search

tomasstolker / species

Toolkit for atmospheric characterization of directly imaged exoplanets
https://species.readthedocs.io
MIT License
22 stars 10 forks source link

BT-NextGen (or similar stellar model) with Fe/H < 0? #74

Closed wbalmer closed 7 months ago

wbalmer commented 10 months ago

Hi Tomas!

Would it be possible to add the low metal BT-NextGen grid points? Or perhaps there is a phoenix stellar grid with a range of metallicity? For instance 51 Eri A is Fe/H ~= -0.1

I can get a single spectrum and ingest it to species (e.g. here https://phoenix.astro.physik.uni-goettingen.de/?page_id=16) , but it would be nice to be able to fit for Fe/H for these hosts.

tomasstolker commented 9 months ago

Hi @wbalmer,

Thanks for the suggestion! I checked the SVO Theoretical Spectra and was reminded that it was not possible to include the metallicity I think. The metallicity grid has only several points and these are different for each log(g) and also have differences in alpha enrichment.

See for example the available grid points for Teff=10000:

btnextgen

Do you know if the grid for which you provided a link is more regularly sampled?

tomasstolker commented 9 months ago

This was actually far from the complete list of spectra at Teff=10000. It was only showing the first 10 of them...

I think that I only had selected spectra for which the alpha enhancement is solar. That is why the grid in species only has spectra for solar and super-solar [Fe/H]. For the spectra with sub-solar metallicities the alpha value are mainly 0.4 with BT-NextGen.

Perhaps it would be best if we create a second BT-NextGen grid for sub-solar metallicities? Although that would fix the alpha value to super-solar.

wbalmer commented 9 months ago

A separate grid for low metallicity, alpha=0.4 seems reasonable given the data at hand? I'll need to read up on what effect the alpha value has on the actual spectrum, I assume enhanced alpha is standard for lower metallicites given that is what is hosted on SVO? but I need to brush up on my stellar evolution and synthesis...

tomasstolker commented 8 months ago

Yes exactly, from the figure on this wikipedia page I also understood that alpha enrichment inversely correlates with metallicity.

From that website: _The diagrams are using abundance ratios that are normalised to the Sun (placing the Sun at (0,0) in the diagram) and allow to easily identify stars of the Galactic stellar high-alpha disk (historically known as the Galactic stellar thick disk) via their typically enhanced in [α/Fe] and stars of the Galactic stellar low-alpha disk (historically known as the Galactic stellar thin disk) having [α/Fe] as low as the Sun. It further allows to identify stars that are likely born at times or in environments very different to the stellar disk, typically metal-poor stars (with low [Fe/H] < -1) which likely belong to the stellar halo or accreted features._

Might indeed be good to check the impact of the alpha parameter on the spectrum. Perhaps for a typical calibration it will not matter, or the effect is mainly in the optical.

I can add the low metallicity grid! Might be in the new year, so please remind me in case I forget 😉.

wbalmer commented 8 months ago

Hi Tomas! Sorry to bother you, I just wanted to bump this to the new year since this would be super helpful to have in the context of our AF Lep b paper this month. No big rush, though, since I have preliminary analyses I can run without having the perfect absolute flux for that spectrum.

tomasstolker commented 8 months ago

Downloading the spectra from SVO was extremely slow because of the high resolution. Instead I have downloaded synphot5_pheonix-models.tar from https://archive.stsci.edu/hlsp/reference-atlases. It doesn't mention the alpha value, but it does include subsolar metallicites.

tomasstolker commented 8 months ago

It seems that the PHOENIX spectra from the REFERENCE-ATLASES have been resampled to a low resolution of several hundred. For calibrating photometry that will be fine, but you want to use this for calibrating a GRAVITY spectra, right? I will download it from SVO instead and will see how far it gets today...

wbalmer commented 8 months ago

Hi Tomas, did that download go okay?

tomasstolker commented 8 months ago

Yes I managed to download the subsolar grid! Turned out that downloading from SVO in several chunks was much faster than downloading all Teff at once. I will try find some time this afternoon to resample and upload it for species.

tomasstolker commented 7 months ago

Hi @wbalmer! I have added the subsolar metallicity grid of BT-NextGen as 'bt-nextgen-subsolar' in commit 05bf3d173a725a575da7a4d88830cde6aae12abe. These spectra all have [alpha/Fe] = 0.4 with the largest metallicity being -0.5. Solar metallicity is not included since those spectra mainly had [alpha/Fe] = 0.0.

The grid was not fully regularly sampled, so you have to be a bit careful when using/fitting spectra close to any of the missing grid points, because those fluxes are set to zero. For a best fit spectrum, it is therefore useful to check with ReadModel what the spectra at the nearest grid points look like.

tomasstolker commented 7 months ago

Hereby an example of spectra from 5000 to 7000 K: spectra_teff

tomasstolker commented 7 months ago

As function of log(g): spectra_logg

tomasstolker commented 7 months ago

And perhaps most important, as function of metallicity. It is a somewhat small effect in the K band, but the variation in slope seems in particular not negligible. spectra_feh

wbalmer commented 7 months ago

Thanks again Tomas, I really appreciate this. I've tested adding and interpolating the grid and everything works great.