>>> We are looking to apply a digital filtering approach to a stage of our
>>> smurf based ORAC-DR pipeline, as this has proven advantageous in
>>> related experiments with IDL. We're wondering if there are one
>>> dimensional convolution routines available in Starlink that do not use
>>> FFT. If so, any documentation and examples would be most helpful.
>>
>> Are you looking for a library or an actual application? And what kind
>> of filter do you have in mind?
>>
>> There are boxcar and tophat filters in smurf/libsmf. In terms of
>> applications there are BLOCK and GAUSMOOTH in KAPPA for example.
>
> Our main goals are baseline removal, and background atmospheric noise
> reduction in FTS based interferograms.
Hello,
For baseline removal, one application is KAPPA MFITTREND. If you take a
look at the REDUCE_FTS_ZPD ORAC-DR recipe, and the
_BASELINE_INTERFEROGRAM_ primitive which it uses, you can see how I
made use of that application. It might be possible for you to use that
primitive directly in your recipe if it does what you want -- or
generalize it if necessary, for example by adding an order parameter.
For noise reduction do you mean smoothing? KAPPA BLOCK could be used for
that. REDUCE_FTS_ZPD uses another primitive _FIT_INTERFEROGRAM_PEAK_
which uses BLOCK to smooth the interferogram to get the initial fit, so
you could take a look at that primitive for an example of how to invoke
BLOCK along the 3rd axis.
> The IDL CONVOLV function has proven to be very effective in removing the
> baseline of the interferograms in the FTS pipeline, by flattening their
> low frequency curvatures. It also effectively reduces background noise.
I don't know what that IDL function is, but what kind of filter were you
convolving with the interferograms? Do you have any example data, or
plots of the interferograms before and after applying your IDL code?
Best regards,
Graham
|