IRAS Explanatory Supplement
V. Data Reduction
E. Overview of Small Extended Source Data Processing
E.3 Source Construction and Hours-Confirmation
Table of Contents | Index | Previous Section | Next Section
In point source processing each pair of seconds-confirming detections was merged into a single detection. This was not the case for extended sources, nor was there any explicit check analogous to the point source hours-confirmation processing. Each individual seconds-confirmed detection was passed to a processor which attempted to piece together detections from the same band and the same hours-confirming coverage into a single source. Sources were considered to be from the same hours-confirming coverage if they were observed within 36 hours of each other. In most cases this process resulted in an hours-confirmed source in the point source processing sense, i.e., the resultant source was made up of two or more pairs of seconds-confirming detections whose detection times differed by a few hours. However, it was also possible for two or more seconds-confirming detections without a subsequent hours-confirming sighting to result in an acceptable source.
The piecing together of sources involved several stages. First, seconds-confirmed detections which lay sufficiently close to one another on the sky and which had detection times within 36 hours were linked together. If a detection failed to link with any other detection, it was rejected and not used in subsequent processing. Each group of linked detections was mapped onto a rectangular grid representing their distribution on the sky. Each pixel in the grid was a 1' square. The size of each detection was just the detector size in the cross-scan direction, and W(k) in the in-scan direction (Eq. (V.E.2)). This grid consisted of a number of cells which may be referred to by the coordinates (K, L). The number of detections, N(K,L), which overlapped each grid cell was found and each cell was assigned a weight, W(K,L), which was proportional to the sum of the intensities in each of the detections overlapping the cell. The intensity in a detection was taken to be the amplitude of the square-wave peak divided by the detector aperture area. The flux from the source seen by each of the detectors was calculated. This flux was then distributed among the grid cells overlapped by the detector according to the weights W(K,L). The total intensity associated with each grid cell was then the sum of the contributions from each overlapping detector, divided by the number of such detectors.
Having assigned an intensity to each grid cell, the cells which were regarded as contributing to the source were identified. The cell with the maximum intensity was located. If this lay close to the boundary of the grid then no source was identified. Otherwise all the cells near to the maximum intensity cell which had an intensity above 1% of the maximum were linked into a single source; if the intensity distribution had a minimum boundary and then a rise beyond the minimum, the linking process was stopped at this minimum.
After all the cells within the source were identified, a check was made to see if most of the source lay on the grid. This was done by requiring the intensity on all four boundaries of the grid to be less than 10% of the maximum intensity on the grid. If this was not the case then "source run-off" was said to have occurred. If, in addition to source run-off, the maximum intensities on either pair of opposite grid boundaries were sufficiently different, indicating the presence of a strong flux gradient, then the source was likely to be a relatively small bump on the side of a larger source, and no source was returned. If source run-off occurred, but there was no evidence of a strong flux gradient across the grid then the source was accepted but a flag was set to indicate that it was possibly a fragment of a larger source.
The flux of the source was taken to be the sum of the intensities of all the grid cells which were linked into the source, multiplied by the grid cell area. The intensity-weighted sum of the number of detections contributing to each grid cell in the source was found. This was called the confirmation level of the source, and gives an indication of how well the various detections contributing to the source were put together to form a single source. Only those sources whose confirmation level was above a threshold of 2 were passed on for further processing. In addition, a source was required to have at least one detection with a signal-to-noise ratio greater than 3.
Sources which passed the confirmation level and signal-to-noise thresholds became the input for subsequent confirmation processing, they were referred to as intermediate small extended sources, and were treated as hours-confirmed sources. The centroid and extent of each of those sources was computed the extent being characterized by the matrix of the second moments of the flux distribution. This matrix was diagonalized to yield the semi-major and semi-minor axes of the source, together with the angle between the semi-major axis and the ecliptic meridian. The number of detections in the source was also carried. A detection was counted if it contributed more than half its own flux to the source.
Table of Contents | Index | Previous Section | Next Section