GR6 Direct Data Release (Changes Relative to GR4/5)

The GALEX GR6 direct imaging data release are to be delivered to MAST
between April and June 2010. The grism data release will occur shortly
thereafter. The most notable changes between GR6 and GR4/5 include
changes to the bandmerged source catalog (-mcat.fits). The Project has 
removed 26 columns which are either redundant or of questionable quality. 
These include the "BEST" and "ISOCOR" fluxes and magnitudes. The Project
has also added 53 columns, mostly FUV measurements using the NUV source 
position.  Significant updates and improvements have been made to the 
calibrations used for the GR6 data reductions. These include:

* Flat Field : New maps were created using millions of source measurements, 
    effectively improving the non-Poisson flux error (on average) from 
    1.2% to 0.8% in the NUV and from 5.5% to 2.5% in the FUV.

* Flux Response Trend with Time : The NUV decrease in sensitivity has been 
    adjusted from 1.5% per year in GR4/5 to 1.3% per year in GR6. The FUV
    decrease has been adjusted from 0.2% per year to 1.0% per year.

* FUV to NUV Position Correction : The FUV photons are now corrected for 
    temperature effects with adjustments of up to 2 arcseconds relative 
    to GR4/5. The average (non-Poisson) positional accuracy (RMS) of the 
    FUV sources has been improved from 0.48 to 0.34 arcseconds.

* Distortion Corrections : Correction maps have been produced as a
    function of STIM separations. Adjustments are made to photon
    positions on the order of +/-1 arcsecond (maximum) and +/-0.15
    arceconds (on average).

* Magnitude Zero-Point Adjustment : GALEX magnitudes for the GR6 release 
    are about +0.043 magnitudes fainter in NUV and +0.033 magnitudes 
    fainter in FUV as compared to GR4/5. This is mainly due to a better 
    CALSPEC HST spectrum for the standard star LDS 749b.

-------------------------------------------------------------------------------------------------------------------
General Release 6 (GR6) CSS Addendum (07/08/2011)

comprised of 12 DIS, 389 MIS and 66 GII tiles, these new tiles added to GR6 are identified in the
listing under the column 'additionalInfo' in http://galex.stsci.edu/doc/GR67.txt as 'new tile for GR6 addendum/CSS'

The Cross-Survey Secondary tile coadds are composed of visits with a range of exposure times and positions. 
To ensure that the coadd intensity maps (*-int.fits.gz) have consistent quality across the tile, a cut is 
made at 20% of the maximum relative response for the tile. Data below that cut is not included in the 
intensity maps but is still present in the photon count and relative response files (*-cnt.fits.gz and *-rr.fits). 
New intensity maps can be generated from these files if the low exposure time data is required.  The cut 
on relative response can sometimes create uneven borders or regions with only FUV data even though there weren't 
any FUV_ONLY visits in the coadd intensity images (FITS and jpeg files).  Note that the sextractor output 
source catalogs (*-mcat.fits) are based on the intensity images.

-------------------------------------------------------------------------------------------------------------------
== Grism Pipeline Changes (Notes for the GR6 Release) ==

For GALEX grism data reduced and released after May 27, 2011, a new grism pipeline will be used which includes a 
new grism flux calibration and improved aspect solution.  The list of columns in the final products, '-xg-gsp.fits' 
and '-pri.fits', remains unchanged.

Two new files '-pht.fits' and '-rsp.fits' will be included in data releases.  These are whole field images of the 
flux (in photons) and the response (in seconds) which are rotated such that the grism dispersion is horizontal 
(along rows).  The FITS WCS header cards are set up in these images such that a RA, Dec position on the image 
represents the 'undeviated wavelength position' for a source with that RA and Dec on the direct image.   
Given the wavelength vs. offset functions and effective area curves from the GALEX documentation, it is possible 
for the user to extract spectra from these images.

The grism aspect solution has been improved.  This reduces the failure rate for the grism pipeline from 8% to 5% 
and eliminates the large (>4 arcsecond) errors that occurred in about 1% of the reductions.

A new grism flat field calibration is applied which reduces flux differences from the same object across the detector 
by about 1-2%.

New grism effective area curves were generated using the white dwarf standard star HZ 21 (same as before).  These
are very close to the previous curves, except that we have now corrected for the FUV 2nd and 3rd order overlap.  
The FUV 2nd order spectra (the primary spectra in the -xg-gsp.fits file) is now corrected only for 2nd order light.  
This means that any 3rd order contamination (which would only appear at the red edge of the FUV spectrum above 1700 Angstroms) 
would appear as additional flux (generally only noticeable for very blue objects).

More details of the GR6 grism pipeline changes will appear in a future document.
-------------------------------------------------------------------------------------------------------------------