Creating colour profiles

To obtain colour profiles and determine the colour gradients we initially had 8 galaxies observed in two filters to create a ($J-K$) profile. The image quality of NGC 2424 was too bad to match sufficient stars to match the two images for subtraction, while the J- and the K'-bands images of UGC 3186 were both simply too faint to obtain a coherent colour profile from which could be determined a gradient. Because of this these two galaxies were rejected.
Colour profiles can only be created when both galaxy images are exactly aligned and have the same pixel size. The K'-band images were matched to the J-band images using common forground stars. The seeing of both images was determined to decide which image would have to be convolved to the image with the higher seeing. This was usually the J-band image, although convolving was not always necessary. Because the J-band images show more and brighter stars as they have a higher resolution, these images was used to mask all the stars and from that create a mask image that could be used for the J- and the K'-band images. This to not be hindered by contamination of the galaxy luminosity by forground star light. To both images and the mask image the rotation and centering of the J-band image was now applied.
Vertical colour profiles were created by subtracting vertical K'-band surface brightness profiles from binned vertical J-band surface brightness profiles. At the subtraction the masked areas were ignored and only those vertical data positions both images had a value for were used. For the combined error calculation a 1$\sigma$ level of the variation in the sky background was used.
We extracted vertical surface brightness profiles at nine positions along the major axis of the sample galaxies, as we wanted to apply equal intervals that approximately corresponded to a factor of the typical scalelengths found with the two disk fits (20-25 arcseconds). The two profiles were binned as done before, the bins growing at an exponential factor $b_f$ from the midplane, to retain an approximately constant overall S/N ratio in the resulting vertical profiles and decrease large irregularities. $b_f$ was chosen to be 1.07, the minimal value that was also used for fitting the surface brightness profiles. No larger number was chosen as it would reduce the amount of datapoints.
To determine the radial positions we used scalelength factors of 0, 0.5, 1, 1.5 and 2, on both sides of the galaxy. We did not use the scalelengths from our 2D Two Disk fits as initial tests showed no radial behaviour of the profiles. Due to the small field of view of the UKIRT telescope it was not possible to go out much further radially so we had to limit ourselves to 2 scalelengths for the radial profiles. The radial width of the bin of each cut was larger than the previous to adjust for the lower S/N at larger radius. The radial binsize was set manually, the size a step of 2 pixels larger for each outer cut.