10 November 2004
Survey ..................... DPS
Telescope .................. ESO/MPG 2.2m
Instrument ................. WFI
Program IDs ................ 169.A-0725; 67.A-0244(A); 164.O-0561
Origin ..................... ESO/EIS
Number of regions .......... 3
Region ..................... DEEP1; DEEP2; DEEP3
Number of Fields ........... 11
Passbands .................. U (2); B; V; R; I (2)
Number of Filters .......... 7
EIS Release Number ......... 23
Version .................... 0.9
Total Data Volume .......... 25.55 Gb
Release Date ............... November 2004
Release prepared by ........ EIS/PSG team, L. F. Olsen
Product Type ............... Final Stacked Images
Number of Stacked Images ... 40
Data Volume ................ 25.55 Gb
Origin ..................... ESO/EIS
Number of regions........... 3
Region ..................... DEEP1; DEEP2; DEEP3
Number of Fields ........... 11
Passbands .................. U (2); B; V; R; I (2)
EIS Release Number.......... 20
Release Date................ 27 October 2004
PRODUCTS IN PREVIOUS RELEASE
Product Type................ Nightly Products
Number of Reduced Images.... 287
Based on ideas submitted by the ESO community and evaluated by ESO's Survey Working Group (SWG), the DPS survey, administered by the ESO Imaging Survey (EIS) and Public Survey Group (PSG) teams, comprises three optical and infrared strategies. The optical part consists of a deep survey covering three regions of 1 square degree on the sky, in U-, B-, V-, R- and I-passbands, using the wide-field imager (WFI) mounted on the ESO/MPG 2.2m telescope at La Silla. Each of the three DEEP regions -- 1, 2, 3 -- is covered by four WFI pointings (in five passbands) -- a, b, c, d -- per region. The regions were selected both to enable observations year-round and because they overlapped with regions of other scientific interest. For instance, DEEP1 was chosen to complement the deep ATESP radio survey carried out with the Australia Telescope Compact Array (ATCA) covering the region surveyed by the ESO Slice Project, while DEEP2 included the CDF-S field. Finally, DEEP3, was chosen in the northern galactic hemisphere, thus providing an almost year-round coverage (see the original proposals for more details). The location and the characteristics of the surveyed regions as well as the planned limiting magnitudes in each passband can be found in the DPS strategy page. One should be aware that in an attempt to improve the performance of the survey, new U and I filters were purchased and used during the course of the observations (see below).
The data contributing to the present release were obtained as part of the ESO Large Programme: 164.O-0561, carried out in visitor mode, and its service mode continuation (30 hours) 169.A-0725 (Principal Investigator: J. Krautter, as chair of the SWG), complemented by V- and R-band observations from the following ESO programme (Principal Investigator: P. Schneider): 67.A-0244, covering the DEEP-1 region.
The present release consists of final stacked images for 11 DPS fields including the data from the above programs accumulated until September 28, 2002. The image stacks were created from the 287 reduced images released in EIS release number 20. The 287 reduced images were converted into 42 stacked images of which 40 are being released.
Of the 11 fields being released seven (1.75 square degrees) are covered in at least 4 passbands, of which 5 in 5 passbands (UBVRI). For these cases, the integration times approach those originally intended, yielding limiting magnitudes close to those requested in the original proposal. It is worth mentioning that the field DEEP-2c have more data than those being presently released due to the contribution of other programs (e.g. GOODS, COMBO-17). A complete release of these is beyond the scope of the present release, which deals with DPS survey data alone.
The accuracy of the astrometric calibration is estimated to be better than 0.2 arcsec. The photometric zeropoints of the final stacks rely on the zeropoints of the reduced images for which the accuracy was estimated to range from 0.05 to 0.1 mag, except for the U-band for which the errors are of the order of 0.2 mag. It is likely that these errors may be reduced in the future when all the effects impacting on the error budget are better understood. The quality of the photometric zeropoints of the final stacked images is estimated from the scatter of the computed magnitude differences between the final stacks and the contributing photometric frames. In general, the mean offsets are within few hundredths of a magnitude with a scatter of about 0.1 mag.
For more information about the terminology and conventions used in this document refer to the WEB README pages.
This is the fourth official (one infrared and 3 optical, see EIS Survey Release page) release of reduced data for DPS. The data set being released was accumulated in visitor mode during ESO observing periods 64 and 69.
