robsharp42
commented
4 months ago Partial solution was pointed out by Chris Lidman,
there are wavelength limits set in the params_ns.json file (which I somehow missed)
It would be probably be helpful to have a way to set the dispersion to. The numbers seem to have been set to the R7000 grating by default (I had R3000 data):
{ "step": "cube_gen", "run": true, "suffix": "08", "args": { "multithread": false, "adr": true, "wmin_set": 5400.0, "wmax_set": 7000.0 }
Very happy with much of what the pipeline is doing right out of the box. Great that it's doing such a nice job with all sorts of data.
I am having issues with the wavelength solution though, and I can't find a switch or flag to change (and I did look, honest). I have data with B3000/R3000. The code is generating two nice data cubes, but the blue one has 2840 pixels and covers what I think is the correct wavelength range, while the red has 1277 pixels and is missing at least the second half of the spectrum.
There is no one simple signal answer for the correct way to represent the spectra, since the resolution is wavelength dependent, and "most of us" want a simple linear common dispersion solution. But both arms of WiFeS intrinsically have ~4k pixels sampling a spectral resolution element of 2-3 pixels. I'd sort of expect output spectra with ~3,500-4,500 pixels depending on if I wanted to preserve spectral resolution or to avoid overly resampling to lower resolution regions.
I also would like to grid all data in a common mode to a common output solution. The old pipeline allowed that (badly).
Am I just missing where to set the output ranges and dispersion keywords?