Exposure and Overhead

by Gustaaf Van Moorsel last modified Jul 07, 2017 by Mark Claussen

Time on Source: Exposure Calculator

The VLA observer is responsible for observing all calibrators required to properly observe as well as calibrate the data after observation in the allocated telescope time. The total time a proposer requests has to include not only the time on the sources of interest, but also time spent observing calibrators, slew time between sources, and various types of setup times. The VLA Exposure Calculator is a web-based tool (https://obs.vla.nrao.edu/ect) to help observers to perform these approximate calculations.

If you have an old JAVA version of the exposure calculator, please discard it and do not use it. Please make sure you use the latest web-based version of the calculator.

For the 18A version of the exposure calculator, there may still  be some messages that appear that are bogus, in particular ones about "Below brightness temperature limit". This message also mentions something about the confusion limit. For A  configuration, even at low frequencies, confusion should not be a problem.  For D configuration, of  course, confusion could be a problem at low frequencies.   There are also times when a message appears about a setup spanning a frequency range larger than the receiver bandwidth, even when it's clear that the setup does not do that. If there are concerns about any messages from the ECT,  or if you have a question about the confusion calculation in D configuration setups, please send a message to the NRAO Helpdesk.

The Exposure Calculator performs the following three types of calculation:

  1. given bandwidth, sky frequency, image weighting, number of polarizations, and RMS noise required the ECT will return time on source
  2. given time on source, bandwidth, image weighting, number of polarizations, and sky frequency the ECT will return RMS noise
  3. given time on source, RMS noise required, image weighting, number of polarizations, and sky frequency the ECT will return bandwidth

The calculator essentially solves the image noise equation given in the Sensitivity section of the VLA Observational Status Summary (OSS).

Running the Calculator

Important:

  • The fields labeled Representative Frequency and then Bandwidth must be entered before the calculator will do anything else.
  • After entering data into a field a carriage return <cr>, a <tab>, or a mouse click outside the input field will submit the data to the calculator.
  • By hovering the mouse over some fields, a tool tip with some helpful information is shown.

The following screenshot shows the various fields, a description of which is included further below.

 

 

Description of the various fields:

  • Array Configuration (A, B, C, or D): does not affect the RMS noise, but does indicate the brightness temperature sensitivity as well as the confusion level that will be reached (see the OSS for further discussion of confusion).
  • Number of Antennas: many use 25 instead of the full total 27 antennas to allow for the contingency that not all telescopes are in working order. For RMS calculations per baseline (e.g., for calibration) enter 2 for the number of antennas. See the Calibration section in the Guide to Observing with the VLA for more information.
  • Polarization Setup: either single or dual polarization products. Typical value would be dual for Stokes I flux (density) measurements.
  • Type of Image Weighting: this is the weighting of the data in the u-v plane during imaging. This affects the RMS noise sensitivity and the beam. Natural weighting is usually used to obtain the best sensitivity and a larger, somewhat worse beam, in terms of angular resolution and sidelobes. Robust (= 0) imaging gives somewhat less sensitivity but a somewhat smaller and better beam as compared to Natural. Use Natural for detection experiments.
  • Representative Frequency: the observing frequency (not the rest frequency) that determines the observing band. The online dopset tool can be used to convert rest frequencies to on-the-sky frequencies at the VLA for certain observing dates.
  • Approximate Beam size: This is the approximate synthesized beam size which is based on the frequency, the configuration selected, and the type of weighting. If the bandwidth divided by the frequency is greater than or equal to 0.25, a range of beam sizes is shown that correspond to the range in frequencies.
  • Digital Samplers: 3-bit or 8-bit can be selected. If the bandwidth is wider than 2048 MHz, the calculator will automatically set the samplers to 3-bit, with 8-bit not selectable. Bandwidths up to 2048 MHz cause the calculator to default to 8-bit samplers, although 3-bit sampling may be selected if desired. Note that, in general, 3-bit is less sensitive to lines for a given time and requires more overhead time. For more information, see the VLA Samplers section in the OSS.
  • Elevation: the RMS noise, especially at high frequency, is worse at lower elevation mostly due to the atmospheric opacity increasing the system noise. We calculate the increase in system noise assuming an average opacity selected by season (see Average Weather below). The elevation goes into the calculation as the standard exponential of the opacity multiplied by the secant of the elevation. The 4 elevation selections for calculation purposes are:
      • zenith, elevation = 90 degrees;
      • high, elevation = 60 degrees;
      • medium, elevation = 40 degrees;
      • low, elevation = 18 degrees.
    • Note that for S-band (2–4 GHz), ground spillover results in higher system temperatures; S-band observations below 20 degrees elevation are not recommended.
  • Average Weather: this entry selects the empirical opacity based on many years of tipping data at the VLA, as discussed in VLA memo 232 and EVLA memo 143. What is actually selected is the 22 GHz opacity; opacities at other frequencies are derived from this. The average weather corresponds to 22 GHz opacities as follows (no diurnal variations in the opacity are accounted for): 
      • Summer: 22 GHz opacity = 0.158
      • Autumn: 22 GHz opacity = 0.07
      • Winter: 22 GHz opacity = 0.045
      • Spring: 22 GHz opacity = 0.091
  • Calculation Type: see the three main operation modes described in the ECT introductory section.
  • Time on Source: input or output field. This is the time on the source of interest, not including overhead for calibration, slewing, etc. Also see the description of the next field, Total Time.
  • Total Time: input or output field. This is the total time to be requested in the proposal, and includes typical additional calibration, slewing, etc., for a single target. Default overhead values, as a function of receiver band and array configuration, have been implemented into the time or noise calculations and are based on a 2-hour scheduling block. As the actual length of the block can vary widely, these values should be viewed as guidelines only. It does not specifically account for the fixed 10-13 minute startup overhead that should be added to every session. Also, as noted in the Overhead section below, the overhead is different (longer) for 3-bit observations as compared to 8-bit observations. The overhead values in the calculator are closer to the 3-bit values. Both Time on Source and Total Time, which includes the single source logistical overhead, are displayed (and can be input) in the calculator. Also, in the Overhead section, we provide guidelines on how overhead time is calculated assuming a 2-hour scheduling block. Time on Source or Total Time takes various input formats, e.g., 20s (for twenty seconds), 10m 10s (for ten minutes, ten seconds) or 2.5h (for 2.5 hours). Time inputs can also be just seconds, minutes, or hours (e.g., 800s, 75m), or 2h 5m 35s (for 2 hours, 5 minutes, 35 seconds). Spaces between units are optional.
  • Bandwidth (Frequency): the bandwidth in frequency units. For continuum use the full usable bandwidth (excluding RFI), up to 2 or 8 GHz. For spectral line observations, one can use the width of a channel or the width of many channels that define the RMS in the science goal. This can be the total anticipated width of a line or a fraction thereof. In any case, please mention and explain the bandwidth value that is used in the Technical Justification

