The Great Observatories Origins Deep Survey (GOODS)
Spitzer Legacy Data Products
First Data Release

1.0 General information

2.0 Observations

2.1 Fields and program IDs
2.2 IRAC observing strategy and AORs


3.0 Data reduction

3.1 SSC pipeline processing
3.2 Frame-level post-BCD processing
3.3 Astrometry and image registration
3.4 Image combination
3.5 Exposure and weight map scaling
3.6 Flag maps
3.7 Differences between GOODS-S (v0.21) and GOODS-N (v0.30) processing


4.0 Data products

4.1 File names
4.2 World coordinate system
4.3 Science images
4.4 Exposure maps
4.5 Weight (inverse variance) maps
4.6 Flag maps
4.7 FITS headers


Tables:

Table 1: Parameters for GOODS IRAC superdeep epoch 1 data
Table 2a: Chronological AOR summary, IRAC GOODS-S epoch 1
Table 2b: Chronological AOR summary, IRAC GOODS-N epoch 1
Table 3: Flux and magnitude conversion factors
Table 4: GOODS IRAC flag map values


Other information:

Complete list of FITS images in the first data product release
Example of GOODS data product FITS header

Figures:

Figure 1: GOODS-S IRAC superdeep epoch 1 field layout
Figure 2: GOODS-N IRAC superdeep epoch 1 field layout

1.0 General Information

This document describes the first release of data products from the Great Observatories Origins Deep Survey (GOODS) Spitzer Space Telescope Legacy Science program.

This first data release (DR1) consists of "best-effort" reductions of data taken with the Infrared Array Camera (IRAC, Fazio et al. 2004) on-board Spitzer. These are images from the first epoch (out of two) of the "superdeep" IRAC observations for each of the two GOODS fields. The superdeep program is sometimes also described as the "deep" program in GOODS literature, and is to be distinguished from the "ultradeep" program that covers a small portion of the GOODS-N field.

The data products are described in detail below, and consist of mosaiced, co-aligned images in all four IRAC channels on both fields, plus associated exposure, weight and flag maps.

The GOODS team is writing a paper which will describe the observations and data (Dickinson et al., in preparation). Please reference this paper when using these data products in published research.

2.0 Observations

Here, we provide a brief description of the IRAC superdeep epoch 1 observations.

2.1 Fields and program IDs

The GOODS Spitzer Legacy program observations cover two fields on the sky. One of these fields (GOODS-N) coincides with the historical Hubble Deep Field North (Williams et al. 1996), while the other (GOODS-S) coincides with the Chandra Deep Field South (Giacconi et al. 2001). These fields have extensive observations at virtually every wavelength accessible from major space- and ground-based observatories, including deep, multicolor data from the Advanced Camera for Surveys (ACS) on the Hubble Space Telescope (Giavalisco et al. 2004).

Generically, the GOODS fields consist of sky regions that are approximately 10 x 16.5 arcmin on the sky. The orientations of these fields were originally chosen to match scheduling constraints for Spitzer observing. Coordinates, position angles, and other important parameters for the epoch 1 IRAC observations of the GOODS fields are summarized in Table 1.

The Spitzer GOODS observations are divided into two separate observing programs, one for each field: program 169 for GOODS-N, and program 194 for GOODS-S. These program numbers cover both the IRAC and MIPS portions of the GOODS observing program.

2.2 IRAC observing strategy and AORs

The IRAC superdeep observations for each field were broken into a series of Astronomical Observation Requests (AORs), each several hours long, that were designed to enable efficient scheduling. 39 AORs were used for the CDF-S IRAC epoch 1 superdeep observations. The HDF-N IRAC epoch 1 AORs were designed to be somewhat longer each to improve scheduling efficiency, resulting in 33 AORs in total.

The AORs for the GOODS IRAC superdeep observations have two targets (one per field), designated "HDF-N-CXO" and "CDF-S". (The "CXO" designation for HDF-N indicates that the pointing was optimized for overlap with the Chandra X-ray Observatory 2 Msec data on the HDF-N field.) The pointing centers for the two fields are given in Table 1, along with the mean position angle (east of north) of the long axis of the field averaged over the duration of the IRAC epoch 1 observations.

