RA le 590 minus330 le Dec le minus280Brazil 3 370 le RA le 420 minus300 le Dec le minus250

the noise source amplitude variations we found that the noisesource had become unstable in the course of time and that its useas a transfer calibrator failed to meet the accuracy we requiredTo filter oddly behaving data we turned to an alternate criterionwhich relied on the distribution of the lowest observed antennatemperature T min

A per time slot (154m max) Non-linear excur-sions of the baseline were found to correspond to low valuesof T min

A so we fixed lower limits for its distribution to improvethe characterization of the systematics that affect the survey

For this purpose we split the Brazilian survey into three sep-arate data sets each comprising about two weeks of data col-lection whose characteristics reflect the seasonality effect be-tween summer and winter but also the long-term performanceof the experiment over a month of winter observations Thefirst of the two Brazilian winter data sets (winter I) coincidedwith the same period of the year as the single Colombian dataset Figure 8 shows the distribution of T min

A for the four datasets All three Brazilian data sets are seen to be well con-tained under a single-peaked distribution centered around 45 Kwhereas the Colombia data are not as smoothly distributed witha main peak below 3 K a secondary peak coinciding with thesingle peak of the Brazilian data sets and a likely third andsmaller peak around 6 K Although a realistic lower limit at 35 Kcould be chosen for the Brazilian data as indicated by the ver-tical arrow in the figure a value of 2 K had to be used for theColombian data set to avoid excessive decimation of the survey

A1 page 8 of 20

C Tello et al The GEM 23 GHz survey

Fig 6 Overview of the survey data processing

The clumpy distribution of T minA in the Colombian data set

does have an immediate effect in the mapping of its survey dataAs Fig 9 shows there is clear evidence of an inhomogeneousbaseline due to scan-induced circular striping On the other handmapping of the Brazilian survey shows a smoother baseline con-trast but a rather different and disturbing systematic shows up inthe form of heavy striping in declination as a result of substan-tial ground pick-up in the northern part of the Cachoeira Paulistasky

43 RFI and ground contamination removal

The two maps in Fig 9 show several bright spots due to man-made interference (RFI) whose excision was accomplished bymeans of a low-pass filter during a four-step iteration processaimed at removing the ground striping In the Colombia mapthere is also the sinusoidal streak left by an artificial satellitewhose transmissions highlight portions of the contaminatingtrack To remove this type of RFI a tedious monitoring of thesurvey time slots was conducted and the affected time slots werecleaned of observations found within the perimeter of the satel-lite streak

The cleaning of the ground striping proved on the otherhand to be a homogeneous baseline-dependant process so itwas applied iteratively to the Colombian and Brazilian data setsin slightly different ways as indicated at the bottom of Fig 6

For both data sets the process started by flagging the presenceof the Sun and Moon whenever their angular separation fromthe observing direction was less than 30 rarr 60 and 6 respec-tively Subsequently a variance map of the antenna temperaturesin the remainder of the survey was obtained but the inhomoge-neous baseline of the Colombian data set required its baselineto be calibrated as described in Sect 5 before its variance mapwas obtained In a second step the observations were comparedto the variance map on a per pixel basis and a new set of TODwas assembled from those observations that did not exceed threetimes the variance in the corresponding pixel A pixel resolutionof 360256 = 140625 was chosen to prevent undersampling forthe size of our data sets In the third iteration step the new andRFI-cleaned TOD were averaged in azimuth to produce meanprofiles of the ground contamination for each observed time slotas long as the observations were distanced more than 30 fromthe Galactic plane to prevent real sky features from appearing ascontaminating signatures in the mean azimuth profile Finally inthe last and fourth step an overall mean azimuth profile for eachof the four data sets was secured by excluding any remainingisolated sky feature This was accomplished with a low-pass fil-ter which looped over the azimuth profiles from all time slots ina given set to exclude profile points with more than three timesthe variance per azimuth bin (05 wide) The loop was repeateduntil no more profile points were excluded From here on thefour-step iteration process was repeated but the subtraction ofthe normalized overall-mean azimuth profile from the previous

A1 page 9 of 20

C Tello et al The GEM 23 GHz survey

Fig 6 Overview of the survey data processing

The clumpy distribution of T minA in the Colombian data set

does have an immediate effect in the mapping of its survey dataAs Fig 9 shows there is clear evidence of an inhomogeneousbaseline due to scan-induced circular striping On the other handmapping of the Brazilian survey shows a smoother baseline con-trast but a rather different and disturbing systematic shows up inthe form of heavy striping in declination as a result of substan-tial ground pick-up in the northern part of the Cachoeira Paulistasky

43 RFI and ground contamination removal

The two maps in Fig 9 show several bright spots due to man-made interference (RFI) whose excision was accomplished bymeans of a low-pass filter during a four-step iteration processaimed at removing the ground striping In the Colombia mapthere is also the sinusoidal streak left by an artificial satellitewhose transmissions highlight portions of the contaminatingtrack To remove this type of RFI a tedious monitoring of thesurvey time slots was conducted and the affected time slots werecleaned of observations found within the perimeter of the satel-lite streak

The cleaning of the ground striping proved on the otherhand to be a homogeneous baseline-dependant process so itwas applied iteratively to the Colombian and Brazilian data setsin slightly different ways as indicated at the bottom of Fig 6

