As we were visually inspecting the cube in Viewing spectra and redshifted lines, we noticed some slices with very bad noise. They will later affect our detection within the cube, so in this section let’s have a fast look at them here. We’ll start by looking at the two cubes within the downloaded FITS file:
$ astscript-fits-view a370-crop.fits
The cube on the left is the same cube we studied before.
The cube on the right (which is called
STAT) shows the variance of each voxel.
Go to slice 3195 and press “Next” to view the subsequent slices.
Initially (for the first 5 or 6 slices), the noise looks reasonable.
But as you pass slice 3206, you will see that the noise becomes very bad in both cubes.
It stays like this until about slice 3238!
As you go through the whole cube, you will notice that these slices are much more frequent in the reddest wavelengths.
These slices are affected by the emission lines from our own atmosphere! The atmosphere’s emission in these wavelengths significantly raises the background level in these slices. As a result, the Poisson noise also increases significantly (see Photon counting noise). During the data reduction, the excess background flux of each slice is removed as the Sky (or the mean of undetected pixels, see Sky value). However, the increased Poisson noise (scatter of pixel values) remains!
To see spectrum of the sky emission lines, simply put a region somewhere in the
STAT cube and generate its spectrum (as we did in Viewing spectra and redshifted lines).
You will clearly see the comb-like shape of atmospheric emission lines and can use this to know where to expect them.