Observations were executed over the course of several days, during IRAC campaign 4 for GOODS-S, and campaign 8 for GOODS-N. The start and end times for the epoch 1 observations are summarized in Table 1; Table 2 gives a detailed list of the AOR labels, AORKEY identifying numbers, and start dates/times for each AOR. The telescope orientation rotates during this time, by approximately 1 degree per day, and hence so does the coverage of the GOODS field. The field rotates around the pointing centers given in Table 1. This rotation blurs the exposure map somewhat, but has useful consequences for the data reduction. Some IRAC array effects are oriented along rows or columns (e.g., "muxbleed" or "column pulldown" effects - see below and the IRAC Data Handbook), and the rotation makes it easier to ameliorate or remove some of these effects over the course of the observing program. The rotation range for each field is given in Table 1.

Table 1 - Parameters for GOODS IRAC superdeep epoch 1 data



Parameter	GOODS-S	GOODS-N
Spitzer program ID	194	169
Target name	CDF-S	HDF-N-CXO
RA (J2000)	03:32:30.37	12:36:54.87
Dec (J2000)	-27:48:16.8	+62:14:19.2
Mean position angle	-14 deg	+48 deg
Range of PAs	5.5 deg	5.7 deg
Start date/time	2004-02-08 12:32:15	2004-05-19 14:20:52
End date/time	2004-02-16 05:15:13	2004-05-26 11:20:14

Table 2 - Chronological AOR summary, IRAC superdeep observations



Within each AOR, the telescope was moved through a mapping pattern so that each IRAC channel covers a mosaic of 2x2 pointings, with total extent approximately 10 arcmin on a side (modulo dithering). IRAC observes simultaneously in all four channels, with channels 1 and 3 (3.6 and 5.8 microns) covering one pointing on the sky, and channels 2 and 4 (4.5 and 8.0 microns) covering another pointing. The two IRAC fields of view are separated by about 6.7 arcmin in the focal plane, and the long axes of the GOODS fields are oriented along the direction separating the two IRAC fields of view. The consequence of our 2x2 mapping pattern is that, in a given observing epoch, the area covered by channels 1+3 and that covered by channels 2+4 have a small region of overlap (about 3 arcmin, modulo "softening" by the dithering pattern). The GOODS-S 4-channel overlap region includes the Hubble Ultradeep Field, while the GOODS-N overlap region partially includes the historical WFPC2/NICMOS Hubble Deep Field North. Figure 1 illustrates the layout of the CDF-S epoch 1 IRAC observations, while Figure 2 shows the same for HDF-N.

After approximately 6 months, when the telescope orientation has rotated by 180 degrees, the fields are observed in a second epoch, when the pointings for the two IRAC fields are swapped relative to the first epoch. In this way, after two epochs, each GOODS field will have complete coverage in all four IRAC channels, with an overlap strip in the middle receiving twice the exposure time of the rest of the field. The exposure time per channel per sky pointing is approximately 23 hours per epoch, and will be double that in the overlap strip.

Figure 1: GOODS-S IRAC superdeep epoch 1 field layout



Figure 2: GOODS-N IRAC superdeep epoch 1 field layout



The observations used a frame time of 200 seconds per exposure in channels 1-3; channel 4 takes 4 x 50s frames in that interval. Within each pointing in the 2x2 mapping pattern, the telescope was dithered through a medium-scale cycling Gaussian random pattern, described in the Spitzer Observer's Manual section 6.2.3.4.1. The maximum extent of this dither pattern is approximately one half the IRAC field of view, i.e., about 2.5 arcmin. Most CDF-S epoch 1 AORs used 11 dithers per map pointing, while most HDF-N epoch 1 AORs used 13 dithers, resulting in 44 and 52 independent sky pointings respectively over the whole map per AOR. For a few AORs, the number of dithers was different, varying the AOR duration in order to provide flexibility for telescope scheduling. The starting dither index for each AOR was incremented by the number of dithers per map pointing, so that each AOR used a different set of dither positions. Because GOODS uses more than the 311 positions that are available in the Gaussian random pattern, the starting index rolls over with a stagger to avoid duplication of AORs' dither sets.

Here, we briefly describe the reduction of the GOODS IRAC superdeep data and the construction of the data products.

