# Velocity Linetype

As mentioned in Section 5.4, many Miriad tasks allow a range of channels to be given by their velocities (rather than their channel numbers). Because these channels often do not correspond to any channels that the correlator might produce, they will be call velocity channels''. Velocity channels are specified using the line parameter. The velocities are given in km/s, in the radio or optical definition, with respect to the rest frame of the dataset. The main reason for velocity channels' is that they are corrected for the Earth's motion - a particular source velocity component will remain in the same velocity channel regardless of whether the observatory Doppler tracks or not.

The velocity specification is given in the form

line=velocity,nchan,start,width,step
or
line=felocity,nchan,start,width,step
where nchan is the number of velocity channels to select, start is the centre velocity of the first channel selected, width is the width of each channel, and step is the velocity step between channels. With the velocity form, the radio velocity definition is assumed, whereas the felocity form uses the optical definition. Note, however, both forms produce channels at equal increments in radio velocity (even the felocity linetype). See Section 5.4 for more information.

For example

line=velocity,10,1.5,1.0,3.0
will specify 10 channels, having velocities (radio definition) centred at 1.5, 4.5, 7.5, etc, km/s, with respect to the rest frame. Each channel has a width of 1 km/s.

If the start, width and step values cause the velocity channels not to map directly to the underlying correlator channels, a weighted sum of correlator channels is used to determine a velocity channel. If the magnitude of the velocity channel width (the width value in km/s) corresponds to correlator channel increment, i.e. if

(where is the correlator channel increment) then the weighted sum is equivalent to linear interpolation.

For telescopes that do not Doppler track, such as the ATCA, it will rarely be the case that velocity channels map directly to correlator channels because of the continual change in the observatory's radial velocity.

There are some caveats when using velocity channels:

• The rms noise in each velocity channel will vary if there is not a one-to-one mapping between these and correlator channels. Also there may be some noise correlation between consecutive channels. The theoretical noise given by Miriad will generally be an overestimate. This is worst for a long synthesis when the channel resolution is 0.5 km/sor finer. In this case, the true theoretical noise will be, on average, a factor of less than that suggested by Miriad.
• The `weighted sum'' method of determining a velocity channel is inferior to more sophisticated interpolation schemes (e.g. the FFT based interpolation in AIPS CVEL).