We demonstrate how two simple pattern matching techniques have allowed the Far Infrared Absolute Spectrophotometer (FIRAS) aboard the Cosmic Background Explorer (COBE) satellite to help us achieve unprecedented photometric accuracy of our spectra while contending with enormous volumes of often heterogeneous data. The first is a robust averaging technique that reduces the volume of data by identifying outlying spectra that in a statistical sense do not ``belong'' to the ensemble being coadded. The second technique is a variant of the venerable CLEAN algorithm, and is used to remove the effect of cosmic ray hits on our detector. The method is an improvement over the conventional CLEAN in that the algorithm continuously updates the choice of CLEAN profile during the CLEANing process.
Astronomical Data Analysis Software and Systems I
- Pub Date: