[nofitty376]

That spectrum is so noisy. How can they infer the blue fit from the (noisy) white points? The data look almost consistent with flat (no detection). And even if there is a detection, it looks like many other models could potentially fit the data...

[bigbillheck]

> How can they infer the blue fit from the (noisy) white points?

By having a detailed model, and modern probabilistic techniques:

The planet’s terminator is modelled as a plane-parallel atmosphere in hydrostatic equilibrium, with uniform chemical composition. The chemical abundances and pressure-temperature (P-T) profile are free parameters in the model. The retrieval framework follows a free chemistry approach, whereby the individual mixing ratio of each chemical species is a free parameter.... Our canonical model comprises of 22 free parameters overall: 11 corresponding to the individual mixing ra- tios of the above chemical species, 6 for the P-T profile, 4 for the clouds/hazes and 1 for the reference pressure Pref , defined as the pressure at a fixed planetary radius of 2.61 R⊕. The Bayesian inference and parameter estimation is conducted using the MultiNest nested sam- pling algorithm (Feroz et al. 2009) implemented through PyMultiNest.

(Sections 2.4 and 3.1 from https://stsci-opo.org/STScI-01HA2G716KS9YGAGVY1WBVFJ8Y.pdf)

> And even if there is a detection, it looks like many other models could potentially fit the data...

Name three.

[gus_massa]

Can I help the GP?

In the paper they analyze 3 models, "no offset", "offset" and "offsetx2". It's strange that they get better fit for CO2 and CH3 en the "offsetx2" model, but in that model the DMS disappears. So there it at least one model.

Also, they analyze common molecules like CO2, CH4, H2O, NH3 and biologically interesting molecules like CH3-S-CH3 (DMS), HCN, CH3-Cl. From the discussion in the paper it looks like the CH3- part is important, so I'd like to see a brute force search with everything that is in https://en.wikipedia.org/wiki/Atmosphere_of_Titan and has a methyl group, like CH3-CCH, CH3-CN. My Chemistry and Astronomy is no so good, so I'd like to add CH3-OH, CH3-NH2, CH3-SH, CH3-CHO and a few more from https://en.wikipedia.org/wiki/List_of_interstellar_and_circu... I removed the ones that are big or has too many oxygen (like CH3-COOH).

[nextaccountic]

what means planet's terminator?

[thenickdude]

The boundary between daytime and nighttime [0]

[0]: https://en.wikipedia.org/wiki/Terminator_(solar)

[gus_massa]

[I'm not an expert is spectroscopy, but let me guess.]

The bump at 4.3um looks real, and it seams to be an standard absorción of CO2. https://www.quora.com/Does-CO2-absorb-all-infrared-frequenci...

The bump at 1.2, 1.4 and 2.4um looks real. I found this showing a peak for CH4 at 2.325um. http://www.astrochem.org/data/CH4H2O.php

[Sorry for the sources, but I'm not an expert is spectroscopy.]

My guess is that they assumed something like

a% * CH4 + b% * CO2 + c% * H2O + others

and get the best fit for a%, b%, c%, ... using the white points. Later, using these numbers they draw the blue line.

The peak for DMS is not clear for my untrained eye, so I can't guess what they did there. (Perhaps it's just the best fit.) It would be nice to see the a graph of the blue line they guessed with DMS and a superimposed red line with and atmosphere with an alternative atmosphere where the DMS is replaced with something uninteresting (N2? H2O? More CH4? I have no idea what is uninteresting here.)

[fritzo]

Right, show a posterior distribution not some hallucinated point estimate.

[V1ndaar]

Check out the paper and not a PR piece [0].

[0]: https://stsci-opo.org/STScI-01HA2G716KS9YGAGVY1WBVFJ8Y.pdf