Discrepancy between synchrotron models s4, s5 and s7 and Planck LFI 30 GHz data

[1cosmologist]

This is an effort to validate the PySM3 synchrotron models by comparing with Planck LFI 30 GHz (NPIPE Full, A and B splits) and WMAP 23 GHz (9year) maps. For this I generate synchrotron maps for models s4, s5, and s7 at 30 GHz at NSIDE=512. A color correction factor of 0.969 is multiplied to the LFI 30GHz, and a factor of 0.427 is multiplied to the WMAP K band. I also use an inverse noise variance weighted computation of the synchrotron map at 30 GHz by Jacques Delabrouille. All maps are reconvolved with a 2 degree beam. The power spectrum is computed for three cases: A. 70% galactic synchrotron mask, B. 33% mask of the northern part of the galactic mask and C. 37% mask of the southern part of galactic mask.

All The EE and BB power spectra computed from models and maps for 70% mask:

For north sky:

For south sky:

Clearly, the models show discrepancies when compared with the data. This discrepency is very much present in the multipole range important for primordial B-mode science.

[brandonshensley]

Thanks @1cosmologist, a lot to dig into here!

First question: where does the factor of 0.969 come from? I.e., can you point me to a paper?

Second question: overall, the agreement looks pretty good to me. Are you most concerned with the BB spectra in the northern sky? Have you done any map-based comparisons?

[seclark]

It would also be helpful to see the masks used for each of these. And a caution for our interpretation that ~1/2 the points in these plots are below the beam scale.

[seclark]

And let me also add my thanks and I'm eager to dig into this together at Monday's hack session!

[1cosmologist]

Thanks @1cosmologist, a lot to dig into here!

First question: where does the factor of 0.969 come from? I.e., can you point me to a paper?

Second question: overall, the agreement looks pretty good to me. Are you most concerned with the BB spectra in the northern sky? Have you done any map-based comparisons?

Hey @brandonshensley, the color correction factor is from the Planck 2013 V: LFI calibration (https://arxiv.org/abs/1505.08022), table 1. Note this bit different from the factor Jacques calculated using the LFI 30 GHz bandpass. We are sticking to the LFI team's value for average color correction factor here.

I have indeed done map based comparisons. The figure below is the relative difference of the polarised intensity s5-LFI30/LFI30.

We also look at scatter of the log(pol intensity) of s5 vs LFI30 at NSIDE=16. We would expect a slope of 1 but there is a clear tilt. The black line is slope 1, red is the actual trend line.

I think the agreement needs to be looked at bit closely, in particular in the patches of interest for CMB experiments. While the general agreement is reasonable, except in the north sky BB, are we OK with the agreement for 20 < ell < 100?

[1cosmologist]

It would also be helpful to see the masks used for each of these. And a caution for our interpretation that ~1/2 the points in these plots are below the beam scale.

Hi @seclark, here are the masks:

I agree that the beam is perhaps too large. It might be helpful to a power spectrum comparison at native resolution. It might be better to get a bandpass integrated synchrotron map for the models. I will add it to my list.

[1cosmologist]

And let me also add my thanks and I'm eager to dig into this together at Monday's hack session!

How long is the hack session planned for? It is a bit early for me, but I will try to join in 30 min or an hour later.

[seclark]

How long is the hack session planned for? It is a bit early for me, but I will try to join in 30 min or an hour later.

It's scheduled for an hour, 7-8am PT. I will have to jump at 8 but discussions could certainly continue longer if people are keen. If you're able to have some overlap in that timeframe that would be great, but I understand it's early!

[1cosmologist]

the color correction factor is from the Planck 2013 V: LFI calibration (https://arxiv.org/abs/1505.08022), table 1.

CORRECTION: https://arxiv.org/abs/1303.5066