For both data sets the process started by flagging the presenceof the Sun and Moon whenever their angular separation fromthe observing direction was less than 30 rarr 60 and 6 respec-tively Subsequently a variance map of the antenna temperaturesin the remainder of the survey was obtained but the inhomoge-neous baseline of the Colombian data set required its baselineto be calibrated as described in Sect 5 before its variance mapwas obtained In a second step the observations were comparedto the variance map on a per pixel basis and a new set of TODwas assembled from those observations that did not exceed threetimes the variance in the corresponding pixel A pixel resolutionof 360256 = 140625 was chosen to prevent undersampling forthe size of our data sets In the third iteration step the new andRFI-cleaned TOD were averaged in azimuth to produce meanprofiles of the ground contamination for each observed time slotas long as the observations were distanced more than 30 fromthe Galactic plane to prevent real sky features from appearing ascontaminating signatures in the mean azimuth profile Finally inthe last and fourth step an overall mean azimuth profile for eachof the four data sets was secured by excluding any remainingisolated sky feature This was accomplished with a low-pass fil-ter which looped over the azimuth profiles from all time slots ina given set to exclude profile points with more than three timesthe variance per azimuth bin (05 wide) The loop was repeateduntil no more profile points were excluded From here on thefour-step iteration process was repeated but the subtraction ofthe normalized overall-mean azimuth profile from the previous

A1 page 9 of 20

AampA 556 A1 (2013)

Fig 7 Images of the Moon before (left panel) and after (right panel) fine-tuning the pointing calibration constants Contour levels are in antennatemperature units

Fig 8 Histograms of observed antenna temperature minima in theColombian and Brazilian data sets Villa de Leyva (dotted line) andCachoeira Paulista (gray line ndash summer gray-filled ndash winter I hatched ndashwinter II)

iteration was subtracted in the first two steps The normalizationconsisted in offsetting the profile down by the amplitude of itslowest profile point In this way the ground contamination wassuccessively eliminated until the overall-mean aziumth profile ofthe last iteration was practically reduced to zero

Figure 10 shows the cumulative overall-mean azimuth pro-files of the four data sets after meeting the convergence criterionThe spiky Brazilian profiles reflect the nature of the striping indeclination but they also highlight the seasonality effect and theneed to assess winter and summer data sets separately As ex-pected the profile of the Colombian data set shows the contami-nation from the ground to be much lower and less variable Stillresidual horizontal striping remained visible in the final iteratedmap and additional destriping was applied in the final prepara-tion of the map as described in the next section

Table 4 Surveyed data composition

Colombia Brazil

Description (h) () (h) ()

Survey total 16285 10000 53211 10000Corrupted data 413 254 TOD mismatch 656 403 Noise source 1141 701 3991 75Sun and Moon 170 1044 7811 1468T min

A exclusion 438 269 4336 815RFI 246 151 346 065

Mapping total 11691 7179 36727 6902

Table 4 summarizes all cuts applied to the data sets as a resultof the selection and cleaning processes described so far A moreconservative minimum angular separation of 60 for the Sun wasused with the Brazilian data sets to eliminate traces of scatteringsidelobes from the three-legged feed support structure as wellas to guarantee a safer margin of secondary sidelobe suppres-sion The angular separation for the Colombian data set at 30reflects the lack of conclusive evidence for sidelobe contamina-tion at larger angular separations due to the reduced sensitivityfrom a smaller data volume Corrupted data and TOD mismatchdid not apply to the Brazilian data sets since an automatic datacollection algorithm was implemented to keep the observationaltime slots of equal length Finally some 68 h of data associatedwith extreme weather conditions severe cases of RFI and unusu-ally large excursions in the readings of the temperature sensorsand of the radiometer signal were flagged for exclusion duringthe observing runs in Colombia and did not enter the process-ing scheme outlined in Fig 6 Consequently they were not ac-counted for as part of the total survey time in Table 4 when com-pared with Table 1

5 Final map preparation and calibration

The final preparation of the GEM 23 GHz survey was di-vided into three major steps First the baseline of the

A1 page 10 of 20

C Tello et al The GEM 23 GHz survey

Fig 9 Maps of unprocessed survey data from selected time slots (T minA limited) in the Villa de Leyva (top panel) and in Brazilrsquos winter I (lower

panel) data sets including the prominent feature introduced by the Sun around 5h of RA in the Colombian map

Fig 10 Cumulative overall-mean azimuth profiles of the ground con-tamination component of the antenna temperature Villa de Leyva(cross-hatched)) and Cachoeira Paulista (thick line ndash summer gray-filled ndash winter I thin line ndash winter II)

ground-subtracted Brazilian data sets was calibrated and theresulting map was merged with the one from the cleanedColombian data set Second the destriping technique devel-oped by Platania et al (2003) was applied to the merged mapto clean residual striping due to ground contamination andbaseline inhomogeneities Third a direct comparison with theRhodesHartRAO survey at 2326 MHz (Jonas et al 1998) wasperformed to absolutely calibrate the GEM survey

51 Baseline calibration

To reduce the circular stripes introduced by the scanning tech-nique in the TOD of the Brazilian data set its baseline wascalibrated following the same prescription as was applied tothe Colombian data set (see Fig 6) This calibration consisted

in resolving the true sky temperature distribution at the levelwhere its coldest contour would intersect at least once everysurvey scan (a 60 wide circular scan centered at the zenith witha 230 times 185 HPBW) In other words the coldest observationof every survey scan would be normalized to a uniform temper-ature or apparent zero-point of our temperature scale across theentire declination band This assumption is to a first-order ap-proximation sufficient to correct the 1 f -noise in the TOD asthe dominant source of baseline inhomogeneity but it also de-fines an effective full-beam brightness temperature scale for thesurvey

