Controlling Parameters

  1. cal or no cal? If cal: differential? do we have a calibration for the appropriate epoch?

Steps

To reduce one session worth of data.

ID points of failure here...

  1. Identify primary calibrator scan (and project)
  2. Read, optionally calibrate to Kelvin, and fit primary calibrator. Fit result is in counts or Kelvin, which determines the units of subsequent conversions as well.
  3. Determine a list of scans to read and fit: all CCB RaLongMap scans except those manually excluded by an excludescans option.
  4. For each scan: read; optionally calibrate to Kevlin; and fit. Failure modes: read fails; fit fails.
  5. For each scan: scale fit results to celestial calibrator units. Note, units must be consistent here between calibrator fit and target source fit (counts or kelvin)

To average results, recipe A:
  1. Compute a running boxcar noise estimate (absolute median deviation, 30 min - 60 min in length; minimum 7 points per buffer say). The criteria for this should probably be very minimal (no pointing or focus flagging, say). Do any channel averaging you want to do before the running boxcar estimate. Failure mode: not enough good points to assign an RMS to a given, good measurement.
  2. Compute a weighted error in the mean for each source, and an error in that mean from the weighted scatter.
This approach will best account for varying weather conditions

To average results, recipe B:
  1. Identify a set of scans to estimate the noise covariance matrix. These should be weak (< 15 mJy) sources, in representative weather conditions.
  2. Compute the noise covariance matrix. Invert it. store both in a self-described, retrievable way. Failure mode: inverse of covariance matrix does not exist.
  3. Fit a power law spectrum to the data, using the covariance matrix. assign error bars to the parameters (mean flux, power law index).
  4. Optionally, use relative noise weights from boxcar calculation in each of noise cov calc'n, and fit to the data. These, I think, can be different sets of weights.

Additional capabilities needed

  1. tweak data inclusion criteria in boxcar calculation
  2. easily set calibrator flux density
  3. check for known failure points
  4. proper covariance and averaging calculation

-- BrianMason - 06 Nov 2007
Topic revision: r1 - 2007-11-05, BrianMason
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback