Teraminutes for the meeting of Jan 13th 2004

 Yesterday 1700 images were validated. Around 160 images still left to process by qualityFITS and 320 that must be re-graded from "C" to correct value on mix3.

 Comment: Yannick thinks that the manual step of QF validation will always have to be done, because it's hard to find the bad images automatically. Some comments from Henry about the counts produced automatically by qualityFITS. We will try to update the QF literature comparison to include counts measured on the CFHTLS deep fields.

SPICA (Laurent)

Some discussion about the use of skyprobe to reject images which have very heavy extinction (passing clouds). Currently skyprobe values are not used by SPICA, based on past recommendations, because most skyprobe were either missing or wrong. This is no longer true: we can have skyprobe data for most images; however, it is very difficult to have all of them in due time... this may be a severe limitation of the automatised photometric calibration process.

 It is not clear how to pipeline process data which have no skyprobe value (there are a large number of these images from 2003).

 Laurent will implement again a skyprobe value based on the median of ±5 skyprobe data. This will minimise random error and wrong +10 value that may happen when the skyrobe detector is not working correcly.

SPICA, release T0002 (Laurent)

 QFits on pix9 still running, but should finish very soon. Then all the images will have been pre-processed (remaining images will be validated today or tomorrow).

 The stacking should start probably at the beginning of next week. Henry, Yannick, Mireille, Jean-Christophe and Gilles will be responsible for scientific verification of the stacks. All these personns must receive the spica end-of-processing message from spica@iap.fr. Henry will not be available at the end of next week (Wednesday - Sunday)

 Yannick asks all persons in charge of T0002 monitoring to write a comment in the spica-interface comment box when they have checked an image produced by spica. This will avoid several personns check the same data. HOWEVER: DO NOT PUT A GRADE, otherwise data will be put on /fc output disk for CADC.

SCamp (Emmanuel)

 A method has found to parallelise SCamp and produce potential speedups of around a factor of two with a machine with four processors. But this solution uses too much memory. More tests need to be done.

 Some field configurations produce NaN values. For example, if there are several fields that are connected together by images obtained with several different filters, the astrometric solution can be found while the photometric rescaling is wrong. Emmanuel will fix this problem. In the meantime the NaN values will be replaced with fixed values.

Swarp (Henry, Emmanuel)

 Swarp still doesn't write the gain correctly into the header. This should be added.

VOtable (Jean-Christophe)

JC is still writing the doc, Henry will read it to check the english. The VOtable Clib developped by Jean-Christophe will be put on Terapix public access next week.

Skywatcher (Jean-Christophe):

Nothing new since last week.

ICD (Laurent, Jean-Christophe):

 Still waiting for comments. Jean-Christophe sent a version to CADC.

 A script has been written which will manage the upload of data to the CADC

Defectix (Anthony):

 MEF and single FITS Megacam loading library finished.

 Problems with images produced in double: DS9 and other software have problems reading very large images in double. This part of the software will be re-written to output images in float.

Hardware (Fred, Emmanuel, Yannick

 ECC should be enabled in mix2 / mix3 after qualityFITS runs have finished (the kernel has been changed on these machines, with the old kernel it was not possible to enable.

Data backup on the digital machines (Gilles):

 Almost all the disks attached to the digital machines have been backed up.

 The machines will still be turned off in the third week of January. Gilles will tell the users.

Other matters

04AC10 for Bill Harris(Mireille)

 Data reduced without any big problems, with the exception of usual problems with the correct back mesh size to use with images which have very large objects.

 Mireille used the new STIFF that turns out to work very well on these data.

Next Tera-meeting: Thursday 20th January 2005