A&A 526, A47 (2011)<strong>tel</strong>-<strong>00726959</strong>, version 1 - 31 Aug 2012Thanks to its sensitivity, this instrument will allow, in its compactconfiguration, line surveys to be carried-out down to theconfusion limit toward a large number of sources. Spectral surveysare thus still in their infancy and will very likely becomeroutine observing modes in the coming years.Spectral surveys covering large frequency bands require specifictools to be analyzed efficiently. In this artic<strong>le</strong>, we presenta software that is intended for the analysis of spectral surveys.In Sect. 2, we briefly describe how such surveys are analyzed.In Sect. 3 we detail how our software was designed and imp<strong>le</strong>mentedto carried-out such an analysis. Finally Sect. 4 concludesthis artic<strong>le</strong> and discuss future developments.2. Spectral surveys analysisThe analysis of a spectral survey usually consists in identifyingthe various lines and in deriving the physical and chemicalproperties of the emitting gas (density, temperature and columndensities of the observed species). The main difficulty in suchidentification is that large mo<strong>le</strong>cu<strong>le</strong>s may have hundreds of linesin the (sub-)millimeter range. These species – such as methanol,methyl formate or dimethyl ether – are often named weeds byspectroscopists. If the lines are too broad, they may overlap andb<strong>le</strong>nd together, which makes the identification of weaker linesdifficult. This is the line confusion limit (Schilke et al. 1997):line identification is not limited by the signal-to-noise of the observations,but by the line b<strong>le</strong>nding.Because of this prob<strong>le</strong>m, extreme care must be takenwhen identifying species from a spectral survey. Herbst &van Dishoeck (2009) summarize the criteria for a firm detectionas follows: “(i) rest frequencies are accura<strong>tel</strong>y known to1:10 7 , either from direct laboratory measurements or from ahigh-precision Hamiltonian model; (ii) observed frequencies ofc<strong>le</strong>an, nonb<strong>le</strong>nded lines agree with rest frequencies for a sing<strong>le</strong>well-determined velocity of the source; if a source has a systematicvelocity field as determined from simp<strong>le</strong> mo<strong>le</strong>cu<strong>le</strong>s, anyvelocity gradient found for lines of a new comp<strong>le</strong>x mo<strong>le</strong>cu<strong>le</strong>cannot be a random function of transition frequency; (iii) allpredicted lines of a mo<strong>le</strong>cu<strong>le</strong> based on an LTE spectrum at awell-defined rotational temperature and appropria<strong>tel</strong>y correctedfor beam dilution are present in the observed spectrum at roughlytheir predicted relative intensities. A sing<strong>le</strong> anticoincidence (thatis, a predicted line missing in the observational data) is a muchstronger criterion for rejection than hundreds of coincidencesare for identification. This last criterion is one of the strongestarguments for comp<strong>le</strong>te line surveys rather than targeted linesearches”.The rest frequencies needed to fulfill criterion (i) are usuallytaken from spectral lines catalogs, such as the Cologne Databasefor Mo<strong>le</strong>cular Spectroscopy (CDMS, Mül<strong>le</strong>r et al. 2001) ortheJPL Mo<strong>le</strong>cular Spectroscopy catalog (Pickett et al. 1998). Forcriterion (ii), we need to compare the consistency of the centroidvelocities of all the line candidates. Finally criterion (iii)requires to perform a model of the predicted emission of thegiven species so that it can be compared with the observations.The traditional technique for this consist in building a rotationaldiagram (Goldsmith & Langer 1999) to see if all detected linesagree with a sing<strong>le</strong> rotational temperature and column density.Alternatively, one can compute synthetic spectrum and compareit directly with the observations – a technique cal<strong>le</strong>d forward fitting(Comito et al. 2005). This approach is also extremely usefulwhen one wants to search for weak lines of a specie among hundredsfrom various weeds: a synthetic spectrum of the emissionof the weeds can be constructed to fit the observed transitions inA47, page 