next up previous contents
Next: 3.3 Expected Count Rates Up: 3.2 Point Source Sensitivity Previous: 3.2.2 Existence of a

  
3.2.3 Detection of Previously Unknown Sources Over a Field

If one wishes to survey a field for X-ray sources, one should consider the number of false sources expected in the region. If one is interested primarily in source detection (Poisson statistics case), one can obtain a rough estimate of the number of false sources by considering the number of independent detect cells in the region of interest. For example, in the inner 10' diameter region, there are approximately 10000 5 '' x 5'' detect cells. Setting the detection significance level at 10-4 would therefore produce, on average, one false source in the region. One must be careful, however, in applying this analysis to source detection algorithms which use sliding detect cells, since adjacent cells are not independent. In such cases, the false source rate may best be determined by simulations. The PROS task QPSIM may be used to simulate QPOE files containing blank HRI fields with given background rates.

In the Gaussian statistics case, there is no simple way, a priori, to estimate the number of false sources from the count rate significance threshold. In this case, simulations are necessary, and were, in fact, performed to set detection thresholds for the ROSAT HRI Level 1 software in the SASS system. It was determined there that for typical HRI exposures, detection thresholds of 2 - 3$ \sigma$ resulted in approximately 1 false source per field in each cell size used. Based on experience with the Einstein Observatory, a more conservative limit, say 3 - 4$ \sigma$ might be appropriate. Again, observers may wish to perform their own simulations if this point is crucial to their feasibility analysis.


next up previous contents
Next: 3.3 Expected Count Rates Up: 3.2 Point Source Sensitivity Previous: 3.2.2 Existence of a
rsdc@cfa.harvard.edu
1999-05-25