There are various methods for extracting columnar data from binary fits tables. In IDL the mrdfits astro lib function reads fits tables into structures with the appropriate tag names. fv is a useful utility for viewing fits tables and can be downloaded here.

unit response (1 cps;m_GALEX = 0)

The completeness and reliability of the GALEX catalogs are functions of the sensitivity limit set, the exposure time, and the background level. The local source density is of course an additional important factor. The exact relationships are still a topic of investigation by the GALEX Science Team. Some benchmark estimates have been established using the initial three months of survey data. Completeness and reliability have been investigated using multiple visits to the same fields, using MIS and DIS observations of AIS fields, and by using artificial subvisits created from MIS visits. Reliability has been studied using combined FUV and NUV catalogs, and by comparing GALEX with other catalogs such as SDSS.

The following plots give preliminary completeness results for exposure times of 400-1600 seconds in the relatively low background Groth DIS region. NUV is left, FUV is right. Note all GALEX magnitudes are AB.

Sensitivity vs. exposure time for low background targets (DIS, which have low diffuse galactic light and zodiacal background) is shown in the Figure below. At these background levels, imaging surveys are background limited for exposures longer than 2 ksec [NUV] and 10 ksec [FUV] respectively. Background levels may be as high as 3-5 times these, with corresponding reduction in the transitional exposure time.

Catalog reliability has been measured by comparing detected GALEX sources with SDSS DR1 sources. GALEX sources without a SDSS DR1 within 6 arcsec radius are considered spurious, and reliability is calculated as R=1-[#GALEX w/ NO SDSS]/[#GALEX]. Reliability of 90% is achieved in the AIS at m_FUV~21 and m_NUV~22. In the MIS, 90% occurs at m_FUV~23.25 and m_NUV~23.25. A small fraction in the MIS NUV sources may indeed have missed detection by SDSS, so this may be an underestimate of the MIS NUV reliability.

Magnitude errors are a determined by the Source Extraction (Sextractor) routine using simply Poisson errors expected from the number of photons in the source and background (i.e, . Errors do not include errors in the determination of the background (which is based on a large

scale smoothed backround image) or on any systematic component.

The fastest way to do this is to use ds9, the image display program freely available from http://chandra-ed.harvard.edu/install.html . Open the image of interest in ds9 (the image must have a J2000 WCS), then go to Region (on main menu), Load Regions, and find the ttttttt_vvvv[f,n]dcat.ds9reg file. GALEX detection ellipses determined by the object size and position angle will be displayed as green ellipses.

Artifacts are in general very distinguishable from real sources notably with visual inspection. A catalog of known artifacts is given here. Most artifacts are associated with bright stars and/or particular regions of the field of view and can therefore be anticipated and removed or ignored. The pipeline currently recognizes and flags many but not all artifacts. Many FUV artifacts are apparently immediately from the lack of a corresponding NUV source which should be present in almost all cases when real FUV sources are detected.

GALEX has a peak spectral response (effective area) of approximately 22 cm² in the FUV and 49 cm² in the NUV. The mean response in the FUV between 1350 and 1800 Angstroms is 13 cm². The mean response in the NUV between 1800 and 2800 Angstroms is 35 cm². Plots of the response of the most significant spectral orders in each band are shown.

The mean dispersion for the FUV in 2nd order is 1.6 Angstroms per arcsecond (range 1.2 to 1.9). The mean dispersion for the NUV in 1st order is 4.0 Angstroms per arcsecond (range 3.3 to 4.3). With a 5 arcsecond FWHM PSF, a point source would yield a FWHM resolution of approximately 8 Angstroms in FUV, and 20 Angstroms in NUV. A plot of the dispersion function for the most significant orders are shown. Note that, in direct image mode, the source would appear at approximately the position of offset=0 in grism mode.

As of 9 December 2003, ERO data products will be served by the Caltech Science operations Center and by gsfc link. By 15 January 2004, the Multimission Archive at Space Telescope (MAST) will have the catalog data on line with a query server and other tools for efficient data retrieval.

