Basic Calibration and Imaging

1
Basic Calibration and Imaging

• (Originally prepared by George Moellenbrock and
  Kumar Golap)

2
CalibrationCalibration Principles

• See Appendix E of Cookbook
• The general synthesis calibration problem in the
  complex visibility domain
• Vobs MfMBGDPETF Vtrue A
• Vtrue the true, ideal visibilities FT(I)
• Vobs the observed, realistic visibilities
• Each complex calibration term corresponds to an
  effect corrupting the observations e.g., for an
  antenna-based term on baseline i-j
• G Gij Gi x Gj

3
Calibration Principles (cont)
Vobs MfMBGDPETF Vtrue A

• Antenna-based terms
• F ionospheric Faraday rotation (TBD)
• T tropospheric effects (gaincal, opacity)
• E collecting area (gaincurve, mosaic)
• P parallactic angle (TBD, toolkit)
• D instrumental polarization (TBD, toolkit)
• G electronic gain (gaincal)
• B bandpass (bandpass)
• A additive noise, including RFI (TBD)
• Baseline-based terms (closure errors)
• M baseline-based gain (blcal)
• Mf baseline-based bandpass (blcal)
• A additive noise
• Right-most terms () may have direction-dependence
  , and so become part of the visibility transform

4
Calibration Principles (cont)

• Calibration is the solution to a Matrix Equation
  (Measurement Equation Hamaker, Bregman, and
  Sault 1996)
• The wavefront is a vector, after all!
• Scalar treatments instructive, sometimes
  practical, but necessarily incomplete (and not
  just for polarization!)
• Terms do not mutually commute, in general
• Factorability of terms not formally assured
• Use orthogonality, timescales, field properties
  to separate
• All terms mutually relative
• Antenna-based terms preserve closure
• Bookkeeping solutions nominally per field, per
  spectral window transfer among fields, spectral
  windows, time is required

5
Calibration Principles (cont)

• Basic Calibration approach
• Choose calibrator fields containing sources of
  known structure (usually points, if available)
• Observe calibrators on appropriate timescales at
  appropriate frequencies with sufficient
  sensitivity for relevant calibration terms
• Solve for relevant calibration terms in the
  appropriate sequence
• Apply derived calibration to science targets and
  image
• Self-calibrate, if possible/desired

6
Calibration in CASA

• The calibration solve is generic
• Vcorrected-obs J Vcorrupted-mod
• Vcorrected-obs partially corrected data
• Vcorrupted-mod partially corrupted model
• J the calibration term under consideration
  for which the data used for solving is
  appropriate
• Every calibration solve is relative to the
  current model and to whatever calibration is
  already in hand
• Solved-for term parameterized as needed

7
Basic Calibration Tasks

• setjy set the calibrator model visibilities
• gaincal solve for time-dependent generic
  antenna-based gains
• bandpass solve for channel-dependent
  (antenna-based) gains
• blcal solve for baseline-based gains
• plotcal plot calibration solutions (and flag
  them)
• fluxscale transfer flux density scale from
  primary calibrator(s)
• applycal apply calibration

8
Basic task parameters

• Dataset specification and data selection
• Solving parameters
• Apply parameters

9
Demo A simple VLA spectral line example

• NGC5921 HI with VLA
• One spectral window, total intensity only (RR and
  LL)
• Requires gain (G) and bandpass (B) calibration
  only
• Two calibrators 3C286 for flux density and
  bandpass 1445099 for time-dependent gain
• Basic Calibration equation
• Vobs BGVmod
• Plausible calibration heuristic
• 1. temporary G on 3C286 Vobs GVmod
• 2. B on 3C286, using G Vobs B(GVmod )
• 3. G on both, using B (B-1Vobs) GVmod
• 4. fluxscale on G
• 5. Apply to NGC5921 Vcor G-1B-1Vobs

10
2. Imaging

• Use clean
• clean(vis"ngc5921_src.split.ms",imagename"ngc592
  1_task",mode"velocity",alg"hogbom",niter6000,ga
  in0.1,threshold"8.0mJy" , mask"",cleanbox,nc
  han38,start"1595km/s",width1,step"-5.5km/s",im
  size256, 256,cell15.0, 15.0,stokes"I",field
  "0",spw"",weighting"briggs",rmode"norm",robust
  0.5)
• Use invert and deconvolve
• invert(vis'ngc5921_src.split.ms',imagename'ngc59
  21_invert', mode'velocity',nchan38,start'1595km
  /s',step'-5.5km/s',field'0',imsize256,256,cel
  l15.,15.,weighting'briggs',rmode'norm',robust
  0.5)
• deconvolve(imagename"ngc5921_invert.dirty",model
  "ngc5921_decon",psf"ngc5921_invert.beam",alg"hog
  bom",niter6000,gain0.1,threshold"8.0mJy")

11
The difference
12
Imaging tasks

• see Chapter 5 of Cookbook
• Invert
• deconvolve
• clean
• single-field cleaning, variety of algorithms
• mosaic
• For multi-field data, uses mosaic uv-gridder
  (uv-plane mosaicing on single image)
• feather - combine single-dish and uvMS
• widefield

13
Some useful parameters

• phasecenter
• phasecenter'5' field id 5
• phasecenter'J2000 19h30m25.6 -20d30m00.5'
• restfreq (rest frequency for velocity)
• restfreq'1.420GHz'
• 'Interactive' clean
• cleanbox'interactive'

14
Demo imaging

• NGC5921 HI with VLA
• calibrated data in the previous demo

15
Exercises (calibration)

• Verify suggested NGC5921 calibration heuristic
• Attempt alternative heuristics, e.g.,
• Vary solution intervals
• G then B (no G with B)
• B(t) only (use 1445099 flux density from above)
• Explore G vs. T heuristics with Jupiter data
• Compare results using plotcal/plotxy