PreCam, a Precursor Observational Campaign for Calibration ...

PreCam, a Precursor Observational Campaignfor Calibration of the Dark Energy Survey

K. Kuehn1,2, S. Kuhlmann1,2, S. Allam3, J. T. Annis3, T. Bailey1,4, E. Balbinot5,6, J. P.Bernstein1, T. Biesiadzinski7, D. L. Burke8, M. Butner9, J. I. B. Camargo6,10, L. A. N. daCosta6,10, D. DePoy11, H. T. Diehl3, J. P. Dietrich7, J. Estrada3, A. Fausti6, B. Gerke8,12,V. Guarino1, H. H. Head9, R. Kessler13, H. Lin3, W. Lorenzon7, M. A. G. Maia6,10, L.

Maki7,14, J. Marshall10, B. Nord7, E. Neilsen3, R. L. C. Ogando6,10, D. Park3,15, J. Peoples3,D. Rastawicki16, J.-P. Rheault10, B. Santiago5,6, M. Schubnell7, P. Seitzer17, J. A. Smith9,

H. Spinka1, A. Sypniewski7, G. Tarle7, D. L. Tucker3,2, A. Walker18, W. Wester3

(the Dark Energy Survey Collaboration)(ANL-HEP-PR-12-60; FERMILAB-PUB-12-393-AE-CD-PPD)

ABSTRACT

PreCam, a precursor observational campaign supporting the Dark Energy Survey (DES),is designed to produce a photometric and astrometric catalog of nearly a hundred thousandstandard stars within the DES footprint, while the PreCam instrument also serves as a prototypetestbed for the Dark Energy Camera (DECam)’s hardware and software. This catalog representsa potential 100-fold increase in Southern Hemisphere photometric standard stars, and thereforewill be an important component in the calibration of the Dark Energy Survey. We provide detailson the PreCam instrument’s design, construction and testing, as well as results from a subsetof the 51 nights of PreCam survey observations on the University of Michigan Department ofAstronomy’s Curtis-Schmidt telescope at Cerro Tololo Inter-American Observatory. We brieflydescribe the preliminary data processing pipeline that has been developed for PreCam data andthe preliminary results of the instrument performance, as well as astrometry and photometry ofa sample of stars previously included in other southern sky surveys.

Subject headings: instrumentation:detectors–methods:observational–techniques:image processing–astrometry–reference systems–surveys

1High Energy Physics Division, Argonne National Lab-oratory, Lemont, IL 60439

2Corresponding Authors; email: [email protected] (KK),[email protected] (SK), [email protected] (DLT)

3Fermi National Accelerator Laboratory, Batavia, IL60510

4Princeton University, Princeton, NJ 085445Instituto de Fısica, UFRGS, Porto Alegre, RS - 91501-

970, Brazil6Laboratorio Interinstitucional de e-Astronomia –

LIneA, Rio de Janeiro, RJ - 20921-400, Brazil7Department of Physics, University of Michigan, Ann

Arbor, MI 481098SLAC National Accelerator Laboratory, Menlo Park,

CA 940259Department of Physics and Astronomy, Austin Peay

State University, Clarksville, TN 3704410Observatorio Nacional, Rio de Janeiro, RJ - 20921-400,

Brazil11Department of Physics & Astronomy, Texas A&M Uni-

versity, College Station, TX 7784312Lawrence Berkeley National Laboratory, Berkeley, CA

9472013Department of Astronomy and Astrophysics, Univer-

sity of Chicago, Chicago, IL 6063714Department of Physics and Astronomy, Wayne State

University, Detroit, MI 4820215Illinois Math and Science Academy, Aurora, IL 6050616Department of Physics, Stanford University, Stanford,

CA 9430517Department of Astronomy, University of Michigan,

1

arX

iv:1

208.

0865

v1 [

astr

o-ph

.IM

] 3

Aug

201

2FERMILAB-PUB-12-393-AE-CD-PPD

Operated by Fermi Research Alliance, LLC under Contract No. DE-AC02-07CH11359 with the United States Department of EnergyOperated by Fermi Research Alliance, LLC under Contract No. DE-AC02-07CH11359 with the United States Department of Energy

1. The Dark Energy Survey in the Con-text of Current Cosmology

The Dark Energy Survey, or DES, will map5000 deg2 of the southern galactic cap to ob-serve more than 108 faint galaxies, more than3000 Type Ia supernovae, and myriad other ob-jects, to determine the nature and temporal evo-lution (if any) of dark energy (Annis et al. 2005;Bernstein et al. 2012). The DES utilizes the DarkEnergy Camera, or DECam (Flaugher et al. 2012),installed on the Blanco telescope at Cerro TololoInteramerican Observatory, to observe each point-ing numerous times in each in five different pass-bands: g, r, i, z, and Y (the first four of whichare similar, but not identical, to the Sloan Dig-ital Sky Survey, or SDSS (Fukugita et al. 1996)filters). The DECam includes 62 2k x 4k pixelscience CCDs (along with additional guide and fo-cus/alignment CCDs) totaling ∼570 megapixels,creating a field of view of ∼3 deg2. A distinguish-ing characteristic of DECam is the choice of 200-micron thick, fully-depleted, n-type, red-sensitiveCCDs for the imager array. These CCDs have amuch better red-sensitivity, with an efficiency ofgreater than 50% per photon at wavelength 1000nm, nearly an order of magnitude better thanstandard thin CCDs (McLeod et al. 1998) at thatwavelength.

