Present: Yannick, Gilles, Mireille, Fred, Jean-Christophe, Laurent, Emmanuel, Henry

Status of the release (Laurent, Yannick) For the very-wide, of the 1600 images 400 have been processed. Assuming we process at this rate, we still need one month.

Need to reduce the number of ldacs to mix3 / mix2 to prevent system swapping during the matrix solution. Need to find a way to determine the optimal number of ldac files.

Completeness plots in the very-wide (Jean-Christophe). For many plots (particularly in the g-band) the completeness graphs were not generated. Henry will try to investigate this.

Network interruptions (Emmanuel) If the network is down SCAMP cannot retrieve the USNO-B catalogue necessary for the astrometric solution. How can we remove this dependence on external resources for the processing? Emmanuel suggests that we use the option in scamp to save the reference catalogue. The problem is to create the list of reference catalogues necessary for processing (or can we simply just copy the USNO-B catalogue locally?). Jean-Christophe will investigate this possibility.

Time required to finish the release from now is around one month of processing. We need to find a more efficient method of processing the data. Can we, for example, separate processing by scamp and Swarp? Laurent will investigate what the various possibilities are.

Photometric comparison between t0001/t0002 images (Henry, Mireille, Yannick) Henry finds a small photometric offset between T0001 and T0002 in all bands (about 0.05 mag. in u,r,i,z and 0.10 in g). Mireille made u/g/r/i/z stellar color-color plots in two W1 fields. The 2 fields are consistent, an offset in visible , likely produced by g-data (0.1 mag. offset). We need to make more colour-colour plots for the deep images (Henry will do this) to investigate this. Mireille will produce the colour-colour plots from the wide. Henry has supplied the colour-colour data from the Besancon model to Mireille. Henry will also investigate the best way to estimate the background for deep stacked images (should we try to implement a two-pass mode to remove the objects before stacking).

Similar checks need to be done with g-r-i Very Wide data. No check done so far. Yannick , Henry, Jean-Christophe and Mireille will check.

Final astrometric solution (Yannick) There are some stacks where the final astrometric solution has an elliptical error ellipse. This happens in the wide W1 only and not in the deep D1. Maybe this is one field which has a bad astrometric solution (suggestion from Emmanuel). Emmanuel and Yannick will check.

Modelling the focal plane (Fred, Emmanuel, Yannick) Fred has generated an 'ideal' focal plane based on all images with a given runid, using the new tool developed by Emmanuel for this task. Works pretty good if there are enough images ( 20 images) to construct a reference focal plane for a given combination of QrunID and filter. (For future releases CFHT will use SCAMP calculate the astrometric solution, so we shouldn't need to do this).

Shear E/B modes (local) (Yannick) comparison between astrometrix and scamp are in good agreement when local stacks are compared. (B-modes are similar in both cases).

Data transfer to CADC (JC): All QF-ins have been transferred to CADC. For the release, no images have yet been transferred, either the T0002 and T0001. JC is current carrying out the verification of very-wide data. JC provide a useful tool that produces real-time statistics on completeness in each filter as function of exposure time. TERAPIX can use it at http://clix.iap.fr/admin/completude/ .

Defectix (Gilles) Defectix was run on 21 images. There many special cases which caused problems (for example, images with at low galactic latitude or with large bright galaxies). Gilles and Yannick will identify what features / problems exist and report these to Anthony. The sample contains:

- 1 CFH12k image - 14 Megacam MEF selected in PI data: big galaxies, millions of stars, various astetoid tracks, strongly saturated stars - 6 Megacam MEF images selected from Wide data and corresponding to the same field, but with 1-4' offsets: test reproducibility of masks. - 1 Wide stack (no MEF) of the 6 MEF : tests the stacked masks against mask of the stacked image.

Hardware (Fred, Emannel, Yannick) Fred has carried out a lot of work on mix4 (ex mix2) to try to resolve the problem of the degraded /raid2 array. Fred physically exchanged the disks between the two raids: the problem didn't follow the disks. It seems the problem comes from the disk support mechanism.

Hardware purchases: Yannick, Fred, Emmanuel) mix5: we will try and order the machine next Tuesday. We will try to decide on the remaining purchases (fibre channel, ftp machine) next week.

Interface follow-up Yannick asks that people send comments and suggestions to Laurent about the interface.

Next tera-meeting: FRIDAY APRIL 1, room 281, starting at 14:30.