Criteria for image selection from astroquery retrieval

[PaddyKavanagh]

In general, astroquery will fetch more than one image per filter per SNR. We need a way to select only one of these for use in the analysis. This seems to be difficult using astroquery itself due to limitations in the SHA form. Possible solution is:

• fetch all files per filter per SNR
• read data/header using astropy
• filter according to criteria using available information in the data/header (e.g., SNR distance from centre, corners,, exposure time, etc.)
• remove any unwanted files

[PaddyKavanagh]

Here is a suggestion on how to go about this:

1. Use target and principal investigator: Since we only want images from the Spitzer SAGE survey of the LMC, the target for all the SAGE observations will be set as 'LMC' in the header (under the 'OBJECT' keyword). Also, we can use the name of the PI to further check for SAGE images (under the 'OBSRVR' keyword). Keep SAGE images and delete others.

This will at least get rid of the pointed observations which are very patchy and not useful for us.

2. Use the coordinates of the centre of the image with the SNR coordinates: We may have more than one SAGE image per SNR so we need to select one. Probably easiest just to pick the one where the centre of the image is closest to the SNR. The positional header keywords are NAXIS (no. of pixels on one dimension), CRVAL (coordinates at the reference pixel), CRPIX (reference pixel) and a few others which aren't important here. It is very likely that the reference pixel will be located at the centre of the image (which can be checked as NAXIS/2 should be about equal to CRPIX) so CRVAL1 and CRVAL2 can be taken as the RA and Dec (in decimal degrees) of the centre of the image. Then just determine the distance of this position to the SNR for each image and keep the closest one.

This can then be the final image which we can slice out a smaller region around the SNR.