The goal of the DES is a factor of 3-5 im-provement in the measured Dark Energy TaskForce Figure of Merit of the dark energy pa-rameters w0 and wa over a Stage II experiment(Albrecht et al. 2006). To achieve this goal, theDES has an all-sky photometric calibration re-quirement of 2% (and a goal of 1%). Becausethe science requirements for the DES are quitestringent, the characteristics of the DECam mustlikewise be precisely controlled. One importantmethod of calibrating the performance of theCCDs (and the entire optical system) during theSurvey is to compare DECam measurements toprevious measurements of known standard stars.Several catalogs of equatorial and Southern Hemi-sphere standard stars exist (Ivezic et al. 2007;Smith et al. 2002; Smith et al. 2013), but the ex-tremely sparse nature of the observations render

Ann Arbor, MI 4810918Cerro Tololo Interamerican Observatory, La Serena,

Chile

them inadequate for the needs of the DES. Fur-thermore, there are almost no standard stars thathave been observed in the Y band. Thus, theefforts to provide calibration standards for theDECam are vital to the success of the overall Sur-vey. The PreCam instrument contains two CCDstaken from among the spare devices made for theDECam itself, making the focal plane similar toa 1/31st-scale version of the DECam focal plane.This allows the PreCam survey observations toprovide many of the necessary calibration stan-dard stars for the DES, especially in the earlyyears of the Survey. This in turn potentiallyallows up to 10% more of the DES time to bededicated to science observations rather than cali-bration measurements, while still maintaining thedesign requirement of 2% photometric accuracythroughout the survey area. Much of the PreCaminstrument’s hardware and software (such as thereadout electronics) is likewise similar to that ofthe DECam, and has also been used for testing theDECam CCD electronics during the reassembly ofthe DECam system at CTIO after shipment.

To accomplish its calibration goals, the PreCamsurvey observes relatively bright (magnitude 14-18) stars predominantly within the DES footprintusing filters very similar to the DES grizY filtersystem (Abbott et al. 2009). For this survey, thePreCam instrument is mounted on the Universityof Michigan Department of Astronomy’s Curtis-Schmidt (C-S) telescope at CTIO. The first majorstep of the PreCam survey is to observe thousandsof existing standard stars that have been cata-logued as part of the SDSS or u′g′r′i′z′ standardstar surveys (Ivezic et al. 2007; Smith et al. 2002;Smith et al. 2013). We then determine the sur-vey’s photometric accuracy by comparing our ob-servations to these other catalogs. We then usefurther PreCam observations of target candidatestandard stars throughout the observed surveygrid (∼105 stars overall) to provide a significantlyexpanded catalog that will be used to improve theoverall photometric accuracy of the DES. On theC-S, the PreCam instrument’s field-of-view is 1.6◦

× 1.6◦ (or 2.56 deg2). The PreCam survey wasdesigned to cover 10% of the DES footprint ina sparse grid in RA and DEC, and each field inthis grid was planned to be observed several timesin each of the five filters. This grid pattern en-sures that the DECam will regularly observe these

2

calibrated standards during routine survey opera-tions, thus obviating much of the need for dedi-cated standard star observations with DECam it-self and thereby saving valuable observing timethat can be devoted to further DES science goals.Figure 1 shows the planned PreCam survey gridoverlaid on the DES footprint. In Section 2 wedescribe the design and construction of the Pre-Cam instrument, as well as preliminary tests ofits performance. In Section 3, we describe Pre-Cam survey operations, while Section 4 describesthe preliminary post-survey processing of the data.Sections 5 covers the results of this analysis, and inSection 6 we discuss the impact of these results inthe context of DES calibrations. In Section 7, weconclude with a look toward the future prospectsfor PreCam.

2. Instrumentation and Bench Tests

The PreCam instrument was primarily de-signed and built at Argonne National Labora-tory, with the data acquisition (DAQ) electronicscomponents contributed by Fermi National Ac-celerator Laboratory. The CCD detectors residein an aluminum pressure vessel designed to oper-ate at ∼10−6 mbar and -100◦ C, as required toreduce thermal noise to acceptable levels withinthe CCDs. In addition to the throughputs forcryogenic cooling and vacuum, the vessel has a4 inch diameter sapphire window with an anti-reflective coating which allowed ∼ 99% transmis-sion of light from the telescope to the CCDs1. Acustom-designed shutter2 is mounted in front ofthe vessel window and is actuated by compressedgas with an opening/closing stroke lasting lessthan 0.25 seconds (with the precise time depend-ing somewhat on the applied gas pressure). Drynitrogen is fed to the shutter and regulated byan electronically-controlled valve that receives theopen signal in coincidence with the recording ofthe observed photoelectrons by the CCD. The ves-sel also has a Vacuum Interface Board (VIB) forsignal readout from the CCDs (Shaw et al. 2012).The CCDs themselves are mounted on an alu-

1Al2O3 dewar window with A/R coating manufacturedby JML Optical Industries, LLC, Rochester, NY 14625(http://jmloptical.com)

2Compressed-gas actuated shutter manufacturedby Packard Shutter Co., Fiddletown, CA, 95629(http://www.packardshutter.com)

minum focal plane immediately behind the vesselwindow and are positioned with an accuracy ofbetter than 1mm, with a spacing between thetwo CCDs of ∼0.5 mm. Kapton cables 30 cm inlength separately connect each CCD to the VIB.The focal plane is secured to a Cu thermal transferblock, which provides thermal contact between theCCDs and the CryoTiger cold probe3. A custom-designed block of G10 plastic secures the thermaltransfer block to the outer aluminum vessel whileminimizing thermal transfer to the vessel walls.Two Pt thermocouples attached to the thermaltransfer block are used to measure the tempera-ture near the CCDs, and two 25 W heaters con-nected to the thermal transfer block complete thefeedback loop for temperature control. While theCryoTiger is designed to operate at full power con-tinuously, a Lakeshore 332 (LS332) device moni-tors the thermocouples and provides power to theheaters to maintain temperature stability4. Toprovide vacuum, a turbopump is attached to thevessel during testing and maintenance, but notoperation5. Additionally, a small quantity (∼ 10g) of charcoal getter is secured to the thermaltransfer block to adsorb any gases not removed bythe vacuum pump6. After vacuum was establishedduring bench tests, outgassing was determined tobe minimal, and the presence of the getter al-lowed the operating pressure (10−6 to 10−5 mbar)to be maintained for more than a month withoutfurther pumping. Outside the vessel, the VIB isconnected to a Monsoon electronics crate contain-ing pre-production DES clock and bias and dataacquisition boards; the Monsoon crate is then con-nected to the control PC with a fiber optic S-Linkinterface (Cardiel-Sas et al. 2008). The entiredata DAQ system uses PANVIEW (a customized“Pixel Acquisition Node” LabVIEW interface),which is in turn controlled by the observer througha prototype of the Survey Image System ProcessIntegration, or SISPI (Honscheid et al. 2008), the

3PolyCold CryoTiger system manufactured by Brooks Au-tomation, Inc., Chelmsford, MA 01824

4LS332 manufactured by LakeShore Electronics, Wester-ville, OH 43226 (http://lakeshore.com)

5HiCUBE Turbomolecular pump manufactured byPfeiffer Vacuum, GmbH, 35614 Asslar, Germany(http://www.pfeiffer-vacuum.com)

6Activated charcoal getter manufactured by Adsorbents &Dessicants Corporation of America (ADCOA), Gardena,CA 90247 (http://www.adcoa.net)

3

Fig. 1.— The PreCam survey grid (black and green points), overlaid on an earlier proposed version ofthe DES footprint (blue shaded region with SN fields as small boxes; the most recent footprint has sinceundergone minor revisions). The grid facilitates the connection of the region of overlap with the SDSS datato the region of overlap with the VHS and SPT data. During standard DES operations, the camera willintersect one of these grid points approximately every 20 minutes throughout the night.

software manager and user interface for the DE-Cam. See Figure 2 for a CAD model of the Pre-Cam instrument and for a photograph of the com-plete PreCam system, as well as Table 1 for thecharacteristics of the instrument.

Bench tests were undertaken at Argonne Na-tional Laboratory from March to July, 2010, todetermine the effectiveness of each individual com-ponent and of the entire integrated system. As al-ready described, vacuum was created by the tur-bopump and maintained by the getter for nearly amonth during the testing period. Once the LS332was programmed with the proper set-points andPt thermocouple resistance values (including theresistance of the readout cables), the temperaturewas stable to within 0.25K. With the PANVIEWsoftware running within SISPI, the CCDs wereread out and combined into a single FITS imagein ∼20 seconds. Other bench tests (e.g., photontransfer curves, see Figure 3) determined the read-noise, dark current, and full well of the CCDs . Atthe conclusion of these tests, the PreCam instru-ment was shipped to CTIO for installation andoperation on the C-S telescope. In the next sec-

tion, we focus on the PreCam survey operationsand data-taking.

3. Operations and Data-Taking

3.1. PreCam Installation and Commis-sioning on the Curtis-Schmidt

We were scheduled for 112 nights of telescopetime (including installation and testing) throughthe University of Michigan Department of Astron-omy on their Curtis-Schmidt telescope at CTIO.The C-S telescope is a 0.61 meter aperture tele-scope of classical Schmidt design, with the CCDsmounted at a Newtonian focus. Its fast optics andwide field of view make it ideally suited for long-duration wide-field observation campaigns such asthe PreCam survey. Figure 4 shows the PreCaminstrument as it is being installed on the C-S tele-scope, along with a close-up view of the instru-ment.

After installation, testing, and commissioning,we had 64 potential nights of on-sky observing.Of these, 51 provided useful science data, whilethe other nights were dedicated to PreCam or C-

4

Fig. 2.— Left: A CAD rendering of the PreCam instrument showing the various components (without theencasing vacuum vessel). Right: Photograph of the PreCam instrument and the bench test setup at ArgonneNational Laboratory, with the shutter installed in front of the dewar window.

CCD Type 2 x DECam 2048 x 4086 pixelsCCD Pixel Scale 1.45 arcsec/pixelDigitization Precision 16 bit/pixelNumber of Readout Ports 4 (2 per CCD)Window Type 4 inch diameter Sapphire, AR CoatedCooler Type and Setpoint CryoTiger, -100◦ CReadout Speed ∼20 sGain 4 ADU/photoelectronDark Noise ∼1 ADU/minReadout Noise ∼1 ADU/pixelFullwell 141K,113K,134K,144K e−/pixelPixel Size 15 µm x 15 µmField of View 1.6◦ x 1.6◦ (<10% vignetting in central 1.0◦ diameter area)

Table 1: PreCam Instrument Characteristics

5

Exposure Time (s)

0 20 40 60 80

(Da

ta­F

it)/

Fit

­0.2

­0.15

­0.1

­0.05

0

0.05

0.1

0.15

0.2

Mean (ADU)

10000 20000 30000 40000

)2

Va

ria

nc

e (

AD

U1000

2000

3000

4000

5000

6000

7000

8000

9000

Slope = 0.251 ADU/e­

Fig. 3.— Left: Linearity of CCD response for various exposure times. Data are consistent with ∼50msshutter actuation time. Exposure times of several tens of seconds ensures linearity. Right: Variance incounts is likewise linear over the region of interest for exposure times.

Fig. 4.— Left: Installation of the PreCam instrument on the University of Michigan Curtis-Schmidt telescopeat CTIO. Right: A close-up of the PreCam instrument fully installed, with associated DAQ electronicsconnected to the Vacuum Interface Board for data readout.

6

S engineering tasks (e.g., fixing a broken domemotor), or encountered various hardware or soft-ware malfunctions (see Data Processing, Section4), or, rarely, poor weather conditions. The test-ing/commissioning tasks completed during Au-gust and September of 2010 included mountingand testing the new camera; installing and testinga new LED-based flat-field system, which was aprototype of the DECal system that is part of theDECam project (Rheault et al. 2012); replacingthe C-S’s undersized secondary mirror with a newmirror and mount; the subsequent re-collimatingof the optics; and interfacing the PreCam in-strument with the C-S Telescope Control System(TCS). After the secondary mirror was replaced itwas discovered that the new mirror did not meetrequirements and we reinstalled the original sec-ondary and mount. The final step of the commis-sioning process involved characterizing the shapeof the focal plane and determining the best fo-cus settings of the C-S mirrors. Although theSchmidt telescope has a curved focal plane, weelected to observe without a field flattener. Cus-tomized software for focus optimization was de-veloped (Allam 2012); output showing the cur-vature of the focal surface is shown in Figure 5.From these observations, filter- and temperature-dependent focus settings were determined and aprocedure for nightly focus determination was de-veloped.

3.2. The Flat-Field System and PreCamFilters

The PreCam flat-field system consists of ascreen placed within the C-S dome and illumi-nated by six sets of LEDs mounted near the topof the C-S telescope. These LEDs emit light atwavelengths of 470, 740, 905, 970, and 1000 nm, aswell as a broad-spectrum “white” light. A seventhLED was included in the flat-field system but wasnot used for PreCam observations. These wereused to ensure uniformity of the system responseto sources of uniform illumination; where varia-tions from uniformity were identified within theobserved data, a standard procedure was appliedto “flatten” the system response so that everypixel of the CCDs effectively would register thesame number of counts for a given source inten-sity. This also allowed us to measure (and correctfor) any possible inaccuracies in collimation of the

optics that would cause non-uniform illuminationacross the focal plane. The flat-field system isalso used to test the effects of the shutter open-ing/closing time on image uniformity. Figure 6shows that such non-uniformities are negligiblefor any image longer than 8 s even without anyshutter map corrections; thus, PreCam images areunaffected by the non-zero shutter actuation time,as they are all 10 s or greater in duration.

