Errors Associated with Observations

Given the expected very low amplitudes of variation in young stars, a few tenths or less of a magnitude in the near-IR (Carpenter et al. 2001), it is important to be aware of common problems intrinsic to the instrument and type of observations, which might introduce errors of this order. To a large extent, many of these problems were elimi- nated by the software pipeline during reduction and calibration. However, others do persist giving rise to false detections which show up in the catalogues. Although they represent a small fraction of the dataset, they need to be filtered out (see Sect. 2.1.3) since only a small fraction of point source detections are expected to be variable.

Figure 2.2: WFCAM coverage of the ρ Ophiuchi Cluster. The WFCAM field of view is seen overlaid on an optical image of the Ophiuchus molecular cloud (Digital Sky Survey). The dark cloud region where star formation is occurring can clearly be seen. Coordinates of each pointing are given in Table 2.1. The nomimal area covered with 4 pointings is ∼0.8 deg2_.

2.1 Near-IR photometry: WFCAM/UKIRT 21

Figure 2.3: Example of persistence (Irwin et al. 2008), an artifact that is common to IR detectors and present in the WFCAM images. The image on the left-hand-side shows a 10s H-band exposure of a bright (6th magnitude), saturated star, whereas on the image on the right-hand-side is an exposure of a different part of the sky taken 15s later which shows a false source, caused by the previous image.

One of the causes of artifacts is persistence, which occurs when the charge from a bright source is not completely removed after a detector reset, causing the appear- ance of a false source in the following exposure (see Fig. 2.3). The brightness of the persistence image depends on the source count rate, filter and number of resets and its size is typically that of the saturated part of the parental image. Another form of persistence occurs for the first frame after a filter change because the detector is exposed to unfiltered light and saturates.

Another problem caused by bright sources is that of cross-talk between the detector channels within a quadrant, which produces a sequence of spurious images in some or all the other 7 channels. These images always appear offset across the channel width by an integer multiple of 128 pixels, and the number of appearances depends on the number of counts of the bright star. These spurious sources appear in the same location relative to the bright star for all the filters and as such they show up as real entries in the catalogue. They are due to capacitive coupling between channel circuits on the detector chip, hence they appear as the derivative of the parent image. A diagram explaining this effect is shown in Fig. 2.4.

Figure 2.4: Artifacts in WFCAM images: cross-talk. Bright stars cause cross-talk between detector channels which appear in the images as the derivative of the parent image.

Bright sources also cause the appearance of a diffraction pattern, which shows up as eight spikes around the bright star (Fig. 2.5). Four diffraction spikes arise from the spider holding the telescope secondary mirror, and four from the spider holding the focal plane in the instrument which is rotated 45◦ with respect to the secondary mirror spider. Spurious sources occur in lines along these diffraction spikes, and the detection algorithm often identifies these as real sources which then are included in the catalogue (see Sect. 2.1.3). Furthermore, stars which fall in the spikes or in the surrounding area are likely not to have reliable photometry.

Finally, cosmic rays that hit the detector during the observations will also appear in the detections catalogues. This effect did not have to be accounted for since with time- series photometry these detections are automatically eliminated once several nights are merged, given the very low probability that such an event occurs exactly in the same detector pixels. However, it could still affect the photometry of a star for a single night, for which a final eye-inspection of the images is always desirable, and has been performed.

WFCAM observers have also reported other artifacts in the images. In some frames the counts for some channel may be offset high or low by a few counts. This effect hap- pens preferentially for shorter wavelength filters given the lower sky counts. The offset

2.1 Near-IR photometry: WFCAM/UKIRT 23

Figure 2.5: WFCAM images show a diffraction pattern for bright stars. Spurious source detection occurs in the lines along the eight spikes. In the top and bottom part of the image, the cross-talk effect can be seen, which shows up asdonnut-shape artifact.

can vary from one frame to the next which means it does not get removed by subtrac- tion of a dark frame. In the worst cases, the bias offsets confuse the object detection algorithm resulting in a large number of spurious sources in the channel. Other unde- sirable effects are the bright moon ghosts, false images which appear when the moon is shinning directly onto the field lens near the top of the telescope. The appearance of the ghosts depends on the angle of inclination of the moon relative to the optical axis. However, none of these effects has been seen in the WFCAM data for this thesis project.

Some of the sources of error presented can be eliminated, or their effects in the data minimized, during the observations, and also by performing careful maintenance of the instrumentation. However, others are intrinsic to the instrument and type of observations and can only be handled a posteriori in the catalogue. The next section will give an insight into the reduction process and also presents the criteria applied afterwards in order to remove as much as possible the spurious entries from the final source list and ensure the reliability of the dataset for this project.