3.1 SSC pipeline processing

The GOODS team started reductions of the data using products generated by the Spitzer Science Center (SSC) Basic Calibrated Data (BCD) pipeline. The CDF-S IRAC epoch 1 BCD data that were initially delivered to the GOODS team were processed with pipeline version S9.1.0, and the GOODS team data products are based on that version. Those data were later re-processed by the SSC before they were placed in the public archive; the archived version was processed with pipeline version S10.5.0, but this was not used for the GOODS v0.21 data products. The HDF-N IRAC data used for the GOODS data products are, instead, identical to the version available through the SSC archive, and were processed by pipeline version S10.0.3. Significant differences between the GOODS-S and GOODS-N processing are discussed further in section 3.7.

A very small number of frames were not correctly processed by the SSC pipeline, and thus no BCD products were available. We hope to recover those frames for future data products when the IRAC data are reprocessed through the S11 pipeline later in 2004.

3.2 Frame-level post-BCD processing

In post-processing of the individual BCD frames, we applied the following steps:

HDR-mode images: Each AOR begins with two short exposures observed in "HDR-mode", with exposure times of 0.4 and 10.4 seconds. These were discarded in all subsequent analysis.

Unit conversion: BCD products were converted from units of MJy/sr back to DN by dividing by the conversion constant given in the FLUXCONV keyword in the BCD image headers, and multiplying by the exposure time given in the EXPTIME header keyword.

Median image subtraction: Within each AOR, a median sky image was constructed from the dithered exposures (typically 44 frames per AOR for the CDF-S, and 52 frames per AOR for the HDF-N). This image was then subtracted from each exposure within the AOR, in order to remove (presumably) additive effects from (1) small residual structure/gradients in the background level, that we presume result, in part, from additive bias effects and/or possibly illumination issues relative to the sky flats used for flatfielding the data, and, (2) long-term persistent signal due to bright sources observed prior to the GOODS AOR (see the IRAC Data Handbook section 4.5 for a discussion). Note that the persistence signal can gradually change with time over the duration of an AOR. In our current, best-effort reductions, no further attempts were made to remove the time-varying component of image persistence. To a large degree, this will be eliminated, or its impact will at least be diluted, by the combination (with outlier rejection) of many dithered exposures. This can be improved in future versions of the GOODS data products. However, we note that unremoved time-variable persistence effects generally have quite low amplitude, and should not significantly limit the data quality of the current generation of GOODS data products.

Column pull-down correction: When a star or cosmic ray in IRAC images for channels 1 and 2 produces a count level of approximately 35,000 DN, there is an additive offset effect on pixels in the column or columns where this occurs, known as "column pull-down" (see the IRAC Data Handbook, Section 4.11). The GOODS team has developed an empirical algorithm for detecting this effect, measuring its amplitude, and applying an additive correction to the affected columns. This was applied to all images.

Note that there are also column- and row- effects in channels 3 and 4 that have smaller amplitudes and somewhat different characteristics (probably the same as the "banding" described in the IRAC Data Handbook, section 4.9). We have not yet attempted to make any such corrections in channels 3 and 4. As a result, some elevated signal is seen along the row direction from bright sources, most noticeably in some of the GOODS channel 4 data (e.g., the "jets" that appear to radiate from the brightest, low redshift galaxies in the CDF-S IRAC channel 4 images). We hope to correct these in future data product releases.

Muxbleed: Another bright source effect is "muxbleed", described in sections 3.2.2 and 4.8 of the IRAC Data Handbook. Muxbleed only affects channels 1 and 2, and is strongest in channel 1. Although the BCD pipeline has a muxbleed correction module, it does not currently do a good job of removing this effect. We attempted to improve this situation by applying a post-facto correction based on an improved algorithm to the data in channel 1 (only), and by taking advantage of the fact that the Spitzer position angle rotates with time over the course of the observations for a given field. When the correction algorithm predicted a residual muxbleed signal amplitude greater 10 DN, we simply masked out the effective pixels, weighting them to zero when combining the dithered data (see below). Because of the field rotation, different pixels relative to the source position are affected in different AORs, and the masked muxbleed trail should be eliminated from the final stack as it rotates off itself from one AOR to another. For pixels where the muxbleed amplitude falls below the specified threshold, the (small) correction was applied without further consideration.

In practice, even this improved correction scheme failed to remove a lot of the muxbleed signal in the GOODS data. The SSC and the GOODS team are now investigating improved muxbleed corrections, and we expect to apply these to future releases of GOODS data products. The GOODS data products released here include flag maps (described below) that indicate the approximate locations of the strongest residual muxbleed features in the images.

