head 1.4; access; symbols; locks http:1.4; strict; comment @# @; 1.4 date 2004.12.02.11.47.56; author MarkAllen; state Exp; branches; next 1.3; 1.3 date 2004.12.01.21.40.16; author MarkAllen; state Exp; branches; next 1.2; 1.2 date 2004.12.01.18.08.18; author MarkAllen; state Exp; branches; next 1.1; 1.1 date 2004.12.01.17.52.28; author MarkAllen; state Exp; branches; next ; desc @none @ 1.4 log @Constructing ELAIS Catalog SEDs @ text @ Constructing ELAIS Catalog SEDs. ======================================================================= Mark Allen (CDS), Dec 02, 2004 These notes are intended to help develop the AVO systems for constructing SEDs from the ELAIS catalog data. They cover transforming data into the correct flux units, putting the SEDs into specview compatible files, and accessing these files via SSA. De-redshifting and normalizing the SEDs are also covered. The files referenced here are all available on the wiki page http://www.euro-vo.org/twiki/bin/view/Avo/ElaisSeds What is an SED? --------------- A Spectral Energy Distribution is like a low resolution spectrum, it is basically a set of flux densities in different wavelength bands. The main task in constructing an SED from a set of measurements in different bands is to put all the measurements in the same units, and to also know the wavelengths of the bands in the same units. Once the fluxes and wavelengths are converted, they can be put into 3 columns (wavelength, flux, flux_error) and plotted with a tool like specview. SEDs of ELAIS catalog objects ----------------------------- The Galaxies case requires SEDs to be constructed from the measurements in the ELAIS final merged band catalog (Rowan-Robinson et al. 2004, MNRAS 351, 1290). This is a fairly simple case because the catalog is simply a set of columns with measurements in the different bands (no cross matching needs to be done). The flux conversions details are described below. The ELAIS Catalog ----------------- The ELAIS catalog is available in its original published form http://astro.imperial.ac.uk/Elais/Data_release/Finalbandmerged/elaiscatf_21a.dat. This is a plain ascii file, but note there are a number of formatting and content problems with this file. The catalog is available in Vizier, where the columns have been marked up with metadata, and it can be accessed in many formats including VOTable. (J/MNRAS/351/1290/catalog) A VOTable version of the catalog, saved from Vizier is available Note that the column names of the Vizier version are different to the original column names in the ascii version. Converting to Flux Units ------------------------- The following table lists all the columns which have photometric measurements, their original and Vizier names, their wavelengths in microns, and the formulae to convert the values into flux densities in mJy. --------------------------------------------------------------------------------------------------------- Measurement orig Vizier wavelength formula (measurement -> mJy) column column (micron) name name --------------------------------------------------------------------------------------------------------- 20cm flux (mJy) sR S20cm 200000. no conversion needed 20cm flux error (mJy) errR e_S20cm no conversion needed 175um flux (mJy) s175 S175um 175.0 no conversion needed 175um flux error (mJy) err175 e_S175um no conversion needed 90um flux (mJy) s90 S90um 90.0 no conversion needed 90um flux error (mJy) err90 e_S90um no conversion needed 15um flux (mJy) s15 S15um 15.0 no conversion needed 15um flux SNR snr15 S/N15 no conversion needed 15um flux error (mJy) --- --- s15/snr15 7um flux (mJy) s7 S6.7um 6.7 no conversion needed 7um flux error (mJy) err7 e_S6.7um no conversion needed J band (mag) amj Jmag 1.25 sj = 1.592 * 10^(-0.4 * amj) * 1.0E6 J band error (mag) errj e_Jmag sj * errj / ( 2.5 * alog10(exp(1.0))) H band (mag) amh Hmag 1.65 sh = 1.024 * 10^(-0.4 * amh) * 1.0E6 H band error (mag) errh e_Hmag sh * errh / ( 2.5 * alog10(exp(1.0))) K band (mag) amk Kmag 2.2 sk = 0.620 * 10^(-0.4 * amk) * 1.0E6 K band error (mag) errk e_Kmag sk * errk / ( 2.5 * alog10(exp(1.0))) U band (mag) am1 Umag 0.3609 sam1 = 1.71080 * 10^(-0.4 * am1) * 1.0E6 U band error (mag) em1 e_Umag sam1 * em1 / ( 2.5 * alog10(exp(1.0)) g' band (mag) am2 g'mag 0.4857 sam2 = 3.87614 * 10^(-0.4 * am2) * 1.0E6 g' band error (mag) em2 e_g'mag sam2 * em2 / ( 2.5 * alog10(exp(1.0))) r' band (mag) am3 r'mag 0.6215 sam3 = 3.14196 * 10^(-0.4 * am3) * 1.0E6 r' band error (mag) em3 e_r'mag sam3 * em3 / ( 2.5 * alog10(exp(1.0))) i' band (mag) am4 i'mag 0.7671 sam4 = 2.51594 * 10^(-0.4 * am4) * 1.0E6 i' band error (mag) em4 e_i'mag sam4 * em4 / ( 2.5 * alog10(exp(1.0))) z band (mag) am5 Zmag 0.909870 sam5 = 2.21820 * 10^(-0.4 * am5) * 1.0E6 z band error (mag) em5 e_Zmag sam5 * em5 / ( 2.5 * alog10(exp(1.0))) --------------------------------------------------------------------------------------------------------- A version of the ELAIS catalog containing the additional columns of fluxes and flux error has been generated (using the formulae given above). elaiscat_21a_mga1.vot Extracting SEDs --------------- Once the photometric measurements are all converted to fluxes in the same units (mJy), the SEDs can be constructed. For example, the first entry in the catalog, object C15_J002818-424303, the SED would be: wavelength flux flux_error (micron) (mJy) (mJy) 15.0000 2.4650000 0.0000000 2.20000 72.109359 1.5275491 1.65000 111.55779 2.0549706 1.25000 144.39196 3.0587680 0.621577 217.37049 0.0000000 0.485727 3.9120048 0.0000000 ( Note that this object was only detected in 6 out of the 13 bands. Also note that some of the flux errors are zero, this is simply because that is what is listed in the catalog. ) The plan for the demo is to use Specview to display these SEDs, so the format of the SED files needs to be compatible with Specview. Currently the possibility of having Specview understand VOtable is being discussed. In the mealtime, we can use the Specview ascii format. An example of this format for the same example object C15_J002818-424303 as shown above is: [demo@@rain SEDs]$ more ELAISC15_J002818-424303_SED.txt ELAISC15_J002818-424303 micron mJy 15.0000 2.4650000 0.0000000 2.20000 72.109359 1.5275491 1.65000 111.55779 2.0549706 1.25000 144.39196 3.0587680 0.621577 217.37049 0.0000000 0.485727 3.9120048 0.0000000 A set of Specview ascii format files for all 3523 objects in the ELAIS catalog are available in SEDs.tar Simple Spectral Access ---------------------- Presuming that the generated SED files would exist in 'MySpace', or somewhere accessible by a URL, the best way to get them into Specview is via the Simple Spectral Access protocol. There is no plan to make an actual server to provide these files on request, but it is easy to use a file containing a SSA response describing the SED file. This is along the lines promoted by Pedro Osuna (see http://pma.standby.vilspa.esa.es:8080/vospec/VOSpecHowTo.html). The SSA response file can be loaded into the AVO prototype, which then allows the files to be directed into Specview, and for the positions of the objects to be used as a catalog in the prototype. (It wont work for VOSpec because the files are Specview format) An example SSA response file describing the Specview ascii format SEDs of the first 1000 objects is available here (ELAIS_first1000_sed_ssa.txt). This describes the SED files that currently reside on astro.u-strasbg.fr/~allen/ELAIS_SEDS/SEDs/ Next Steps : De-redshifting and Normalization --------------------------------------------- The science scenario requires that the observed SEDs be de-redshifted (i.e. have each point in the wavelength axis changed to wave/(1+z)), and then have the SEDs normalized at a given wavelength. This is required in order to make comparisons with the models. Specview has the capability of interactively performing these steps on individual SEDs loaded into the Specview. It is not practical to do this interactively for a large number of objects. It would be good if these steps could also be done with "scripts acting on Myspace files", so that these manipulations can be done on the whole set of SEDs, also allowing the results to be saved. The steps have actually already been applied to the whole sample using IDL. The results, consisting of