Scheduling EVN spectral line observations

This page describes special considerations for planning and scheduling a spectral line observation to be correlated at JIVE and processed in AIPS.

Considerations for Spectral Line Observations

Frequency Setup

The EVN has standard band pass filters of 16, 8, 4, 2, 0.5 and 0.125 MHz. Other filters (1 MHz, 62.5 kHz) should be available at the stations as plug-ins, but have not been used much. Bands below 2 MHz are observed in oversampling mode. The EVN MkIV data processor at JIVE is able to deal with this mode, but currently this results in a small loss (4%) of sensitivity. The current limit is an oversampling factor of 4, so 4Msample/s recording can provide down to 0.5MHz BW. The velocity range of individual line sources should be taken into account when choosing the frequency setup. Obervations are performed with fixed frequencies, and Doppler corrections are to be determined in the scheduling process. SCHED automatically calculates appropriate observing frequencies if, in the keyin file, the source velocities and the rest-frame frequencies of the lines are specified (see Spectral line observation setup in SCHED). Note that target lines should be located close to the center of the frequency band because the band edges have lower sensitivity and can have large phase offsets. In particular, an absorption-line observation needs emission/absorption-free frequency ranges bracketting the absorption lines in the same frequency band. In addition, calibration using continuum calibrators will benefit from increased bandwidth, and so the bandwidth should be as large as possible. However, the 16-MHz bands in the EVN have rather poor characteristics and should be avoided. If a line source has a wide velocity coverage, overlapping a part of the frequency (velocity) coverage between two frequency bands may be a good idea to see a part of the line components in the two bands. In that case, however, the data reduction of the individual bands should be done independently in AIPS.

Radio Frequency Interferance (RFI)

The useful available band width is limited by the RFI distribution, especially at L-band. By looking at total-power spectra of previous L-band observations, which are available on the EVN Pipeline Feedback page, an indication of the RFI environment may be found.

Spectroscopy

The number of spectral channels should be large enough to avoid artificial effects (e.g. RFI, unexpected spurs, etc.) and to divide the true lines into more than a few spectral channels. Currently the EVN MkIV data processor at JIVE uses exclusively uniform weighting. Signal sampling with 2 bits per sample is recommended for obtaining higher sensitivity in each of the spectral channels. You should specify in the schedule file that phase cal signals be turned OFF.

Fringe Finders and Bandpass Calibrators

As always, fringe finders are required for every observation in order to determine the station clock delay offsets and drift rates. As some spectral line observations will use relatively narrow band widths, special care should be taken so that the fringe finders can be detected on all baselines. In addition, the same considerations hold for bandpass calibrators. When several line sources are observed with different frequency setups in one observation, AIPS requires bandpass calibrators to be observed with the same setups, so independently for each target. Thus the fringe finders and bandpass calibrators should be carefully selected, especially in observing bands with low sensitivity. For example, at 22GHz, the following sources are recommended as bandpass calibrators: NRAO150, 0420-014, 4C39.25, 3C273B, 3C279, NRAO530, 2145+067, 3C454.3 .

Phase, Delay and Delay-Rate Calibrators

Relative offsets of delays, rates and phases among frequency bands and polarisations can be removed by fringe fitting of the continuum calibrators independently for individual IF bands and polarisations. Because only phase and rate offsets, not delay offsets, can be determined from maser sources, delay calibrators should be separately inserted. To do this, the continuum calibrators should be in the same part of the sky and strong enough to be detected in each of frequency bands within a coherence time and should be observed every hour or less. Again, to facilitate transparent processing in AIPS, each frequency setup should have its own delay calibrators.

Amplitude Calibration

The EVN telescopes regularly measure system temperatures using noise diodes. At present, some telescopes can measure the temperatures only in time gaps between observing scans. Therefore, individual oberving scans are recommended to be reasonably short (e.g. 11 min). The data are now corrected with an improved 2-bit van Vleck correction to account for the statistics of high/low bits for each IF's data stream at each station. Thus the AIPS task ACCOR shouldn't be run. It should be okay to use auto-correlations for bandplass corrections or to use the task ACFIT.

Phase-referencing Observations

This observation mode probably requires, especially for a lower frequency band, the detection of the continuum sources in a narrow band within the coherence time. In principle, it is possible to choose different bandwidths by setting different band pass filters and a single sampler rate. This is done to observe continuum and line sources with wide and narrow bands, respectively, at the same time. This is accomplished with the SCHED parameters BBFILTER and BW. However, it is not straightforward to deal with such data in AIPS. At higher frequency, the cycling time should be no longer to 2 minutes (see VLBA Scientific Memo 24 in more detail). For the EVN, the minimum cycling time may be about 90 seconds for nodding large telescopes in a source separation of 2 degrees (at elevation lower than 60 degrees), in which 1/3 of observation time is consumed for slewing telescopes.

JIVE correlator configuration for spectral line observations

The maximum output rate of the EVN correlator at JIVE is currently 6 MB/s, which allows a read-out of the correlation at full resolution with a visibility integration time of 0.25 second (or 0.125s for configurations that use half the correlator). If requested during the proposal stage and approved by the EVN PC, projects can be correlated in multiple passes to increase the total output rate by obtaining data from a subset of frequency bands or polarisations in each pass. In AIPS, such data sets should be processed independently because the time tags on the data cannot be trusted to be identical.

All these modes can have 0.25-second integration time. However, this integration time should be used only when there are scientific arguments (field of view) to do so, as this results in enormous data sets (about 7GB per hour of observation at the current maximum output rate). Usually 4-second integration time is recommended.