Sep. 20 2012 Spectral line submillimetre observations
Observations in the submillimetre wavelengths are in principle not different from those made at millimetre wavelengths. There are however, three significant differences: 1. The detectors are more challenging 2. The observatories are at higher elevations and the most significant 3. The atmosphere is more variable and more opaque
Properties of the noise: White noise The value of the signal at any moment is random, i.e. not related to signal before or after it. The noise has usually Gaussian distribution.
Properties of the noise: 1/f noise The 1/f noise spectral energy density is proportional to the reciprocal of the frequency.
Why care for the noise spectrum? Why does the noise spectrum matter? White noise integrates down, 1/f does not! When observing the signal sampling (integration) time must be kept so short that the dominating source of noise is in the white noise regime.
The signal spectral density as a function of frequency. Two weather conditions 1/f noise white noise
The signal RMS integrates down as a function of time t as t Allan plot White noise 1/f noise
The observations in (sub)mm region are done most of the time in the on off mode. In this mode a measurement of the background is subtracted from the measurement of the source. For this to work the on and off measurements have to be done during a time period when the emission from sky is approximately stable, i.e. within the time when white noise dominates. Second requirement is that the emission from the off position consists of nothing else than the source background emission, i.e. there is no additional source.
The easiest way to fulfill these requirements is to chop fast in azimuth using either mirrors near the receiver or by wobling the secondary mirror. When mirrors are used the on off switching can be done rapidly and the off position can be 10' to 20' away. When the switching is done using the secondary mirror the switching can be done only at a rate of ~2Hz or less.
Problems of beam switching/wobling. 1: The off position is at the same elevation as the source and therefore the off position rotates on the sky with the time (alt az telescope). 2: The optical path of the on and off measurements is different which produces standing waves when doing spectral line observations. Remedy is to use dual beam switching. For details see the Observational astronomy II lectures.
Baseline problems in beamswitching single dual
Baseline problems in position switching The receiver sensitivity is not constant in frequency. If the power received from the on and off positions is different (e.g. because of a different elevation) the on minus off spectrum will be offset from 0 and (usually) has ripples. Remedy: always try to have the elevation offset between the on and off positions as small as possible.
One possibility is to use two or more off positions and to optimize the off position integration times so that the average off elevation is the same as that of the on position. As the positions of the off positions relative to on change with lst the relative integration times have to be adjusted. Having more than one off position per on increases the time used for the observations significantly.
The spectrum rms T (can be calculated with the radiometer formula: System temperature Tsys (K) Integration time t (s) Detection bandwidth Beff (Hz) Constant C is 1 for frequency switched data, 2 for position and beam switched data. Beff usually ~twice the spectrometer channel width for anything fourier related
Tsys in the radiometer formula is the equivalent SSB system temperature. For DSB receivers the equivalent Tsys is ~twice the SSB Tsys. The inverse square Tsys is used as the weight when adding up spectra.
The drawback of traditional spectral line observing methods is the large time overhead. The overhead (i.e. time used for other than integrating on the source and on the off positions) is many times the dominating part of the observations. When observing a single position for a long time the situation is not bad if e'g. beam switching is used. However, when mapping a large region (n by m positions) with short integration times the overhead ( dead time ) can be very large.
The so called on the fly (OTF) mapping minimizes the overhead. In OTF the map positions are not integrated one by one but the telescope scans in az el or eg. RA DEC over the source and spectra are recorded once a second or faster. Off position can be observed at regular intervals or alternatively, the spectra recorded off the source may be used. OTF is presented in Mangum et al. 2007, A&A 474, 679
Examples of scanning patterns: Linear scan Spiral scan Hypocycloid scan
Thumb rules for OTF: Nyquist sampling: The spectra must be recorded at ~(beam FWHM)/2.4 spacing maximum in both scanning directions. To minimize the beam smearing one should sample 4.5 spectra every beam FWHM. The off measurement must be obtained while white noise dominates
The observed OTF data will be regridded off line after the observations. Different convolution functions can be used but the spatial resolution of the resulting map depends strongly on the chosen function. The most simple function is the pill box but it has undesirable properties. The sinc function is the best but it dies off slowly and is therefore calculation inefficient. A compromise is a sinc multiplied by a narrow gaussian
Convolution functions Fourier transforms of the convolution functions The spatial frequency response of a single dish telescope is effectively multiplied by the Fourier transform of the convolving function.
Convolution functions. Z is the distance to the grid position centre. Eta tells how much the data oversamples the convolution function. R=3 is FWHM
HH92 point to point 18 OTF 18 OTF C O(1 0) C O (3 2) continuum 115 GHz 329 GHz 0.85mm APEX 12m APEX 12m Amherst 15m
The temperature and the density of interstellar material varies strongly. ISM can be probed by observing atomic and molecular and transitions. The probe or trace element is chosen depending on the density and temperature of the ISM component in question.
The submillimetre spectrum of warm molecular clouds associated with high mass star formation are rich in spectral lines. Besides warm the cores in these clouds are are also dense. Dark clouds and the molecular clouds associated with low mass star formation are cold and not as dense as the high mass star formation sites. The submillimetre spectrum of these clouds is poor.