Spitzer Frontier Field data products
v2.0 February 12, 2016
These data products are the first release of the directors discretionary time (DDT) and public archival data available for the Frontier Fields clusters as of Dec 1, 2015. The data product use an enhanced pipeline based on MOPEX/APEX that is described in brief at the end of this readme.
The naming convention is:
OBJECT.INSTRUMENT.CHANNEL. extension
Where INSTRUMENT.CHANNEL corresponds to:
- IRAC.1 IRAC 3.6um data
- IRAC.2 IRAC 4.5um data
- IRAC.3 IRAC 5.8um data
- IRAC.4 IRAC 8.0um data
- MIPS.1 MIPS 24um data
- mosaic.fits Mean mosaic. Most users will measure photometry on this product.
- unc.fits Mean combined uncertainty map which should be used for estimating errors.
- cov.fits Mean coverage map weighted by exposure time indicating the number of images that went into each pixel, used to determine gain.
- std.fits Measured standard deviation at each pixel.
- median_mosaic.fits Median combined mosaic. These images do a better job of rejecting moving objects than the median mosaic.
- median_mosaic_unc.fits Median of the input uncertanty files.
- Same files with _lin in their name for MACS0717 use linear interpolation rather than drizzling due to the low coverage in the extended region around MACS0717.
All images are in units of MJy/sr. The IRAC images have a pixel scale of 0.6" and the MIPS data of 2.45". The zero-points for the AB system are 21.5814 for IRAC and 18.5265 for MIPS.
Pipeline Improvements from V1.0
The major differences are that the background subtraction/matching is significantly better and the subtraction of bright stars to improve photometry near the cluster cores. The process of bright star subtraction will be improved in future releases, in the wider area around MACS0717 in particular.
All clusters are drizzled with a pixfrac of 0.01 in IRAC Channel 1 and 2, and a linear interpolation in Channel 3,4 where data exist. This reflects the different coverage levels in the different bands. For MACS0717 two mosaics are provided for Channels 1 and 2. The first covers only the deep area and is drizzled with a pixfrac =0.01 like the other clusters. The second set have _lin in their name and use a linear interpolation to cover the wider area where the coverage is too low for a pixfrac=0.01.
Here are the detailed improvements to the reduction vs. the previous releases.
- A `First Frame Effect' correction has been applied to remove pattern noise due to the electronic resets in IRAC.
- Bright stars have been subtracted
- The individual frames have been further corrected with a "Delta-Dark" that accounts for latent images and background differences between the master dark and the local dark.
- The backgrounds were matched by iteratively masking objects until the true median of the background was determined.
- The column pull-down correction has been improved, and when a correction was not possible masking has been improved.
- The unc images are now scaled to the correct noise level to account for the fact that the bias pedestal level is not corrected in the warm mission pipeline, leading to an over-estimate of the noise in the pipeline products.