Scanamorphos *


     version 24 available (July 2014)

               main changes in v23 and v24:

  • The PACS distortion flatfield (see item (j) on the tips page) is now taken into account. It was previously unavailable to all map-makers except HIPE-PhotProject. This changes the flux calibration by ~ 7% at 160 µm (and by ~ 1% at 70 and 100 µm). Please note that when the coverage varies significantly from place to place, it is not enough to rescale the final maps by a constant factor, but they have to be regenerated.
  • Data with very inhomogeneous coverage maps (in particular PACS observations in the red band taken after a matrix was damaged, i.e. starting with OD 1375) are now dealt with correctly.
  • The destriping of PACS observations of large and diffuse fields is substantially improved.
  • The subtraction of the average drift on short timescales can no longer introduce low-level noise.
  • One no longer has to worry about the ordering of the scans on input.

    Please look at the top of the tips page for some important advice.
     companion paper and supporting documents available here
     user guide decoupled from the paper available here



Go to the records page to see a list of changes, and read the updated README file and user guide.

Visit the tips page to look for answers to some usage questions.

Read the paragraph on backgrounds in Herschel maps if you ever wondered about them.


      If you wonder about the shape of some point sources in PACS maps:
          On the PACS Calibration Web, you will find a very useful document describing the structure and possible
          distortions of the point spread functions (in particular the elongation and undershooting at high scan speed).

     Likewise, on the SPIRE Calibration Web, you will find up-to-date documentation about the average beams.



  documentation and reference 
  download and installation 
  inquiries 
  acknowledgements 




Scanamorphos is an IDL software to build maps from scan observations made with bolometer arrays, in particular with the PACS and SPIRE photometers onboard the Herschel space telescope (wavelength range of operation: 70 to 500 µm). The prototype software has been developed on SPIRE simulated data and on real data from P-Artemis, an instrument of the same design as one of the PACS subarrays, but operating on the ground (mounted on APEX). After the launch and performance verification of Herschel, it has been extensively tested on both SPIRE and PACS flight data.

Scanamorphos is chiefly designed to remove the low-frequency noise, both thermal and non-thermal, causing additive brightness drifts. To this end, it makes use of the redundancy built in the observations (each portion of the sky with nominal coverage is sampled by multiple bolometers at multiple times) to derive the brightness drifts directly from the data themselves. This means that Scanamorphos does not depend on any noise model, and is thus immune to mismatches between models and reality, that may occur because the noise is not perfectly calibrated, or does not satisfy the model assumptions, or varies in time, or is altered by the pre-processing (from level 0 to level 1 in the Herschel jargon). Furthermore, the algorithm is adaptable to other bolometer arrays, by modifying the few instrument-specific parameters.

Scanamorphos is flux-conserving and restores sources on all spatial scales, ranging from point sources to extended structures with scales just below the map size. (Of course, there is no hope of faithfully restoring brightness gradients on scales larger than the map; to handle these, strong assumptions have to be made.)

Here are the main functionalities:
  • subtraction of linear baselines (brightness drifts with timescales larger than the scan leg duration)
  • subtraction of the average brightness drift (i.e. thermal drift), on timescales smaller than the scan leg duration
  • subtraction of the individual brightness drift of each bolometer (i.e. flicker noise), also on timescales smaller than the scan leg duration
  • detection and masking of glitches left by the pipeline
  • detection and masking of brightness discontinuities (jumps) in PACS data
  • projection of signal, error, drift and weight maps on a spatial grid specified by the user, employing a more sophisticated technique than implemented in the SPIRE pipeline (the latter performs a closest-neighbor projection, introducing large biases between beam center and projection center), but assuming axisymmetric bolometers (which is true for SPIRE, but not for PACS).
The processing is fully automated, but ample visualization of intermediate results at various steps is enabled.


If you use Scanamorphos to map your SPIRE data, you might need point response functions (PRF) also mapped with Scanamorphos. PRF built in March 2010 from Uranus observations are available from this page. They will be updated when I find the time.



* portmanteau word composed of "scan" and "anamorphosis"
   anamorphosis: reversible transformation of an image by a mathematical or optical operator
   (from the Greek , implying the idea of a transposition or return; and : shape)



~o~


documentation and reference:


Please cite the companion paper when using Scanamorphos:

     Roussel, H.  2013, PASP 125, 1126   (was  arXiv:1205.2576  for more than a year)
     available as a postscript file :   scanam_29jul13.ps.gz
     (please note that the file contains many large figures, and is 82 Mb when g-unzipped)

It contains a detailed description of the algorithm, illustrations of the various processing steps drawn from SDP (Science Demonstration Phase) data, that are public, and the results of two PACS simulations.


A document comparing PACS and MIPS surface brightnesses for very extended and diffuse emission
(made in response to the sixth referee report):   report_cirrus_pacs_mips.pdf

My contribution to the SF2A 2012 meeting:   2012sf2a.conf..0559R.pdf


The decoupled user guide is available as a pdf file:   scanam_manual.pdf

Additional information is given in the tips page.


~~o~~


download and installation:



Scanamorphos is distributed under the terms of the GNU General Public License. You will need recent versions of IDL and the IDL Astronomy library.
Unpack the software archive below, which will create a directory named  SCANAMORPHOS/ .

    scanamorphos_23july14.tar

Follow the instructions contained in the  README  file under this directory, and read the documentation. The HIPE scripts and Scanamorphos inputs (ascii files and IDL commands) used to process the observations illustrated in the companion paper can be found in the  SCANAMORPHOS/examples/  directory.


~~~o~~~


inquiries, bug reports, suggestions for improvements or new functionalities:



If you have questions or comments about any aspect of Scanamorphos, please contact me by electronic mail (roussel at iap dot fr). If you wish to be notified of updates, simply include "register scanamorphos" in the subject field.

Scanamorphos is distributed to the astronomical community in the hope that it will be useful. But since the software could not be tested in any possible configuration, you may encounter bugs now and then. In this event, please give me a precise account (with the inputs and full error message), so that they can be fixed in patches and future releases.


~~~~o~~~~


acknowledgements:


The procedures in charge of reading and formatting the input data make use of some code written by Marc Sauvage and Pierre Chanial (CEA, AIM, France). Their help is gratefully acknowledged.

This work also benefited from working within the SPIRE Scan Map Pipeline Validation Team (a subgroup of the Instrument Control Center), whose members are collectively thanked.

Finally, I warmly thank the colleagues, too many to cite all, who allowed me to test the code on PACS observations from their programs, previously reduced up to level 1, and experimented with the first public version or tested the successive releases, reporting bugs and/or suggesting improvements.




Hélène Roussel
Institut d'Astrophysique de Paris - Université Pierre et Marie Curie
July 2014