White paper  -  Edwin Valentijn  -  OmegaCEN / OmegaCAM / AstroWise 7 June 2005

Optimizing OmegaCAM/VST Surveys: Tiling and Pixelating the Sky
A plan for coordination of coordinates for both Survey data taking and processing

On average  OmegaCAM will observe ~20 fields per night, which  for say 300 nights  per year equals to ~6.000 fields /year, which roughly corresponds to the  equivalent of the full area of the Southern sky in ~3.5 years of observing. After a number of years of observing there will be many overlaps between observations taken for different projects, and clearly there is a lot to gain by organizing the fields in advance.
(footnote: for the forseen 10 year lifetime of VST/OmegaCAM this corresponds to observing the equivalent of the whole Southern hemisphere in 3 different bands with 0.20 arcsec pixels)

The AstroWise system will ingest the majority of these observations in  federated archives distributed over at least  NL, D, F and I.
The archive will contain the raw observational data together with processed, astrometrically and photometrically calibrated, science images, which will be published directly from the archives (datawarehouses)  to the VO.
The science images are often taken with multiple exposures (dithers) on the telescope.  In the image pipeline these images are re-gridded  and sub-sequently co-added.

Obviously, with this enormous amount of observations there is a lot to gain by co-ordinating:
For OmegaCAM we propose  a standard set of field centers for the observations (1),  also to be used for the reference coordinate system (2) and hence the pixel-grid (3).

A preliminary tiling and pixel grid scheme has been defined for OmegaCAM (Class PlateSystem.py) which involves  22.717 Square degree fields for the Southern hemisphere with on average about 5% area overlap between adjacent fields (see figures skygrid_*.ps).
Fields are positioned in 95 strips of constant  Declination,  separation in RA is optimized to maximize continuity in declination direction, which is the more  complex part of the definition.
In the re-gridding process,  science images are projected with the traditional gnonomic tangential projection with pixel size 0.2 arcsec, with projection centers identical to observing field centers. This defines a complete sky pixel grid.

Benefits of adopting such a scheme, both for  Survey preparation and for data reduction are:
-- Of course observers do have the option to override the tiling when there are good scientific reasons to do so.
-- We are in contact with VISTA on this scheme, but  unfortunately the larger field of view of Vista does not really  match.
-- Putting integer number of squares covering the surface of a sphere is an art in itself. Further improvements and insights are very welcome!

Formal ESO status is:

A few lines on the tool itself Platesystem.py :
  1.  it can be used to design a rectangular grid (tiles) on the sky with free parameters field of view, nominal overlap and minimal overlap;  it plots and prints the grid (tiles) - the script is clever in making the grid smooth and avoiding big jumps.
  2. The default values actually correspond to the tiling we have tentatively adopted for OmegaCAM.
Once the Class is instantiated it can be applied in different ways:
At the bottom of the script text some examples are included which are executed when running the script as is.

The script does nothing in automatic OB creating for large numbers of observations, and we are very interested in co-operating with
VISTA for  that part.