In the FUV channel, diffuse galactic light (DGL) from dust scattered starlight dominates the background. DGL varies from 300 ph cm² s¹ A¹ sr¹ (PU) to > 2000 PU, depending largely on the dust column density (and to some extent on the local stellar radiation field), which is well correlated with extinction and HI column density. Some contribution to the FUV background comes from H₂ fluorescence, HII 2

photon emission, and line emission from ionized gas in the 10^{4 -}10⁶ K interstellar medium. In the NUV channel, zodiacal light dominates the background, with a substantial contribution from DGL as well. Detector background is very low in comparison (<1%), except in local “hot spots” which are masked in pipeline processing. Nightglow from residual atmosphere at 700 km altitude produces a modest amount of background in both channels, which increases at the beginning and ending of any eclipse as the zenith angle increases (see eclipse profile here).

GALEX can only observe when it is in the earth’s shadow, or eclipse, because on the day side of the orbit atmospherically scattered sunlight and airglow would swamp and might damage the detectors (especially the NUV detector). The GALEX field of view is 1.25 degrees in diameter. Even the small amount of residual atmosphere at the 700 km GALEX orbital altitude scatters significant flux into the telescope. FUSE, (the Far Ultraviolet Spectroscopic Explorer) also has to contend with atmospherically scattered sunlight and airglow, but its field of view covers about 100,000 times less sky, so much less of the scattered light enters the spectrograph. There are other details in the way the two instruments operate that make GALEX more susceptible to atmospherically scattered sunlight background. These include their wavelengths. FUSE operates at 905-1195 Angstroms. GALEX operates at 1350-2800 Angstroms, closer to the peak of the sun’s illumination. Also, during data reduction is not possible to remove atmospheric lines from the slitless spectroscopy GALEX uses, in contrast to the slit spectroscopy of FUSE.

The ERO imaging data quality should be quite good. The satellite performance has been stable, and the pipeline has been revised twice to account for on orbit performance. We anticipate that further improvements in the pipeline may improve the PSF and flat field moderately, and may handle detector artifacts more automatically. A principle deficiency in the pipeline is the handling of the most extended sources, which tend to be shredded. Faint sources near the confusion limit will probably require refined processing to achieve the theoretical sensitivity limits. The ERO spectroscopic products have received substantially less attention, and we expect significant improvements in these in future releases.

Visit the GALEX public web site, http://www.GALEX.caltech.edu/, or the NASA Goddard Space Flight Center (GSFC) GALEX Guest Investigator program web site, http://galexgi.gsfc.nasa.gov/docs/galex/ . Once the GALEX Guest Investigator program NASA Research Announcement (NRA) is released in mid January, 2004, Goddard Space Flight center will provide a help desk. Contact the help desk by email at GALEX.helpdesk@GALEXgi.gsfc.nasa.gov, or by phone at (301) 286 3623. Before then you can call GALEX Mission Scientist Susan Neff at GSFC, (301) 286 5137. (You may also send suggestions to add to this list of Frequently Asked Questions (FAQ) to: askGALEX at srl.caltech.edu .

Where is a web site with the same (or more) information as the GALEX Guest Investigator program CD so friends or colleagues can learn about GALEX: http://GALEXgi.gsfc.nasa.gov/ .

A gnomonic projection maps a sphere onto a plane by projecting all points on the sphere radially from the sphere’s center

onto a plane that is tangent to the sphere. This projection distorts both angle and area but it has the useful property that all great circles are projected into straight lines. One can see this by noting that all great circles lie in planes containing the center of the sphere and that the projection will therefore be the line of intersection between the plane containing the great circle and the plane of the projected map. For more detail see for instance: http://mathworld.wolfram.com/GnomonicProjection.html