When applied to the Colombian data set the precision of thebaseline calibration was set by the level of the fluctuations in theground profile of Fig 10 or 62 plusmn 16 mK Without this a prioricalibration the ground profile fluctuations would have increasedto 1357 plusmn 424 mK and would have compromised any subse-quent baseline calibration For comparison the ground profilefluctuations in the Brazilian data averaged to 355 plusmn 55 mK inthe winter I 350 plusmn 56 mK in the winter II and 362 plusmn 57 mK inthe summer data sets before their baseline was calibrated Thisdifference in implementing the baseline calibration did not al-low the Colombian and Brazilian surveys to be merged alongtheir entire region of overlap (minus23 ltsim δ ltsim +6) Instead anoffset of 148 plusmn 24 mK was applied to the Colombian data ac-cording to the difference between the two surveys along thenorthern boundary (best signal-to-noise ratio) of the Braziliansurvey for α ltsim 17 h and around an overlapping rectangular re-gion sim11 wide for α gtsim 17 h The merging of the two surveysin this overlap region allowed resiliant traces of ground strip-ing in the Brazilian survey to be reduced Ultimately the base-line calibration relied on the uniformity in the sampling of thesurveyed sky area With our observational technique of zenith-centered circular scans the sky is sampled preferentially towardthe edges of the declination band This intrinsic bias diminisheswith the number of observations by statistical virtue alone andthus contributed to the baseline differences between the twomaps Of course cuts in the TOD such as those indicated inTable 4 also affected the sampling uniformity and introducedadditional baseline inhomogeneities

A1 page 11 of 20

AampA 556 A1 (2013)

Fig 11 Map of residual striping obtained with the FFT filtering technique of Platania et al (2003) on the ground-corrected merged map

52 Cleaning of residual striping

The merged map showed traces of residual striping whoseRMS level could be estimated at 58 mK after filtering the zero-frequency component of the map in Fourier space The FFT de-striping technique of Platania et al (2003) was subsequently ap-plied to produce a cleaner map The technique was adapted froma method applied by Schlegel et al (1998) to destripe the IRASmap and uses source extraction and stripe identification thresh-olds in 30 partially overlapping patches of 32 times 32 pixels forthe GEM map This small number of pixels per patch and thecorresponding large angular dimension was suitable to addressthe variability of the striping pattern in the merged map and re-sulted in the stripe image shown in Fig 11 with an RMS levelof 22 mK The procedure was not applied to five patches (151of the map) containing the Galactic plane because of steep signalgradients and a clean GEM map was obtained by subtracting thestripe image from the merged map

The destriping procedure was applied in Platania et al(2003) to the 408 MHz full-sky map of Haslam et al (1982)and to the 1420 MHz (Reich 1982 Reich amp Reich 1986) and2326 MHz (Jonas et al 1998) surveys Given the importanceof the latter for the absolute calibration of the GEM map wecompare in Table 5 the effect of the destriping on the GEMand Rhodes maps in terms of the percentage temperature vari-ations δT of the pixels Nearly half of the destriped pixels in theGEM map did not exceed variations of more than 5 whereasin the Rhodes map the corresponding percentage of pixels washigher (asymp70) Therefore we examined the significance of thevariations larger than 5 for the GEM map and Table 5 alsoshows the distribution of the percentage variations in two sep-arate temperature regimes GEM-I of negative residuals (ΔT lt0 K) and GEM-II of positive residuals (ΔT gt 0 K) Despite thehigher contribution of pixels with positive residuals with varia-tions δT gt 10 their average residual was only 27 plusmn 17 mKwhereas the smaller fraction of pixels with negative residu-als in the same range averaged to a significantly higher neg-ative residual of minus51 plusmn 44 mK For variations in the 5 ltδT le 10 range the average residuals were 13 plusmn 11 mK andminus18plusmn 13 mK Altogether this tendency of the negative residualsto require larger corrections than the positive residuals could im-ply that the ground subtraction slightly overcorrected this type ofcontamination

Table 5 Distribution of percentage temperature variations δT in de-striped pixels of the GEM and RhodesHartRAO surveys

Survey lt1 1 lt δT lt 5 5 lt δT lt 10 gt10() () () ()

GEM-I 65 221 139 60GEM-II 59 172 120 164GEM 124 393 259 224Rhodes 183 506 225 86

53 Absolute calibration

The absolute baseline calibration was obtained by means of aT-T plot between destriped versions of the GEM and Rhodesmaps The Rhodes survey from which the isotropic componentsof the CMB and the diffuse extragalactic background had beensubtracted was convolved with a Gaussian profile with a FWHMof 140prime to match the GEM survey The overlap between the twosurveys covers the declination range minus51 ltsim δ ltsim +30 afterexcluding boundary systematics Using a celestial grid with aresolution of 360256 = 140625 produced 13 132 paired pix-els whose T-T plot is shown in Fig 12 along with a linearfit (98 correlation) for calibrating the true zero-point of theGEM survey baseline Notwithstanding the spectral implicationsdue to the small deviation of the linear coefficient from unity(+143plusmn16 mK Kminus1 is equivalent to a temperature ratio betweentwo synchrotron components had the GEM band been centeredat 231405 plusmn 045 MHz) we can alternatively estimate the zero-point correction from the straight mean of the pixel-by-pixeltemperature differences at 3125 plusmn 0092 K Similarly the dis-tribution of these pixel differences is nearly normal as shown inFig 13 with a Gaussian envelope centered at 31356plusmn 00019Kand standard deviation of 655 plusmn 19 mK An important featureof the destriping procedure is that it has only a negligible ef-fect on the temperature scale of the map and to illustrate thiswe included in Fig 13 the Gaussian envelope and its underlyingdistribution of pixel differences for the ground-corrected mapbefore the destriping We estimate the error in the zero-levelof our survey at 103 mK by combining in quadrature the stan-dard deviation of the pixel differences with the error of 80 mK inthe absolute zero-level of the Rhodes survey according to theircalibration with the absolute sky temperature measurements ofBersanelli et al (1994) at 2 GHz from the South Pole

A1 page 12 of 20

C Tello et al The GEM 23 GHz survey

Fig 12 T-T plot between the destriped GEM survey at 2300 MHz andthe RhodesHartRAO survey at 2326 MHz

Fig 13 Distribution of pixel-by-pixel temperature differences be-tween GEM and Rhodes maps before (ground-corrected) and after thedestriping