The present release consists of 40 fully calibrated ESO/MPG 2.2m WFI stacked images in U- (10), B- (6), V- (8), R- (9) and I- (7) passbands for the 11 fields observed. The 40 stacked images were created from the 287 reduced images released in October 2004 in EIS release number 20 and more details about them can be found in the related documentation.
The EIS survey infrastructure is still under development, and since the last release a number of new features have been added, such as
Official EIS products can be retrieved via two alternate routes, both originating at the EIS home page. These procedures are described elsewhere (see Retrieving EIS Products). It is worth reiterating that to request data users have to be registered with the ESO Science Archive.
In the case of the present release the "data release information section" is followed by a section detailing the "contents" of the release, listing:
For each survey product the EIS system prepares an extensive ``product'' log from which the process log and the configuration file used can be easily accessed and inspected. This infrastructure provides the means of ensuring the uniqueness of the ingested products. It also provides the information required to understand differences between versions of the same product.
Even though, the product logs are still not in their final form, the current version of the product logs is included in this release to help the users to select suitable products for their specific projects.
Typically, the product logs comprise six distinct sections represented in the rendering of the HTML. All sections have a ``Product Identification'' sub-section which, among other things, identifies the user that created the product, the type and main attributes like passband and exposure time. These sections consist of:
Even though not fully completed, the logs available in the present release serve to illustrate the type of information the system will be able to provide survey users. In fact, it is envisioned that the logs would be included as part of the images, possibly as an extension.
For wide-field instruments, image properties are computed within a cutout selected near the center of the image. This is done, in order to speed up the process. Unfortunately, the computation of properties such as seeing and limiting magnitude occasionally lead to unreasonable values which are then reported in the product logs. In these cases, the attributes are re-computed during quality control. However, the present infrastructure does not yet allow the product logs to be updated. In the present release there are four such cases, for which the limiting magnitude failed during the un-supervised reduction and a value of -99 is reported in the log.
For the time being, only a preliminary implementation of this infrastructure is available.
In the upper right corner of the data release WEB page there are two buttons (completeness and properties) which link to plots showing the completeness of the dataset being released relative to the original strategy. It also presents different visualizations of the attributes associated with the images being released (for details see Data Quality Assessment section below). A more detailed description is provided below.
Since the infrastructure is still under development, currently plots are being produced without adequate description. It is foreseen that these plots will be embedded into HTML files providing captions and statistics.
Since WFI was first offered as a common user instrument at the ESO/MPG 2.2m telescope in 1999, the available filter set and naming conventions have evolved somewhat. To avoid confusions it is instructive to state explicitly the filter set used when conducting the present survey. The current La Silla Science Operations naming convention assigns all filters which are permanently available at the WFI an ESO identity number. If the filter does not have a special name (U, B, V etc.) then the assigned name is "CWL/FWHM" where CWL is the truncated central wavelength (nm) and FWHM is the truncated full-width at half-maximum (nm). If the filter has a special name, the FWHM of the filter (in nm) is appended to the special name. A full description of the filter names can be found at the WFI Filter Characteristics page of the La Silla Science Operations.
The present survey makes use of the following WFI filters:
|Passband||ESO Filter Name||ESO Identity Number|
Due to the on-going nature of the survey, the passband range, depth and completeness of the products in the present release is not homogeneous. In fact, the completeness of each field may vary, as some fields have not yet been completely observed. This is shown in the completeness plot accessible from the release WEB page. This figure shows for each field two histograms of time versus passband, the latter represented by the ESO filter identity number. The hatched histogram represents the total integration time required by the original strategy (in seconds). The full colored histograms represent the sum of the integration time (TOT-EXPT keyword in the header) of all stacked images in a particular filter. The color code adopted for the filters (passbands) here and elsewhere is as follows: U (#877; blue); U (#841 magenta); B (black); V (green); R (red), I (#879; magenta); and I (#845; yellow). Note that due to the lack of prior information about the new U-filter (#877) and the strong desire to reach the proposed limiting magnitude in U-band, observations taken in less than ideal conditions were repeated also leading to total integration times exceeding those of the original proposal.
As explained elsewhere (see EIS Definitions and Conventions) the conventions adopted by EIS to identify a pointing both internally and externally (e.g. OB name, object name, target name, field name) and their graphical representations on the EIS WEB pages have evolved considerably as new procedures were introduced, making it impossible to strictly maintain consistency.
Field names can normally be found in the OBJECT keyword of the FITS header. Since they are, normally, inherited from ``Observation Blocks'' (OBs) which have been prepared by multiple team members, the naming convention may not be homogeneous, and thus it is recommended that the value of the OBJECT keyword should not be, in general, used to rename files.
In the particular case of this survey one should be warned that the contents of the OBJECT keyword do not necessarily correspond to the name convention listed above. For instance, for the DEEP-2c pointing one also finds object names given by AXAF-FX-YY where X represents a passband and YY is the OB identifier.
This release consists of ``stacked'' images, thus for each field and filter there is only one image combining all contributing ``reduced'' images.
A first release of optical data for this survey (stacked images for DEEP2c) was carried out 7 March 2001 and described in detail in Arnouts et al. (2001). Since then several improvements have been introduced in the code and are briefly mentioned below.
The stacked images were produced from the nightly reduced images available through release number 20. These images were astrometrically and photometrically calibrated as described in ``that release''. Here the procedures applied to the reduced images for creating the stacks are described.
Each field is the co-addition of a number of reduced images grouped in a number of "Stack Blocks" (SB). The SBs were constructed based on the pointing, instrument, and filter of the reduced images. Further consistency checks include the astrometric reference grid as detailed below.
The production of the stacked images consists of the following steps:
Inspection of the images after stacking indicated that in one case the final stacked image was significantly degraded by the inclusion of an image (graded B) with high noise amplitude. Therefore, this image was not included in the production of the released stack. The reason for this problem is being investigated and may lead to the inclusion of additional constraints for the automatic rejection algorithm.
The astrometry for the SBs completely rely on the astrometric calibration of the reduced images. As described above, in the creation of the SBs a number of conditions are applied to ensure consistent astrometric calibrations for all contributing images.
The astrometric conditions for including an image into an SB is that the distance between the center of the reduced image relative to the others is smaller than 0.25 times the field-of-view. Furthermore, the adopted reference grid (reference coordinate, pixel scale, orientation, and type of projection) has to be the same for all the contributing images.
The WCS of the stacked images is reported in the image header using the CD-matrix notation. The projection adopted is TAN and the orientation is north up and east to the left.
The reduced images were calibrated to the Vega magnitude system based on observations of Landolt (1992) standard stars as described in release number 20. Based on these photometric calibrations the photometric zeropoints of the stacked images were derived, as described below.
Stacks with one photometric frame
For SBs where only one of the input images were taken in a photometric night, this photometrically calibrated image is used as a reference. Firstly, the flux in this reference image is scaled to the flux level that would be obtained at zero airmass based on the extinction found in the photometric solution. Then the other contributing images are scaled to that same flux level corresponding to that at zero airmass. The scaling factor is obtained by comparing the object magnitudes. After this scaling all the scaled images are averaged as described above. The zeropoint of the resulting stack is the zeropoint at zero airmass of the photometric reference image.
Stacks with more than one photometric frame
For SBs with more than one photometric frame a reference image is created by first scaling all the photometric frames to the flux level at zero airmass based on the extinction for their respective photometric solutions. Then, by computing the weighted average of all of these scaled images. All input images (photometric and non-photometric) are scaled to the same flux level as the reference image. These scaled images are averaged (weighted) to yield the stacked image. The zeropoint of this image is the weighted average of the zeropoints at zero airmass of the photometric input images.
Stacks with no photometric frames
Occasionally, none of the reduced images in an SB are photometric, possibly for different reasons (no standards in the night, observations in non-photometric nights). In such cases, all images are scaled to the flux level of an arbitrarily chosen reference image within the stack. The zeropoint of the final stacked image is then the zeropoint of the adopted reference image. For these cases, the convention adopted is same as that of the non-photometric reduced images.
Image Product Calibration
According to the above procedures the zeropoints of each stack was determined and the image specific zeropoint was added to the header of the images. Therefore, the Vega magnitude is given by
mag (Vega) = -2.5*log(flux) + ZP
where the flux is the number of counts directly measurable on the images (note that the stacked images are normalized to 1 sec).