The GALEX spectroscopic mode employs slitless spectroscopy. This provides an image of the sky in which every object is spread out into a spectrum similar to what one sees when viewing the sky directly through a prism or transmission grating. GALEX uses a transmission grism, a ruled prism. This has the advantage of providing spectra for all the objects in the large GALEX field of view but the disadvantage of overlapping spectra in crowded fields. For this reason we take spectra of each part of the sky at different spectral position angles on the sky. This permits us to remove the confusion caused by overlapping spectra. An image strip is a two-dimensional spectrum. As in slit spectroscopy, one dimension is the spectral dimension and the other is the spatial dimension. In slitless spectroscopy however, the spectral dimension is also a spatial dimension, thus a single point on a GALEX grism-mode image represents various wavelengths depending on the source position in the sky, along the spectral-dispersion dimension.

The image strip is a portion of the (corrected) two-dimensional detector image (in grism mode) representing the locations where photons from the spectrally-dispersed source reach the detector. The image strip size is usually about 80 by 600 arcseconds (default scale is 1 arcsecond/pixel), but this size can vary depending on the source flux. The image strip is long enough to include the primary grism orders (1,2 for NUV and 2,3 for FUV) and is wide enough to include all the background used for background subtraction during spectral extraction. The source is centered on the middle row of the image in the dimension orthogonal to the dispersion. The blue end of the spectrum is to the left or lower column numbers. Multiple grism orders may be present--usually 1st&2nd for NUV, and 2nd&3rd for FUV. An image strip, as described above, contains photons from a single visit during a single eclipse, at a single grism position angle. Image strips can also contain photons from multiple visits at various positions angles, by taking either a sum or a median at each pixel in the individual image strips. The image values are scaled integers with the zero point and scale factor given in the header for each source. Negative values in the image strips indicate masked pixels which are ignored in the final spectral extraction.

The following two images show image strips from a single source (5917) from NUV (top) and FUV (bottom). The extracted spectrum is given in the bottom panel.

The GALEX data pipeline converts GALEX satellite telemetry data and any necessary corollary data into calibrated images and catalogs. The GALEX Science Operations Center (SOC) receives data from the satellite and ingestpipe unpacks it into time tagged photon lists, instrument/SC housekeeping and satellite aspect information. From these data sets, orbpipe generates images, spectra and source catalogs. An astrometric module corrects the photon positions for detector and optical distortions and determines an aspect solution using star positions from the time tagged photon data. A photometric module accumulates the photons into count and intensity maps and extracts sources from images. A spectroscopic module uses image source catalog inputs to extract spectra of these sources from the multiple slitless grism observations.

The data are initially being flat fielded using ground measurements of the system throughput, and this yields relative photometry on the order of 25% repeatability. Work is currently being done to refine the flat field based on much higher resolution flight data, and we expect this to improve the relative photometry substantially.

The GALEX detectors have a non-linear response at high count rates due to both local effects attributable to the intensifiers and global effects due to the electronics. The global effects are corrected in the pipeline, and amount to a correction as high as 40% for the highest allowable global rates (100,000 cps). The local linear range of count rates has also been tested on the ground for the two detectors. We found that for isolated stars, the FUV detector is linear to about 100 cps (m~14) and the NUV detector is linear to about 1000 cps (m~12.5). This difference is attributable to the proximity focus method of the NUV detector, which spreads the PSF out over a larger area on the intensifier surface, reducing the current density. We now have a wealth of flight data and are using it to refine the linearity calibration across each detector field of view.

Yes, the PSF will change under intense illumination as the intensifiers exhibit a phenomenon known as "gain sag" whereby the central region of bright star images will be flattened and then eventually cored out as the intensity of the star increases.

The point spread function or psf is determined by the microchannel plate detector PSF, as well as the GALEX (Ritchey Chrétien) optical design and the as built tolerance errors. The detector psf is determined by the position digitization process, which is analog and subject to random noise. The psf varies across the image due primarily to gain variations (lower gain regions having a broader psf). Other effects that affect the wings of the psf include surface roughness of the optical surfaces, ghosts from multiple reflections in refractive optical elements, and grazing reflections from baffles or struts in the optical beam path. Most optical design aberrations cause the psf to vary radially over the field of view, but those associated with the dichroic beamsplitter cause variation along the satellite X axis, which can vary in sky coordinates, depending on the satellite orientation around the telescope optical axis. Thus, in general a given source in a repeated observation of the same part of the sky will have a different psf if the satellite orientation is different around the telescope optical axis.