2 of 5an iterative fashion. Once the brightest lines have been mode<strong>le</strong>d,one can compare the synthetic spectrum to the observed one tolook for lines from <strong>le</strong>ss abundant species (see Belloche et al.2008, for an examp<strong>le</strong> of this technique). Of course, this also allowsthe physical and chemical properties of the emitting gas tobe derived.Since spectral surveys may contain thousands of lines, theyrequire specific tools to be efficiently analyzed. Two packageshave been developed for that purpose. The first of them,XCLASS (Schilke et al. 2001), is an extension of the widelyused CLASS data reduction software, which is part of Gildas.XCLASS contains a spectral line database which is built fromthe CDMS and JPL catalogs. Technically, it uses the MySQLdatabase server which must be instal<strong>le</strong>d on the user computer.This database may be updated manually, by replacingthe database fi<strong>le</strong> by the one provided by the program authors.XCLASS allows the user to look for lines corresponding to agiven frequency in its catalog, but also to make a model at theLTE of the observed spectra. XCLASS has been successfullyused to reduce several spectral surveys obtained with the CSOand the IRAM-30 m (Schilke et al. 2001; Comito et al. 2005;Belloche et al. 2008). However, XCLASS is based on an obso<strong>le</strong>teversion of CLASS, which is not maintained anymore.Indeed, the CLASS internal structures was largely rewritten in2005–2006 to adapt to the chal<strong>le</strong>nges of data reductions comingwith the recent generation of receivers (Hily-Blant et al.2005). The second package, CASSIS, has been developed primarilyto analyze Herschel-HIFI spectral surveys, although itcan be used to analyze surveys from ground based <strong>tel</strong>escopesas well. CASSIS itself does not have data reduction capabilities;therefore data must first be reduced in another software such asCLASS or HIPE (Ott et al., in prep.) before analysis in CASSIS.CASSIS uses a database which is built from the CDMS and theJPL catalog; in recent CASSIS versions, this database (SQLite)is embedded in the program so that an external database server isno longer required. Like XCLASS, CASSIS allows the forwardfittingof a spectrum, but also the search for the various transitionsof a given specie.3. Weeds design and imp<strong>le</strong>mentation3.1. General designWeeds has been designed specifically to analyze spectral surveys,following the approach presented in Sect. 2. Although itsdevelopment was inspired by the XCLASS and CASSIS packages,it is different in several aspects. Weeds is an extension ofthe current version of the CLASS software, and is mostly writtenin Python language, except for a few command written in theGildas command interpreter (SIC) language. To do this, Weedsuses the new possibility offered by GILDAS to inter<strong>le</strong>ave Pythonand SIC in the same session (Bardeau et al. 2010). In particular,the variab<strong>le</strong> contents are shared between Python and SIC. Pythonhas several advantages over other languages for developing suchextensions. It benefits from a large library of modu<strong>le</strong>s that allowcomp<strong>le</strong>x tasks – such as making a query in a VO-compliantdatabase, see Sect. 3.2 – to be done relatively easily. Although itis interpreted, it is still computationally efficient, because criticalmodu<strong>le</strong>s (e.g. the modu<strong>le</strong> for array computations that we usefor the spectra modeling, see Sect. 3.4) are written in compi<strong>le</strong>dlanguages such as C or Fortran. Weeds is distributed with Gildassince April 2010. The source code is freely availab<strong>le</strong> from theIRAM website 1 . A user manual is also availab<strong>le</strong> on that page.1 http://iram.fr/IRAMFR/GILDAS/