Scaled-down versions of filters very similar tothose used by the DECam were also incorporatedinto the PreCam instrument, and they performedas expected (hereafter the PreCam filters are re-ferred to with the subscript pc). The measuredwavelength-dependent transmission of the filtersconvolved with atmospheric absorption and CCDquantum efficiency is shown in Figure 7, while Fig-ure 8 shows the color terms for a selection of ob-served stars, relative to the SDSS and UKIDSSfilters, compared to those from the Pickles stellarlibrary (Pickles 1998). The excellent agreementbetween the expected and observed performanceof the filters is thus confirmed. A comparison ofthe PreCam filters to those actually manufacturedfor and installed on the DECam also will be per-formed to ensure that any differences between thetwo sets of filters are accounted for prior to theapplication of PreCam data to DES calibrations.

3.3. Observing Strategy and the Observ-ing Tactician

During primary data-taking (November 2010 toJanuary 2011), each night of observing was pre-ceded by collection of bias (0 s duration) and darkimages. Around sunset (when there is minimalstray light illuminating the interior of the C-Sdome), flat-field images were also acquired usingthe previously-described system. Once the skywas dark enough (i.e. after astronomical twilight),we made pointing and focus determination obser-vations before proceeding to regular observations.

As part of the PreCam observing procedure, aprototype of the DES Observing Tactician, or Ob-sTac (Annis & Neilsen 2012), was implementedand tested; this automated target selection andscheduling program was used for the vast majorityof the science data-taking, leading to significantimprovements in performance that will be carriedover to the DES itself. ObsTac has two primaryelements: 1) a database that contains tables of

7

Fig. 5.— Red/dark points show the Full Width at Half Maximum (FWHM) for stars as a function of positionon the focal plane. A flat plane means that a given focus setting will provide comparable FWHM for theentire focal plane, whereas a curved surface means that objects appear larger in some regions of the focalplane. These data show that, without a field flattener, only a portion of the focal plane is at the best focusfor a given image (4096 pixels ≈ 1.6◦).

Fig. 6.— Shutter actuation effects as a function of exposure time, from 0 s (top left) to 16 s (lower right).The nonuniformities clearly disappear by the 8 s image (second row, rightmost image). This shows that anyexposure greater than 8 s—including all PreCam observations—will not be significantly impacted by theshutter actuation time.

8

Fig. 7.— Measured transmission as a function of wavelength for PreCam grizY (solid lines) and SDSSgriz (dashed lines) filters, convolved with atmospheric transmission and CCD quantum efficiency. Noteparticularly the increased effectiveness of the PreCam system at the redder end of the spectrum.

Fig. 8.— PreCam photometry compared to SDSS photometry for g, r, i, z, and Y bands as a functionof SDSS/UKIDSS color (grey/light points). Because of the different filter and CCD response of PreCamcompared to SDSS/UKIDSS, a transformation is required to place the observations on identical systems.With an additional constant (zero-point) offset applied, the linear trends smoothly overlap the Picklestheoretical stellar library (red/dark points). While Red Giant Branch stars appear among the theoreticalpoints for the r-band, there do not appear to be any such stars among the observed data. Both the zero-point offset and the linear color terms for each filter are applied in the final comparison of PreCam andSDSS/UKIDSS photometry.

9

field positions, desired exposure times for each fil-ter, exposures already completed or planned, andother data that influence target selection; and 2) aSISPI service, the ObsTac server, that returns thespecification for an observation (pointing, expo-sure time, filter, and earliest and latest acceptabletimes of observation) upon receiving a request foran observation at a given time.

When called, the ObsTac server determines ifthere are any exposures needed with time restric-tions (calibration fields, for example), and if thereare, it returns the relevant data for such an ex-posure. Otherwise, it returns the highest priorityexposure of the survey that is observable, where“observable” takes into account the airmass andphase and separation from the moon. The algo-rithm used to calculate the priority was modifiedseveral times over the course of PreCam observing,but setting time and sky brightness were generallythe determining factors. Fields that set first weretargeted first, as they may not have been availablelater in the survey. Exposures in bluer filters werepreferred in dark sky conditions, because redderfilters can better tolerate brighter sky conditions.In standard observing with the DES, there willalso be additional supporting services—for exam-ple, a simple SISPI service will monitor the lengthof the observing queue, and fill it when necessarythrough calls to the ObsTac server.

For the PreCam survey, however, not all as-pects of the ObsTac infrastructure were used; in-stead, we performed simulations of the night to beobserved, recording which exposures were selectedby ObsTac at any given time. This record wasthen provided to the PreCam observing software,which read and executed the automating observ-ing plan. Because the plan was generated for thewhole night ahead of time, ObsTac could not au-tomatically accommodate changes in observing inreal time. In particular, the actual times of obser-vation occasionally differed from those planned byObsTac, resulting in more large airmass observa-tions than originally expected. Because airmass-dependent corrections as well as FWHM selectioncriteria were applied to the observed data, suchhigh-airmass observations are not expected to neg-atively impact our final results (apart from per-haps reducing the total number of usable stan-dards stars that are incorporated into our cata-log).

Based on the ObsTac observing plan, the Pre-Cam observing software interacted with the C-STCS to move the telescope to the desired observ-ing position, changed the filter to the desired set-ting, opened and closed the shutter, and interfacedwith the DAQ system that recorded and read outeach image. Once each image was completed, theobserving script selected the next object in the ob-serving plan and repeated the observing processuntil all objects were observed, morning twilightprevented further observing, or the observer man-ually interrupted the observing sequence. Guidedby this automated process, we obtained 11020 im-ages during the PreCam survey. The majority ofthese images were located along SDSS Stripe 82and throughout the eastern half of the DES foot-print, though we also obtained some images out-side that region. PreCam survey images containjust under 10 million identified objects, thoughthis includes multiple observations of the sameobjects. All observations are divided into threedistinct categories: calibration exposures (biases,dark frames, dome flats), standard fields (10 s im-ages of known bright standards), and target fields(including both known and candidate standardstars). Standard observations were repeated in ev-ery filter once every hour throughout the observ-ing nights in order to ensure the internal consis-tency of the PreCam dataset. Furthermore, theseobservations were used to determine the averagezero-point offset for data taken with each filter oneach night. With these zero-points offsets applied,the photometric accuracy is greatly improved (seeResults, below).

