In carrying out a very wide field survey we face many challenges that we need to overcome. The commercial lens that we use for PASS0 suffers from optical vignetting and spherical aberration. Vignetting reduces the S/N towards the edges of the images and spherical aberration deforms the PSFs of the stars towards the edges of the images.
In Figure 2.4 we show a 2D polynomial fit to the sky background of a typical PASS0 image taken at f-stop of 1.2 which clearly illustrates the magnitude of the vignetting. Ideally the sky background should be flat. This vignetting is also visible in the image shown at the top in Fig- ure 2.5 which was taken with an f-stop of 1.2. In the same image we can see that the star PSFs are being deformed towards the edge of the image, which is a bigger effect than the short star trails in the 20 second exposures.
The easiest way to reduce the problems with the vignetting and lens aberrations is to step up the f-stop. The image shown at the bottom in Figure 2.5 has been taken with an f-stop of 2, and now the vignetting amplitude is greatly reduced and the star PSFs are not distorted towards the edges of the CCD. However, by stepping down from an f-stop of 1.2 to an f-stop of 2, the lens diameter decreases by a factor of 0.6, and consequently the S/N we can achieve in a fixed exposure time falls by a factor of 0.36. If we can develop methods to cope with the vignetting and PSF deformations when doing the image reductions and data photometry, then it is clear that we should still choose an f-stop of 1.2 to maximise our S/N.
Our approach to calibrating properly the lens vignetting involves obtaining a set of high quality flatfields. Twilight sky flats are not suitable because they produce a large sky background gradient since the sky is not sufficiently uniform over the PASS0 large field of view. Also, since twilight is restricted to a short time period of about half an hour, only a limited number of flat fields may be obtained. Finally, stars appear in the twilight flat field images since the lens is a fast lens, and these stars will interfere in constructing a high quality master flat field. Our solution has been to construct a large white screen of light diffuser material which we use as a source of uniform
52 CHAPTER 2. DESIGN OF THE PASS0 EXPERIMENT
Figure 2.4: Plot of a 2D polynomial fit to the sky background of a PASS0 image taken at f-stop 1.2. The effect of the lens vignetting is to create a sky background that reaches a peak in the middle of the CCD and falls off towards the edge of the CCD. Ideally the sky background should be flat.
illumination. Many flat fields may be taken at once using this device which allows us to construct a very high S/N master flat field of a truly uniform source. Note that flat fields are all taken at the same CCD temperature for consistency and this temperature is below 20 degrees Celsius.
Next we consider the photometry method that we intend to use for the PASS0 image data. Aperture photometry is possibly the simplest method and we make some initial tests on a WASP0 image of 60 second exposure using the AP10 CCD. The first challenge was to design the correct aperture for a trailing star and we did this by redesigning the aperture used in the WASP0 pipeline as shown in Figure 2.6. It is clear from the image that many apertures overlap, a situation that is worsened by the trailing. For PASS0 with a much larger pixel scale, this problem will be much worse.
Figure 2.5: Top: PASS0 20 second image using a f-stop of 1.2. Bottom: PASS0 20 second image using a f-stop of 2.
54 CHAPTER 2. DESIGN OF THE PASS0 EXPERIMENT
Figure 2.6: A WASP0 60 second image with example apertures to be used for aperture photometry. The 180 mm lens FOV is 9◦
×9◦
with a diameter of 6.30 cm and aperture f/2.8. The WASP0 camera uses the AP10 CCD.
At this point we see that the image subtraction method (see Chapter 3) can reduce the problem of the blending of stars since all constant stars are removed in the difference images, and measuring the difference images of variable stars is then a much cleaner process. Only blend stars that lie within the PSF of the star being considered will continue to affect the lightcurve. Also, by combining the image subtraction method with an optimal PSF scaling on the difference images in order to measure the differential fluxes, we can achieve a better signal-to-noise ratio than aperture photometry (because aperture photometry is not optimal). It is clear that for the experiment to work we will need to create a pipeline that uses difference imaging.