S. Maret et al.: Weeds: a CLASS extension for the analysis of millimeterand sub-millimeter spectral surveys<strong>tel</strong>-<strong>00726959</strong>, version 1 - 31 Aug 2012Because Weeds is an extension of CLASS, it can be used toanalyze any data format that CLASS supports. In practice, theCLASS data format is used by many ground-based <strong>tel</strong>escopes(e.g. IRAM-30 m, CSO and APEX). Data from other <strong>tel</strong>escopescan be converted to FITS format and imported into CLASS aswell. For examp<strong>le</strong>, Herschel-HIFI can be imported into CLASSthrough the FITS fil<strong>le</strong>r delivered by the HIPE data reductionsoftware (Delforge et al., in prep.). In order to analyze data inWeeds, the data must have been calibrated and reduced first. Thereduction usually consists in flagging the bad channels, averagingthe scan covering the same frequency range together, andremoving a polynomial baseline. If the data were obtained withdoub<strong>le</strong> sideband (DSB) receiver, sideband deconvolution mightbe needed in order to produce a SSB spectrum. This requires aspecial observing technique, i.e. a number of overlapping spectrawith shifted local oscillator frequency. Deconvolution canthen be performed in CLASS using the algorithm developed byComito & Schilke (2002). Thus data reduction and analysis canbe done within the same environment.3.2. Spectral line catalogs queriesAs mentioned above, line identification requires repeated queriesto spectral lines catalogs, such as the CDMS or the JPL. UnlikeXCLASS and CASSIS – who require a custom catalog instal<strong>le</strong>don the user’s computer – Weeds performs queries in spectral linedatabases through the Internet 2 . This has the advantage of not requiringany update of a custom catalog: changes in the database,such as species addition or line frequency corrections or updates,are readily availab<strong>le</strong> in Weeds. In order to make queries inspectral lines catalogs, we have imp<strong>le</strong>mented the VO-compliantSimp<strong>le</strong> Line Access Protocol (SLAP, Salgado et al. 2009) inWeeds. This protocol allows spectral line databases queries tobe made in a standardized way; any database that imp<strong>le</strong>ments theprotocol can be accessed by Weeds. Because it is a VO standard,it is likely that more and more spectral line database will use it inthe future. Nonethe<strong>le</strong>ss, as of this writing only the CDMS is accessib<strong>le</strong>using that protocol, through an interface at the Paris VOObservatory (Moreau et al. 2008). Therefore, in order to accessthe JPL catalog from Weeds, we have imp<strong>le</strong>mented queries inthe specific protocol which is used by this database. The CDMScan be accessed through its own protocol as well.For the moment, only one database can be used at a time; itis not possib<strong>le</strong> to combine the catalogs, i.e. to use species someout the JPL and some out the CDMS. In the future, the VAMDCproject 3 will provide a sing<strong>le</strong>, unified database, including stateof-artspectroscopic data from both the CDMS and the JPL catalogs.We plan on imp<strong>le</strong>menting an access to this database fromWeeds as soon as it it re<strong>le</strong>ased.From the user point of view, Weeds provides a commandto search for lines corresponding to a given frequency rangein a spectral line catalog. The user can se<strong>le</strong>ct a region on thespectrum displayed in CLASS, and the command prints all thelines from the catalog around the region se<strong>le</strong>cted. The lines canbe filtered out on the basis of the species they belong to, theirEinstein coefficient, or their upper <strong>le</strong>vel energy. For doub<strong>le</strong> sidebandspectra, a command option allows the search for lines fromthe image band.2 However, Weeds can make a cache of part or an entire catalog, sothat it can be used later with no Internet connection.3 http://www.vamdc.org/3.3. Lines browsing/identificationTo secure the detection of a species in a spectral survey, oneneeds, according to criterion (ii) to search for all the transitionsof that species in the entire frequency range covered by the survey.One also needs to measure the velocity of each line to checkthat they correspond to a sing<strong>le</strong> velocity. Weeds allows the userto browse through a survey rapidly. For this, Weeds has a commandto search for all the lines of a given specie that fall in thefrequency range covered by the survey. The command prints thelines in the terminal, but also builds an internal index containingall these lines, that we can order either by increasing frequencyor increasing upper <strong>le</strong>vel energy. Another command allow