Discrepancies between DESI AGN/QSO Summary Catalog and EMFit Catalog

Question

Hello!

I have noticed some discrepancies between the flux measurements in the DESI DR1 AGN/QSO summary catalog and the EMFit catalog. One initial issue I noticed was that the NII6548 emission line is missing from the AGN/QSO summary catalog, while the OI6300 emission line is missing from the EMFit catalog; in order to compute complete BPT emission line ratios for AGNs, one needs to rely on both of these catalogs (for example, I use the EMFit catalog for the NII and SII BPT diagrams, and the AGN/QSO summary catalog for the OI diagram). 

This led to the main issue at the heart of this post: the emission line fluxes differ, in some cases substantially, between the AGN/QSO summary catalog and the EMFit catalog. To illustrate this, I have below a truncated list of fluxes for some sources in a sample I am working with, where the first list shows the OIII5007 flux from the AGN/QSO summary catalog and the second list shows the OIII5007 flux from the EMFit catalog: 

OIII5007 from AGN/QSO Summary Catalog:

[14.979571, 3.8091967, 82.84453, 11.1245365, 233.59866, 23.382256, …

OIII5007 from EMFIT:

[15.370550, 13.788657, 86.307097, 15.539612, 165.716087, 21.7445915, …

As you can see, while some of the differences are small, others are discrepant by a factor of ~2-3. While the flux measurements for these catalogs were derived differently, it is surprising to see such large discrepancies. And these discrepancies do lead to differences in emission line ratios (for example, the log(OIII/Hbeta) ratio is different between the EMFit and AGN/QSO catalogs). 

Is there any recommendation as to which catalog/flux measurements we should prefer between the two catalogs? And is it "safe", for example, to use the OIII, Hbeta, and Halpha from the EMFit catalog while using the OI6300 information from the AGN/QSO catalog? 

Thanks!

Answer 1

Hi - 

Thank you for the question. The primary reason for the discrepancy is the way the two algorithms handle the fitting. The emission line fitting included in the DESI DR1 AGN/QSO catalog comes from FastSpecFit – this includes a wide coverage of the spectrum and emission lines, whereas the EmFit catalog focuses only on the low-redshift [NII] and [SII] BPT lines. There is a plan to include the [OI] doublet in EmFit, which will be released with the upcoming version.

EmFit does a careful work in fitting narrow and broad components simultaneously - it includes second components for all lines, that can either be extra narrow component in a double-peaked emission lines or outflow components. For the Balmer lines, there can additionally be a broad component (meaning it can reach three components if there are both an outflow and a BL). FastSpecFit does include narrow and broad components in the Balmer lines, but it does not allow outflow/second components. 

Additionally, the definition of “broad component” is different in the two catalogs. In the DESI DR1 AGN/QSO catalog, broad lines are selected as having a FWHM > 1200 km/s, so a source with broad H-alpha component of FWHM = 1000 km/s would not be classified as a broad-line AGN by the DESI DR1 AGN/QSO catalog, but it would be in EmFit. 

As mentioned above, the discrepancy also comes from double-peaked emission-lines, where the two peaks are separated in EmFit, but not by FastSpecFit. For plotting the BPT diagrams, it is recommended to add the fluxes from these individual components. The selection of these sources from EmFit is described in Appendix A of the paper: https://ui.adsabs.harvard.edu/abs/2025ApJ...982...10P/abstract – We are working on adding the necessary FLAG to separate these sources in the EmFit catalog.

A suggested exercise is to select the emission lines with no second or broad component in the emission lines from EmFit and compare them with the DESI DR1 AGN/QSO catalog: this will give a small scatter around the 1:1 relation.

Regarding the lack of [NII]6548 emission-line in the AGN/QSO catalog - FastSpecFit does fit both the emission-lines in the [NII]6548,6584 doublet, but only the second line [NII]6584 is used in the [NII]-BPT diagram - this is the primary reason to include only this line in the DESI DR1 AGN/QSO catalog.

Regarding the [OI]6300 emission line, we are planning to add that to EmFit soon. However, given that it is a faint line compared to the other BPT lines, one might expect that the presence of an extra component should not affect it by a lot. So, in principle, one can use EmFit for other lines and [OI] from the AGN/QSO summary catalog.

Hope this helps.

Best wishes,

Raga, Mar, Stéphanie