SHARC-2 350 micron observations of distant submillimeter-selected galaxies and techniques for the optimal analysis and observing of weak signals
New 350 micron data constrain accurately the thermal far-infrared spectral energy distributions (SEDs) for 12 distant submillimeter selected galaxies (SMGs). The results confirm that the linear radio to far-infrared correlation, established for local infrared galaxies, holds out to high redshifts z...
| Summary: | New 350 micron data constrain accurately the thermal far-infrared spectral energy distributions (SEDs) for 12 distant submillimeter selected galaxies (SMGs). The results confirm that the linear radio to far-infrared correlation, established for local infrared galaxies, holds out to high redshifts z ~ 1--3. The low correlation constant q ~ 2.14 is more indicative of star formation than AGN-fueled dust heating. The sample exhibits an apparent luminosity--temperature relation (L_FIR ~ T_d^2.89), possibly owing to selection effects. As a result, photometric redshifts in the radio or far-infrared may not be viable, but expressions may relate the observed quantities for current flux and volume limited SMG samples. These suggest that SED estimation may be possible, for objects similarly selected, based on a single radio or far-infrared flux measurement.
The detection of these faint objects (~10 mJy at 350 micron) from the ground is complicated by a bright (~1000 Jy) and highly variable (~10 Jy RMS in 10 minutes of integration) atmosphere with a 1/f^2 noise spectrum and by instrumental 1/f noise. To reach optimum sensitivities, a careful analysis of the data is required, and well-chosen observing strategies are helpful. The principal techniques that aid the extraction of weak signals from colored noise are presented. Close to optimal analysis is implemented effectively by the CRUSH software. Both the computing and storage requirement of the implementation scales linearly with the size of the data set, making this approach superior to the computationally expensive alternatives for handling the very large data volumes expected from future instruments. |
|---|