V.D.2.f Triple-Detection Mode (Edge Detections)

IRAS Explanatory Supplement
V. Data Reduction
D. Point Source Confirmation
D.2 Overview of Seconds-Confirmation
D.2.f Triple-Detection Mode (Edge Detections)

Chapter Contents | Introduction | Authors | References
Table of Contents | Index | Previous Section | Next Section

Figure V.D.2 The confusion processing described in the text attempted to deal with the various combinations of possibly multiple sources seen on more than one detector.
larger largest

Some scan paths available to inertially fixed point source images traversed three detector slots in the same wavelength band as portrayed in Fig. V.D.2a. Only some combinations of detector slots were consistent with this possibility. When three detections were found to be confirmable on the basis of position, the slots involved were tested for compatibility with this triple-detection mode. If they were consistent, then the source position could be localized to within the small region of cross-scan overlap, and a small uniform uncertainty in the cross-scan position of the source resulted. Many sources in the catalog have at least one edge detection, resulting in correspondingly small cross-scan uncertainties.

If the slots were not consistent with an edge detection, then the situation was diagnosed as confused (see below). An example of three detections which were not consistent with the triple-detection mode is shown in Fig. V.D.2b. Whenever a legitimate triple-detection pattern occurred the possibility that two or more point source images were confused on the central detector had to be considered. Figure V.D.2c shows two point-source image paths which are close together in the scan direction but separated in the cross-scan direction, so that the central detection is a confused response to the two. The additional cross-scan coverage of the triple-detection mode increased the probability that such confusion would occur. The only thing that could be done to avoid accepting this case as a single point source was to require the edge-overlap detections to be of roughly comparable brightness, so that if they were actually two different sources, one would expect the central detector to see about twice as much flux. This would tend to cause the flux test to be failed, so that the triple-detection mode would be rejected, and confusion processing would be invoked. A faint confused source might not upset the flux test, and so relative faintness on an edge-overlap detector itself was made to be a reason for disqualification of the triple-detection mode. On the other hand if only one source were involved but it barely grazed one of the edge-overlap slots, there would be no great loss in disallowing it, as it contained little information anyway, although the detection could cause confusion again at hours-confirmation.

These considerations led to requiring the ratio of fainter to brighter edge-overlap detection to be greater than 0.01 in order to continue to entertain the possibility of a triple detection mode. Failure to pass this test caused the situation to be processed for confusion processing.

Next a flux test was applied to the brighter of the two edge-overlap detections and the full-hit detection. This was an n test on the log-flux discrepancy and had a threshold of 4.4 (i.e., reject one real source per hundred thousand). If this test failed, confusion processing was applied; otherwise parameter refinement was performed for position and flux. The fainter edge-overlap detection was not used for flux refinement, and the brighter was used only if the ratio of the fainter flux to the full-hit flux was less than 0.1, as this indicated that the brighter edge-overlap detection's slot should have intercepted essentially all of the flux. To reduce the effects of spurious weak detections, the fainter edge-overlap detection could contribute to the position refinement only by having a signal-to-noise ratio above 5 and a detection correlation coefficient above 0.96. If it qualified then the pairwise position refinement described below was performed first for pair consisting of it and the full-hit detection, so that the slot-extension logic (see below) could be activated for this pair. The brighter edge-overlap detection contributed position and flux information according to the same rules as the double-detection mode; these are discussed in the parameter refinement sections below.

Chapter Contents | Introduction | Authors | References
Table of Contents | Index | Previous Section | Next Section