Facilities > VLA > Documentation > Manuals > OPT > On-the-fly Mapping and Mosaicking

On-the-fly Mapping and Mosaicking

by Gustaaf Van Moorsel last modified Sep 21, 2017 by Lorant Sjouwerman

On the fly mapping or on the fly mosaicking (OTF, OTFM) scans are different from other scan modes as the data is actually taken while the telescope pointing moves between two points on the sky. This move is done in a linear fashion with a constant slewing speed with respect to the sky. That is, the starting and ending RA and Dec. (J2000) coordinates are taken from the OPT scan, and the telescope slews in the given stripe duration (see below) at a constant speed linearly in RA and linearly in Dec. Thus whereas the scan in east to west and west to east in principle require different absolute telescope slewing rates seen by the operator, the slewing rate on the sky is not affected by the direction of the slew. The scan duration, number of phase center steps and number of visibility dumps per integration determine the interpolation mapping grid and map sensitivity. Note that these OTF slews are using the Mercator projection of the coordinates in RA and Dec and therefore might have map deformation implications for OTF close to the (north) celestial pole.

For the remainder of this section it is assumed that the observer has consulted the guidelines given in the mosaic page, and that the phase center grid and integration time per pointing are known.

As OTF modes typically are used to cover a large area of sky in a short time, the default resource integration times may not be sufficiently short. How to set up an OTF resource is described further below.

OTFM in the OPT

To observe in OTF mode the OPT needs to know both the beginning and end points of the OTF scan in RA and Dec. (J2000). The scan ranging from the beginning to the end point is referred to as a "strip" or "stripe". Per strip these two boundary target positions should be defined in the source list (SCT) as two separate positions. The intermediate phase centers for the mapping between the beginning and end points are determined by the on-line system as described below. These intermediate points are not displayed in the OPT scan list (reports tab), only the boundary positions. Of course these boundary target points can be entered in the SCT manually, but for any larger (rectangular, hexagonal, etc) grid mosaic, by far the easiest way to do this is to generate the grid in (e.g.) RA for moving in constant Dec. using a text editor and to import this grid with the text import tools given elsewhere. Some examples are given below; end-points other than in equatorial coordinates will be converted to J2000 RA and Dec. internally by the SCT.

Scan Mode "On The Fly Mosaicking" reveals the required fields to schedule an OTF strip; see the screen shot below. These fields are not only the beginning and ending position of the strip (as defined above), but also some specific settings for defining the strip. The "number of steps" and "integrations per step" require user input. Furthermore, the values for "integration time" and "stripe duration" are derived from the resource selected and the number-of-steps-plus-one times the number-of-integrations-per-step. The "plus-one" is added to the number of steps to account for edge effects (actually an extra iteration at the start that gets flagged in the data). Note that the "stripe duration" is not the same as the "scan timing duration". The latter should include at least 10-15 seconds of overhead to position the telescopes to the beginning of the strip (i.e., to include turn around and settle time) and thus somewhat depends on the angular distance to slew between the observed strips.

Number of steps (call this variable N):
The strip is divided linearly in N-1 equal parts between the RA end-points, and similarly in Dec, and thus contains N pairs of (RA,Dec), including the end points. These coordinates will constitute the phase centers used during the strip observation and end up as the "sky target positions" (or mapping grid points) of the data taken along the strip in the visibility data set. Each of these RA,Dec coordinates (fields) can be individually imaged if desired, or selectively included in joint deconvolution algorithms.
Integrations per step (variable M):
This is the number of time intervals associated with each phase center, each RA,Dec coordinate above. If individually imaged, the total integration time per field is M*Tvis (see next) and thus a measure for the image sensitivity (but see the mosaicking page for more accurate details).
Integration time (Tvis):
This is the integration time per visibility as specified in the resource selected in the resource field.
Stripe duration (Tstrip):
The time it will take to complete this strip observation. That is including the startup integration for the strip (i.e., N+1 steps), but excluding the overall startup of slewing the telescopes to the beginning of the strip (i.e., the slew to the strip starting position). This number is calculated as Tstrip = (N+1)*M*Tvis .
Scan timing (Tscan):
The total time of the scan, which is the Stripe Duration plus the slewing overhead. The latter needs about 10-15 seconds extra overhead if it is the continuation of a previous strip pattern, e.g. a reversal of the motion at the next strip in the same area of the sky, but otherwise perhaps may need many extra minutes of slew if it is the first strip of a new mosaic. It is possible to absorb the starting slew in a normal scan on the beginning position for the appropriate duration and give it an intent that is easily recognizable for flagging but that would add to overhead. This number thus should be at least Tscan > Tstrip + 10sec. See the scan import example below.
RA direction:
Moving between two positions on the sky can be done in several ways: the shortest way, and something else. The default is to choose the shortest direction in RA (and always the shortest direction in Dec). However, the scan moving direction in RA can be explicitly selected using the radio buttons here. Of course, when going the long way instead of the short route, the on-the-sky slew rate calculation below should use the strip angle of the longer path.