3.4. The Quick-Reduce Pipeline

During observations, a prototype of the QuickReduce (QR) pipeline developed for the DECamby the DES-Brazil team was deployed at CTIOand adapted to handle data from the PreCam sur-vey. The QR pipeline carries out the followingsteps: 1) automatically reads the most recent im-age stored on disk; 2) applies standard correctionsto this image (overscan, bias subtraction and flat-field correction); 3) extracts a catalog using SourceExtractor (Bertin & Arnouts 1996); 4) computesthe mean of the measured FWHM and evaluatesthe distortion of the Point Spread Function (PSF)of the sources classified as stars on each CCD; 5)compares these values to user-specified values to

10

determine whether the image complies with thequality criteria established; and 6) produces a setof web pages summarizing the results and showinga JPEG representation of the images reduced. TheQR pipeline also allows displaying the raw and re-duced fits files using, for instance, DS9. While theuse of the QR pipeline during PreCam observingwas limited, the experience was extremely valu-able in the development and testing of the finalsystem that has been incorporated into SISPI foruse with the DES.

At the end of each night, the raw PreCam datawere also transferred to the DES tertiary datastorage site in Rio de Janeiro, Brazil, and sub-sequently to Fermilab using a tool named BitsAround the World (BAW) developed by the DES-Brazil group as a substitute for the official CTIOData Transfer System (DTS), which was still un-der development at the time. These data weresubsequently used in the development and test-ing of a pipeline for regular, automated, completereduction of PreCam and (soon-to-be-acquired)DES data. Although the pipeline was not usedto reduce all of the accumulated PreCam data,it was applied to several nights of data, demon-strating that approximately eight hours of datacan be successfully reduced and the results trans-ferred during the following daytime period, en-abling a rapid feedback mechanism for guiding ob-servations on subsequent nights, a critical compo-nent of the DES Observing Strategy.

Table 2 shows the total number of science andcalibration images obtained per filter (after ini-tial data quality selection), along with the numberof stars identified in these images, while Figure 9shows the completed coverage map overlaid on theDES footprint, as well as filter-specific coveragemaps.

4. Preliminary Data Processing

Once the data were collected, we performedstandard bias correction and flat-fielding of all theimages (at -100◦C, the dark current was negli-gible and thus was not subtracted from the im-ages), using master bias and master flat-field im-ages that were derived from all useful images ofthese types. After these standard steps, furtherprocessing was required due to unique circum-stances confronted during PreCam observations.

The full details of the final processing and analysissteps will be detailed in a subsequent publication(Allam et al. 2013); here we describe the prelim-inary analysis applied to a representative sample(∼10%, described in Table 3) of the data collected.

4.1. PreCam-Specific Image Processing:Streaking and Banding Corrections

One significant problem encountered duringdata-taking was horizontal streaking and band-ing within the images. This was eventually tracedto microscopic damage within the cables connect-ing the VIB to the Monsoon crate. Design dif-ferences preclude this problem from occurring onthe DECam; meanwhile, in preparation for futurePreCam data-taking, we have prepared a repairedcable as well as a strain-relief system that preventsthe weight of the cable from damaging the sensi-tive connections as the telescope is moved. Pre-Cam data that have already been collected werescanned for the presence of streaking or banding;nearly 40% of the images showed some signs ofstreaking. Because this induced common-modenoise was observed to be cyclic and to exhibit thesame pattern in each amplifier for each CCD, astraightforward software algorithm comparing thepixel-by-pixel counts in the four amplifiers permit-ted the identification (and removal) of the bandingand streaking effects (see Figure 10 for an exam-ple of an image with significant streaking, andthe subsequent corrected image). Photometricand astrometric accuracy of the vast majority ofthese images was restored with this process; only3.6% of the images were finally considered “unre-coverable” due to the presence of significant anduncorrectable streaking. Meanwhile, the residualuncertainty in photometric accuracy introduced

Date (UT) Filter(s) Used NTargetExposures

21011215 i, z 40, 2920110107 g, z 194, 320110108 r, z 56, 1020110112 i, z 19 , 1920110113 i 17420110117 i 207

Table 3: Observation log for the 10% subsampleanalyzed here, detailing the number of target ex-posures taken, the date, and the filter used.

11

Filter Target Field Exposure Duration (s) NumberTargetExposures NumberStd.Exposures NumberStarsgpc 36 1888 700 959,085rpc 51 1868 700 1,881,279ipc 65 3152 700 3,881,716zpc 162 269 700 1,816,290Ypc 73 343 700 890,181

Table 2: Number of exposures of target fields and standard fields (observed with 10 s exposures), as well asthe number of stars identified from both types of exposures in the PreCam observations, for each PreCam(pc) filter.

Fig. 9.— Coverage maps showing the areas of the PreCam grid observed for each of the five filters duringSeason One of operations. Points outside of the grid are known standard star fields also observed during thePreCam survey. We are currently considering a second season of observations to complete the grid in all fivefilters.

12

by this process is small (mean variation of 0.01mag) for any given image; therefore, thousandsof additional images were retained in our dataset,and ultimately will contribute to the final PreCamStandard Star Catalog.