    At lower frequencies, RFI can limit the amount of effective bandwidth for observing. The calculator does not take this into account in its calculations, so it is up the observer to insert a reasonable bandwidth for continuum observation calculations. The maximum affected bandwidth for the lower frequency bands (L through Ku-band) is given below, and also as messages in the calculator:

    • L-band (1–2 GHz):  maximum affected bandwidth 40% (i.e., 600 MHz is a reasonable effective total continuum bandwidth)
    • S-band (2–4 GHz):  maximum affected bandwidth 25% (i.e., 1500 MHz is a reasonable effective total continuum bandwidth)
    • C-band (4–8 GHz): maximum affected bandwidth 15% (i.e., up to 3.4 GHz is a reasonable effective total continuum bandwidth when using 3-bit samplers)
    • X-band (8–12 GHz): maximum affected bandwidth 15% (i.e., up to 3.4 GHz is a reasonable effective total continuum bandwidth when using 3-bit samplers)
    • Ku-band (12–18 GHz) maximum affected bandwidth 12% (i.e., up to 5.28 GHz is a reasonable effective total continuum bandwidth when using 3-bit samplers).
  • Bandwidth (Velocity): the bandwidth in velocity units (assuming the line is at redshift zero). For continuum, use the full usable bandwidth (excluding RFI) up to 2 or 8 GHz in the field above (Bandwidth Frequency) instead of this field. For line observations, one can use either the field above in frequency units or this field in velocity units (at z=0). See Bandwidth Frequency above for what to enter.
  • RMS Brightness Temperature: the conversion to RMS brightness temperature from RMS flux density depends on the size and shape of the synthesized beam. Since details of the actual (u,v)-coverage are unknown, we have to make certain reasonable assumptions about the beam shape; here we assume that the beam is Gaussian and round. This means the derived RMS brightness temperature is approximate only. More details about this conversion can be found on our mJy/beam - Kelvin conversion page.
  • Confusion Level:  Given the array configuration (i.e. synthesized beam), and the sky frequency, the confusion level is calculated.  This confusion level is displayed here.
  • HI Column Density: this feature is only for science projects targeting neutral Hydrogen and is shown in the calculator when the frequency is at L-band and < 1500 MHz. The calculation assumes a rectangular line shape of width given in the bandwidth entry and calculates an HI column density based on the RMS value.
  • Help: clicking this button brings you back to this page.
  • Save: a screen capture to a PDF file is available (the Save button at the very bottom of the calculator). This PDF can be uploaded to the proposal's Technical Justification in the Proposal Submission Tool (PST). Before uploading, please check the PDF file for any errors (which may be caused by timeout of the web based exposure calculator tool) and compare the numbers in the PDF with the text input fields in the proposal Technical Justification and possibly the Scientific Justification.