Our estimate agrees with the expected accuracy of the Moon-calibrated temperature scale in Sect 32 where we quotedan error lt5 that the cold sky baseline calibration translatesinto lt156 mK Similarly the difference between the appar-ent Moon calibration-dependant zero-point and its true valueis 108 mK

The clean and calibrated GEM survey at 23 GHz is pre-sented in Fig 14 The brightness temperature scale of the sur-vey is tied up to the resolution-limited approach of the cold skybaseline calibration and the full-beam definition of the Rhodessurvey We can obtain a conversion factor between the full-beam estimates of the two surveys by linearly correlating theGEM survey with the unconvolved version of the Rhodes surveyThis gives a conversion factor 08151 plusmn 00031 which matchesthe 08641 for the emission ratio in the direction of the Galacticcenter reasonably well given the 5 uncertainty of the Rhodestemperature scale and the compounded error in the zero-level

accuracy of the GEM survey In Figs 15 and 16 a set of pre-defined contour levels shows a direct comparison between thesetwo surveys which enables some common features to be easilyidentified In the next section we address some differences thatemerge in the morphology of their large-scale structure of theradio continuum

6 Discussion

In Sect 22 we introduced the GEM project as an experimentconceived to improve the mapping of the spectral and large-scale properties of Galactic synchrotron emission This improve-ment would be centered on overcoming two major undesired ra-dioastronomical shortcomings that affect existing ground-basedsurveys the mutual consistency of their baselines and the ob-servational bias of striping effects In this section we discusspreliminary results based on the GEM 23 GHz survey whichwhen compared to the already published HartRAORhodes sur-vey in the same frequency band indicate a promising trend to-ward achieving these goals

Our discussion will be developed in the framework of a con-cept introduced by Bennett et al (2003) for defining foregroundmasks in studies of the CMB with the WMAP data productsTheir basic assumption was that the asymmetry in a histogramof the sky temperature distribution is due to Galactic foregroundemission In their Fig 1 they show how the sky temperature pro-file for the first-year results of the WMAP mission in the K-bandmay be decomposed into a symmetric distribution about the peakand a remainder Even neglecting the effect of point sources thisremainder is recognized as a temperature excess caused by theintervening presence of our Galaxy

61 All-sky surveys

Since the K-band survey of the WMAP mission has a sig-nificant synchrotron component of Galactic emission we in-cluded in the present discussion the 408 MHz all-sky surveyof Haslam et al (1982) with its notably dominant synchrotroncharacter We chose the one-degree smoothed version of the lat-est co-added total intensity map in the K-band1 as well as adestriped version of the 408 MHz map (Platania et al 2003)Their equal-area Mollweide projections in celestial coordinatesare displayed in Fig 17 This choice of coordinates enhancesthe declination-dependant nature of ground-based surveys andto make the results comparable across all surveys they wereconvolved to a resolution of 23 for consistency with the GEMsurvey Figure 18 shows the histograms of their correspondingtemperature distributions Their profiles are characterized by asimilar two-component separation which we applied to the colorrendering of the maps such that their shades of gray pixelsrepresent the transition between the symmetric distribution oflow-temperature pixels in black and the excess foreground emis-sion within the bounds of the temperature scale of the mapsConversely pixels found in the range between the lower limit ofthe remainder and the upper limit of the symmetric distributionare displayed in gray

The increasingly higher contrast of synchrotron radiation inGalactic emission toward the lower frequencies appears to be thedetermining factor of the appearance in the morphology of thetransition region At 408 MHz the transition beyond the Galacticridge marks a well-defined boundary between the high-latitude

1 httplambdagsfcnasagov

A1 page 13 of 20

AampA 556 A1 (2013)

Fig 14 Mollweide projection of the 23 GHz GEM survey in Galactic coordinates

Fig 15 Mollweide projection in Galactic coordinates of the large-scale structure of the radio continuum in the 23 GHz GEM survey accordingto selected contours at the level of detection of well-known radio sources and Galactic features (a) Virgo A and 3C273 contours between 023and 043 K (b) Orion A B and Rosette nebula contours between 043 and 063 K (c) Centaurus A and outer countours of the Cygnus regionbetween 064 and 103 K (d) inner contours of the Cygnus region between 103 and 131 K and (e) innermost contours of the Gum nebula

Fig 16 Mollweide projection in Galactic coordinates of the large-scale structure of the radio continuum in the RhodesHartRAO surveyat 2326 MHz according to the same selected contours as in Fig 15

A1 page 14 of 20

C Tello et al The GEM 23 GHz survey

Fig 17 Mollweide projection in celestial coordinates of the transition (gray pixels) between low- and high-emission regions in the all-sky surveysof the WMAP K-band at 23 GHz and of the Haslam map at 408 MHz

Fig 18 Histograms of the sky temperature distributions in the maps displayed in Fig 17

A1 page 15 of 20

AampA 556 A1 (2013)

o

Fig 19 Mollweide projection in celestial coordinates of the transition regions for partial-sky coverage in the GEM strip (upper left) WMAPK-band (upper right) 408 MHz (lower left) Rhodes and (lower right) GEM

areas of low Galactic emission and the excess foreground emis-sion from the Galactic plane and bulge as well as from struc-tures like the north Galactic spur At 23 GHz the transition re-gion in the WMAP K-band survey breaks up into several patchesof gray pixels that blend without any sharp boundary into thelow-emission region of the symmetric component From theseconsiderations we accordingly expect that at intermediate fre-quencies the two-component scenario should be a constant fea-ture of the large-scale structure of the radio continuum and thatthe transition between high- and low-emission regions shoulddisperse into the latter with increasing frequency

