MSCRED: Mosaic Processing
Frank Valdes wrote on Jun 26, 1998
Q: I have a question regarding processing of the mosaic data. Would it be better to regrid all of the individual images before flat fielding (including regridding the flat field image) If not, then I would expect (from reading the user guide) that my image would not look flat before regridding? A: It is NOT a good idea to flat field after regridding. The discussion in the guide can be confusing. What the guide says is when you flat field the data then, almost by definition, the sky will be flat. This means when you regrid you need to have flux conservation turned off to preserve this flat sky in the final regridded image. The point of the discussion in the guide is that if you plan to do photometry on the unregridded data then flat fielding is not strictly correct because in reality the sky is not flat in a wide field image due to a variable projected pixel size on the sky. In this case you would need another task to correct what the flat field does (i.e. make the sky flat) to make the photometric counts per arcsec be ok. This task doesn't yet exist. If one was to do this and then regrid, the sky would be flattened again in the regridded image by turning on the flux conservation since this will then regrid to constant pixel sizes projected on the sky. All this does not apply to your data because 1) there is no task yet to fix what the flat field does and 2) if you are talking about regridding then you don't really care if the flat fielded data before regridding is photometrically correct. Cheers, Frank
Frank Valdes wrote on Jun 26, 1998
Q: When one has flatfielded an image, do we expect that the value of the sky, for example, should be the same from chip to chip? By taking several measurements of the sky for each individual chip, it seems that the flatfielding has reduced the difference of the chip to chip measurements. Does any of the routines in the processing take into account the difference from chip to chip by doing some sort of scaling so that one ends up with the same sky level independent of the chips? A: The flat fielding, specifically sky flats, should bring all CCDs to a common sky level in principle. In practice this is not perfect. There is nothing else in the current guide discussion. We have recently experimented with do a sky subtraction using a low order (a plane) to take the sky to zero and remove some sky gradients, which are also not dealt with in any other way. This looks like it works well. So in a future guide this will be discussed. Basically what you need to do is use IMSURFIT (probably with MSCCMD) with a low order and using large median boxes, say 100x100. I may add a specific task to MSCRED for this purpose.
Last post on Jun 26, 1998