4.2. PreCam-Specific Image Processing:Shutter Corrections

Repeated actuation of the shutter led to fail-ure of the shutter blades on several occasions.Ultimately, some of the blades completely broke,thereby preventing the shutter from fully closing(and obviously impeding our data-taking efforts).Prior to complete failure, however, the shutterblades would occasionally “stick” in a partially-open configuration. As with the streaking andbanding problems, DECam’s design differencesprevent such a problem from occurring during theDark Energy Survey. Within the PreCam dataset,only a small fraction (∼3.5%) of the images wereidentified as suffering from shutter problems, andthe vast majority of these are still useful for de-termining stellar photometry, as a) only a smallregion of the full image is affected, and b) localbackground-subtraction algorithms can correct forthe increased noise in these regions with negligi-ble degradation of the resulting photometric accu-racy. Additionally, the shutter actuation systemhas been modified with a slow-release valve for thecompressed gas that extends the lifetime of theshutter blades by approximately a factor of threerelative to the original shutter system. Whilethis may not completely eliminate shutter prob-lems during future operations, it improves shutterperformance considerably and, coupled with thenear-real-time image analysis provided by the QRsystem to identify broken or stuck shutter blades,will prevent significant loss of observing time dur-ing any future PreCam observing campaigns. Theeffect of the increased shutter closing time uponthe uniformity of focal plane illumination is stillbeing quantified, but because it only increases theblade actuation time by a factor of ∼2, it is notexpected to impact PreCam exposures of 10 s orlonger.

4.3. PreCam-Specific Image Processing:Illumination Correction

Finally, minor electronics issues associated withthe CCDs and readout electronics arose during

data processing. Specifically, “dipoles” becameapparent in the partially-processed images. Theseoccur when charge is improperly separated be-tween (vertically) adjacent pixels, and one pixel isobserved with a significantly greater backgroundvalue than its immediate neighbor. These andother spurious variations are eliminated both inthe production versions of the DES hardware andin the upgraded PreCam hardware; for all PreCamdata already obtained these variations are elimi-nated by Illumination Correction processing (seeFigure 11). Illumination Correction is effectivelyan additional application of flat-field corrections,incorporating a pixel-by-pixel multiplicative fac-tor based on the mean of all target field exposuresfor each filter during a given night of observing(as opposed to the initial flat-fielding, which wasbased on master flat images constructed from ded-icated dome flat exposures taken throughout theduration of the survey). Instrument-induced vari-ation between pairs of adjacent pixels is correctedin this fashion, from ∼16% to a level consistentwith normal background variation, and any stel-lar profiles affected by dipoles are marginally im-proved (due to more accurate distribution of stel-lar photon counts). However, the improvement tothe photometry of known standard stars is negli-gible solely by removing the dipoles, because (ingeneral) the values for both pixels was already in-corporated into any stellar photometry measure-ments, since all the light that came from a star(even that portion of it translated by one pixel)was still associated with that star. However, forthe longer exposures (ipc-, zpc-, and Ypc-band im-ages), Illumination Correction removed significantadditional variations in illumination not correctedby the initial flat-fielding process, thus improvingthe final photometric accuracy of the stars foundin images in those three bands. Because the pho-tometric accuracy of stars in the gpc- and rpc-bandimages was not improved by the Illumination Cor-rection, this step was not applied to images takenin these filters.

4.4. Star Flat Corrections and CatalogGeneration

After the streaking/banding, shutter, and illu-mination variation issues are resolved, as a finalprocessing step, residual instrumental effects areremoved by means of applying a star flat correc-

13

Fig. 10.— Left: A representative PreCam image with significant streaking that would interfere with thephotometric determination of the stars in that field. Right: The same image after streaking removal,exhibiting desired behavior for both background regions and stars.

Fig. 11.— Left: A segment of a representative PreCam image before illumination correction, showingvertically-oriented light and dark spots (“dipoles”) inside the green circle. Right: The same region of theimage after Illumination Correction, showing that the dipoles have been removed.

14

tion (see Figure 12). These star flat images arecreated by comparing, over the course of a night,the magnitudes of known standard stars with theirpreviously measured magnitudes from, e.g., SDSSor other catalogs. This yields a fine-scale (of or-der 100 sub-regions per image) zero-point correc-tion that is applied to each image in addition tothe global zero-point for the entire image. Afterthis processing, a catalog is developed for eachimage using Source Extractor. The preliminaryPreCam Southern Star Catalog is derived fromthis catalog, after the application of additional se-lection criteria; specifically, the Source Extractor-determined Stellarity >0.95 and FWHM <4 pixels(based upon a 12 arcsec diameter circular aper-ture). In the next section we describe results fromthe ∼10% subsample of the PreCam stellar catalogto which we have applied this preliminary analysis.

5. Results

Preliminary measurements of target candi-date standard stars in the PreCam images arecompared to the 2MASS Point Source catalog(Skrutskie et al. 2006) for astrometry, while forphotometry they are compared to the SDSS DR7catalog (Ivezic et al. 2007) as well as the Equa-torial and Southern extensions of the u′g′r′i′z′

catalogs (Smith et al. 2013). For the astrome-try, we use a set of python scripts to perform a“first pass” fit, which is good to ∼2 arcseconds(RMS). We are currently refining these scriptsand, with proper treatment of the field curvatureand using independent astrometric solutions forthe two PreCam CCDs, we have achieved astro-metric accuracy good to ∼0.3 arcseconds (RMS),or about 1

5 of a pixel. This is sufficient to provideunique identification of any stars that will be in-cluded in the final version of the PreCam StandardStar Catalog. For the photometry, existing stan-dard stars are matched if the PreCam-determinedposition was within 3 arcseconds of the knownposition, and then nightly zero-point offsets andairmass-dependent color corrections are applied tothe matched stars prior to determining the finalphotometric accuracy. A representative sampleof the results taken from ∼10% of the observa-tions, or six nights out of the total of 51, showsthat PreCam measures stellar photometry witha mean accuracy of 3-5% (depending on filter)relative to SDSS results, and better than that

(∼2%) for Southern u′g′r′i′z′ stars brighter thanmagnitude 15 (see Figures 13 and 14). For thefinal PreCam catalog with improved data analy-sis and averaging of multiple observations of eachfield, our photometry is expected to achieve therequired 2% accuracy (Allam et al. 2013).

In addition to the production of a prelimi-nary standard star catalog based upon previousstandards, PreCam provides a preview of thescience that will be conducted with the DES.SN2010lr, a Type Ia supernova discovered by theCatalina Real-Time Transient Survey, or CRTS(Drake et al. 2011), was observed by PreCam aswell, and we show a preliminary light curve of theSN in Figure 15. The observed magnitude near 18is in agreement with photometric measurementsfrom the CRTS. Spectroscopic measurements by(Prieto et al. 2011) lead these authors to claimthat this SN is most similar to SN1998bu withpeak magnitude occurring around December 20(Day 15 of the figure). Finally, it is worth not-ing that SN2010lr was seen (after the fact) inPreCam data from 2010-12-05, a date well beforethat of the initial CRTS discovery image. Thisinitial detection was made at a magnitude signif-icantly below that for which the PreCam surveyis designed, thus the standard analysis proceduresdescribed above were not applied; instead, afterdata processing a single cut using Source Extrac-tor’s DETECT THRESH of 1.5 was made to generatethe object catalogs that include the supernova.Therefore, we expect the photometric accuracyof this SN measurement to be significantly worsethan for any of the standard stars in our catalog;nevertheless, the clear detection of the SN in thePreCam data is an encouraging sign of what willbe accomplished with the significantly larger andmore sensitive Dark Energy Survey.