Background subtraction After the processing steps described above, a robust modal sky estimator was applied to each BCD image, and the net sky level was subtracted (as a constant) from each image.

3.3 Astrometry and image registration

Next, we derived an internally consistent astrometric solution for each IRAC image. The procedure was executed for each AOR independently. Sources were detected in each IRAC image within an AOR, using the pointing-refined world coordinate solution generated by the BCD pipeline. The pixel positions of these sources were cross-matched between IRAC frames, and were also matched to external catalogs of objects in the GOODS fields, using a catalog limited in the K-band from GOODS ground-based data, and with positions determined to a high degree of accuracy from the GOODS HST/ACS image mosaics. A global astrometric solution for all frames in each AOR was then derived, allowing for image translations, rotations, and geometric distortion described by a cubic polynomial equation.

3.4 Image combination

Images within each AOR were combined into a mosaic using procedures akin to the "multidrizzle" method used for HST/ACS GOODS data processing. The core routines for projecting pixels from the detector plane to an output image plane use the "drizzle" method of Fruchter & Hook (2002).

Using the astrometric solutions derived above, images in each AOR were initially projected to a common tangent plane on the sky to form a data cube, using a drizzle scale=0.7 and pixfrac=1.0 (see Fruchter & Hook 2002 for an explanation of the drizzle parameters). At each sky pixel, a noise model was computed that accounts for the photon statistics from the sky plus source signals and the detector readout noise and dark current, as well as a local gradient term that boosts the effective local variance in regions (e.g., near the centers of bright sources) to account for the possibility of small registration uncertainties or PSF variations from frame to frame. These parameters were carefully tuned by experimentation to optimize the rejection of outlier pixels, e.g., cosmic rays and detector defects, while minimizing over-rejection at the positions of sources. However, we still experience a modest degree of over-rejection at the cores of very bright objects (peak values greater than 20000 DN) in channels 1 and 2 in the GOODS-S v0.21 images. (Further refinements in the processing for the GOODS-N v0.30 data have reduced the degree of over-rejection.) The pixel values from each image were compared to the median for all images at that sky position. The differences between these values and the median were compared to the noise model, and outliers beyond a specified threshold were masked, so that they could be assigned zero weight in the final image combination. Pixels that fall near the outskirts of the field covered by a given AOR, where only a few frames contribute to the sum, are automatically excluded because the outlier rejection becomes ineffective when the redundancy is small.

After the outlier pixels have been identified, the images are then re-drizzled to form a final mosaic image per AOR, with the outliers identified above assigned zero weight in the combination. The final mosaics were drizzled onto an output grid of pixels with a uniform size of 0.600 arcsec/pixel, oriented on the sky according to the world coordinate standards defined for all GOODS data products (discussed in more detail below). The drizzling was done using the "point kernel", which ensures that each input detector pixel per image contributes only to a single output pixel in the drizzled mosaic. In this way, the noise values in adjacent pixels in the GOODS IRAC data products are uncorrelated with one another. This is different from most other drizzled data products with which users may be familiar, e.g., the Hubble Deep Field WFPC2 and NICMOS data sets, or the GOODS ACS mosaics. Those data products used finite drizzling kernels, resulting in significant noise correlation between adjacent pixels. The very large number of dithered exposures for the GOODS IRAC data sets allow us to take advantage of the point kernel, which ensures uncorrelated pixels (simplifying data analysis), and which maximizes the net angular resolution of the final drizzled images.

Finally, the drizzled stacks per AOR were combined together with exposure time weighting to produce the final GOODS mosaics.

3.5 Exposure and weight map scaling

The output weight maps from the drizzling process used to construct the science mosaics have units of exposure time multiplied by the ratio of the drizzled pixel solid angle (exactly 0.36 square arcsec) to the original detector pixel solid angle (approximately 1.49 square arcsec). The exposure maps that are provided with this data release have been rescaled to remove this pixel solid angle ratio, so that they will represent the actual IRAC integration time at each point on the sky. Because of geometric distortion in the IRAC images, the original detector pixel solid angle varies slightly over the field of view, so a constant rescaling factor is not strictly correct. However, this difference is only a few percent at most over the field of view, and has been neglected here.