theuser to examine each of the line candidate one by one, to seeif the line is detected or not. This command makes a zoom ona small frequency region around the (expected) line, and alsosets the velocity sca<strong>le</strong> with respect to the rest frequency of theline. A vertical mark is also drawn on the displayed spectrum atthe source velocity, so that we can easily determine if the lineis detected or not. A Gaussian fit of the observed line may beperformed to determine the velocity of each line.3.4. Spectra modelingOnce several transitions of a given specie have been found, oneneeds to check if the relative intensities of these componentsagree with a sing<strong>le</strong> excitation temperature (criterion (iii)). In addition,one needs to make sure that non-detected lines are consistentwith the excitation temperature derived from other species– or in other words, that no lines are “missing”. For this Weedsallows the user to compute a synthetic spectrum that can be compareddirectly with the observations (forward-fitting). Followingthe approach used in XCLASS and described in Comito et al.(2005) the synthetic spectrum is computed assuming that theemission arises from one or several components at the LTE.Although this approximation is simplistic – it is well known thatin the inters<strong>tel</strong>lar medium species are often out of local thermodynamicequilibrium, and many sources are known to havedensity and temperature gradients – yet such a zeroth-order approachis often extremely useful to identify lines, as mentionedabove. Once the lines have been identified, a more realistic modeling,taking into account non-LTE excitation effects as well asthe source structure, can be carried-out.Under these assumptions, and after baseline subtraction, thebrightness temperature of a given species as a function of the restfrequency ν is given by:T B (ν) = η [ ( )] (J ν (T ex ) − J ) ν Tbg 1 − e−τ(ν)(1)where η is beam dilution factor, which, for a source with aGaussian brightness profi<strong>le</strong> and a Gaussian beam, is equal to:θ 2 sη =(2)θs 2 + θt2where θ s and θ t the source and <strong>tel</strong>escope beam FWHM sizes,respectively. For a sake of simplicity, the latter is assumed to begiven by the diffraction limit 4θ t = 1.22 c(3)νD4 (Sub-)millimeter <strong>tel</strong>escopes usually have tapers that limit the powerreceived in side-lobes. Because of this, the <strong>tel</strong>escope beam size maybe different that of a purely diffraction limited antenna of the same diameter.However, the difference between the two is usually small: at100 GHz, the measured FWHM of the IRAM-30 m is 24.6 ′′ , whi<strong>le</strong>Eq. (3) gives 25.2 ′′ .A47, page 3 of 5
- Page 1 and 2:
UNIVERSITÉ PIERRE ET MARIE CURIEHA
- Page 3 and 4:
tel-00726959, version 1 - 31 Aug 20
- Page 5 and 6:
Table des matières1 Rapport de sou
- Page 7 and 8:
Rapport après soutenanceHabilitati
- Page 9 and 10:
Chapitre 2Curriculum vitaetel-00726
- Page 11 and 12:
2.7 ANIMATION ET DIFFUSION DE LA CU
- Page 13 and 14:
2.8 PARCOURS 131992-1993 ÉCOLE NOR
- Page 15 and 16:
Chapitre 3Copyright: IRAM/PdBIIntro
- Page 17 and 18:
tel-00726959, version 1 - 31 Aug 20
- Page 19 and 20:
4.2 ETUDES DIRECTES EN ÉMISSION 19
- Page 21 and 22:
4.4 LA LUMINOSITY CO PAR MOLÉCULE
- Page 23 and 24:
356 E. Falgarone et al.: Extreme ve
- Page 25 and 26:
358 E. Falgarone et al.: Extreme ve
- Page 27 and 28:
360 E. Falgarone et al.: Extreme ve
- Page 29 and 30:
362 E. Falgarone et al.: Extreme ve
- Page 31 and 32:
364 E. Falgarone et al.: Extreme ve
- Page 33 and 34:
366 E. Falgarone et al.: Extreme ve
- Page 35 and 36:
368 E. Falgarone et al.: Extreme ve
- Page 37 and 38:
A&A 518, A45 (2010)1001010⌠⌡ τ
- Page 39 and 40:
A&A 518, A45 (2010)If X HCO + is as
- Page 41 and 42:
A&A 518, A45 (2010)tel-00726959, ve
- Page 43 and 44:
A&A 518, A45 (2010)tel-00726959, ve
- Page 45 and 46:
A&A 518, A45 (2010)Table E.2. Data
- Page 47 and 48:
A&A 541, A58 (2012)tel-00726959, ve
- Page 49 and 50:
A&A 541, A58 (2012)tel-00726959, ve
- Page 51 and 52:
A&A 541, A58 (2012)0.08 0.10.20.12
- Page 53 and 54:
A&A 541, A58 (2012)b=240.11 0.1 0.1
- Page 55 and 56:
0.080.09 0.1A&A 541, A58 (2012)0.14
- Page 57 and 58:
A&A 541, A58 (2012)43˚00'0.3 0.44
- Page 59 and 60:
A&A 541, A58 (2012)68˚00'1.1 0.91.
