VLA Atmospheric Phase Interferometer

What is it?

It's the VLA's real-time atmospheric seeing monitor.

For: real-time decision making when observing with the VLA at high frequency.

Shows: phase vs time (and lots more) for a single baseline.

Hardware:

A single-baseline interferometer
Aperture = 2x 1.5-m dishes
Baseline = 300 m
Frequency = 11.7 GHz
Location: near stations E4 and E8, parallel to E arm, near the center of the Y
Source = geostationary satellite beacon
Data: phase time series showing the atmospheric phase noise.
See Radford, Reiland & Shillue, 1996, PASP, 108, 441

Data Display:

As of July 1st the VLA API is no longer accessable via telnet. You must 'ssh' to sitemon.vla.nrao.edu as the sitemon account, for example:

ssh -l sitemon sitemon.vla.nrao.edu

If you have problems or questions concerning accessing the API please send email to helpdesk@aoc.nrao.edu. Questions concerning the API itself should be sent to pvanbusk@aoc.nrao.edu.

About the Displays:

RMS Phase Time Series

Strip chart showing rms phase and saturation rms phase, over the previous 24 h
White vertical line marks the current time, which marches rightwards till midnight UT then wraps back to the LHS.
Can also display power-law slope, and crossing time... switch on these from the Properties and Inputs page, brought up using the right mouse button in the window.
RMS phase is computed over the previous 10-min of data after subtracting a quadratic baseline. Saturation RMS phase is derived from the model fit to the root phase structure function, shown in the next plots. The RMS phase should be root-two smaller than the saturation rms phase.
This plot shows you the long-term trend in the RMS phase. Usually the seeing is stable at night, and RMS phase rises rapidly after sunrise.

Phase Time Series

Strip chart showing raw phase & amplitude over the previous 24 h, measured by the interferometer.
White vertical line marks the current time, marching rightward till midnight UT then wrapping back to the LHS.
Phase wraps are caused by satellite motion.
Small-scale fluctuations around the large fringe rate are caused mostly by the troposphere. The magnitude of this phase noise provides you with a measure of the seeing stability.

Root Phase Structure Function

Quantifies the phase noise, every 10 min from the most recent 10 min of phase time series.
It is a lag spectrum, showing the power of the phase fluctuations as a function of timescale, from 1 s to 10 min.
Usually, one sees power increase toward longer timescales, up to the timescale for a parcel of troposphere to blow across the 300-m baseline (~ 60 s). Beyond that time, the power remains constant at the 'saturation rms phase' value.
The saturation rms is your best measure of seeing
Also plotted is a model fit to the data.
Parameters derived from the model fit are the saturation rms phase, powerlaw slope, and crossing time. The values are shown for the previous 24 h in the first plot, above.
Recommended Cal Time is the calibration cycle time required under present seeing conditions to achieve the user-specified residual RMS phase after calibrating at your observing wavelength. You can key in the wavelength and desired residual phase from the Properties and Inputs page, brought up using the right mouse button in the RH window.
Recommended Cal Time is calculated using the equation for t_cyc, below.

Properties and Inputs

Bring up this page using right mouse button in any window. (The Input part is only available from the RH window.)
Accepts input of wavelength and residual rms phase (when brought up from the RH window). These are used for calculating recommended calibration cycle times (Root Phase Structure Function plot) and for deriving the coherence function (Coherence Function plot; next).
No Paging switches off the 1-min display changes.
Plot -- Phase Time Series (etc) switches on/off traces from the various plots.

Coherence Function

Shows the loss of coherence that would be experienced for integration times between 10 s and 10 min and baselines between 300 m and 3 km.
It uses the previous 10 min of interferometer phase measurements, subtracts a 2nd-order polynomial to remove the long-term drift due to satellite motion, scales the phases from 11.7 GHz to the user-supplied wavelength, then vector averages the data using time intervals from 10 s to 10 min. This gives you the coherence expected on the 300-m baseline of the API. For predicting longer baselines, the phase time series is scaled by (baseline/300 m)^(measured power-law slope) for baselines to 1 km, and then by (baseline/1000 m)^(measured power-law slope / 2) for baselines beyond 1 km, before computing the vector averages. This should be good to 1 km, but gets hairy beyond that depending on behaviour of the structure function that we don't measure. The break at 1 km is due to the scale height of the turbulent layer; turbulence is 3D for scales smaller than 1 km and 2D for scales larger than 1 km.
To change the wavelength, key in a wavelength on the Properties and Inputs page, brought up using the right mouse button in the window.

Histogram of RMS Phase

Histogram showing the cumulative distribution of saturation rms phase over the previous 24 h.
Red bar highlights the present value of saturation rms phase.
In this example, the present seeing conditions are worse than for 88% of the time over the last 24 h.

Weather Info 1

Strip chart showing temperature, dew point, and air pressure (1 hectopascal = 1 mbar) over the previous 24 h

Weather Info 2

Strip chart showing wind speed and direction over the previous 24 h

Right mouse button in each window brings up display options.
Each window cycles through its displays, changing each minute.
Historical data for the last month are available; select the date of interest from the menu brought up using the right mouse button. The other windows will continue to display real-time data until you select a date in them as well.
In Historical Data Display, the structure function plot will display the fit lines from the last 10 min of the selected day, but will continue to display the present structure function since the functions themselves are not logged; probably not so useful. Likewise, the coherence function will continue to display current data because the plots are not logged. It's plausible to recompute the plots, but that'll involve some more work; probably best held over to a complete historical analysis package.

Using the Data:

Saturation rms phase is the most useful quantity characterizing the seeing. It is derived every 10 min from the most recent 10-min of phase time series. It's the rms phase measured on long timescales (> ~ 30 s).

Calibration cycle time is probably the quantity you require for observing. A recommended switching time can be derived from the saturation rms phase using the following formula.

t_cyc	= 2 t_corn (0.53 _rms / _sat)^1/
Where ...
t_cyc	= recommended calibration cycle time / s
t_corn	= corner time / s (derived by interferometer)
	= observing wavelength / cm
_rms	= residual rms phase noise after calibration / d
_sat	= saturation rms / d (derived by interferometer)
	= power-law slope of phase structure fn (derived by interferometer)

Thus, given a desire to calibrate the VLA data and achieve a residual rms phase after calibration of _rms, and given the present measured atmospheric seeing conditions, characterized by the saturation rms, corner time, and power-law slope, then you will need to cycle to the calibrator source at least once every t_cyc seconds.

You must decide what residual rms phase is acceptable. One possible criterion is the required image dynamic range, DNR ~ 81 N / _rms, where N is the number of antennas. Another possible criterion is the decrease of visibility amplitude due to lack of coherence in a given averaging time: amplitude = e^-rms/2, where _rms is the rms of the time series in radians (NB, not the saturation rms). This will be most relevant for observations employing self-calibration. For example, _rms = 10 d leads to a dynamic range limit of ~ 100, or a loss of coherence of 10%. For simple detection experiments, _rms values up to a radian are acceptable.

Links and Details:

Modified on Friday, 26-Sep-2008 12:09:05 MDT by Chris Carilli