The blcal task determines baseline-based time- and/or frequency-dependent gains for all baselines in the data set. Such solutions are in contrast to gaincal and bandpass solutions which are antenna-based and better constrained.

In general, solving for and applying baseline-based calibration can be a very dangerous thing to do, since such non-closing corrections can fundamentally alter the otherwise unique source structure information obtained by an interferometer. Use of blcal should be approached with great care, after all antenna-based calibration options have been exhausted, and then only on long timescales, to ensure that the solution doesn't absorb true---or reinforce false---source structure. You must be sure you have an excellent model for the source (better than the magnitude of the baseline-dependent errors). In any case, blcal will, if used, usually mark the endpoint of a calibration scheme, reinforcing the current source model, and rendering any additional antenna-based calibration (e.g., selfcal) less reliable. As such, it could be viewed as a mostly cosmetic last step in calibration.

The blcal task uses all of the same parameters as gaincal and bandpass, which the exception of gaintype and bandtype, respectively. See "Solving for Calibration" for general information about calibration solving parameters.

Controlling frequency-dependence in blcal: freqdep

The parameter freqdep controls whether or not a channel-dependent solution should be obtained. If freqdep=True, a channelized solution (like bandpass, but baseline-based) will be obtained; otherwise the solution will be unchannelized (like gaincal, but baseline-based).