62 Partial sky coverage

The Rhodes and GEM surveys given their partial-sky cov-erage are not readily comparable with the all-sky surveysGround-based experiments with their sky-scanning limitationsare confined to bands of declination centered on the zenithof the observational site A quick inspection of the surveysin Fig 17 shows that any declination band will undoubtedlydisplay the two-component separation but their character maychange with the range of Galactic Longitude that is interceptedThe Rhodes and GEM surveys were observed from observa-tional sites with a privileged view of the central regions of theGalaxy Consequently we secured the common declination bandin the range of minus51 lt δ lt +9 for comparison with theWMAP K-band and 408 MHz surveys Below we refer to it asthe GEM strip

The upper panels in Fig 19 show the effect of restricting thesky coverage of the WMAP K-band and 408 MHz surveys to theGEM strip Their histograms in the upper two panels of Fig 20show the expected two-component distribution Moreover sincethe lower limit of the remainder and the upper limit of the sym-metric component are precisely the same as those found for theall-sky versions the gray-shaded pixels representing the transi-tion between the two components in Fig 19 correspond to theexact same pixels in Fig 17 We may therefore conclude thatsince the general properties of the morphology of the transi-tion region in the WMAP K-band and 408 MHz surveys are not

affected by the particular choice of sky coverage in the GEMstrip the Rhodes and GEM surveys can be tested for these samegeneral properties

The results are shown in the lower panels of Figs 19 and 20Because the Rhodes and GEM surveys are mapping the sky es-sentially at the same frequency the morphology of their transi-tion zones should not be significantly different from each otherYet these results tell a different story

The main difference is the double-peaked distribution ofthe symmetric component in the temperature histogram of theRhodes survey whereas for the GEM survey its low-emissionregion displays a single-peaked temperature profile similar tothose in the histograms of the WMAP K-band and 408 MHz sur-veys In addition the close proximity between the lower limitof the remainder and the upper limit of the symmetric compo-nent in the histogram of the Rhodes survey results in an excep-tionally narrow transition region between the two componentsThis causes the considerable reduction in gray-shaded pixels thatwould otherwise mark the transition region The GEM surveyon the other hand shows an increase in the width of the tran-sition region compared with 408 MHz that suggests an interme-diate stage toward the patchy distribution seen in the WMAPK-band morphology of the transition region A deeper appreci-ation of these implications lies beyond the scope of the presentdiscussion but it certainly bears relevance to the distinction thatsynchrotron and free-free mechanisms can have in the charac-terization of Galactic emission An improvement in the under-standing of the morphology of the large-scale structure of theradio continuum in high-latitude regions is a major goal of theGEM project

63 Temperature spectral index

We close our discussion with a brief treatment of the tempera-ture spectral index between the above set of surveys as a func-tion of the two-component temperature distribution In Table 6we compare the mean spectral index between the surveys listedin the first and second columns over the emission regions ofthe 408 MHz and WMAP K-band surveys Between these two

A1 page 16 of 20

C Tello et al The GEM 23 GHz survey

Fig 20 Histograms of the sky temperature distributions in the partial sky maps displayed in Fig 19 showing their separation into symmetric andexcess temperature components

Table 6 Spectral index in the 408 MHz and WMAP K-band emission regions

Survey Emission region in Survey 2

1 2 High Transition Low

WMAP K-banda 408 MHza minus272 plusmn 019 minus285 plusmn 015 minus290 plusmn 014

WMAP K-bandb minus272 plusmn 019 minus281 plusmn 017 minus288 plusmn 014RhodesHartRAO 408 MHzb minus267 plusmn 011 minus266 plusmn 010 minus261 plusmn 010GEM minus272 plusmn 020 minus269 plusmn 018 minus256 plusmn 019

408 MHza WMAP K-banda minus272 plusmn 016 minus290 plusmn 007 minus302 plusmn 009

408 MHzb minus272 plusmn 016 minus288 plusmn 017 minus301 plusmn 010RhodesHartRAO WMAP K-bandb minus276 plusmn 024 minus308 plusmn 013 minus332 plusmn 018GEM minus277 plusmn 023 minus308 plusmn 013 minus333 plusmn 020

Notes (a) All-sky coverage (b) Sky coverage in the GEM strip

surveys the mean spectral index decreases from the high- tothe low-emission regions as expected from a Galactic latitudedependence given the confinement of relativistic eletrons withlower energies and therefore lower losses to scale heights closerto the Galactic plane and their sources but also due to the flat-ter spectral index of HII regions The mean spectral index of thehigh-emission region is essentially insensitive to the survey usedto determine it However for the transition and low-emission re-gions the mean spectral index decreases more sharply in the re-gions defined for the WMAP K-band survey The same behaviorcan be seen when the partial-sky coverage in the GEM strip isused instead

The steepening of the synchrotron spectral index with fre-quency is also apparent for the GEM and Rhodes surveys Their

mean spectral index with respect to the 408 MHz is practicallyconstant across the two emission regions and their transitionFor the GEM survey this constant behavior is not readily ver-ified unless the mean spectral indices are calculated with respectto the emission regions defined by the temperature distributionof the GEM survey itself For this the estimates for the high-transition- and low-emission regions become minus266 plusmn 024minus256plusmn017 and minus262plusmn014 respectively Thus the constancyof the mean spectral index between 408 MHz and the 2300 MHzcan be given directly for the entire GEM strip irrespective ofemission region and results in minus264 plusmn 011 according to theRhodes survey and minus263 plusmn 021 according to the GEM sur-vey Guzman et al (2011) have also recently reported a tem-perature spectral index of minus25 to minus26 over most of the sky

A1 page 17 of 20

AampA 556 A1 (2013)

