Data Analysis - IRAC

Overview

Here you will find links to the information required to reduce and analyze your IRAC data. Your Spitzer data can be downloaded from the archive through Leopard. We assume that you are starting with the Basic Calibrated Data (BCD) products.

User's Guides

IRAC "Pocket Guide" (pdf)

IRAC Data Handbook

Spitzer Observer's Manual (SOM) - see especially Chapter 6: IRAC.

IRAC Photometry page - detailed guide to IRAC photometry. See the bottom of this page for a brief overview.

Data Reduction Cookbooks - these pages are still under construction but may be helpful to get a feel of how to mosaic the BCDs and extract source flux densities.

Spitzer Synthetic Photometry - guide to predicting the calibrated flux densities that would be reported in the various Spitzer bands (e.g. for deriving photometric redshifts).

MOPEX Online Manual - online documentation for MOPEX - the SSC software package for photometric reduction and analysis.

IRAC Interest Group - a forum for IRAC users to discuss data analysis issues.

Contributed Software page - useful reduction software contributed by the Spitzer community. Users are advised to check this page before writing their own data reduction programs.

Software for IRAC data reduction

The SSC supported software for the reduction and analysis of imaging data (IRAC, MIPS and IRS Peak-Up imaging), specifically outlier rejection, mosaicking and source extraction, is called MOPEX (MOsaicking and Point source EXtractor). A GUI version of MOPEX has just been released, but we continue to support the command line version as well. Both versions are distributed together and are fully described on the MOPEX/APEX page. Sample namelists for IRAC data reduction can be found on the same page. Documentation can be found on the MOPEX Online Manual page, and an example walk-through of how to reduce an IRAC data set can be found on the Data Reduction and Analysis Cookbooks page.

Users should note that, while IRAC data can be reduced with any FITS data analysis tool, geometric distortion is important in Spitzer imaging, and this information is stored in the header rather than in an external distortion table.

Filenaming Convention and Pipeline Products

Basic Calibrated Data downloaded from the archive consist of many data products for each frame. Here you can find a description of all the files and explanations of the filenaming conventions, including mask bit definitions and recommended fatal bit patterns to use with your mask files. The basic files that you will use for your data reduction are your data files (*bcd.fits), uncertainty files (*bunc.fits), bad pixel mask files corresponding to each data file (*bimsk.fits; aka DCE Status Mask files) and the permanently damaged pixel mask (*pmask.fits, found in the $mopex_installation/cal/ directory; see the Calibration Files section below for more details).

Data Features/Caveats

The Data Features page summarizes the most common IRAC image features, including image artifacts and radiation hits. We describe the features, show images which have representative examples of them, and provide a recommended mitigation method for their removal from the data.

For the most up-to-date information on data features and mitigation efforts, see the IRAC Instrument page.

Calibration Files

Array-Location-Dependent Photometric Corrections for Compact Sources with Stellar Spectral Slopes

IRAC Basic Calibrated Data (BCD) are corrected for pixel-to-pixel gain variations, a process commonly known as "flat-fielding". This flat-fielding does a good job of correcting the extended zodiacal background or extended objects with a similar spectral slope, but is incorrect for objects with spectral energy distributions that are falling through the IRAC passbands (i.e. most stars and galaxies). This effect has been directly measured and the amplitude is sizeable - up to 10% peak-to-peak in some cases. This is larger than any other source of uncertainty in the IRAC calibration and should be corrected using the array-location-dependent images. See the IRAC Photometry page for details on how to apply the correction.

IRAC Pmasks

The IRAC pmask files flag quasi-static bad pixels in the arrays. They are updated every 2-6 months depending on solar activity. Please read the included "pmasks.README" file carefully and note which set of pmasks are applicable to the date when your observations were taken.

IRAC Preset PRF Files

N.B. We recommend that users be very careful when performing PRF-fitting on IRAC data. Many users may prefer to use aperture photometry in IRAC data until PRF-fitting is better understood. For uncrowded fields, aperture photometry will give better results. PRF fitting is only beneficial in crowded regions.

IRAC Pixel Solid Angle Correction Images

Required if you are performing photometry on the BCDs, or for creating the array location dependence mosaic.

Basic Data Reduction Steps

Once you have downloaded your data, a basic data reduction will include some or all of the following steps to mosaic the BCDs and extract the photometry. For a more detailed description of the IRAC data reduction steps, or for information on doing photometry on the individual BCDs, see the IRAC Photometry page.

Note that much of this information is applicable only to point sources. See the IRAC Extended Source Calibration page for rcommendations on reducing and analysing IRAC data of extended sources.

