Working with MinXSS and DAXSS data?

[ianan]

Describe the feature

Given that we have a version working with Chandrayaan-2/XSM, it would be good if we could get the MinXSS-1/2 and DAXSS loaded and fitted in sunkit-spex. They are currently using OSPEX and (Py)XSPEC, so already some Python things...

This should be relatively "straightforward" as:

• The response files are standard ARF and RMF which we already can handle, i.e. https://lasp.colorado.edu/minxss/data/response-functions/
• The spectrogram are a single file per mission covering containing all the data available online in both IDL sav and ncdf formats. These have the MinXSS-1/2, not sure about DAXSS. So just need to load one file, and then extract out the correct info for the time of the flare of interest and background (for instance pre-flare).
• We have (or nearly have) a thermal model where we can vary abundances, which is crucial for X-ray spectra over this energy range with this energy resolution.

They have IDL and Python examples online showing how to load/plot/fit the data as well as a Colab notebook showing how to get the spectral data and responses to setup and run PyXSPEC.

Proposed solution

No response

[settwi]

I think we will want the CHIANTI grids to go to significantly lower energy, like 0.1 keV. I was trying to generate tables with Fiasco about a year ago, i got through the lines and stopped before the continuum. i could upload that somewhere and we could resume?

[samaloney]

I think there's few aspects to this

1. Get the data and relevant meta data into our Spectrum object
2. Get the response data into our matrix model (combine arf and rmf)

Spoke to some of the MinXSS/DAXSS team

• For initial work can just not fit below 1keV so can use what we currently have
• Need to add more abundances (Nickel ....)
• Need gain and offset model (varies in the orbit/mission)

Initial look at the data

• Spectra have 1024 channels but ARF and RMF only have 1000 (on the count side)
• Last two bin edges are repeated which gives a dE of 0 and we end up with nans
• First count bin energy is negative
• In the L1 data have fully calibrated spectral irradiance - could just fit photon model and gain + offset to that?