How the ROSAT PSPC survey was generated
This document describes the steps used in generating the
SkyView ROSAT PSPC survey.
Description of the Survey
The SkyView ROSAT PSPC survey is a mosaic of public PSPC observations
available at the High Energy Science Archive Research Center. It includes
data from all PSPC observations with the following exceptions: data taken
with the Boron filter, data taken prior to 1991-1-1, and observations for
which there was no good exposure. [Currently there are also 6 observations
not included because the number of photons in the observations makes the
mosaicking program overflow memory limitations.] A total of 4174
observations are included in the mosaic. 274 verification period
observations are excluded as well as 40 observations for which there is no
The mosaic comprises 600x600 pixel tiles. Each pixel is 15" square
so that the coverage of an individual tile is approximately
2.5° square. Each tile is a gnomonic projection about the center
of the tile, i.e., the intersection of the four central pixels.
The tiles are arranged on the sphere in 'rows' at declinations separated
by 2°. Each row has tiles separated in right ascension by no
more the 2°. Thus tiles should have an overlap of at least 30 arc
minutes (or 120 pixels). The total number of tiles needed for complete
sky coverage would be 10359.
For each observation a determination was made of the tiles to which
it would contribute and then the counts and exposure were added
to cumulative counts and exposure maps for the tile. When all observations
had been processed an intensity map was created by dividing the counts
maps by the exposure.
Three different mosaics where created by using different cut-off
radii for included observations.
The three cut-off radii
used were 2°, 1° and 0.6°. For these
three cut-offs data were found for 5998, 4540 and 3984
For each tile the contributing observations
were recorded and are included in the tile headers.
Virtually all tiles have only partial coverage.
Total sky coverage is estimated at 14% for the 2°
survey, 4% for the 1° survey and 2% for the 0.6° survey.
Coverage in 'interesting' regions of the sky is typically much higher.
Using the ROSAT surveys in SkyView
SkyView uses the tiles as a standard SkyView survey.
The 1° and 2° intensity maps are available directly through
the forms interfaces. Ten surveys total are available to users
either from the results of previous search request, or through
the Batch interface: Intensity, counts and exposure maps for
the 0.6°, 1° and 2° calculations, and the old PSPC intensity
survey (which should be similar to the 2° maps with substantially
SkyView does not know anything about which parts of a tile
may have been observed. It may return a blank image the user requests
an area near but not overlapping other observations.
How was the mosaicking done?
This section describes the algorithms used in mosaicking the data.
Throughout we have used the term tile to refer to the mosaic fragments,
while observation refers to the ROSAT observation being
added to the mosaic.
Phase 1: Generating the counts and exposure maps
The system looped recursively over the directory structure containing
the PSPC observations. Whenever the job was interruped it was restarted
using an observation catalog to ensure that data were not
For each directory:
If the directory was for an observation in the observation catalog
the directory was skipped.
If the directory was for an observation with the boron filter
(an 'f' as the second letter), the directory was skiped.
A list of problem directories was checked. If the directory was
in the problem list then it was skipped. The problem list
currently comprises: wp201754a01
The directory containing was checked to ensure that
both an exposure map and events file were present. If not the
directory was ignored.
The exposure map file was read. If the exposure map was null,
the observation was added to observation catalog and it was noted that
this observation had a null exposure. No further processing was done.
The date of the observation was taken from the exposure header.
If the date was prior to the start of 1991, then the observation
was added to the observation catalog and noted as a verification
observation. No further processing was done.
The table of tile centers was used to find which tiles might overlap
Any existing counts and exposure maps for the affected tiles were copied
into a working area from the archival area.
The counts data was read in and the counts were deprojected
into J2000 RA and Dec coordinates. Only counts
within the medium-energy band were included. Data from channels
10-235 were included in the maps. This underestimates the contribution
from observations before 1991-12-14 slightly.
[Currently if the number of events is greater than 3,000,000, then
the directory is skipped -- but not added to the catalog].
For each tile:
The RA and Dec centers of the pixels of the tile were
calculated and then projected into the coordinate system
used in the exposure map.
The exposure at the tile pixel centers were calculated
using a bi-linear interpolation of the observation exposure.
For each cut-off radius (in descending order):
The tile pixel centers that were within the cut-off
radius were tabulated. Other pixels were discarded.
If an existing exposure map existed for this tile and radius
it was read in otherwise a null map was created for the tile.
The exposure for the selected pixels was added.
The new or updated exposure map was written out.
A list of the tile pixels that were updated was returned.
The events were now projected into tile pixel coordinates.
For each cut-off radius (in descending order):
Any events which projected into pixels which were
not returned as being included in the corresponding
exposure map were discarded.
If an existing counts map existed for this tile and radius
it was read in otherwise a null map was created.
For each event the corresponding tile pixel was
incremented by one.
The new or updated count map was written out.
If the updating of exposure and counts maps had been successful:
the maps from the working area were copied to the
The directory was added to the observation catalog.
Phase 2: Generating the Intensity Maps
Once a complete set of counts and exposure maps was generated,
a loop over the counts maps was initiated. Each counts
and exposure map was read and the ratio -- the intensity map
was computed. The value of the intensity map at all points
where the exposure was 0 was set to -1. A FITS header
was created which included the projection information for the
map and information about which observations contributed
to each image.
A counts, exposure and intensity FITS file
was then produced using this header with an appropriate BUNIT's
field added. The output files were sorted in 9 directories.
Direct all questions regarding SkyView to our Feedback Page
SkyView has been developed with generous support from the NASA AISR
and ADP programs (P.I. Thomas A. McGlynn)
under the auspices of the High Energy
Astrophysics Science Archive Research Center (HEASARC) at the NASA/
Astrophysics Science Division.
We gratefully acknowledge the support of NASA and contributors
of SkyView surveys.