1 / 23

Astrometry @ TERAPIX: SCAMP

Astrometry @ TERAPIX: SCAMP. E.Bertin (IAP & Obs. de Paris/LERMA). Automatic astrometric and photometric calibration with SCAMP. SCAMP and the TERAPIX software suite The CFHTLS Object centroiding Field centering The astrometric solution In practice Performance summary

newman
Download Presentation

Astrometry @ TERAPIX: SCAMP

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Astrometry @ TERAPIX:SCAMP E.Bertin (IAP & Obs. de Paris/LERMA) Astro-WISE workshop 11/2005

  2. Automatic astrometric and photometric calibration with SCAMP • SCAMP and the TERAPIX software suite • The CFHTLS • Object centroiding • Field centering • The astrometric solution • In practice • Performance summary • Forthcoming developments Astro-WISE workshop 11/2005

  3. A complete image compositing system Exposures Pixel weighting Defectix, Eye,Weightwatcher Source extraction SExtractor Astrometry Photometry SCAMP Warping and stacking SWarp Final images Astro-WISE workshop 11/2005

  4. The CFHTLS • Motivated the development of SCAMP • >20TB of science image data • Each individual MEGACAM exposure is a 1º mosaic of 36 x 9Mpixel CCD frames and contains about 104-106 detections. • 1500 sq.deg. covered in up to 5 bands • In a given region of the sky anything from 1 to 2000 exposures can overlap • Astrometric and photometric calibrations must accomodate various sky coverage strategies • Scientific goals require an accurate relative positioning of images and photometry homogeneous at the % level Astro-WISE workshop 11/2005

  5. Centroid measurements • Calibration performed with SExtractor catalogs • New centroiding procedure in SExtractor: • X_IMAGE,Y_IMAGE« isophotal » measurements are fast but their accuracy is poor • XWIN_IMAGE, YWIN_IMAGEGaussian-weighted centroiding:. • FWHM of the Gaussian set to twice the half-light radius • Improves astrometric accuracy by ~3x on detections with high S/N. • 5 iterations in average • As accurate as PSF-fitting on background-noise limited images • Works well with galaxies X_IMAGE XWIN_IMAGE XPSF_IMAGE Astro-WISE workshop 11/2005

  6. Astrometry: pattern matching • Requires a reference astrometric catalog (currently GSC, USNO, or UCAC, 2MASS or SDSS) and a first guess of field center coordinates, pixel scale and image orientation • Two steps: • Find position angle and pixel scale using pairs of detections • Find coordinate shift using detections Astro-WISE workshop 11/2005

  7. Astrometry: finding position angle and scale • Match source pairs in the reference and extracted catalog in pos.angle-log(distance) space using bandpass-filtered cross-correlation (e.g. Kaiser et al. 1999) • Must pay attention to folding on the scale axis • In general, a 7’15’ image can be correctly oriented and scaled within a 1sq.deg box. Astro-WISE workshop 11/2005

  8. Astrometry: coordinate shift • Match source pairs in the reference and extracted catalog in projected coordinate space using cross-correlation • Possible image flip • Modulation due to source clustering and catalog boundaries must be minimized with data-windowing and bandpass-filtering • Typically, a 7’15’ image can be correctly positioned within a box of a few sq.deg. • Slight improvement by adding magnitude dependency. Astro-WISE workshop 11/2005

  9. A global solution • Astrometric distortions require a 3rd order polynomial in projected coordinates ξ • 20 free parameters per CCD • Naive approach: fit the distortion coefficients for each exposure using a reference catalog (GSC, USNO,…) • Simple and fast but too sensitive to inaccuracies in the reference catalog.especially when a little more than 20 stars are cross-identified on a CCD. • Global solution: fit the distorsion coefficients by additionally minimizing the distances between overlapping detections. • For every source s on overlapping exposures a and b minimize • Manages to reach the theoretical instrumental accuracy • Implemented in many current astrometric reduction packages (e.g. Deul 1995, Kaiser et al. 1999, Radovich 2002), Astro-WISE workshop 11/2005

  10. A global solution • MEGACAM: 36x20 = 720 free parameters per exposure • Quickly leads to impractically large normal equation matrices • Iterative approach necessary • Too many free parameters: robustness problems arise because of a lack of sources or confusion in some fields • For a given instrument (and a given filter combination), one may assume that the distorsion pattern does not vary measurably over some period of time • We must allow the linear part of the distorsion pattern to vary globally from exposure to exposure because of differential atmospheric refraction and flexures • 720.ninstru + 6.(nexp- ninstru) free parameters • Requires an intermediary transformation to a common re-projection • stereographic projection chosen because it maps disks to disks. • Jacobians of the re-projections are involved Astro-WISE workshop 11/2005

