README for TNO Search Field Deep Images G. Bernstein 12/22/03 INTRODUCTION The "TNO Search Field" images are the sidereally summed images of a search for trans-Neptunian objects (TNOs) using the Wide Field Camera of the HST/ACS instrument. The observations were taken under Cycle 11 program GO-9433, "The Size Distribution of Kuiper Belt Bodies," with G. Bernstein as PI. There is no proprietary period for these data. The TNO search consisted of approximately 96x400 s exposures in the F606W filter for each of six contiguous ACS fields of view. The exposures were spread over a 15-day period in Jan-Feb 2003, with a complex sequence optimized for the discovery of the slowly-moving TNOs. The TNO detection process requires the subtraction of all sidereally fixed sources (stars and galaxies) from each image, followed by summation and point-source detection along all candidate TNO orbits. The images available herein are the sidereal-image templates that were subtracted from the search-field exposures. These were produced by averaging, with sidereal registration, all of the available exposure time from the program. FIELD CHARACTERISTICS The TNO search field is approximately 6'x10', centered near 14h 07M 53.3s -11d 21' 38" (J2000). The CCD axes lie close to the local ecliptic cardinal directions. The six principal pointings of the HST are labelled A-F, aligned roughly as follows: --------------- | | | | E | F | ^ | | | | | | | N -------------- Ecliptic W-> | | | | C | D | | | | | | | -------------- | | | | A | B | | | | | | | -------------- The pointings are selected so that there are no gaps between the six fields. At each pointing, a pseudo-random dither pattern was executed to improve the sampling characteristics of the final image. The dither amplitude was not, however, broad enough to fill in the gap between the two CCDS on the WFC, so there are small, nearly vertical stripes running down each column for which there is no data. The combined image files are produced on a grid that is free of distortion in the tangent-projected plane of the sky, and oriented to the local ecliptic axes. Because of the severe distortions in the WFC, the outer edges of the mosaic are somewhat ragged, and the files contained herein circumscribe the valid image areas. Areas with no valid image data are filled with zero in the SCI extension and have zero weight in the INVVAR image. CONTENT OF THE FILES There are seven FITS files in this distribution. The first is h_tnoALL50_img.fits mosaic of the entire survey area, on 0.050" pixels (8460 x 12265 pixels, roughly 1.24 GB) and there are six more finely pixellized images: h_tnoA25_img.fits subregion "A" of the mosaic, on 0.025" pixels h_tnoB25_img.fits subregion "B" of the mosaic, on 0.025" pixels h_tnoC25_img.fits subregion "C" of the mosaic, on 0.025" pixels h_tnoD25_img.fits subregion "D" of the mosaic, on 0.025" pixels h_tnoE25_img.fits subregion "E" of the mosaic, on 0.025" pixels h_tnoF25_img.fits subregion "F" of the mosaic, on 0.025" pixels (each ~8500 x 8500 pixels, ~830 MB). The six subareas overlap by 50-100 pixels at their edges. Each file is a multi-extension FITS file with a null primary extension, and three image extensions: # name type content 1 SCI float science data (flux, counts per 0.050" pixel per second) 2 INVVAR float inverse variance of flux (a.k.a. weight) 3 COUNT int number of exposures contributing to the image The header information in the files is very restricted. The world coordinate system is specified and is highly accurate (see ASTROMETRY section) because the images are combined on to a distortion-free grid. PRODUCTION OF THE IMAGES The individual exposures were extracted from the HST archive, with the standard on-the-fly bias and flat-field corrections. Catalogs of objects in each individual exposure are produced with the SExtractor code (Bertin & Arnouts, 1996), giving centroids for all objects. The pixel-coordinate centroids were mapped to the local tangent plane using WFC/F606W distortion maps provided by J. Anderson. The HST orientation information from the image headers was used to make an initial map to ICRS coordinates for each exposure, which is sufficiently accurate to permit matching of identical objects that appear in multiple exposures. We then re-register all 584 constituent exposures to a common astrometric frame by allowing every exposure a free translation and linear transformation (6 degrees of freedom per image) to the common frame, and adjusting these degrees of freedom to minimize the scatter in centroids of the objects matched between exposures. The entire mosaic is registered to the ICRS frame by the inclusion, in the minimization, of a set of stars with astrometric positions provided by N. Zacharias, from observations using the UCAC astrograph and further images provided by J. A. Smith from the CTIO 0.9-meter imager. Also included in the fit are objects from a series of 10-second ACS exposures of the TNO search fields, to bridge the dynamic-range gap between the astrometric standards and the HST objects. Once the astrometric solutions for all exposures are obtained, the image-combination algorithm is a relatively simple variant of the usual Drizzle process. First a sky level is determined for each CCD readout quadrant, and subtracted from the source images. An updated list of hot columns was produced by examining the data, and these have their weight set to zero on the source images. Then a grid of destination pixels is set up in the tangent-plane projection about the nominal field center, with 0.050" (0.025") pixels, as desired. For each of the destination pixels, we execute the following steps: 1) Find all pixels in all individual (source) exposures for which the pixel center is within +-0.025" (+-0.013") of the target pixel center in both x and y directions. 2) Extract measured flux values and uncertainties for all the contributing source pixels that have non-zero weight. 3) Delete outlying values (e.g. cosmic rays) with sigma-clipping algorithm. 4) If the number of valid source images is below chosen threshold, place zero values in the SCI and INVVAR images. 5) Create weighted mean of remaining valid source pixels, and place the mean, inverse-variance of mean, and number of contributing images into the output image. This is similar to executing the Drizzle algorithm with zero "drop size." There is no attempt to apportion the source-pixel flux between destination pixels, we just view each output pixel as the average of all the surface-brightness measurements that were made within the source pixel's extent. The virtue of this method is that unneeded PSF smoothing is avoided, and there is little or no correlation between adjacent pixels of the output image. For the 0.025"-scale images in particular, the PSF and noise properties of the output images should be very well behaved. Of course the noise per pixel in the 0.025" images is higher than in the 0.050" images because each has only ~1/4 as many contributing source images. But this reflects the true nature of finer sampling. PHOTOMETRY The units of the SCI images are, nominally, electrons per second per 0.050"-square pixel. We pass on the photometric zeropoints provided for WFC/F606W by the on-the-fly processing software. No attempt has been made to refine these. These result in a magnitude zero point (in the ST system for F606W filter) of 26.67 for 1 ADU in the 0.050" image. The zeropoints for the 0.025" images differ by a factor 4 (in flux), because we have placed the source pixels onto destination pixels that are only 1/4 the solid angle of the nominal source pixel, so they have MAGZERO=28.18. ASTROMETRY Because the mosaic has been registered to UCAC-derived stellar positions, we expect the absolute accuracy to be ~10 mas. Orbit-fitting residuals suggest additional astrometric uncertainties of order ~5 mas, which is similar to the uncertainties in the WFC distortion corrections derived by Anderson. The astrometric information is encoded into the FITS headers. CREDIT AND FURTHER INFORMATION At present the only published reference material on the TNO search field is the paper providing the scientific results of the program: Bernstein et al, astro-ph/0308467 (submitted to AJ). Use of the TNO search field images should cite this paper for now. Questions about the images may be referred to garyb_at_physics.upenn.edu. At present we have not even produced catalogs of these fields, so anyone who does this or produces useful information that should be added to this site should contact the PI.