- Page 61 and 62:
A&A 541, A58 (2012)α(J2000)-19H57m
- Page 63 and 64:
A&A 541, A58 (2012)tel-00726959, ve
- Page 65 and 66:
A&A 541, A58 (2012)Jenkins, E. B.,
- Page 67 and 68:
A&A 541, A58 (2012)Galactic Latitud
- Page 69 and 70:
tel-00726959, version 1 - 31 Aug 20
- Page 71 and 72:
5.2 UNE PHYSIQUE BIEN CONTRAINTE ET
- Page 73 and 74:
5.3 PERSPECTIVES : DES RELEVÉS DE
- Page 75 and 76:
A&A 435, 885-899 (2005)DOI: 10.1051
- Page 77 and 78:
tel-00726959, version 1 - 31 Aug 20
- Page 79 and 80:
J. Pety et al.: Are PAHs precursors
- Page 81 and 82:
J. Pety et al.: Are PAHs precursors
- Page 83 and 84:
J. Pety et al.: Are PAHs precursors
- Page 85 and 86:
J. Pety et al.: Are PAHs precursors
- Page 87 and 88:
J. Pety et al.: Are PAHs precursors
- Page 89 and 90:
J. Pety et al.: Are PAHs precursors
- Page 91 and 92:
566 J. R. Goicoechea et al.: Low su
- Page 93 and 94:
568 J. R. Goicoechea et al.: Low su
- Page 95 and 96:
570 J. R. Goicoechea et al.: Low su
- Page 97 and 98:
572 J. R. Goicoechea et al.: Low su
- Page 99 and 100:
574 J. R. Goicoechea et al.: Low su
- Page 101 and 102:
576 J. R. Goicoechea et al.: Low su
- Page 103 and 104:
578 J. R. Goicoechea et al.: Low su
- Page 105 and 106:
580 J. R. Goicoechea et al.: Low su
- Page 107 and 108:
tel-00726959, version 1 - 31 Aug 20
- Page 109 and 110:
J. R. Goicoechea et al.: Low sulfur
- Page 111 and 112:
L42J. Pety et al.: Deuterium fracti
- Page 113 and 114:
L44J. Pety et al.: Deuterium fracti
- Page 115 and 116:
772 J. R. Goicoechea et al.: The io
- Page 117 and 118:
774 J. R. Goicoechea et al.: The io
- Page 119 and 120: 776 J. R. Goicoechea et al.: The io
- Page 121 and 122: 778 J. R. Goicoechea et al.: The io
- Page 123 and 124: 780 J. R. Goicoechea et al.: The io
- Page 125 and 126: 782 J. R. Goicoechea et al.: The io
- Page 127 and 128: A&A 494, 977-985 (2009)DOI: 10.1051
- Page 129 and 130: M. Gerin et al.: HCO mapping of the
- Page 131 and 132: M. Gerin et al.: HCO mapping of the
- Page 133 and 134: M. Gerin et al.: HCO mapping of the
- Page 135 and 136: M. Gerin et al.: HCO mapping of the
- Page 137 and 138: Table 1. Observation parameters for
- Page 139 and 140: A&A 534, A49 (2011)tel-00726959, ve
- Page 141 and 142: A&A 534, A49 (2011)tel-00726959, ve
- Page 143 and 144: A&A 534, A49 (2011)tel-00726959, ve
- Page 145 and 146: tel-00726959, version 1 - 31 Aug 20
- Page 147 and 148: tel-00726959, version 1 - 31 Aug 20
- Page 149 and 150: tel-00726959, version 1 - 31 Aug 20
- Page 151 and 152: tel-00726959, version 1 - 31 Aug 20
- Page 153 and 154: tel-00726959, version 1 - 31 Aug 20
