IRAS Explanatory Supplement
V. Data Reduction
D. Point Source Confirmation
D.2 Overview of Seconds-Confirmation
D.2.h In-Band Seconds-Confirmation Position Refinement
Table of Contents | Index | Previous Section | Next Section
Position refinement was performed for detection pairs as follows. The in-scan position angle and its uncertainty were obtained from those of the two detections by applying Gaussian estimation. The cross-scan position angle and its uncertainty were obtained by computing the nominal intersection of the slot extents as mapped onto the sky (i.e., including spacecraft limit cycle motion), and then including a "slot extension" safety term if the ratio of the fainter detection to the brighter was below 0.8 in the 12, 25, and 60 µm bands, and in all cases for the 100 µm band. This slot extension was obtained from a lookup table for each band which gave the amount of extension as a function of the flux ratio. The edge of the fainter-detection slot defining the net cross-scan range is the edge which was extended. For flux ratios greater than about 0.6, this extension was negligible. For a flux ratio of 0.4, the extensions in the 12, 25, 60, and 100 µm bands were 13", 13", 28", and 34", respectively. For a flux ratio of 0.2, the extensions are 20", 20", 43", and 65", respectively. In cases involving only detectors at the cross-scan boundary of the entire survey array, a slot extension to allow for the possible passage of the image outside of the detector area was added; the values used for this were 45", 89", 138", and 206" in the 12, 25, 60, and 100 µm bands, respectively.
The error due to the net remaining slot overlap was modeled as a uniformly distributed random variable from this point on. This completed the computation of the position angles and their uncertainties in the uncorrelated-error frame of reference. The correlated errors (i.e., the relatively slowly varying absolute pointing errors) were taken to be those associated with the drop-dead source. These were modeled as Gaussian random variables, but were not yet convolved with the other uncertainties because they did not enter the band-merging problem (see Section V.D.3). Instead they were merely carried along so that they could be included after the band-merging was completed.
The twist angle for the refined position was simply interpolated linearly in time from those of the two detections being processed. No uncertainty was carried for the twist angle because the reconstruction error could have been an order of magnitude larger than the requirement without significantly affecting the position information; the reconstruction appears to have been well within its requirements.
Table of Contents | Index | Previous Section | Next Section