The on-the-sky slew rate is determined by the strip angle (in arcmin or arcsec) and the stripe duration (in minutes or seconds), where the maximum tested slew rate in S-band currently is of the order of 10'/s (10 arcmin per second, which is the same as 10 degrees per minute): slew rate = (Change in angle)/(N*M*Tvis) .

As for standard scan modes, the data rate is determined by the resource details. That is, as determined by the amount of correlator capacity used and the dump time (Tvis). The resource summary in the RCT will show this number when creating the resource, and the reports summary page top table will also summarize it. Note that for the typically short integration times used in OTF, the data rate quickly will approach or exceed the limits set for General Observing Capability modes.

Creating an OTF resource

The main difference between a normal observing scan resource and an OTF resource is that typically the visibility dumps should occur at a rate higher than the potential smearing effects on the data. It is thus advisable to dump at least a few times while the telescopes move a primary beam on the sky, say Tvis(sec) <~ 0.1 x freq(GHz) / slew-rate(arcmin/sec), but this is dependent on the required sensitivity and science goal.

The resource can be created as described in the OPT RCT section, with the visibility integration time entered in the first tab. However, care should be taken not to exceed the current data rates (shown in the box in the RCT), which happens easily with demanding correlator setups as well as with shorter integration times.

Current constraints:

  • Phase centers (N) cannot be changed faster than at 0.5 second intervals.
  • Visibility integration (dump, Tvis) times must be at least 250 ms (which would be the integration time in the RCT).
  • Resource data rates may not exceed the limits for (Resident) Shared Risk (25 MB/s, 60 MB/s).
  • Limited slew rates in Azimuth (40 deg/min) and Elevation (20 deg/min) will pose a limit on the maximum slew rate in RA and Dec, depending on geometry (and LST) of the observation. It is best to stay on the safe side of these limits.

Other notes:

  • Slewing at high elevations should be minimized due to potential tracking and dish deformation issues close to zenith. That is, for OTF targets close to 34 deg Dec., map it well before or after transit, i.e. close to rise or set!
  • Times for dumps are in UTC (which is very close to LST for Tvis << 1 min), scan times are in LST.
  • Individual antennas may start and stop scanning a strip up to 0.5 seconds late with respect to the rest of the array. This will introduce a pointing offset of up to 0.5(sec) x slew-rate for that antenna. To limit imaging problems without flagging that antenna as bad (where it may not be known which antenna it actually is), one would want to keep that offset smaller than about a quarter of the primary beam calculated at the highest frequency in use when planning for potential slew-rates.
  • See mosaicking page for Tvis and other mosaicking hints.


Suppose we observe in X band and follow the mosaicking web page to get the sampling distances, i.e., we assume an observing frequency of 9.33 GHz and hexagonal packing at sqrt(3) which is almost the same as using the upper part of the band and sqrt(2). The primary beam is thus 4.5 arcmin (270 arcsec) and we use a strip separation of 2.6 arcmin (156 arcsec). Remember that (ΔRA in angle) is (ΔRA in time) times 15 × cos(Dec), e.g. 4m at 60d = 4m x 15'/m x cos(60d) = 30 arcmin.

The minimalistic SCT input files then may look like this (see text-file input):

1a) A 20-arcmin length, 4-row (10' wide) rectangular area scanned in RA (constant Dec) centered on (J2000) RA=12:00:00, Dec=00:00:00

row01east ; ; ; ; 11:59:20; +00:03:54; ; ; ; ;
row01west; ; ; ; 12:00:40; +00:03:54; ; ; ; ;
row02east ; ; ; ; 11:59:20; +00:01:18; ; ; ; ;
row02west; ; ; ; 12:00:40; +00:01:18; ; ; ; ;
row03east ; ; ; ; 11:59:20; -00:01:18; ; ; ; ;
row03west; ; ; ; 12:00:40; -00:01:18; ; ; ; ;
row04east ; ; ; ; 11:59:20; -00:03:54; ; ; ; ;
row04west; ; ; ; 12:00:40; -00:03:54; ; ; ; ;