1. If you are only concerned with quick photometry, to uncertainties >15%, and you have bright sources, you can use the SSC-provided Post-BCD mosaics, and just carry out points #6, #7 and #11 in this list to extract the photometry. We do not recommend using the Post-BCD mosaics for science. Users should always create their own mosaics when extracting photometry for publication.


2. The first frame of every commanded sequence of observations should be discarded, as it may have a different bias offset to the rest of the observations. The effect is most prominently seen in Channel 3. See the IRAC Data Handbook (pdf), section 3.2.1 for more details.


3. You may want to clean up your BCDs using Sean Carey's code for IRAC artifact mitigation, distributed as Contributed Software.


4. Perform background correction of your BCDs using Overlap in MOPEX. This levels the backgrounds to ensure that the final mosaic is not "patchy". It does not affect the photometry. Overlap does not set the background to zero. If you wish to subtract the background entirely at this point, then you will have to do it yourself.


5. Mosaic your BCDs using Mosaic in MOPEX. Check the coverage mosaics to ensure that the outlier rejection has not been over/under-aggressive (see the IRAC Photometry page for more details.)


6. Perform photometry on your mosaic using either APEX (distributed as part of MOPEX) or the software package of your choice. IRAC NOTE -- USE WITH CAUTION: The SSC's MOPEX package allows PRF-fitting with a preset PRF; and it can also be used to derive PRFs from the data. We recommend that users be very careful when performing PRF-fitting on IRAC data. Many users may prefer to use aperture photometry in IRAC data until PRF-fitting is better understood. Also note that PRF Estimate should not be used with IRAC data, which is undersampled and has intrapixel variability. PRF Estimate can only be used with well-sampled data.


7. The calibration of IRAC is done with apertures of 12.2" (i.e. 10 pixels for the default pixel scale). In order to properly calibrate your photometry, you therefore need to correct from the aperture you used to extract the photometry out to the calibration aperture. See the IRAC Calibration page (or the table in Section 5.5.1 of the IRAC Data Handbook (pdf)) for standard aperture corrections. An analogous scaling factor is also required for the results of PRF fitting.


8. If you are observing sources, with spectral energy distributions that fall through the IRAC passbands (this is true for most stars and low-redshift galaxies) then you need to correct for array location dependence (also see 'Calibration Files' above for more details). This is the dominant source of uncertainty in IRAC photometry, with an amplitude up to 10%. To do this, you need to create an array location dependence mosaic (see the IRAC photometry page for more details):
   
   
   • Firstly, divide the array location correction images by the pixel solid angle correction images.
   • Secondly, create one correction image file for each BCD file, by copying the correction image into a new file and adding the header information from the BCD to the top of the file.
   • Finally, mosaic these correction images using the same mask files as for your BCD mosaic to create an array location dependent photometric correction mosaic.
   
   The correction for your sources can then be applied by multiplying your photometry by the correction value at the corresponding source position on the correction mosaic. If you are carrying out photometry on the individual BCDs then the correction can be applied by multiplying your photometry by the corresponding pixel value in the correction frame.


9. Observers should apply the color correction for their sources from Chapter 5 of the IRAC Data Handbook (pdf) This is a correction of up to 3%. This is an extra correction, in addition to the array location dependence correction. The nominal wavelengths for the IRAC channels are 3.550, 4.493, 5.731 and 7.872 microns for channels 1-4 respectively.


10. A pixel-phase correction should be applied to the measured Channel 1 flux densities (see Chapter 5 of the IRAC Data Handbook (pdf)). This corrects for the difference in sensitivity between the center and edges of each pixel and is up to a 4% effect. No correction is currently recommended for channels 2-4. Users should note that the center of a given pixel is at (0, 0), not (0.5, 0.5).


11. The units of photometry from APEX are automatically given in microJy. If you are using another package, e.g. IRAF, the fluxes will be output in the BCD units of MJy/sr. In order to correct to microJy, you should convert to steradian per arcsecond, and then multiply by the pixel area. The default pixels in the mosaics are 1.22" x 1.22". The nominal pixel widths for the BCDs are 1.221", 1.213", 1.222" and 1.220" for channels 1-4 respectively. The conversion factor for the default mosaicked pixels is therefore:
    
    1 MJy/sr = (1E12 microJy)/(4.254517E10 arcsec**2) x 1.22 arcsec x 1.22 arcsec = 34.9837 microJy.
    
    To compute magnitudes in the Vega system, magnitude = 2.5*log10(f(0)/f), where the zero-magnitudes, f(0), are listed at http://ssc.spitzer.caltech.edu/irac/calib/.