rms measurement from mbspect

[Rhysy]

Hi,

I'm trying to recreate certain miriad routines in IDL. One of these is to measure the rms of an HI spectrum extracted via mbspect, which reports the "clipped rms", a much better estimate than just using the whole spectrum. I can't find any documentation on how this clipping is done - all miriad reports is how many out of the total number of channels are used. I've tried guessing which channels these might be, i.e. those with the lowest flux, or those with the lowest difference from the mean, but I always come back with an rms estimate that's too low.

Many thanks,

Rhys