Lya / Ha flux ratios

[jacopo-chevallard]

@YuichiHarikane noticed that Beagle predicts a broad range of Lya / Ha ratios as a function of ionization parameter and metallicity.

The table below reports the Lya/Ha ratio, and Ha/Hb ratios, for a model with constant SFH and age of 10 Myr

logU	12+log(O/H)	Lya/Ha	Ha/Hb
-3	6.71	12.31	2.95
-2	6.71	10.40	2.91
-1	6.71	4.86	2.86
-3	7.32	11.21	2.92
-2	7.32	7.04	2.88
-1	7.32	2.34	2.85
-3	7.91	8.40	2.90
-2	7.91	3.30	2.87
-1	7.91	1.03	2.84
-3	8.52	4.51	2.94
-2	8.52	1.17	2.92
-1	8.52	0.48	2.86
-3	8.93	2.65	3.06
-2	8.93	0.66	3.01
-1	8.93	0.45	2.93

We should check more carefully how CLOUDY deals with Lya, but by discussing with @scharlot and @feltre, we think that the reason is dust inside HII regions. A physically-motivated photoionization model, like the one we adopt, includes dust grains in HII regions, and the resonant scattering of Lya photons make the Lya optical depth much larger then the Ha one, and therefore Lya photons have much larger probabilities to be absorbed by dust.

This would explain why increasing logU and/or increasing metallicity would lower the Lya/Ha ratio, why leaving the Ha/Hb ratio near the case B value (Ha and Hb are not affected by resonant scattering effects).

[feltre]

FYI,

I believe that Hazy2 (part 2 of the CLOUDY manual) can help us understanding this issue.

Sec. 1.10.7 Optical depths you can see that different optical depths are computed (and accounted for I believe) for different H lines

Sec. 2.1 Line intensities in a dusty cloud "The code uses the albedo of the gas at the wavelength of the line to predict this reflected portion and the full optical depths computed on previous iterations." --> this is also why we use commad

And finally Sec. 9.2.5

H 1 4861 and H 1 1216, are the total intensities of Hb and La, as predicted by the multi-level H atom. These intensities are the results of calculations that include all collisional, radiative, and optical depth effects.

Still have to put everything into contest but it sounds like that the optical depth is computed differently for all the different lines.

[YuichiHarikane]

Hi @jacopo-chevallard and @feltre ,

Thank you for your discussion. So Lya photons are absorbed by the dust grains in the HII region, even if we set tau_Veff=0, right? Also, some Lya/Ha values are ~10-12, larger than the case B value (~8.7). Do you know the reason?

[YuichiHarikane]

And can we still use the Lya outputs (e.g., EW) from BEAGLE as the upper limits? Like Section 4 in Stark+17?

Best, Yuichi

[jacopo-chevallard]

So Lya photons are absorbed by the dust grains in the HII region, even if we set tau_Veff=0, right?

Yes, since this happens within the Stromgren sphere, while dust attenuation prescriptions (e.g. Charlot & Fall) only account for attenuation in birth clouds (after photons escaped the Stromgren sphere) and in the diffuse ISM.

Also, some Lya/Ha values are ~10-12, larger than the case B value (~8.7). Do you know the reason?

We are currently checking this !

[jacopo-chevallard]

And can we still use the Lya outputs (e.g., EW) from BEAGLE as the upper limits? Like Section 4 in Stark+17?

Yes I would say so, but beware that the EW computed automatically by Beagle (if you don't specify a SPECTRAL INDICES CONFIGURATION file) may not be accurate (we removed them from the output file in more recent versions of Beagle) since the pseudo-continuum windows are computed in an automatic way.