we have an order 5 spline interpolation done on the spectrum to correct for residual wavelength calibration errors measured e.g. by the velocity of OH sky lines.
As discussed this type of interpolation can be damaging to the line parameters adjusted in ORCS.
A better solution would be to use the same type of FFT-based interpolation with high oversampling that is done in ORBS. This could be done
1) from the current real only spectrum, assuming perfect phase correction
2) By modifying ORBS to output a complex spectrum and use both real and imaginary parts in the interpolation and resampling.
A last option would be to incorporate the velocity correction into the model, and not attempt any interpolation at this stage.
On this line:
https://github.com/thomasorb/orcs/blob/15c0c92e00012b7e9fb95c805fccd94e1b499f4d/orcs/core.py#L497
and:
https://github.com/thomasorb/orcs/blob/15c0c92e00012b7e9fb95c805fccd94e1b499f4d/orcs/core.py#L197
we have an order 5 spline interpolation done on the spectrum to correct for residual wavelength calibration errors measured e.g. by the velocity of OH sky lines.
As discussed this type of interpolation can be damaging to the line parameters adjusted in ORCS. A better solution would be to use the same type of FFT-based interpolation with high oversampling that is done in ORBS. This could be done
1) from the current real only spectrum, assuming perfect phase correction
2) By modifying ORBS to output a complex spectrum and use both real and imaginary parts in the interpolation and resampling.
A last option would be to incorporate the velocity correction into the model, and not attempt any interpolation at this stage.