Skip to content

PSE Point Source Extraction & Photometry

Jump to bottom
user29A edited this page Jul 4, 2022 · 6 revisions

The algorithm for PSE (point source extraction) done by CCDLAB is described in this paper:

https://iopscience.iop.org/article/10.1088/1538-3873/ab7ee8

Reading the section on PSE will provide full clarity on what the options seen in the CCDLAB PSE tab are about.

But for specific CCDLAB instructions:

  1. In practice, one can set the “Pixel Value” threshold to 0, and use only the “Kernel Sum” threshold. One may also leave the Max threshold at maximum value.
  2. One would iterate or “play” with different values of the Kernel Sum threshold until one determines that they are finding all sources.
  3. On the Processing tab, there is a Filter button with an option to convolve a Gaussian of a specified FWHM, in pixels. But only use this if absolutely necessary. It does use a normalized Gaussian for convolution filtering, and so the total counts of a source over its kernel are not affected, but individual pixels would be.
  4. If one right-clicks on the “PSE_1” drop-down which appears, one may create additional PSE’s for comparison, if desired, etc.
  5. One may also view tables of the PSE results, and if the table is double-clicked on an entry CCDLAB will show you where it is.

Of course, this only identifies the point sources, and the photometric extraction is only for the specified kernel. For more accurate photometry, one may wish to perform more accurate extraction once the source are identified. This can be done in two ways:

  1. If one has a small number of sources with clean surroundings, one may use the “COG” tab below the PSE tab in order to perform Curve of Growth or aperture photometry on sources.

  • a) Place the ROI (region of interest subwindow) over a source, go to COG tab, check “View COG”, and press “Do COG”. One will see a plot of the aperture photometry.

  • b) Play with the “Fit the Last N Points” number until a good fit is produced for the background.

  • c) The intercept of the fit equation are the total counts for the source. Use this for photometric conversion to magnitude as per the Tandon photometric calibration paper.

  • d) I should update CCDLAB to perform COG over a series of PSE sources automatically, assuming that the sources are “clean”.

  1. One may also perform least-square function fitting of either Gaussian or Moffat functions.

  • a) One would check “Fit Sources”, and then choose the function they want. This will automatically fit all PSE sources. Again, if the sources aren’t all “clean”, then adjacent sources can be a problem.

  • b) One also must calibrate the function fit photometry against the COG photometry of clean sources, as function fitting and COG aperture photometry do not produce the same results, and the transformation to magnitudes are based upon aperture photometry as per the Tandon paper.

  1. If one has a cluster of point sources of up to 12 sources within the ROI window, one can press the “Fit” button in the “Fit” tab of the “Sub Image Region of Interest” area, and CCDLAB will fit a compound function (user selected either Moffat or Gaussian) to all sources simultaneously within the ROI window which have a PSE indicator on them. This allows fitting of adjacent sources, i.e., sources whose wings are overlapping or too close to each other for clean COG fitting.

As you know, photometry can be very simple for nice clean sources with no background contamination from nearby sources, but then it quickly gets very complicated for crowded sources, etc.

Clone this wiki locally