Meteor trails in the single frames made streaks in the Atlas Images causing spurious detections in PIXPHOT which result in false point and extended sources, and photometric errors in real point and extended sources which are on or near the meteor track. Since the meteor tracks seldom appear in more than one or two frames, it is possible to blank the affected pixels in those frames and just use the clean frames for imaging and photometry. In the V2 processing, single frame detections (from FREXAS) that were identified as unconfirmed in adjacent frames by POSFRM/PFPREP ('solos') and over a threshold were identified as artifacts and the affected pixels in the single frames were blanked to remove radiation hits and 'hot' pixels. Since a meteor trail also results in sporadic detections in the single frames, a side effect of the solo processing was to remove random portions of meteor tracks, resulting in multicolor stripes in the 3-color images from the V2 processing.
For 2MAPPS V3 and the final reprocessing, it was decided to specifically blank the meteor tracks in the single frames prior to coaddition. This was accomplished by identifying linear features in the identified solos in the single frames, and blanking a strip of pixels all the way across the frame prior to image construction and source extraction. Since most bright meteor streaks leave easily identified linear features in the single frame detection solos files (which are created prior to combining the frames in the pipeline), this was an effective technique. Identified linear features were blanked from one edge of the affected frame to the other in all three bands. There are on the order of 10-20 solos/frame in most of the sky, so searching in the solos file is much more efficient than searching all the single frame detections or comparing the pixels in overlapping frames. The solos from the 3 bands could be combined to increase the number of detections defining the linear features. This technique also seredipidously removed a number of other artifacts. Many fainter meteor streaks were not identified by this technique, but most of these were too faint to cause spurious detections.
Streaks were identified by searching the solos for each frame pairwise, for each pair separated by 10 pixels or more, counting all the solos within 2 and 3 pixels of the line defined by that pair. A candidate streak was identified if there were:
8 or more solos within 2 pixels of a line
15 or more solos within 3 pixels of a line
The candidate with the highest count was identified as a streak, the
associated solos were marked as used, and the seach was continued in the
remaining solos until less than 8 solos remained. For each identified streak,
a line was blanked completely across the frame with a width computed as
follows:
width = 6*(count/8) (pixels)which results in a minimum blanking width of 6 camera pixels (12 arcsec).
Because of problems with slow moving meteors and satellites, if a streak was identified in 3 or more sequential frames, it was only blanked over the identified length to avoid blanking all the coverage for any part of the sky. This requirement could result in incomplete blanking in a small number of cases.
There were a total of 28504 survey frames with one or more streaks identified by this technique and blanked in all 3 bands. Meteor streak removal problems are addressed below.
The following plots show a typical meteor track in an Atlas Image, and the corresponding plot of the solos in frame coordinates:
meteor track in Atlas Image
meteor track in solos file-
The following image pair shows a meteor trail with detections marked with green crosses on the left, and the same image on the right when the meteor trail is blanked in a single frame in all three bands prior to coaddition.
The following image pairs show some examples of images with meteor streaks from the 2MAPPS V2 processing compared with the sucessfully blanked tracks from the 2MAPPS V3 test processing (click on image for full-size version).
Subtle tracks that can be detected by eye in 3 color images remain in many images, although most tracks bright enough to cause false detections were sucessfully removed. There are known to be a few bright tracks that failed to produce enough qualifying solo detections to trigger streak removal. Some meteors left short streaks at the edges of frames that did not meet the minimum (20 arcsec) length to be identified as a meteor streak. There were also some anomalous cases where internal reflections from very bright objects or artifacts from ladybugs crawling on the dewar window were partially blanked leaving edges in the Atlas Images.
Known anomalies remaining in the coadded images are summarized on this page.
anomolous blanking due
to off-axis bright star (Beta Pegasi)
Although this type of artifact could be removed by blanking the entire frame, a reliable means of identifying such frames automatically was not devised in time to implement it for the V3 processing.
This page last updated on Jan 3, 2003.
Gene Kopan - gene at ipac.caltech.edu