6. Discussion

PreCam, a precursor observational campaignfor calibration of the Dark Energy Survey, wasdesigned to provide accurate photometric and as-trometric measurements of stars in a sparse gridthroughout the DES footprint, while the PreCaminstrument also served as a testbed for variouscomponents of the Dark Energy Camera. We suc-cessfully completed observations of roughly halfof the desired target fields comprising the Pre-

15

Fig. 12.— A representative star flat image derived from known standard star observations for a given filteron a given night. When applied in a manner similar to the standard flat-field images, the star flat was shownto further correct the residual variations in photometric accuracy.

Cam survey grid by early 2011, and preliminaryresults based on the analysis of a subset of thedata show that we are approaching the stringentrequirements for calibration of the DES. Duringconstruction and operations we tested the pre-production DES DAQ system, (spare) CCDs forthe DECam, ObsTac and the QR system, as wellas various other components. As described above,many different issues were overcome during thecommissioning and observations of PreCam, in-cluding a secondary mirror of insufficient opticalquality, banding and streaking within the data,shutter failures, and data transport issues. Nev-ertheless, we were able to collect data during 51out of 64 possible nights (∼80% duty cycle), andthe vast majority of those data (well in excessof 90%) provides us with useful measurements ofthousands of previously-measured standard starsand up to a hundred thousand potential new stan-dard stars to a magnitude limit if ipc∼17. All ofthese will be incorporated into the forthcomingPreCam Standard Star Catalog and used as inputsto the DECam calibration procedures. Additionalobservations with the PreCam instrument to com-plete the proposed grid of new standard stars, us-ing the Curtis-Schmidt or another telescope, areunder consideration but are not approved at thistime.

7. Conclusions

In an effort to provide a catalog of standardstars for improved calibration of the Dark EnergySurvey, we designed, built, tested, installed, com-missioned, and operated the PreCam instrumentto collect observations over 51 days in 2010 and2011. We performed PreCam survey observationsin a sparse grid throughout the Dark Energy Sur-vey footprint; photometric accuracy as comparedto the SDSS catalog is between 3.0% and 5.0% forthe sample of data described in this paper, and 2%as compared to the brighter stars of the u′g′r′i′z′

standard star catalog extensions. Final analysis ofthese data should allow us to reach the 2% pho-tometric accuracy for nearly a hundred thousandstars suitable for DES standard star calibrations,and perhaps even obtain the desired 1% accuracy.The full PreCam Southern Hemisphere StandardStar Catalog will be released after complete analy-sis of the full PreCam dataset (Allam et al. 2013),but application of the standard star calibrationsto the DES data will begin with the first light ofDES, currently scheduled for September 2012.

The submitted manuscript has been created byUChicago Argonne, LLC, Operator of ArgonneNational Laboratory (“Argonne”). Argonne, aU.S. Department of Energy Office of Science labo-ratory, is operated under Contract No. DE-AC02-06CH11357. The U.S. Government retains for it-

16

Fig. 13.— Distribution of photometric accuracy of a representative set of stars observed in gpc-, rpc-, ipc-,and zpc-band images relative to SDSS measurements. Nearly all stars measured here are between 14th and17th magnitude. The results show that the fit to the preliminary single-epoch accuracy has a sigma between3%and 5% for these four filters. The final processing and analysis algorithm based on the averaging ofmultiple images of each field is expected to improve these results to between 1% and 2%.

17

Fig. 14.— Distribution of photometric accuracy of a representative set of stars observed in gpc-, rpc-, ipc-,and zpc-band images relative to the Equatorial and Southern extensions of the u′g′r′i′z′ standard star catalogobserved with the U.S. Naval Observatory (USNO) 40 inch telescope. Nearly all stars measured here arebetween 10th and 15th magnitude. The results show that the fit to the preliminary single-epoch accuracyhas a sigma between 2% and 3.5% for these four filters. The final processing and analysis algorithm basedon the averaging of multiple images of each field is expected to improve these results to between 1% and 2%.

18

Fig. 15.— PreCam (pc) and Catalina Real-Time Transient Survey (CRTS) lightcurves of supernovaSN2010lr, showing pre-detection limits and post-detection magnitudes. Uncertainties in the PreCam mea-surements are estimated by taking the difference between the Source Extractor-determined MAG AUTOand MAG APER (with an aperture of 10 arcsec). The presumed start around December 5 is inferred fromthe later observations of the CRTS, and the lines extending to the lower edge of the plot on or before thisdate are pre-detection (5σ) upper limits set by PreCam. As described in the text, the first detection useda distinct analysis procedure with only a 1.5σ detection threshold, thus we expect it to be significantly lessaccurate than any of the photometry results comprising our standard star catalog.

19

self, and others acting on its behalf, a paid-upnonexclusive, irrevocable worldwide license in saidarticle to reproduce, prepare derivative works, dis-tribute copies to the public, and perform publiclyand display publicly, by or on behalf of the Gov-ernment.

This paper has gone through internal reviewby the DES collaboration. Funding for the DESProjects has been provided by the U.S. Depart-ment of Energy, the U.S. National Science Foun-dation, the Ministry of Science and Educationof Spain, the Science and Technology FacilitiesCouncil of the United Kingdom, the Higher Edu-cation Funding Council for England, the NationalCenter for Supercomputing Applications at theUniversity of Illinois at Urbana-Champaign, theKavli Institute of Cosmological Physics at theUniversity of Chicago, Financiadora de Estudos eProjetos, Fundacao Carlos Chagas Filho de Am-paro a Pesquisa do Estado do Rio de Janeiro,Conselho Nacional de Desenvolvimento Cientıficoe Tecnologico (CNPq - Brazil) and the Min-isterio da Ciencia, Tecnologia, e Inovacao (MCTI -Brazil), the Deutsche Forschungsgemeinschaft andthe Collaborating Institutions in the Dark EnergySurvey.