The IRAC images are essentially background limited, and therefore we would expect the RMS shot noise per pixel should be inversely proportional to the square root of the exposure time. We have conducted tests using split data sets to verify that this is indeed the case to a good degree of accuracy. We have therefore constructed noise maps (provided here as inverse variance images) by applying an empirical, constant multiplicative scaling factor to the exposure map images. We compared image values from subsets of the data at each sky pixel position, and measured the statistics of their variance. Point kernel drizzling was used in order to preserve the uncorrelated pixel statistics. Regions around sources were excluded in order to avoid the shot noise contribution from sources, as well as enhanced variance due to small image misalignments and PSF under-sampling issues. The resulting noise measurements were used to compute the rescaling constant to normalize the weight maps to inverse variance.

It is important to note that the inverse variance maps represent only the shot noise component of the image noise at the sky background level of the images, i.e., the sky noise per pixel that would be present if there were no astronomical sources in the image. They do not include the Poisson shot noise from the sources themselves, nor any measure of photometric uncertainty due to image crowding or "confusion noise".

3.6 Flag maps

We have constructed flag maps that may be useful when making object catalogs and analyzing the GOODS IRAC images. These maps identify regions where there are and are not data in a given channel, where the exposure time is low (i.e., around the edges), and (for channels 1 and 2) where there is significant residual muxbleed in the images that may affect source photometry or introduce spurious source detections. The regions flagged for muxbleed were defined by visual inspection, and should be considered indicative only.

The flag images are bit maps, i.e., integers that represent the sum of bit values, each of which indicates a different flag conditions. The flag maps are described in more detail in section 4.6, and the flag values are given in Table 4.

Note that regions with Flag = 2 (i.e. < 20% of the typical exposure time) still have integration times up to 4.6 hours - hardly shallow by Spitzer/IRAC standards! However, those regions will have a fairly steep gradient in their exposure time and local noise amplitude.

We caution that the lower exposure regions of the images can suffer from other cosmetic defects. In particular, a change in the processing methodology for the GOODS-N v0.30 images compared to that used for GOODS-S v0.21 led to some substantial improvements in astrometry and alignment, but also to a mild degradation in cosmic ray rejection in low-weight regions of the image. Therefore, some residual cosmic rays appear in the GOODS-N images in the areas with flag values = 1 and 2, and users should be aware of this when analyzing the images. There are very few residual cosmic rays in the GOODS-S data.

3.7 Differences between GOODS-S (v0.21) and GOODS-N (v0.30) processing

There are several subtle differences between the v0.21 processing of the GOODS-S images, and the v0.30 processing used for the GOODS-N data. For most users, these differences will be unimportant and can be neglected. However, we summarize some of the main differences here.

• SSC pipeline versions: As described in section 3.1, the GOODS-S v0.21 reductions used BCD data processed by SSC pipeline version S9.1.0, while the GOODS-N v0.30 reductions used BCD data from S10.0.3. Differences between the SSC pipeline versions are logged at the SSC IRAC pipeline changes web page. Perhaps the most significant difference between the S9.1.0 and S10.0.3 pipelines is the implementation of automatic correction for scattered zodiacal light (see the IRAC Data Handbook, section 4.7). Stray zodiacal light affects all IRAC images as an additive background term with predictable spatial structure. Prior to pipeline version S9.5, this stray light pattern was (incorrectly) removed from on-orbit observations by multiplicative correction when the images were divided by dark sky flat fields. In images processed this way (e.g., the GOODS-S S9.1.0 data), there will be photometric errors for sources that fall in regions affected by this stray light. Subsequent pipeline versions, such as S10.0.3 which was used for the GOODS-N data, remove the stray light by subtraction from both the flat fields and the science images, resulting in photometrically correct images. The extensive dithering used for the GOODS observations should substantially reduce the amplitude of this photometric effect when many frames are averaged together to form the mosaic stacks. Although we do not believe that this effect should pose a significant photometric problem for the GOODS-S data, we have not quantified its amplitude. Future re-reductions of the GOODS-S data will start with BCD images processed through later versions of the SSC pipeline, and will thus eliminate this problem.

