Fitting the Fringe Map

To remove fringing from the images, the fringe pattern must be scaled and subtracted from the image. The level of fringing depends on the sky brightness, lunar illumination, and cloud cover, and an analytical calculation for the fringe level is not possible. At the moment we have implemented a simple linear fit to find the fringe scale.

The image pixel value is plotted against the fringe map pixel value for the central subregion of the image, as in Figure 7.4. Some of the image pixels contain starlight, so the image values are cropped at the ${3\sigma}$ level, creating the sharp cutoff at the top of Figure 7.4. Under ideal circumstances, the brighter sky values correspond to the higher fringe map values, and a simple linear fit determines the fringe scale. If there is a strong sky gradient across the image, as when the moon is bright and nearby, this can produce a false fit. At the same time, the bright broadband moonlight tends to wash out the fringing pattern. Our fringe subtraction technique might need to be improved to achieve our desired photometric accuracy for faint stars.

Figure 7.4: A plot of typical image values vs. fringe map values. The higher image values typically correspond to higher fringe map values. The slope of the line gives the fringe scaling factor, and the fringe map is then subtracted from the image.