- Page 155 and 156: tel-00726959, version 1 - 31 Aug 20
- Page 157 and 158: ¡¢£¤¥¢¦¢ §¨ ©
- Page 159 and 160: tel-00726959, version 1 - 31 Aug 20
- Page 161 and 162: 7.2 MODE INTERFÉROMÉTRIQUE 161tel
- Page 163 and 164: 7.3 LE FUTUR DE LA RADIO-ASTRONOMIE
- Page 165 and 166: CLASS evolution: I. Improved OTF su
- Page 167 and 168: CLASS evolution: I. Improved OTF su
- Page 169: A&A 526, A47 (2011)DOI: 10.1051/000
- Page 173 and 174: S. Maret et al.: Weeds: a CLASS ext
- Page 175 and 176: IRAM-30m EMIR time/sensitivity esti
- Page 177 and 178: IRAM-30m EMIR time/sensitivity esti
- Page 179 and 180: IRAM-30m HERA time/sensitivity esti
- Page 181 and 182: ¬þ«ðZñÛùþ˜ï³þ«ðWMõÑ
- Page 183 and 184: ¨-A¨-, #f¡§¨\¤qŠ’‹¼€(
- Page 185 and 186: ¤H F!";
- Page 187 and 188: #O.u\J0$!&%©ÉQ©6¦©¨-.u& (HÌ>
- Page 189 and 190: ¨*.G©6E¦cǦ©¨&ẗØn¨"!-‡
- Page 191 and 192: ¢¡¤£¦¥¨§ P© ¡ £¦¥¨§
- Page 193 and 194: '¥0vB"(-7£O.'¥!5#&(-7)7)(:9.*d#J
- Page 195 and 196: ¥¤¡ b@‚¥_ƒO@02,.(-EFJbe O@#V
- Page 197 and 198: 02J$#J+'10-ï 3 02EFJ+')EKJ5L5†R(
- Page 199 and 200: A&A 517, A12 (2010)tel-00726959, ve
- Page 201 and 202: A&A 517, A12 (2010)tel-00726959, ve
- Page 203 and 204: A&A 517, A12 (2010)tel-00726959, ve
- Page 205 and 206: A&A 517, A12 (2010)tel-00726959, ve
- Page 207 and 208: A&A 517, A12 (2010)tel-00726959, ve
- Page 209 and 210: A&A 517, A12 (2010)tel-00726959, ve
- Page 211 and 212: A&A 517, A12 (2010)tel-00726959, ve
- Page 213 and 214: A&A 517, A12 (2010)tel-00726959, ve
- Page 215 and 216: A&A 517, A12 (2010)tel-00726959, ve
- Page 217 and 218: Table C.1. Definition of the symbol
- Page 219 and 220: IRAM Memo 2011-2WIFISYN:The GILDAS
- Page 221 and 222:
WIFISYN3. practiceWIFISYN3. practic
- Page 223 and 224:
WIFISYNA. implementation planWIFISY
- Page 225 and 226:
tel-00726959, version 1 - 31 Aug 20
- Page 227 and 228:
8.3 PERSPECTIVES 227tel-00726959, v
- Page 229 and 230:
3 REQUIREMENTS 44 CHANGES FOR END-U
- Page 231 and 232:
5 CHANGES FOR PROGRAMMERS 125 CHANG
- Page 233 and 234:
A EXHAUSTIVE DESCRIPTION OF THE CHA
- Page 235 and 236:
Chapitre 9Copyright: Stéphane Guis
- Page 237 and 238:
9.3 ACTIVITÉS 2008-2011 237tel-007
- Page 239 and 240:
Contribution de l'Action Spécique
- Page 241 and 242:
A. des multi-pixels à bure : une s
- Page 243 and 244:
5. besoins en services annexes, bé
- Page 245 and 246:
Articles publiés dans des revues
- Page 247 and 248:
ARTICLES PUBLIÉS DANS DES REVUES
- Page 249 and 250:
Mémos IRAM et ALMAtel-00726959, ve
- Page 251 and 252:
Actes de colloques nationaux et int
- Page 253 and 254:
ACTES DE COLLOQUES NATIONAUX ET INT
- Page 255 and 256:
tel-00726959, version 1 - 31 Aug 20