IRAS Explanatory Supplement
XI. Known Processing Anomalies
D. Errors in Cross-Scan Uncertainties Related to Failed Detectors
Table of Contents | Index | Previous Section | Next Section
Several changes were made in the course of the processing to ameliorate the effects of failed detectors. While all of the changes Resulted in overall improvements, the last change introduced an error as well. The last change handled the situation when an edge detection occurred opposite a failed detector and when the fluxes of the detections differed by more than a factor of two, so that the detections failed to seconds-confirmed. Since only one detection could band-merge with the rest of the source, an improvement was made to select always the brighter detection and to delete the fainter one. An error in the software assigned a wildly incorrect cross-scan uniform uncertainty to the remaining detection one-third of the time.
This change still resulted in a net improvement in the quality of the sources. The erroneous cross-scan uncertainties were immediately eliminated if band-merging was successful. A problem was created when the faulty detection was merged with another real source that had transited the focal plane at precisely the same time. No restriction on cross-scan position existed because of the impossibly large cross-scan uniform uncertainty associated with the detection. The resulting incorrectly band-merged source failed to have upper limits from being assigned to unobserved (or unmerged) bands. Thus, the Working Survey Data Base (WSDB) contains some hours-confirmation with some bands with FSTAT = 0, and detectors all 0 (see Section V.D.8). A detailed comparison of the processing of the data with and without this error showed that no sources with signal-to-noise ratio greater than 9 were lost due to this error.
Table of Contents | Index | Previous Section | Next Section