• Astrometric alignment: The GOODS-N v0.30 processing uses an improved methodology for astrometry and image combination compared to that employed for GOODS-S v0.21. As a result, the GOODS-N IRAC data have smaller position-dependent systematic residuals in their alignment to the world coordinate frame defined by GOODS ACS and ground-based data.

• Cosmic ray rejection: As a consequence of software changes between v0.21 and v0.30 processing, the rejection of cosmic rays in regions with low exposure weighting is somewhat poorer for GOODS-N than for GOODS-S. Some unremoved cosmic rays are still present in low weight regions, although the rejection is very good in regions with the full exposure time. Regions with lower exposure time can easily be identified and screened by using the flag maps.

• Muxbleed and pulldown masking: Essentially the same algorithms were used for correcting column pull-down and muxbleed (see section 3.2) in the channel 1 and 2 data for both fields. However, for the GOODS-N v0.30 processing, we more aggressively masked affected pixels during the image combination process. Since more pixels were masked in GOODS-N than GOODS-S data, the net retained exposure times for channels 1 and 2 are somewhat smaller in GOODS-N than in GOODS-S.

• Noise calculation and weight map scaling: As described in section 3.5, we directly measured the image noise at the background level, and then scaled the weight maps to represent the absolute inverse variance of the background. The exact procedure used to do this was different for the GOODS-S and GOODS-N data. For GOODS-S, the RMS frame-to-frame variation in the values of pixels contributing to a given sky pixel was measured (per AOR), and the uncertainty in the combined image was assumed to be RMS/sqrt(N) where N is the number of frames combined. For GOODS-N, we instead constructed two "half data sets", each using half of the total set of images on the field, aligned to the same sky projection. We took the difference between the two half-data sets, divided by 2, and measured the RMS in this difference image. The latter approach resulted in a slightly higher background noise estimate for GOODS-N, which we believe to be more realistic since it does not assume sqrt(N) noise scaling.

4.0 Data Products

The first release of GOODS data products consists of FITS images of the IRAC superdeep data for both GOODS fields. Our understanding of IRAC instrument behavior and data processing is continuing to evolve, as are the software pipelines and the calibration of the instrument. This first release consists of "best-effort" data products available at this time, and will eventually be superseded by reprocessed versions in a future data release. The version numbers for these data products, based on GOODS internal nomenclature, are v0.21 for GOODS-S, and v0.30 for GOODS-N.

4.1 File names

File names for these GOODS data products include the following components, separated by underscores ("_"):

1) GOODS field ("n", "s")
2) Instrument (here, always "irac")
3) Channel (here, "1", "2", "3", "4")
4) Data set and epoch (here, always "s1" for "superdeep epoch 1")
5) Release version (here, "v0.21" for GOODS-S, "v0.30" for GOODS-N)
6) Image type ("sci" = science image; "exp" = exposure map; "wht" = weight, "flg" = flag map)


As an example, the GOODS-S IRAC channel 1 superdeep epoch 1 science image (version 0.21) is named "s_irac_1_s1_v0.21_sci.fits".

• Complete list of FITS images in the first data product release.

4.2 World coordinate system

All GOODS imaging data products are generated using a common scheme for world coordinates and pixel projection, which we briefly describe here. The images are projected on a tangent plane, with a the tangent point (CRVAL1,2) selected to be near the center of each field (GOODS-N, GOODS-S). They are aligned with north up (+y) and east left (-x). The pixel scales for GOODS imaging data products from different telescopes and instruments are always chosen to be integer multiples of one another. For the IRAC GOODS images, this scale is 0".600/pixel, which is approximately (but not exactly) half the native IRAC pixel scale. (Other scales for GOODS public-release data sets include 0".15/pixel for the ESO/VLT ISAAC CDF-S data, and 0".03/pixel for the HST/ACS Treasury Program images.) The pixel position (CRPIX1,2) that corresponds to the tangent point (CRVAL1,2) is always set to be a half-integer value. In this way, GOODS imaging data products from different telescopes and instruments can always be mapped to one another by simple integer rebinning, if desired.