  11. Astrometry: observing strategies  Astro-WISE workshop 11/2005

  12. Astrometry: improving accuracy • For large time intervals between exposures, one may leave source proper motions and parallaxes as free parameters (e.g. Eichhorn & Russell 1976) • Differential Chromatic Refraction • Atmospheric • Chromatic aberrations • Intrinsic sources of astrometric errors • Variability of the intra-pixel response profile from pixel to pixel • Mostly affect IR arrays • On modern CCDs, repeatability of centroiding with properly sampled stars is ~ 1/300th of a pixel over the array (e.g. Yano et al. 2004) • Step-and-repeat pixel size error on some large CCDs (Shaklan & Pravdo 1994): typically 0.5m (a few hundredth of a pixel) each 512 or 1024 pixel • Equivalent to a few mas • MEGACAM CCDs do not have this problem Astro-WISE workshop 11/2005

  13. Differential Chromatic Refraction • For a star with spectral index , observed at zenithal distance z in a filter of bandwidth w (in microns) centered on wavelength 0 (in arcsec): • For CFHTLS, w0.1m • At z=45 deg, ∆z varies from ~20mas (z band) to ~150mas (u band). • Most all-sky catalogs are not corrected for DCR! • Correction for differential chromatic refraction available in SCAMP • 2 different photometric instruments are required at least • Available in the merged catalog ∆ and ∆ as a function of u-g at airmass ~1.5 Astro-WISE workshop 11/2005

  14. Proper motions • Proper motion estimates • At least 2 different epochs are required • Available in the merged catalog Comparison with USNO-B1 proper motions: 20 D3 exposures in r over a period of 15 months Astro-WISE workshop 11/2005

  15. In practice • Command line: scamp *.cat • When read by SCAMP, SExtractor binary catalogs are automatically organized • By astrometric context (FITS keyword list) • By photometric context (FITS keyword list) • By group of fields on the sky (initial WCS information) • Specific handling of mosaics • An astrometric reference catalog is automatically downloaded from the CDS around each group of fields • CDSclient (F.Ochsenbein) • 2MASS, GSC, UCAC, USNO catalogs • Can be stored and retrieved locally • Output: • The solutions can be computed for most WCS projections (WCSlib by M.Calabretta) • Astrometric and photometric solutions written as “pseudo-FITS” headers • A merged catalog (with proper motions) • PlPlot “Check-plots” (PNG,JPEG,PS,…) • XML summary file with statistics • XSL style-sheet for humans Astro-WISE workshop 11/2005

  16. “Check-plots” Astro-WISE workshop 11/2005

  17. Differential Chromatic Refraction • For a star with spectral index , observed at zenithal distance z in a filter of bandwidth w (in microns) centered on wavelength 0 (in arcsec): • For CFHTLS, w0.1m • At z=45 deg, ∆z varies from ~20mas (z band) to ~150mas (u band). • Most all-sky catalogs are not corrected for DCR! • Correction for differential chromatic refraction available in SCAMP • 2 different photometric instruments are required at least • Available in the merged catalog ∆ and ∆ as a function of u-g at airmass ~1.5 Astro-WISE workshop 11/2005

  18. Proper motions • Proper motion estimates • At least 2 different epochs are required • Available in the merged catalog Comparison with USNO-B1 proper motions: 20 D3 exposures in r over a period of 15 months Astro-WISE workshop 11/2005

  19. Current Performance • Speed • Based on FFTW, BLAS/LAPack • Multi-threaded • Efficient cross-identification engine (millions of detections/s) • Full processing typically takes 4s / MEGACAM field (36 CCDs) on a quadri-Opteron@2.4 GHz • Memory footprint: 200 bytes / detection • Accuracy • Star field image simulations: chi2/d.o.f. close to 1 • MEGACAM: internal pairwise residuals down to ~6 mas rms for uncrowded, high S/N sources in selected CFHTLS exposures. 20mas rms is the more typical value • There seems to be a large-scale (~10’) distorsion component variable from exposure to exposure at the level of a few mas RMS • “Anomalous refraction”? Astro-WISE workshop 11/2005

  20. Other instruments: single array Tautenbug (Schmidt) Planet Astro-WISE workshop 11/2005

  21. Other instruments: mosaics CFH12k WIRCAM+2MASS Astro-WISE workshop 11/2005

  22. The (near) future • Public release December’05 • Does photometric calibration, too • Documentation is on the way • Upgrade to the latest version of WCSlib • Compliancy with the most recent WCS prescriptions for describing distorsion patterns • Astrometric calibration web-service • Distributed TERAPIX web-service engine (J.-C.Malapert) Astro-WISE workshop 11/2005

  23. terapix.iap.fr Astro-WISE workshop 11/2005

More Related