the de-redshifted, normalized SEDs of ALL the objects in the ELAIS catalog, in Specview ascii format are available in SEDs_norm.tar . An SSA response file describing a subset (of the reddest) objects is available in ELAIS_norm_sed_ssa.txt IDL procedure : elais_seds.pro ------------------------------ The IDL procedure that was written and used to generate all the results described here, is packaged up in a tar file elais_seds_IDL.tar with the few other supporting .pro files. The main file is elais_seds.pro. It is not really intended to be used by anyone, but it may be useful as a guide for how to do all the transformations. For example, the normalization of the SEDs requires linearly interpolating between the two SED points that bracket the normalization wavelength (and testing that this crtieria is met!). Caveats, limitations etc. -------------------------- The flux conversions as described here neglect the aperture corrections that should be applied to the optical U to Z band measurements. This should not be critical at this point. The SEDs generated with the method described here have been compared in detail with SEDs plotted in the original Rowan-Robinson et al. paper. The JHK fluxes calculated here appear to be overestimated compared to the fluxes plotted in the paper. Discussion with the authors (Rowan Robinson, Babbege) has not resolved this issue, and they confirmed we have the correct transformation formule etc. We have neglected including information about the band-widths. When plotting SEDs it would be good to show the bandwidth in some way. In the Jan 2003 demo we did this by showing the bandwidth as an error bar along the spectral axis. Since specview doesn't appear to have the capability to show the bandwidth I haven't followed this up. I've presumed that SED files generated in a workflow would be stored in 'MySpace' or on some URL accessible location. The MySpace browser in the AVO prototype does not allow direct loading of spectra files (it does for images and catalogs) but requires that these spectra are described by a SSA file. Loading of SSA files through the MySpace browser has been tested, but in these cases the actual SED files were located on a URL location. So, one question is, can files in MySpace actually be described in a SSA file, and then be accessed via the prototype? One current limitation with loading SSA files into the prototype is that the size is limited. In fact the SSA file of all the SEDs wont load (but the smaller ones like the 1st 1000 sources will). this can be gotten around in a number of ways, one is to have SSA files for individual sources, or subsets of sources. I don't see this being a major problem for the demo. This document deals with the ELAIS SEDs, but much of it will also apply to the spectra generated by the various models. They also can be put into a specview format, and made available via SSA. @ 1.3 log @Constructing ELAIS Catalog SEDs @ text @d4 1 a4 1 Mark Allen (CDS), Dec 01, 2004 d49 1 a49 1 including VOTable. @ 1.2 log @Constructing ELAIS Catalog SEDs @ text @d3 1 a3 1 ================================ d14 4 d53 1 a53 1 Note that the column names of the Vizer version are different d59 1 a59 1 measurements, thier original and Vizier names, their wavelengths d137 1 a137 1 is being discussed. In the meatime, we can use the Specview ascii d179 1 a179 1 Next Steps : De-redshifting and Normalisation d183 1 a183 1 and then have the SEDs normalised at a given wavelength. This is d214 1 a214 1 For example, the normalisation of the SEDs requires linearly d216 1 a216 1 normalisation wavelength (and testing that this crtieria d220 44 @ 1.1 log @Constructing ELAIS Catalog SEDs @ text @d106 1 a106 1 has been generated (using the formulae given above), and is available at ... d151 1 a151 1 ELAIS catalog are available here ... d170 3 a172 2 of the first 1000 objects is available here. This describes the SED files that currently reside on astro.u-strasbg.fr/~allen/ELAIS_SEDS/SEDs/ d194 1 a194 1 format are available here .... d197 1 a197 1 is available here. d203 1 a203 1 results described here, is packaged up in a tar file d205 1 d215 1 @