Fig 21 Mollweide projection in celestial coordinates of the temperature spectral index maps for the GEM (top) and Rhodes (bottom) surveys withrespect to the 408 MHz (left) and WMAP K-band (right) surveys in the GEM strip

between 45 MHz and 408 MHz On the other hand when themean spectral index of the GEM and Rhodes surveys is obtainedwith respect to the WMAP K-band the steepening of the spectralindex increases toward the low-emission region as expected forrelativistic electrons of higher energies propagating to the largerscale heights that characterize these regions at high Galactic lat-itudes This steepening is not entirely due to synchrotron lossesand has been shown to have a dependance on the missing sub-traction of a monopole component in the lower frequency data(Kogut et al 2012)

Deriving the spectral and spatial distribution of the syn-chrotron spectral index fundamentally depends on the calibra-tion of an homogeneous baseline in each of the compared sur-veys This dependance is very sensitive to the scanning strategyof the observations Despite significant improvements applyingdestriping techniques it is apparent from the results in Table 6that higher dispersions are associated with mean spectral indecesbetween surveys with different scanning strategy than betweensurveys with a similar one The spatial distributions of the tem-perature spectral index shown in Fig 21 have been rendered infalse color shades to enhance the striping pattern of declinationscans in the Rhodes and 408 MHz surveys Stripes in declinationare mainly due to differential ground pick-up through the an-tenna sidelobes a systematic effect that in the GEM survey wassignificantly reduced by the choice of zenith-centered circularscans and the double-shielded configuration of the experimentalset-up

The spectral index map between the GEM and the 408 MHzsurveys shows a systematic drop in a rather large region in an ap-parent southern extension of the Galactic bulge however A sim-ilar effect is seen in the corresponding map for the Rhodes sur-vey over a rectangular region centered approximately on Cen ACuriously the apparent southern extension of the Galactic bulgedoes not show up as an unexpected symmetrical feature in thespectral index map between the GEM and WMAP K-band sur-veys where its distribution follows similar patterns on both sidesof the Galactic plane near the Galactic center This kind of sym-metry is absent from the Rhodes and 408 MHz spectral indexmaps with respect to the WMAP K-band survey For now wecan only expect that a more thorough revision of the method

used to calibrate the baseline of the GEM survey can possiblyressolve discrepancies of this nature As we noted in Sect 52the cleaning of residual destriping revealed a tendency of thenegative residuals to require larger corrections than the positiveresiduals which could imply that the ground subtraction slightlyovercorrected this type of contamination

7 Conclusions and future prospects

The preparation of low-frequency surveys of the radio contin-uum from the ground poses a considerable challenge to scientificendeavours given the need to combine the stability of instru-ment performance with observational constraints over long peri-ods of time Any mismatch in this process introduces systematiceffects which radio-astronomers aim to keep under control bydesigning experiments that can address the desired balance be-tween the proposed science and enduring logistics The GEMproject has evolved in this scenario for nearly two decades withthe ultimate goal to improve our understanding of the large-scalestructure of Galactic synchrotron emission in total intensity andpolarization between 408 MHz and 10 GHz signaling the ever-increasing impact that foreground contamination has on CMBstudies and its implications for the astrophysics of the ISM

We have presented our efforts at 23 GHz by combiningtotal intensity observations obtained with a portable 55-mradiotelescope in Colombia and Brazil to produce a synchrotron-dominated template of Galactic emission with 66 sky cover-age from δ = minus5173 to δ = +3478 with an average HPBWof 231 times 185 The main focus of our analysis has been a thor-ough assessment of the systematics that affected the observa-tions in particular the ground contamination The resulting sur-vey shows that its sky temperature distribution into regions oflow and high emission is consistent with the appearance of atransition region as seen in the Haslam 408 MHz and the WMAPK-band surveys The distribution of the temperature spectral in-dex using the GEM survey does not show a spatial dependenceon these regions when calculated against the Haslam 408 MHzsurvey but it steepens significantly from high- to low-emissionregions with respect to the WMAP K-band survey in agree-ment with energy losses associated with relativistic electrons

A1 page 18 of 20

C Tello et al The GEM 23 GHz survey

and their confinement volumes within the Galaxy Proper ac-counting of the synchrotron spectral steepening between thesesurveys can help in establishing useful constraints on the frac-tional contribution of synchrotron and free-free components inthe WMAP K-band survey and therefore place upper limitson residual emission components such as spinning dust (Telloet al in prep)

The results presented in this article summarize one of themain phases in the developing of the GEM project which startedwith the total intensity 408 MHz survey of Torres et al (1996)and is currently dedicated to polarization and total intensity mea-surements of the Galactic foreground at 5 GHz The prepara-tion of spectral index maps between the four working frequen-cies of the project (408 MHz 1465 MHz 23 GHz and 5 GHz)promises yet to reveal an astrophysically interesting scenario forincreasing our knowledge and understanding of the nature andcomposition of Galactic emission

Acknowledgements We are enormously grateful to several generations of stu-dents and technicians at LBNL and INPE for whom GEM was a rich learningexperience We are particularly indebted to John Gibson Alexandre MR AlvesLuiz Arantes and Luiz Antonio Reitano for their dedicated commitment toJon Aymon Tony Banday Justin Jonas and Andrew Jaffe for helpful ad-vice and to SLBINPE for logistics support in Cachoeira Paulista The de-striped version of the RhodesHartRAO map was reproduced courtesy of TonyBanday The GEM project in Brazil was supported by FAPESP through grants9703861-2 9706794-4 and 0006770-2 TV acknowledges support from CNPqthrough grants 3052192004-9 3036372007-2 4843782007-4 3081132010-1 5062692010-8 MB acknowledges the support of the NATO CollaborativeGrant CRG960175 ST acknowledges the support provided by Colcienciasfunding of the GEM project in Colombia through project 2228-05-103-96 con-tract No 221-96 DB acknowledges support from FCT ndash Portugal and POCIthrough an SFRHBPD grant and project grants POCTIFNU422632001 andPOCICTE-AST572092004 Last but not least we would like to acknowledgethe refereersquos comments without which the significance of this article would nothave been fully appreciated