The Collaborating Institutions are Argonne Na-tional Laboratory, the University of California atSanta Cruz, the University of Cambridge, Centrode Investigaciones Energeticas, Medioambientalesy Tecnologicas-Madrid, the University of Chicago,University College London, DES-Brazil, Fermi-lab, the University of Edinburgh, the Universityof Illinois at Urbana-Champaign, the Institut deCiencies de l’Espai (IEEC/CSIC), the Institutde Fisica d’Altes Energies, the Lawrence Berke-ley National Laboratory, the Ludwig-MaximiliansUniversitat and the associated Excellence Clus-ter Universe, the University of Michigan, the Na-tional Optical Astronomy Observatory, the Uni-versity of Nottingham, the Ohio State Univer-sity, the University of Pennsylvania, the Univer-sity of Portsmouth, SLAC, Stanford University,the University of Sussex, and Texas A&M Uni-versity. Fermilab is operated by Fermi ResearchAlliance, LLC under Contract No. DE-AC02-07CH11359 with the United States Department ofEnergy. SLAC is operated by Stanford Universityunder contract No. DE-AC02-76SF00515.

The authors recognize the efforts of the Fermi-

lab engineers who were instrumental in designing,producing, and testing pre-production DES detec-tor hardware for the PreCam instrument, partic-ularly Theresa Shaw and Walter Stuermer. Welikewise thank former members of the DES-Brazilgroup, Leandro Martelli and Bruno Rossetto, fortheir support of the PreCam efforts. The au-thors are also grateful for the significant contri-butions of the professional staff of Cerro TololoInteramerican Observatory, including Marco Bon-ati, Gale Brehmer, Jorge Briones, Oscar Saa, andthe TELOPS staff, especially during installation,commissioning, and operation of the PreCam in-strument.

The U-M Curtis-Schmidt telescope is dedicatedto optical observations of space debris, in a pro-gram funded by grants to the University of Michi-gan from the NASA Orbital Debris Program Of-fice. P. Seitzer (Principal Investigator) thanks theOffice for their long term and continuing support.In particular the Debris Program funded upgradesto the Curtis-Schmidt which made an automatedsurvey project like PreCam possible.

This paper makes use of data from the SloanDigital Sky Survey. Funding for the SDSSand SDSS-II has been provided by the AlfredP. Sloan Foundation, the Participating Insti-tutions, the National Science Foundation, theU.S. Department of Energy, the National Aero-nautics and Space Administration, the JapaneseMonbukagakusho, the Max Planck Society, andthe Higher Education Funding Council for Eng-land. The SDSS Web Site ishttp://www.sdss.org/[www.sdss.org].

The SDSS is managed by the AstrophysicalResearch Consortium for the Participating Insti-tutions. The Participating Institutions are theAmerican Museum of Natural History, Astro-physical Institute Potsdam, University of Basel,University of Cambridge, Case Western ReserveUniversity, University of Chicago, Drexel Univer-sity, Fermilab, the Institute for Advanced Study,the Japan Participation Group, Johns HopkinsUniversity, the Joint Institute for Nuclear As-trophysics, the Kavli Institute for Particle As-trophysics and Cosmology, the Korean Scien-tist Group, the Chinese Academy of Sciences(LAMOST), Los Alamos National Laboratory,the Max-Planck-Institute for Astronomy (MPIA),the Max-Planck-Institute for Astrophysics (MPA),

20

New Mexico State University, Ohio State Uni-versity, University of Pittsburgh, University ofPortsmouth, Princeton University, the UnitedStates Naval Observatory, and the University ofWashington.

Finally, this publication makes use of dataproducts from the Two Micron All Sky Survey,which is a joint project of the University of Mas-sachusetts and the Infrared Processing and Anal-ysis Center/California Institute of Technology,funded by the National Aeronautics and SpaceAdministration and the National Science Founda-tion.

REFERENCES

Abbott, T., et al., available online athttp://www.noao.edu/meetings/decam/media/DECam Technical specifications.pdf

Albrecht, A., et al., Report of the Dark En-ergy Task Force; available online at arXiv:astro-ph/0609591v1

Allam, S. (in prep.)

Allam, S., et al. (in prep.)

Annis, J., et al., White Paper Submitted to theDark Energy Task Force; available online atarXiv:astro-ph/0510195 (2005)

For more information regarding the DES, seehttp://www.darkenergysurvey.org

Annis, J. and E. Neilsen (in prep.)

Cardiel-Sas, L., et al., Proc SPIE 7014 (2008)70146P. See also Castilla, J. et al., Proc SPIE7735 (2010) 7735-311

Bernstein, J.P. et al., Astrophysical Journal 753(2012) 152

Bertin, E. and S. Arnouts, Astronomy & Astro-physics Supplement 317 (1996) 393

Drake, A., et al. Central Bureau for ElectronicTelegrams No. 2614, 1 January 2011

For further information on CRTS, see alsoDrake et al., Astrophysical Journal 696 (2008)870

Flaugher, B., et al. Proc SPIE 8446-35 (2012) (inpress)

Fukugita, M., et al., Astron. J. 111 (1996) 1748

Honscheid, K., et al., Proceedings of SPIE 7019-11(2008)

Ivezic, Z., et al., Astron. J 134 (2007) 973

McLeod, B.A., et al., Proc. SPIE 3355 (1998) 477-486

Pickles, A. J., PASP 110, (199) 8638

Prieto, J., et al., Central Bureau for ElectronicTelegrams No. 2620, 4 January 2011

Rheault,J. P., et al., Proc SPIE 8446-94 (2012) (inpress)

Schlegel, D.J., D.P. Finkbeiner, & M. Davis, As-trophysical Journal, 500, 525 (20 June 1998)

Shaw, T., et al., Proc. SPIE 8453-100 (2012) (inpress), and references therein

Skrutskie, M. F., et al., Astron.J. 131:1163 (2006)

Smith, J. A., et al., Astron.J. 123:2121-2144(2002)

Smith, J. A., et al. (inprep.). See also http://www-star.fnal.gov/Southern ugriz/index.html

This 2-column preprint was prepared with the AAS LATEXmacros v5.2.

21