.. _Interpolate.json#/:

Interpolate
===========

The interpolate step replaces flagged values by interpolating them using "neighbouring"  samples (samples close in time and frequency). It calculates the Gaussian weighted sum over  non-flagged samples, with a sigma parameter of one timestep/one channel. The flags are removed after interpolation.  This is in particular useful in combination with averaging; by replacing flagged values before averaging, the output  visibilities will more accurately represent the true sky. This step was aimed to solve frequency structure from  flagging/averaging for the EoR experiment, but might be useful in other cases as a more accurate averaging step.  Details are published in `Offringa, Mertens and Koopmans (2018) <https://arxiv.org/abs/1901.04752>`__.


.. _Interpolate.json#/inputs:

inputs
******


.. _Interpolate.json#/inputs/type:

type
++++

:type: ``string``

:doc: Case-insensitive step type; must be 'interpolate' `.`


.. _Interpolate.json#/inputs/windowsize:

windowsize
++++++++++

:type: ``int``

:default: ``15``

:doc: Size of the window over which a value is interpolated. Should be odd `.`