ReferencesAllen C W amp Gum C S 1950 Austr J Sci Res A 3 224Altenhoff W J Downes D Pauls T amp Schraml J 1979 AampAS 85 691Alvarez H Aparici J May J amp Olmos F 1997 AampAS 124 315Alvarez H Aparici J May J amp Reich P 2000 AampA 355 863Baars J W M Genzel R Pauliny-Toth I I K amp Witzel A 1977 AampA 61

99Banday AJ amp Wolfendale A W 1991 MNRAS 248 705Bartlett J G amp Stebbins A 1991 ApJ 371 8Beck R 2007 in Sky Polarisation at Far-Infrared to Radio Wavelengths The

Galactic Screen before the Cosmic Microwave Background eds M-AMiville-Deschecircnes amp F Boulanger EAS Pub Ser 23 19

Bennett C L Smoot G F Hinshaw G et al 1992 ApJ 396 L7Bennett C L Hill R S Hinshaw G et al 2003 ApJS 148 97Bennett C L Larson D Weiland J L et al 2012 ApJ submitted

[arXiv12125225]Berkhuijsen E M 1972 AampAS 5 263Bersanelli M Bensadoun M De Amici G et al 1994 ApJ 424 517Bonaldi A Bedini L Salerno E et al 2006 MNRAS 373 271Burigana C Danese L amp De Zotti G F 1991 AampA 246 49Cane H V 1978 Austr J Phys 31 561Cardoso J-F Martin M Delabrouille J et al 2008 IEEE J Selected Topics

in Signal Processing 2 735Childers J Bersanelli M Figueiredo N et al 2005 ApJS 158 124Danese L amp Partridge R B 1989 ApJ 342 604Das S Marriage T A Ade P A R et al 2011 ApJ 729 62Davies R D Watson R A amp Gutieacuterrez C M 1996 MNRAS 278 925Davies R D Dickinson C Banday A J et al 2006 MNRAS 370 1125De Amici G Torres S Bensadoun M et al 1994 ApampSS 214 151de Bernardis P Ade P A R Bock J J et al 2000 Nature 404 955de Oliveira-Costa A Tegmark M Davies R D et al 2004 ApJ 606 L89Delabrouille J Cardoso J-F amp Patanchon G 2003 MNRAS 346 1089Delabrouille J Cardoso J-F Le Jeune M et al 2009 AampA 493 835Dickinson C Davies R D Allison J R et al 2009 ApJ 690 1585

Dobler G 2012 ApJ 750 17Dobler G amp Finkbeiner D P 2008 ApJ 680 1222Draine B T amp Lazarian A 1998 ApJ 494 L19Draine B T amp Lazarian A 1999 ApJ 512 740Droumlge F amp Priester W 1956 ZAp 40 236Duncan A R Stewart R T Haynes R F amp Jones K L 1995 MNRAS 227

36Duumlnner R Hasselfield M Marriage T A et al 2013 ApJ 762 10Eriksen H K Banday A J Goacuterski K M amp Lilje P B 2004 ApJ 612 633Eriksen H K Dickinson C Lawrence C R et al 2006 ApJ 641 665Eriksen H K Jewel J B Dickinson C et al 2008 ApJ 676 10Fauvet L Maciacuteas-Peacuterez J F Jaffe T R et al 2012 AampA 540 122Fernaacutendez-Cobos R Vielva P Barreiro R B amp Martiacutenez-Gonzaacutelez E 2012

MNRAS 420 2162Ferreira I S 2008 PhD Thesis Instituto Nacional de Pesquisas Espaciais

BrazilFinkbeiner D P 2004 ApJ 614 186Fixsen D J Kogut A Levin S et al 2011 ApJ 734 5Fonseca R Barbosa D Cupido L et al 2006 New Astron 11 551Fuumlrst E Reich W Reich P amp Reif K 1990b AampAS 85 691Gao X Y Reich W Han J L et al 2010 AampA 515 A64Gold B Odegard N Weiland J L et al 2011 ApJS 192 15Guzman A E May J Alvarez H amp Maeda K 2011 AampA 525 A138Handa T Sofue Y Nakai N et al 1987 PASJ 39 709Haslam C G T Klein U Salter C J et al 1981 AampAS 100 209Haslam C G T Salter C J Stoffel H amp Wilson W E 1982 AampAS 47 1Haynes R F Caswell J L amp Simons L W J 1978 Austr J Phys Supp 45

1Hinshaw G Nolta M R Bennett C L et al 2007 ApJS 170 288Hinshaw G Weiland J Hill R S et al 2009 ApJS 180 225Hinshaw G Larson D Komatsu E et al 2012 ApJS submitted

[arXiv12125226]Hobson M P Jones A W Lasenby A N amp Bouchet F R 1998 MNRAS

300 1Jansky K G 1932 Proc I R E 20 1920Jansky K G 1933 Proc I R E 21 1387Jansky K G 1935 Proc I R E 23 1157Jonas J L Baart E E amp Nicolson G D 1998 MNRAS 297 977Keisler R Reichardt C L Aird K A et al 2011 ApJ 743 28Kiepenheuer K O 1950 Phys Rev A 79 738Kogut A 2012 ApJ 753 110Kogut A Fixsen D J Levin S M et al 2011 ApJ 734 4Krotikov V D amp Pelyushenko S A 1987 Soviet Ast 31(2) 216Lagache G 2003 AampA 405 813Landecker T L amp Wielebinsky R 1970 Austr J Phys Supp 16 1Lawson K D Mayer C J Osborne J L amp Parkinson M L 1987 MNRAS

225 307Leach S M Cardoso J-F Baccigalupi C et al 2008 AampA 491 597Leonardi R Williams B Bersanelli M et al 2006 New Astron Rev 50 977Maeda K Alvarez H Aparici J et al 2000 AampAS 140 145Mangum J G 1993 PASP 105 117Maino D Farusi A Baccigalupi C et al 2002 MNRAS 334 53Mather J C Fixsen D J Shafer R A et al 1999 ApJ 512 511Mathewson D S Broten N W amp Cole D J 1965 Austr J Phys 18 665Mayer C H McCullough T P amp Sloanaker R M 1957 ApJ 126 468McClure-Griffiths N M Dickey J M Gaensler B M et al 2005 ApJS 158

178Meinhold P R Bersanelli M Childers J et al 2005 ApJS 158 101Mejiacutea J Bersanelli M Burigana C et al 2005 ApJS 158 109Mills B Y 1959 PASP 71 267Milogradov-Turin J amp Smith F G 1973 MNRAS 161 269OrsquoDwyer I Bersanelli M Childers J et al 2005 ApJS 158 93Ott M Witzel A Quirrenbach A et al 1994 AampA 284 331Page L Hinshaw G Komatsu E et al 2007 ApJS 170 335Planck Collaboration 2011a AampA 536 A1Planck Collaboration 2011b AampA 536 A20Planck Collaboration 2013a AampA 554 A139Planck Collaboration 2013b AampA submitted [arXiv13035062v1]Planck Collaboration 2013c AampA submitted [arXiv13035072v1]Planck Collaboration 2013d AampA submitted [arXiv13035073v2]Platania P Burigana C Maino D et al 2003 AampA 410 847QUIET Collaboration Bischoff C Brizius A et al 2011 ApJ 741 111QUIET Collaboration Araujo D Bischoff C et al 2012 ApJ 760 145Reber G 1940 Proc I R E 28 68Reber G 1944 ApJ 100 279Reber G 1948 Proc I R E 36 1215Reich W 1982 AampAS 48 219Reich P amp Reich W 1986 AampAS 63 205

A1 page 19 of 20

AampA 556 A1 (2013)

Reich W Fuumlrst E Haslam C G T et al 1984 AampAS 58 197Reich W Reich P amp Fuumlrst E 1990a AampAS 83 539Reich W Fuumlrst E Reich P amp Reif K 1990b AampAS 85 633Reich P Reich W amp Fuumlrst E 1997 AampAS 126 413Reich P Testori J C amp Reich W 2001 AampA 376 861Reichardt C L Shaw L Zahn O et al 2012 ApJ 755 70Reichborn-Kjennerud B Aboobaker A M Ade P et al 2010 in

Millimeter Submillimeter and Far-Infrared Detectors and Instrumentationfor Astronomy V eds W S Holland amp J Zmuidzinas SPIE Conf Ser 7741

Ricciardi S Bonaldi A Natoli P et al 2010 MNRAS 406 1644Rodrigues R 2000 MSc Thesis Instituto Nacional de Pesquisas Espaciais

BrazilRoger R S Costain C H Landecker T L et al 1999 AampAS 137 7Schlegel D J Finkbeiner D P amp Davis M 1998 ApJ 500 525Seiffert M Fixsen D J Kogut A et al 2011 ApJ 734 6Smoot G F 1999 in Microwave Foregrounds eds A de Oliveira-Costa amp M

Tegmark ASP Conf Ser 181 61Smoot G F Bennett C L Kogut A et al 1992 ApJ 396 L1Stil J M Taylor A R Dickey J M et al 2006 AJ 132 1158Stompor R Leach S M Stivoli F amp Baccilgalupi C 2009 MNRAS 392

216Tauber J A Mandolesi N Puget J et al 2010 AampA 520 A1

Taylor A R Gibson S J Peracaula M et al 2003 AJ 125 314Tello C 1997 PhD Thesis Instituto Nacional de Pesquisas Espaciais BrazilTello C Villela T Wuensche C A et al 1999 Rad Sci 34 575Tello C Villela T Smoot G F et al 2000 AampAS 145 495Tello C Villela T Smoot G F et al 2005 in New cosmological data and the

values of the fundamental parameters eds A Lasenby amp A Wilkinson ASPConf Ser 201 138

Testori J C Reich P amp Bava J A et al 2001 AampA 368 1123Testori J C Reich P amp Reich W 2004 in The Magnetized Interstellar

Medium eds B Uyaniker W Reich amp R Wielebinski 57Torres S Cantildeon V Casas R et al 1996 ApampSS 240 225Turtle A J amp Baldwin J E 1962 MNRAS 124 459Watson R A Rebolo R Rubintildeo-Martiacuten J A et al 2005 ApJ 624 L89Westerhout G Seeder C L Brouw W N amp Tinbergen J 1962

Bull Astron Inst Netherlands 16 187Wielebinski R Shakeshaft J R amp Pauliny-Toth I I K 1962 Observatory

82 158Wolleben M Landecker T L Reich W amp Wielebinski R 2006 AampA 448

411Yates K W 1968 Austr J Phys 21 167Yates K W Wielebinsky R amp Landecker T L 1967 Austr J Phys 20 595Ysard N Miville-Deschecircnes M A amp Verstraete L 2010 AampA 509 L1

A1 page 20 of 20

• Introduction
• Large-scale Galactic emission and the GEM project
• • CMB contamination
  • The Galactic Emission Mapping project
  • • The experiment at 23GHz
    • • Data acquisition system
      • Radiometric characterization
      • • Data description and processing
        • • Time stamping and pointing calibration
          • Total power systematics
          • RFI and ground contamination removal
          • • Final map preparation and calibration
            • • Baseline calibration
              • Cleaning of residual striping
              • Absolute calibration
              • • Discussion
                • • All-sky surveys
                  • Partial sky coverage
                  • Temperature spectral index
                  • • Conclusions and future prospects
                    • References