In the case of a non-photometric night, a default value for the zeropoint is adopted and the error in zeropoint is set to -1. For nights without observation of fields containing standard stars, a default zeropoint and an error of -2 are assigned to the image. Finally, during quality assessment of the data, calibrated images with zeropoints that deviate significantly from a reference value have the zeropoint in the header changed to a default value and its error set to -3. Note that when a default value is assigned all images will have the same zeropoint in the header regardless of the airmass at which they were observed.
In the case of stacks, the actual value adopted for the default zeropoint of the reduced images is only relevant in cases where no photometric frames are available for a given SB. The final photometric accuracy, however, will depend on the type of solutions and number of independent nights contributing to the final stack.
The table below shows for each pointing the best type of solution available for each passband.
|#||Region||EIS Field Name||default||1-par||2-par||3-par|
|1||DEEP1||DEEP1a||U B||U R I|
|2||DEEP1||DEEP1b||B||U V R I|
|5||DEEP2||DEEP2b||B||U V R I|
|6||DEEP2||DEEP2c||U V I||U|
|8||DEEP3||DEEP3a||U U R I||B V|
|9||DEEP3||DEEP3b||V||R||U B I|
|10||DEEP3||DEEP3c||R||U B V I|
More details concerning the calibration of the reduced images contributing to the products being presently released can be found in the readme of release number 20
The reduced images still showed a number of cosmic ray hits, because the construction of reduction blocks (RB) was optimized for removing cosmic ray features in the final stacks. This priority in most cases led to the creation of 3 RBs per observed field and filter. Therefore, in most cases the stack blocks (SB) consist of at least 3 input frames, allowing for the use of a sigma-clipping procedure to remove cosmic ray hits from the final stacked image. The chosen sigma clipping is adapted to the number of input images, such that for SBs with less than 5 images a one-sigma clipping is used; SBs with 5 to 10 images use a two-sigma clipping; and SBs with more than 10 images use a three-sigma clipping.
The performance of the automatic satellite track masking algorithm (using the Hough transform) has been proven to be efficient in removing both bright and faint tracks. The frequency for the appearance of these tracks is about 0.9 satellite-tracks/hour, of which only 1/10 are of the bright type. The most extreme case is 3 satellite tracks of varying intensity in a single exposure. The regions affected by satellite tracks in the original images were flagged in the weight images and thus are properly removed from the stacked image. However, in the regions where a satellite track was found in one of the contributing images the noise is slightly higher, as will also be reflected in the weight image.
Before being released the images were examined by eye and graded by the EIS team, with the grade range being from A (best) to D (worst). This grade refers only to the visual aspect of the data (e.g. background, cosmetics). In the future the grade will also include a larger set of quality control parameters (QCP) and will be available for each survey product via the corresponding product log.
Out of the 42 stacked images covering the selected DPS fields, 30 were graded A, 6 B, 4 C, and 2 D, the latter not being released. In addition to the grade a comment may be associated with the visual appearance of the image. The table below presents all cases where comments were made. The table, ordered by increasing grade, lists: in column (1) entry number; in column (2) the EIS field name; in column (3) the passband; in column (4) the grade given by the visual inspection; and in column (5) the associated comment.
|#||Region||EIS Field Name||Passband||Grade||Comment|
|2||DEEP3||DEEP3d||B#842||D||single 300 sec image|
|6||DEEP3||DEEP3c||I#845||C||fringing - Poor background subtraction near bright stars|
|7||DEEP1||DEEP1a||U#841||B||visible electronic noise|
|8||DEEP1||DEEP1b||U#877||B||visible electronic noise|
|9||DEEP2||DEEP2c||U#841||B||visible electronic noise|
|10||DEEP3||DEEP3a||U#841||B||visible electronic noise|
|11||DEEP3||DEEP3a||U#877||B||visible electronic noise|
|12||DEEP3||DEEP3b||I#845||B||fringing near bright star in lower left (SE)|
|16||DEEP1||DEEP1b||I#845||A||low level fringing|
|22||DEEP2||DEEP2a||R#844||A||poor background subtraction around the bright star on the lower left corner|
|24||DEEP2||DEEP2b||V#843||A||residual traces of the inter-chip gaps visible|
|30||DEEP3||DEEP3a||R#844||A||stray light reflections in the upper and lower left corners|
|31||DEEP3||DEEP3a||V#843||A||stray light reflections in the upper and lower left corners|
|32||DEEP3||DEEP3a||B#842||A||stray light reflections in the upper and lower left corners|
|37||DEEP3||DEEP3c||R#844||A||poor background subtraction near bright stars|
|41||DEEP3||DEEP3d||I#879||A||low level fringing - stray light reflection on the lower right corner|
|42||DEEP3||DEEP3d||V#843||A||stray light reflection on the lower right corner|
One of the images was discarded because it was a single B-band of DEEP-3d image of 5 minutes integration time, and the other an I-band image of DEEP-3a produced by a single RB, comprising 6 exposures, with strong fringing. It is also worth noting that all images graded C are I-band, fro which the procedure of de-fringing was less than ideal.