1b) A 20-arcmin length, 4-row (10' wide) rectangular area scanned in RA (constant Dec) centered on (J2000) RA=12:00:00, Dec=+60:00:00

row01east ; ; ; ; 11:58:40; +60:03:54; ; ; ; ;
row01west; ; ; ; 12:01:20; +60:03:54; ; ; ; ;
row02east ; ; ; ; 11:58:40; +60:01:18; ; ; ; ;
row02west; ; ; ; 12:01:20; +60:01:18; ; ; ; ;
row03east ; ; ; ; 11:58:40; +59:58:42; ; ; ; ;
row03west; ; ; ; 12:01:20; +59:58:42; ; ; ; ;
row04east ; ; ; ; 11:58:40; +59:56:06; ; ; ; ;
row04west; ; ; ; 12:01:20; +59:56:06; ; ; ; ;

2) A 7-point (3-row, ~8' diameter) hexagonal scanned in RA (constant Dec) centered on (J2000) RA=12:00:00, Dec=00:00:00

row01east ; ; ; ; 11:59:49.6; +00:02:36; ; ; ; ;
row01west; ; ; ; 12:00:10.4; +00:02:36; ; ; ; ;
row02east ; ; ; ; 11:59:39.2; +00:00:00; ; ; ; ;
row02west; ; ; ; 12:00:20.8; +00:00:00; ; ; ; ;
row03east ; ; ; ; 11:59:49.6; -00:02:36; ; ; ; ;
row03west; ; ; ; 12:00:10.4; -00:02:36; ; ; ; ;

3) A 14-point (5-row, ~13' diameter) hexagonal scanned in RA (constant Dec) centered on (J2000) RA=12:00:00, Dec=+60:00:00

row01east ; ; ; ; 11:59:39.2; +60:05:12; ; ; ; ;
row01west; ; ; ; 12:00:20.8; +60:05:12; ; ; ; ;
row02east ; ; ; ; 11:59:18.4; +60:02:36; ; ; ; ;
row02west; ; ; ; 12:00:41.6; +60:02:36; ; ; ; ;
row03east ; ; ; ; 11:58:57.6; +60:00:00; ; ; ; ;
row03west; ; ; ; 12:01:02.4; +60:00:00; ; ; ; ;
row04east ; ; ; ; 11:59:18.4; +59:57:24; ; ; ; ;
row04west; ; ; ; 12:00:41.6; +59:57:24; ; ; ; ;
row05east ; ; ; ; 11:59:39.2; +59:54:48; ; ; ; ;
row05west; ; ; ; 12:00:20.8; +59:54:48; ; ; ; ;

Example Scheduling Block

An example of a scheduling block that can be created with a text editor and imported in the OPT may look like this (see explanation of the fields):

# example SB for OTF on W51
# rectangular OTF in S band, 8 strips, 0.5s
# define catalogs in which Sources and Resources can be found by name
# Scheduling Block details
SCHED-BLOCK; OTF_W51_S; Dynamic; 1; 2014-06-07; 16:00:00-21:00:00, ; 25.0; D; 225.0; 35.0; N; N; w=100.0,p=60.0; ;
# startup scans on calibrators
STD; Dummy; 3C286_S; S16f_otf_0.5; DUR; 1m; cw; N; N; N; N; SetAtnGain,; ;
STD;3C286;3C286_S; S16f_otf_0.5; DUR; 9m; cw; N; N; N; N; CalFlux, CalGain, CalBP; ;
# Slew to calibrator close to mapping area
# slew from calibrator to beginning point of mapping row 1
# first 4 rows of map
STD;J1925_S; J1925_S; S16f_otf_0.5; DUR; 6m; ; N; N; N; N; CalGain,; ;
STD;Setup01; row01left; S16f_otf_0.5; DUR;35s; ; N; N; N; N; ObsTgt,; ;
OTFM; row01; row01left; row01rght; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row02; row02rght; row02left; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row03; row03left; row03rght; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row04; row04rght; row04left; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
# Slew to calibrator
# slew from calibrator to beginning point of mapping row 5
# next 4 rows of map
# final calibrator scan
STD;J1925_S; J1925_S; S16f_otf_0.5; DUR;95s; ; N; N; N; N; CalGain,; ;
STD;Setup05; row05left; S16f_otf_0.5; DUR;35s; ; N; N; N; N; ObsTgt,; ;
OTFM; row05; row05left; row05rght; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row06; row06rght; row06left; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row07; row07left; row07rght; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
OTFM; row08; row08rght; row08left; S16f_otf_0.5; DUR; 144s; 32; 8; 0; ; N; N; N; N; ;
STD;J1925_S; J1925_S; S16f_otf_0.5; DUR;95s; ; N; N; N; N; CalGain,; ;