Training a GP on the sky lines makes intuitive sense: We know the sky background is going to have correlations in wavelength due to a sea of small emission lines, any one of which may not have significant SNR, but collectively exhibit a nuisance signal to remove.
Implementing the GP is subtle because it's the first time we'll want to isolate just the sky-exposed area, so we'll need to mask the out-of-slit region with emask
Training a GP on the sky lines makes intuitive sense: We know the sky background is going to have correlations in wavelength due to a sea of small emission lines, any one of which may not have significant SNR, but collectively exhibit a nuisance signal to remove.
Implementing the GP is subtle because it's the first time we'll want to isolate just the sky-exposed area, so we'll need to mask the out-of-slit region with
emask