 

A Special Case: P-Band

The exposure calculator has been updated for P band (236–492 MHz) observations. Elevation and seasonal differences have been turned off for this frequency. The noise calculation for P-band, however, is done assuming reasonably high Galactic latitude (greater than 60 degrees). At low Galactic latitudes, the Galactic sky background dominates the system temperature and the exposure calculator does not take this into account. We have made some initial tests on how much time over what the exposure calculator reports is needed for low Galactic latitudes. These are rough estimates: for Galactic latitudes below 30 degrees one should increase the time request over the exposure calculator by a factor ~2; for latitudes between 30 and 60 degrees by a factor ~1.2. For observations at the Galactic Center (a special place), the time might need to be increased by a large factor ~30. We are continuing to try to improve the time estimates and noise calculations for P-band. Questions about P-band observing should be directed to the NRAO Helpdesk.

 

Overhead and Total Time

Every proposal needs to specify the total amount of time requested which includes setup scans, slewing, and observations of calibrators. Using the information supplied in the input fields, the exposure calculator first derives a Time on Source. It then uses the array configuration and the observing band to make a best estimate of the required overhead to arrive at a Total Time for a phase-referencing type of observation to reach the sensitivity requested. The sum of the Total Time with the fixed scheduling block (session) overhead, and any additional overheads described below, is the observing time request that should be specified in the proposal. N.B. The calculations of total overhead depend upon the length of the scheduling block; such calculations done later in this section and in the exposure calculator assume a two-hour block.

Types of Overhead

The main types of overhead are:

Fixed.  The start-up scan sequence needs to be at least 10 minutes (lower frequencies) or 13 minutes (when reference pointing is used) for setup scans and for slew time from the previous (unknown) pointing. Another type of fixed overhead is that there usually needs to be one flux/bandpass calibrator scan per observation. This fixed overhead obviously affects the shorter scheduling blocks (0.5 and 1 hour) more than the longer ones.

Fractional.  This is largely determined by periodic complex gain calibrator scans, slew time in between target source and complex gain calibrator, and slew time between target sources (if more than one). High frequency observations require more frequent source/calibrator sequences and therefore require more fractional overhead. They also require reference pointing scans, typically once every hour.

Thus, for any frequency observing, the overhead involves:

  • A 10- (no reference pointing) or 13- (reference pointing) minute block of setup scans at the start of the observation to ensure to get on source;
  • A flux/bandpass calibrator scan of duration 5–10 minutes, depending on its brightness and position with respect to the target fields;
  • Complex gain calibrator scans, each long enough to detect it (typically ~1 minute duration), and often enough for phase coherence, e.g., once every 30–60 minutes for the low frequencies (see the Low Frequency Strategy in the Guide to Observing with the VLA) and as fast as every 2 minutes at the higher frequencies (see below);
  • Slew time between source changes, that is, twice during a cycle time;
  • a 30 second requantizer scan whenever a scan uses a 3-bit resource different from the resource used in the previous scan (whether 3-bit or 8-bit).

For high frequency observing, in addition to all points above, we also have:

  • A 3–4 minute reference pointing scan once every hour or with each large (>20deg) angular displacement in the sky;
  • More frequent complex gain calibrator scans, from once every minute (fast switching) to once every ~10–15 minutes when relying on self-calibration (see the High Frequency Strategy in the Guide to Observing with the VLA);
  • Increased slewing overhead because of more frequent source changes.

Then there are special cases which require even more overhead, e.g., polarization calibration, or multiple bandpass calibrator scans.

Estimating Overhead

Based on the array configuration and the observing band, the Exposure calculator multiplies the Time on Source with a multiplication factor (factor in the table below) to arrive at a total time, according to the following table:

Table 8.1: Source Scan Length Total Time
A,B configurationC,D configuration
band(s) factor source scan length
factor source scan length
4,P 1.154 50 1.154 50
L,S,C,X 1.261 30 1.261 30
Ku 1.885 10 1.885 10
K,Ka,Q 2.747 2 1.885 10

These numbers were determined by creating realistic scheduling blocks with the scan length on source as shown in the table. A number of assumptions were made:

  • Scheduling block duration: 2 hours. Overhead will decrease slightly for longer scheduling blocks, but can increase substantially for shorter blocks.
  • Source scan duration as shown in the table, calibrator scan duration 1 minute.
  • 3-bit observing for the higher frequency bands, and 8-bit observing for the lower frequency bands.
  • Referenced pointing for Ku-band. If observed without using referenced pointing, 1.5 may be a better value.

Note: every case is different, and the entries in the table (and therefore the Total Time reported by the Exposure Calculator) are only guidelines. Our recommendation, especially for higher frequencies, is to determine your overhead empirically by creating a realistic test scheduling block in the OPT with the on-source time to achieve the science goal. Experiment with different LST start times and use the most reasonable total time reported by the OPT.

Data Volume

Using the total estimated observing time (on source + overhead) and the data rate which is given in the default resources in the PST or in GOST, the total data volume of the proposed observations can be computed. The VLA OSS provides more details on the data rates and limits.