The ‘Photometric Calibration’ button does two things:
- It runs source detection and photometry software (SourceExtractor) over the image. This detects all stars, measures their fluxes and converts them to instrumental magnitudes. These magnitudes need to be calibrated into apparent magnitudes, or at least an instrumental magnitude system which is the same for all images.
- Thus, we need to calibrate the photometry. For this purpose, we have for each target and filter a ‘calibration master frame’. Most of the calibration process is completely automatic, and you, the user, doesn’t need to do much, except checking if the result is ok or not. This is done via two graphs which are displayed when the calibration has finished. Like with the astrometry, this process can take a while, in particular if there are a large number of stars in the image.
The software first matches up all the stars in the image with the stars in the calibration catalogue. Only stars which are judged to be well detected are considered. A plot is then generated which shows for each matched star the difference between the instrumental magnitude in your image and the magnitude the star has in the reference catalogue. This difference is plotted against the instrumental magnitude of your star. This is the top of the two plots.
The software then tries to find a trend line which describes how this difference depends on the magnitude. This is over-plotted as a blue line in the top plot. If the data was correctly calibrated, this blue line should be at a value of zero for all magnitudes. To achieve this the software fits a function to the blue line. This then needs to be subtracted from the magnitude difference. The result is shown in the bottom plot on the result page.
The bottom plot shows as red points the same stars as in the top plot. But this time the difference of their instrumental magnitudes and the magnitude in the reference catalogue is plotted, considering the fitted correction function. Hence, now all these points should scatter around zero along the y-axis. The x-axis now also shows the calibrated magnitude and not the instrumental magnitude anymore.
There are two further green dotted lines which indicate the typical scatter of the red points from the y-value of zero. Typically about 1/3 of the red points should lie above the top green line or below the bottom green line. The smaller black dots in the bottom graph are all other stars in the image which had some issues with the photometry (i.e. they are near the image edge or have a nearby neighbour, or don’t look like a star). They do however, typically behave the same as the ‘perfect’ stars – red dots.
The task of the user is now to either accept, reject/delete or re-calibrate the image.
When to accept an image (Calibration graphs appear correct – green):
Simply put, when everything looks ok – see Figure 1. Basically the red points in the bottom graph scatter around zero. There is more scatter for fainter objects than for bright objects. The scatter for the bright stars (left) is no bigger than 0.1mag. There is not a significant number of black points to the right of the green vertical line – this line indicates the maximum magnitude considered in the calibration. There is no significant gap between the green vertical line and the red points.
When to re-calibrate an image:
Simply put, when the data looks like it should be accepted, but some things could be improved. This can either be if there is a large gap between the vertical green line and the red points – e.g. Figure 2; or there is a significant number of black points to the right of the green vertical line – e.g. Figure 3. In those cases you need to adjust the value for the ‘MAX_USE’ parameter and press “Retry calibration with the above values”. The MAX_USE parameter basically indicates where the vertical green line is. The software tries to determine it’s position but gets it wrong sometimes.
Hence, if there is a gap between the red points and the green line, use a value which corresponds to the magnitude in the top graph where the points ‘thin out’. Similarly, if there are many black points to the right of the green line in the bottom graph, use a MAX_USE value that is a bit larger than the point in the top graph where the green line is drawn. For the example in Figure 3 you could use MAX_USE=19.3 and then you get the output shown in Figure 4. This would be an acceptable fit.
You can repeat the re-calibration as often as you like with different parameters until it looks ok. But usually, after a bit of practice, only one or at most two re-calibrations are needed.
When do you reject the calibration and ‘Reprocess file’:
Simply put, when the data looks very bad and/or has a lot of scatter, such as in Figure 5. There are two possible reasons for such data:
i) Observations were taken under bad conditions, i.e. extremely variable and/or thick clouds. In these cases, when the scatter of the points is more than 0.1mag everywhere, the data is unusable and should not be included in the database – unless:
ii) For some reason the software has mis-identified the filter the data has been taken with. This can happen if for some reason the FITS header entry is wrong, or if you manually entered the wrong filter. We have done this on purpose for Figure 5. We pretended it is an R-Band image but it actually is an I-Band image. If we reprocess the data, with the correct filter, then the result immediately looks like Figure 6. This will, of course need some re-calibration and with using MAX_USE=19.3 it will look like Figure 7 and would then be acceptable.
When to reject the calibration and to delete an image:
Simply put, when all of the above described things don’t work and the data looks very noisy with lots of scatter whatever you do. If you are unsure if an image is acceptable or not, simply do not press any of the four buttons, but click on “Process Files” on the top of the website to get back to your table of images to be processed. Please get in touch stating the image ID number (the one at the front of the filename given to it by our software – first column in the table of images) and we will have a look at it and send you feedback.