Data Releases

In this page, you'll find:

Common Features

Overview

The DLS contains 7 "random" fields, each 2 degrees on a side, called F1 through F7. "Random" means that they were not selected to contain any particular target, but they were selected to avoid bright stars and areas with high Galactic extinction. See the coordinates page for a complete list of coordinates and extinctions. By the end of the survey, we will release the full co-added 2x2 degree field images and catalogs, but in order to release data early, we have defined a complete subset as follows.

We are observing with Mosaic, a camera with a 35' field of view, so we split each subfield into a 3x3 array of roughly Mosaic-size subfields, called p11 through p33 ("p" is for "pointing"). "Completion" means that we have reached 12,000 seconds exposure time (20x600 seconds) in B, V, and z' filters and 18,000 seconds exposure time (20x900 seconds) in R (and of course, verified the quality of the data). Because of dithering (we dither by about 3'), the edges of the subfields are not as deep. This will eventually be filled in by overlapping observations from other subfields, but for now we are releasing the central 8192x8192 which is close to full depth. The Mosaic pixel size is 0.257 arcsec, so the 8192x8192 images are 35.1 arcmin on a side, or 1230 sq. arcmin.

We observe in R when the seeing is good (0.9 arcsec or less FWHM), so for all subfields the R image has the best resolution in addition to being deeper (seeing averages ~1.2 arcsec FWHM for non-R images).

On August 2001, we had our first data release consisting of 2 sub-fields. Since then, we had 2 more data releases to the community, adding to a total of 19 sub-fields.

Images

The images have been processed through flatfielding using standard IRAF tasks (MSCRED), and registered and combined with our own software. The gory details are posted on another page, but briefly, the steps are astrometric zero-pointing, overscan/bias/trim, pupil ghost removal if KPNO data, flattening with dome flats, defringing if z' data, flattening with superflats, masking of any scattered light, sky subtraction of each image (to ensure a smooth sky in the combined image), registration, photometric offset determination, and combine by averaging with 3-sigma clip.

Each subfield has 4 FITS images, for each of the 4 filters BVRz'. F1p22 also has I data, a stack of 10x600 second exposures. The size is a convenient 8192x8192, or 35.1' on a side (0.257 arcsec per pixel). The 4 FITS files are registered, and have a World Coordinate System (for converting x,y to RA,DEC) built in. To save space and bandwidth, the images have been integerized and then gzipped. The resulting files are 50-60 MB each, versus 270 MB for an uncompressed floating point image, and little is lost by integerizing, as the sky noise is 5-10 counts anyway. The FITS header contains information such as WCS, and magnitude zero point. Click here for full documentation on FITS header contents.

Note: CTIO data (southern hemisphere, including F3, F4 and F5) is oriented with east up, north to the right when displayed with IRAF/ximtool, SAOImage, etc, while KPNO data (northern hemisphere, including F1 and F2) is east down, north to the left. Any kind of computation using the WCS will take care of this for you, of course.

JPEGs are also available. The B and the R images have been combined into a "true-color" image. They have been binned 2x2 because 8k JPEGs tend to crash browsers. It is great fun to pan around this image with your browser!

Catalogs

Catalogs are produced by SExtractor (version 2.1.6). More details about the cataloging procedure, as well as the "default.*" parameter files used are available from each release's webpage.

The DLS catalogs have a header that starts with "# fiat 1.0". All comment lines start with "#" so that the files are reasonably easy to manipulate with awk or similar tools, but we have a set of fiat tools designed specifically for typical operations on catalog data.

For ALL catalogs, ALL MAGNITUDES ARE IN THE VEGA SYSTEM. To bring the z magnitudes back to its original AB calibration add 0.54mags (z_AB = z_VEGA + 0.54).

Downloads

The downloads are in the form of full images (fits and jpg format) and catalogs. If you would like to download selected bits of images, see the image cutout page (catalog cutouts soming soon).

We also offer an interactive browser for our color images, which allows you to find the RA and DEC of any object you see.



Third Release

(December 2003)
Subfield Catalog FITS Images JPEG
(from BVR)		 JPEG
(from Rz)
F2p13 F2p13.cat.gz B , V , R , z' F2p13BVR.jpg
(~ 3.0 MB)		 F2p13Rz.jpg
(~ 2.5 MB)
F2p21 F2p21.cat.gz B , V , R , z' F2p21BVR.jpg
(~ 3.0 MB)		 F2p21Rz.jpg
(~ 2.5 MB)
F2p31 F2p31.cat.gz B , V , R , z' F2p31BVR.jpg
(~ 3.0 MB)		 F2p31Rz.jpg
(~ 2.5 MB)
F2p32 F2p32.cat.gz B , V , R , z' F2p32BVR.jpg
(~ 3.0 MB)		 F2p32Rz.jpg
(~ 2.5 MB)
F2p33 F2p33.cat.gz B , V , R , z' F2p33BVR.jpg
(~ 3.0 MB)		 F2p33Rz.jpg
(~ 2.5 MB)
F4p12 F4p12.cat.gz B , V , R , z' F4p12BVR.jpg
(~ 3.0 MB)		 F4p12Rz.jpg
(~ 2.5 MB)
F4p13 F4p13.cat.gz B , V , R , z' F4p13BVR.jpg
(~ 3.0 MB)		 F4p13Rz.jpg
(~ 2.5 MB)
F4p23 F4p23.cat.gz B , V , R , z' F4p23BVR.jpg
(~ 3.0 MB)		 F4p23Rz.jpg
(~ 2.5 MB)
F4p31 F4p31.cat.gz B , V , R , z' F4p31BVR.jpg
(~ 3.0 MB)		 F4p31Rz.jpg
(~ 2.5 MB)
F4p32 F4p32.cat.gz B , V , R , z' F4p32BVR.jpg
(~ 3.0 MB)		 F4p32Rz.jpg
(~ 2.5 MB)

Second Release

(September 2002)
Subfield Catalog FITS Images JPEG
(from BVR)		 JPEG
(from Rz)
F2p11 F2p11.cat.gz B , V , R , z' F2p11BVR.jpg
(~ 3.0 MB)		 F2p11Rz.jpg
(~ 2.5 MB)
F2p12 F2p12.cat.gz B , V , R , z' F2p12BVR.jpg
(~ 3.0 MB)		 F2p12Rz.jpg
(~ 2.5 MB)
F4p11 F4p11.cat.gz B , V , R , z' F4p11BVR.jpg
(~ 3.0 MB)		 F4p11Rz.jpg
(~ 2.5 MB)
F4p21 F4p21.cat.gz B , V , R , z' F4p21BVR.jpg
(~ 3.0 MB)		 F4p21Rz.jpg
(~ 2.5 MB)
F4p33 F4p33.cat.gz B , V , R , z' F4p33BVR.jpg
(~ 3.0 MB)		 F4p33Rz.jpg
(~ 2.5 MB)

First Release

(August 2001)
Subfield Catalog FITS Images JPEG
(from BVR)		 JPEG
(from Rz)
F1p22 F1p22.cat.gz B, V, R, I, z' F1p22BVR.jpg
(~ 3.0 MB)		 F1p22Rz.jpg
(~ 2.5 MB)
F4p22 F4p22.cat.gz B, V, R, z' F4p22BVR.jpg
(~ 3.0 MB)		 F4p22Rz.jpg
(~ 2.5 MB)



Deep Lens Survey: Data Release FAQ

What is the magnitude zeropoint of the images? For the first data release, use the MAGZERO keyword. In the second data release, this keyword was initially omitted. The images were updated promptly, but if you happen to have a copy of the images with incomplete headers, use the table of zeropoints on the data release page.


What is the GAIN of the images? Each image is a mean of 20 exposures, so the effective gain is 20 times the original gain. The headers of most images were not updated with this information and simply inherited the original gain. The original gain is nominally 2.3 for F3, F4,and F5, and 3.0 for F1 and F2, but it depends slightly on which of the 8 CCDs you are talking about. Thus you see a variation from 2.3 to 2.6 within F4p11, because the different combined images inherited gains from different CCDs for reasons we don't need to go into here. Since each part of sky has been covered by several different CCDs, you might want to take an average like 2.45 (times twenty).


What standards do you observe for photometric calibration? Do you have any estimates for the color term for stars/galaxies? We use mostly the Landolt fields SA92, SA95, SA96, SA98, SA101, SA104, and SA104. The BVR magnitudes we take from Landolt and the z' calibration we take from the Sloan listed magnitudes and then convert it (approximately) to the Vega system. Until now, we have not used color terms. We are currently defining the transformation equations between our system and the Landolt system.


Am I correct in assuming that the data are atmospheric extinction corrected? Yes in the sense that all you need to do to get calibrated magnitudes is add the zero point. These deep images are a combination of many (20) exposures and in adding them together we already take into account the airmass variance.


It appears that the catalogs have been trimmed of diffraction spikes, etc around the bright stars. At the same time, catalogs are missing real objects in crowded regions. Cleaning junk around bright stars was done in an automatic way which is pretty effective but not guaranteed. We found that requiring an ellipto errorcode of 0 gets rid of all that junk, so we filtered for that before posting to the public. There are no manually-determined exclusion zones. Descriptions of the ellipto algorithm can be found in Bernstein and Jarvis, AJ 123, 583, (2002).

However, we have since recognized that this strict criterion for measuring shapes cleanly results in too many lost objects in crowded regions. We plan to remedy this in the future by taking ellipto out of the loop for general-purpose catalogs, and instead cleaning around bright stars with exclusion zones and/or SEXtractor flags.