Source Fitting and Positions

• Miriad has a couple of tasks available for finding positions of sources (see also cgcurs in Chapter 17). The simplest of these is maxfit. It fits a 2-D parabola to the image intensities in a 3 by 3 pixel box centred on the brightest pixel in a specified region and reports the fitted location and intensity.

  MAXFIT
  in=ic4296.icln	Input image
  region=relpix,box(-20,-20,20,20)(1,10)	Search region
  log=ic4296.maxfit	Write results to log file

• If you would like to do something slightly more sophisticated than just find a good value for the peak and location of your source, you might like to fit a Gaussian to it with imfit. This task can also fit levels and disks.

  In this example, we fit a Gaussian to a region of an image by inputting the region in a file. The file was generated interactively by cgcurs (see Chapter 17); you delineate a polygonal region around the source of interest with a cursor. The quality of your fit can improve significantly by selecting a snug region, and/or using the clip keyword to keep pixel values with little signal (e.g. sidelobes and noise) out of the fitting process.

  IMFIT
  in=ic4296.icln	Input image
  region=@cgcurs.region	Fit region
  clip=0.1	Only use pixels greater than 0.1
  	in the fitting process
  object=gaussian	Fit a single gaussian
  spar	Unset to let task set initial guess
  fix	Unset to let everything float
  out=ic4296.model	Write residuals of fit out to a dataset
  options=residual	Specify residuals

  Task imfit can fit multiple components. In this case, you will need to set good initial estimates of the source parameters with the spar keyword, and good values for region and clip become more important.

• There is a task called gaufit for fitting Gaussians to profiles from cubes. You are on your own here.

• Tasks often report positional information in absolute image pixel coordinates, but you may often find you would like to convert to some other units. For example, a task has told you that there is something of interest at a certain pixel in your image, but you would like to know its RA, DEC, and velocity. You can use impos to convert for you. This task enables you to convert coordinates from one system to another. The keyword type specifies the input units of the specified coordinate, given by the keyword coord. The coordinate is then converted to all useful types (e.g. pixel to absolute and offset coordinate).

  Note that if you specify a coordinate in km/s then the keyword stype indicates what convention that coordinate is in (radio or optical). You can specify the spectral-axis coordinate in any of frequency and the two velocity conventions, regardless of what the header says the spectral axis is; all spectral axis conversions are done as needed. In addition, any spectral-axis coordinate is converted to each of frequency, and optical and radio convention velocities (if possible).

  See the help file for all the choices.

  IMPOS
  in=ic4296.icln	Input image
  coord=-20.3,234.5,2400.123	Input coordinate
  type=arcsec,abspix,abskms	Input coordinate types
  stype=radio	Velocity is in radio conventions