Since the release of the GOODS HST/ACS v1.0 data products on 29 August 2003, we have found that the absolute astrometry for the GOODS-N field is slightly offset in declination from the reference frame defined by VLA 20 cm source positions from Richards (2000). This difference is approximately

dec(VLA) - dec(GOODS-N) = -0.38 arcsec
with negligible offset in right ascension. We have not yet adjusted the coordinate reference frame of the GOODS-N IRAC data products to account for this, in order to keep the Spitzer and Hubble GOODS data products on the same world coordinate system and pixel grid. We intend to adjust the coordinate reference systems for both the Spitzer and Hubble GOODS data products in future releases. There is currently no indication of any similar offset for the GOODS-South data products, although the radio data in the south are not as useful for this purpose as are those for the HDF-N.

4.3 Science images

The pixel intensities for GOODS IRAC data products are given in units of DN per second, derived from the original SSC BCD products (which have units of MJy/sr) using the FLUXCONV BCD header keyword (see section 3.2). The IRAC Data Handbook, version 1.0, Table 5.2, provides the best current determination of the flux conversion factors for each channel as derived by the SSC. We summarize this information here in Table 3, providing the flux densities in micro-Janskys and the AB magnitudes that correspond to a count rate of 1 DN/sec. This information is also recorded in the image headers in the keywords FLUXCONV and MAGZERO (see section 4.7). For reference, we also list the detector gain (electrons/DN) in each channel; note that the effective gain for the GOODS mosaics (which are normalized to DN/sec) varies with the exposure time as a function of position.

Although the IRAC point spread functions have smaller FWHM, and hence better angular resolution, than had been anticipated before launch, they nevertheless place a substantial fraction of light at large radii away from the center of a source. Careful attention to aperture corrections is therefore needed in order to properly measure source fluxes from any IRAC data set, particularly for extremely deep, crowded images like those from GOODS. In a future data release, we will provide object catalogs and discuss source photometry in detail.

Table 3 - Flux and magnitude conversion factors



Channel	Wavelength	FLUXCONV uJy/(DN/s)	MAGZERO AB for 1 DN/s	Detector gain (e/DN)
1	3.6 microns	3.922	22.416	3.3
2	4.5 microns	4.808	22.195	3.71
3	5.8 microns	20.833	20.603	3.8
4	8.0 microns	7.042	21.781	3.8

4.4 Exposure maps

The exposure maps represent the IRAC integration time in seconds at each position on the sky in the co-added image mosaics, after rejection and masking of outlier pixels (e.g., cosmic rays, pixel defects, muxbleed, etc.). Fine-scale granularity from pixel to pixel in the exposure maps is a consequence of the process of drizzling the images onto a sub-sampled pixel grid using the point kernel, as described in section 3.4.

4.5 Weight (inverse variance) maps

The weight maps represent the inverse square of the RMS pixel-to-pixel noise (in DN/s) at the background level of the images. The construction of these maps is described in section 3.5. This represent the shot noise component due to the sky background and instrument noise only, and does not include Poisson noise from sources, nor any measure of photometric uncertainty due to source crowding or confusion.

4.6 Flag maps

The flag maps identify regions of the images with and without data in a given channel, with reduced exposure time, or where there may be residual muxbleed that could affect source detection or photometry. The flag images are bit maps, i.e., integers that represent the sum of bit values, each of which indicates a different flag conditions.

Table 4 describes the flag values, where the "bit number" starts at 0, and the "flag value" is the equivalent integer value for that bit setting. Bits not described in the table are currently unused for flag settings. These bit values will often appear in combination. For example, regions with < 20% of the modal exposure time (bit 1, flag value 2) also have < 50% of the modal exposure time (but 0, flag value 1). Therefore, those pixels will have flag values of 2 + 1 = 3. Regions with no data will have flag values 64 + 2 + 1 = 67. Regions with residual muxbleed (flag = 16) and also < 50% modal exposure time (flag = 1) will have flag = 16 + 1 = 17.

Note that the regions flagged for muxbleed were defined by eye, and should not be regarded as definitive or complete, but may serve as a useful warning in regions where muxbleed could affect object catalogs. We hope to improve muxbleed removal and flagging in future data releases.

Table 4 - GOODS IRAC flag map values



Bit number	Flag value	Condition
0	0	>50% of the modal exposure time
0	1	<50% of the modal exposure time
1	2	<20% of the modal exposure time
4	16	Region with significant residual muxbleed (Ch1 & Ch2 only)
6	64	No data (zero retained exposure time)

4.7 FITS headers

The FITS headers of the GOODS data product images incorporate various useful and relevant information about the images.

Example of GOODS data product FITS header