1. Image mosaicing.
We use cBCD images (S18.7.0), that are corrected for various artifacts
(muxbleed, muxstripe, and pulldown) mainly due to bright stars, to create
the final mosaic images with MOPEX.  For the IRAC 5.6 and 8.0 micron 
images, and for the MIPS 24 micron image, we performed a median dark 
subtraction on each AOR before mosaicing in order to minimize background 
gradients.

2. Source extraction.
2.1. IRAC
For the 4 IRAC bands, we perform source extraction and band-merging using
SExtractor with an association method. The 3.6-micron image is used
to derive the reference catalog. Sources that are detected in the 
3.6-micron image and that have any association in the other three 
bands within a 2.5 pixel radius (i.e. 1.5 arcsec) we regard as real sources.
We employ a Mexican-hat filter to improve the detection rate of confused 
sources. However, using this method can increase false detection rate as well.
Thus, in order to minimize possible false detections, the catalog only includes
aperture photometry (within a 1.9-arcsec aperture radius) for those sources 
that have a signal-to-noise ratio (snr) above 8 (2.0 and 2.7 micro-Jy at 3.6 
and 4.5 microns respectively) in either 3.6 or 4.5 microns, and a snr above 5 
(1.3 and 1.7 micro-Jy, respectively) in both 3.6 and 4.5 microns.

2.2. MIPS
We do NOT adopt the association method for extraction of sources from the 
MIPS 24-micron image because it is possible that dusty, high-redshift,
star-forming galaxies can be missed using this method. The catalog contains 
aperture photometry within an aperture of radius 7.5 arcsec. Again, we 
adopt the Mexican hat filter and apply a minimum flux-density cut (of S(24) > 
300 micro-Jy, equivalent to ~ 15 sigma) for generating the catalog.

3. Error Analysis
Initially, we use uncertainty maps generated by MOPEX to calculate flux errors
in SExtractor. However, since SExtractor does not take into account correlated 
noise, to estimate the photometric uncertainty we have corrected the 
uncertainty based on the method of Gawiser et al. (2006).
We measure variances of aperture photometries for a range of aperture radius
in ~100 blank regions, that are free of sources. Based on that experiment,
we are able to estimate the empirical uncertainty as a function of aperture
radius. On average, this method yields uncertainties on aperture photometry 
for an aperture radius of 1.9 arcsec, of 0.25, 0.35, 1.7, and 2.1 micro-Jy 
for the IRAC four bands respectively. For the MIPS 24-micron aperture 
photometry, this method yields an uncertainty of 20 micro-Jy for the adopted 
aperture radius of 7.5 arcsec.