Skip to content

Tests: Quantified Flux Mis assignment

kelockhart edited this page Jan 18, 2017 · 8 revisions

The problem

During reduction, the DRP is mis-assigning flux from the correct spaxel+spectral channel to the adjacent spaxels at +/- 32 spectral channels. On the raw frame, these spaxels with mis-assigned flux are above or below the correct spaxel (these spaxels are staggered by 32 spectral channels, hence the spectral channel mis-assignment). Also, in some cases, mis-assigned flux appears in the correct spectral channel but in an adjacent spaxel.

This has been observed in data taken in 2012 and later.

Quantifying the problem

The easiest way to quantify how much flux is being mis-assigned is to use an arc lamp frame where all lenslets are masked except for one column (a.k.a. a single-column arc; a single-column sky frame could also work, but generally the SNR is too low to properly measure the effect).

In the single-column arc, integrate the flux in an emission line in a given illuminated spaxel, over the width of the emission line. This is the properly assigned flux. Then integrate the flux in the adjacent (dark) spaxels, at +/- 32 spectral channels and at the original channel. (We assume that all flux is accounted for in these locations.) This is the mis-assigned flux. The percent of correctly assigned flux is taken as the ratio of the properly assigned flux to the sum of the properly assigned flux and the mis-assigned flux. If no flux is mis-assigned, this percentage should be roughly 100%, within the errors.

As an example, a single-column arc (Kbb/50 mas) taken in April 2015 showed that roughly 71% of the flux was properly assigned. A single-column arc (also Kbb/50 mas) taken in March 2016 (after a detector upgrade) showed that roughly 78% of the flux was properly assigned, using the same emission line as for the April 2015 calculation.

Example spaxels (illuminated plus the dark spaxels on either side) in a single column arc lamp from April 2015:

The code

A short IDL script, misflux_single.pro, is provided to run the above test. Provide the reduced single-column arc cube and the refchannel, or the spectral channel at the center of the emission line to be tested. An optional keyword, /plotflux, can be turned on to show how the routine is integrating the properly assigned flux.