DETECTION LIMITS IN THE FAR-INFRARED

The advent of far infrared arrays will change fundamentally the means of analyzing observations in this spectral region. Sources much fainte r than traditional ''confusion Limits'' will be extracted from images by using computer algorithms similar to CLEAN or DAOPHOT. We have cond ucted numerical experiments to evaluate these techniques and show that they will permit long integrations (similar to 10,OOC sec at 60 mu m, similar to 200 sec at 100 mu m) to achieve nearly photon-background-l imited performance and hence very deep detection Limits. The dominant noise sources - photon noise, confusion by distant galaxies, and confu sion by IR cirrus - scale with nearly the same power of the telescope aperture. As a result, the integration times required to reach confusi on limits are nearly aperture-independent.