The table below shows for each filter the grade breakdown. The table gives: in column (1) the passband; in column (2) the ESO filter number; in column (3) the number of images; in columns (4)-(7) the ratio of images in a given grade to the total number of images taken with that filter.
The table shows that most of the I-band are graded C, in contrast to those of the XMM survey, indicating the importance of the observing strategy in properly removing the fringing. Also note that, in contrast to the reduced images, the U-bands are mostly graded A.
|Passband||Filter||Number of products||A||B||C||D|
Another way to display the grade information is to show the breakdown by filter for images of a given grade, as shown in the following table. The table lists: in column (1) the grade; in column (2) the total number of images with a given grade; in columns (3)-(8) the fraction of images of a given grade and filter relative to the total number of images with that grade.
|Grade||Number of Images||U#877||U#841||B#842||V#843||R#844||I#879||I#845|
In the release of the reduced images the quality of the photometric solutions was assessed. Here the main concern is the accuracy of the zeropoints assigned to the stacked images. Since the zeropoints of the contributing photometric frames is well documented (see EIS release number 20), the comparison of object magnitudes between the contributing images and the ones measured in the stack is used to characterize the quality of the photometric calibration. For each stack the average and scatter of the magnitude offsets between the photometric frames and the final stack are computed. The distributions of these quantities show that the magnitude offsets are in general small and the scatter is of the order of 0.1 mag and comparable to the estimated accuracy of the individual zeropoints.
An important feature of the EIS survey system is that it attempts to provide a backbone infrastructure to enable the user to have a global view of the dataset being released both in terms of the completeness of the survey as well as the quality of the survey products. This is achieved by the production of plots accessible through the release WEB page and the quantitative information (QCP) available in future product logs, so as to make the selection of products with the desired attributes possible. This infrastructure will be complemented by a proper script that will allow the user to select products based on configurable parameters.
The attributes of the released images are summarized in the following figures, available from the WEB release page:
Comparison with other authors
The reduced data for the XMM survey (in the form of stacks) were compared to results obtained using the GaBoDS pipeline (Schirmer et al. 2003) and Erben et al. (2004) developed by the Bonn group. Extensive and repeated comparisons were made between SExtractor-produced catalogs from the images generated by these two independent systems. Initial discrepancies, resulting from the different techniques used (e.g. cosmic ray removal, gain-harmonization) were resolved, leading at the end to results in excellent agreement.
Anticipated future EIS/DPS releases include:
This release is the seventh of 2004, and number 23 since March 1998 (of which two were software releases). It is the second release of final stacked products created using the framework of the EIS Survey System which provides the tools required for the administration of the data including versioning capabilities.
With this relase one can also inspect the number of data and software requests made to the EIS team since May 2001. This can be found at the EIS Survey Release page Unfortunately, information prior to this date is scanty and more difficult to report even though a summary was presented to the STC (see DOCUMENTS pages of the EIS WEB)
With this release EIS completes the release of data from the DPS/WFI survey, pending revisions. This will be complemented by the SOFI infrared data.
Arnouts, S. et al. , 2001, A&A, 179, 436
Erben, T. et al. , 2004, in preparation
Koch, A., Grebel, E. K., Odenkirchen, M., Caldwell, J. A. R., 2004, Astronomische Nachrichten, 325, 299
Landolt, A. U., 1992 Astronomical Journal 104, 340
Manfroid, J., Selman, F., Jones, H., 2001 The Messenger 104, 16
Schirmer, M., et al. , 2003, A&A, 207, 869
Vandame, B. et al. , 2004, in preparation
Vandame, B., 2004, PhD thesis, in preparation