The main limitation on the astrometric repeatability is our knowledge of the fine grained distortion map for each detector. As fields are observed at different roll angles, errors in the distortion map, especially at the field edges, may creep into the astrometric solution. Flight data is currently being used to refine the ground data generated distortion map, and several special observations may be planned for the purpose of observing a large number of FUV stars in order to complete this task. Currently, the astrometric precision is of order 1", but there may be isolated areas at the edge of the field where the error is significantly larger.

GALEX orbits the earth every 98.6 minutes (November 2003). Approximately 1/3 of this time is spent in eclipse, defined as the sun being below the depressed horizon. The actual time available for an eclipse observation is less, and is determined by a combination of observational constraints (sun angle, zenith angle, location of SAA, moon angle) and the observation initiation sequence, which starts with a slew from solar pointing solar arrays to the final target pointing and roll angle (twist). During this time the high voltage is ramped from the intermediate value to nominal levels (which takes 2 minutes). Ramping can only start after the satellite enters the umbra.

The microchannel plate (MCP) detectors that GALEX uses have intrinsically low red leak so they reject longer wavelength light that is outside the nominal bandpass. This is important in the ultraviolet since the sky is much brighter in the visible (redward) than in the UV. To avoid red leaks, CCDs require special filters that are difficult to make and prone to pinholes. In addition, CCDs require cooling, which greatly exacerbates the difficult contamination control necessary for ultraviolet instruments. Next, MCP detectors detect and time tag each photon. This permitted us to save in development cost by using looser satellite pointing requirements and reconstructing the image using software after the data is telemetered to the ground. Were data taken on a CCD detector with the same satellite pointing, the image would be blurred. Finally, the GALEX detector active area is 65 mm in diameter, ideal for this survey mission. CCDs are available in neither the requisite size nor shape, and CCD mosaics have gaps.

GALEX began normal operations in August 2003. We plan to complete
the baseline mission surveys by September 2007, following which we have
proposed to NASA to begin extended mission surveys through September 2010.
The Multi-mission Archive at Space Telescope (MAST) is releasing to the
public the second GALEX public data release, GR2, in stages:
  o GR2a) Released 04 May 2006, NGS (Nearby Galaxy Survey), MIS (Medium
Imaging Survey), DIS (Deep Imaging Survey)
  o GR2b) Released 02 June 2006, AIS (All-sky Imaging Survey)
  o GR2c) Release planned for July 2006, MSS (Medium Spectroscopic Survey)
GALEX Guest Investigator (GI) Program Cycle 1 observations began October
2004. Cycle 2 observations began October 2005. Beginning January 2007 the
GI allocation increases from 33% to 50%. The Cycle 3 GI proposal deadline
is 7 July 2006. For information on GALEX Guest Investigator Program, visit
http://GALEXgi.gsfc.nasa.gov. MAST releases Guest Investigator data to the
public upon completion of the 6-month proprietary period for each data set.

Yes. The three GALEX public data releases, GR2 , GR1, and ERO, are
in the public domain. Relative to GR1, the GR2 data are of improved
quality, coverage, and depth. GR2 is therefore preferred to GR1 as the
basis for science analysis. The Early Release Observation (ERO) data
quality is not appropriate for science; the ERO data are available for
historical reference only.

There is a notch in FUV images due to a high detector background emission “plume” at one edge of the detector. Depending on the roll angle a target is observed it may appear anywhere on the perimeter. The FUV detector also displays some areas of low gain and efficiency. The pipeline masks images in regions were the relative efficiency falls below 0.2 which produces the FUV scallop.