Tags:
create new tag
view all tags

Information and status page for Herschel footprint re-generation

This page compiles the information for the footprint generation using the astrometry information from the latest bulk-reprocessed data (HIPE 14.x).

Ancillary Files

The header and pointing parameter files extracted for each obsid have been stored at the following temporary area in the Herschel legacy repository:

http://archives.esac.esa.int/hsa/legacy/other/.footprint/

Headers and pointing material are in two different sub-directories ( headerFiles and pointingFiles). Note that the PACS-P pointing stuff is really big (100 Gb). In the PACS-P and PACS-S pointing archives there are also about 1100 obsids for which the generation of pointing information failed - I am assuming that this is because they are FAILED observations that did not generate the necessary products where to extract the needed information.

Also added is a text file with some "release" notes, explaining the inputs files that were used. The following remarks apply to the generation of the above files:

PARALLEL mode:

  • I have assumed that the header info only needs to be generated based on the PACS instrument (as the parallel_header.py script does) and that the SPIRE info would be the same. As a matter of fact, if one used parallel_header.py and fetch the obsid with instrument=SPIRE, the script does not work
  • there was no script specific to parallel mode for the pointing information. I have assumed that this is covered in the respective obsid lists for SPIRE-P and PACS-P and therefore is part of the pointing files for those sub-instruments.

SPIRE-S raster maps:

  • there was no obsid list specific to SPIRE-S raster maps, however I was given a dedicated script (spire_spectro_raster_pointing.py) that assumes availability of spire_spectro_raster.list. I have assumed that those obsids are already contained in the spire.spectro list

Malformed footprints from previous release

In the previous release of the footprint, we identified 666 malformed footprints:

  • 506 photometer observations
    • 432 of them (332 PACS/SPIRE parallel, and 100 SPIRE large maps) had footprints already from a previous release, that did not show any problem
    • the other 74 correspond to obsids that got footprints generated for the first time
  • 159 PACS-S observations, all being raster maps
  • 1 HIFI observation (1342203745)
The following approach was followed so far:

  • Regarding the photometer footprints, an example of the problem encountered is shown in the following plot (in white: non-bogus previous version, red: new version, bogus). At the moment, in the HSA we have rolled back to the previous footprint version for those were footprint existed from an earlier version, while for the 74 new obsids we used the simple circle footprint approximation.

  • Regarding the HIFI anomaly, a manual fix was applied and the footprint was provided to the HSA with this fix.
  • Regarding the PACS-S cases, a work-around was applied, consisting in treating each footprint as a rectangle. This turned out to give a good solution of about 3/4 of the concerned obsids (case 1 in bullet list below). For the rest (cases 2 to 4), there are different circumstances that could not optimally been treated with the simple rectangle approach. The four circumstances that apply are illustrated in file Alternative-FP-PACS-S-Raster.pdf
    • case 1 covers the above 3/4 of obsids where I believe the footprint is wrong, and the simple rectangle is a perfect solution
    • case 2 covers a proposed generic solution for circumstances of disjoint regions, where your footprints are correct but in essence
      cannot be ingested. This leads to over-estimated area for geometrical searches
    • case 3 covers the cases where the simple rectangle is not 100% perfect, but the impact on the geometrical searches is limited. For those the malformed footprints are also correct, we just have those small holes in the middle, that prevent the ingestion
    • case 4 covers the cases where my simple rectangle solution is not working sufficiently well. In my slides 4 and 5 are examples which basically are identical to those in case 3, except their geometry makes the simple rectangle a less acceptable solution. Slide 6 are examples similar to case 2, except that again the geometry makes my simple-minded rectangle merge a bad solution.


-- David Teyssier - 2017-01-19

Comments

Topic attachments
I Attachment History Action Size Date Who Comment
PDFpdf Alternative-FP-PACS-S-Raster.pdf r1 manage 570.3 K 2017-01-19 - 23:09 DavidTeyssier  
PNGpng SPIRE_PhotoLarge.png r1 manage 1831.7 K 2017-01-20 - 09:31 DavidTeyssier  
PNGpng SPIRE_parallel-3.png r1 manage 1479.8 K 2017-01-20 - 09:31 DavidTeyssier  
Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r2 - 2017-01-20 - DavidTeyssier
 
This site is powered by the TWiki collaboration platform Powered by Perl