Upper limits on photometric measurements?

[wbalmer]

Hi Tomas!

Does species have the capability to include upper limits from photometric non-detections? I suppose these are tricky to include in the likelihood estimation since they are typically estimated inhomogenously (3 or 5 sigma upper limits, various descriptions of noise, etc). Still, it might be useful to have some framework for doing so, in the case where the lack of the planet flux in one band might be informative (such as disequilibrium chemistry suppressing a particular wavelength range). I can think of a few directly imaged systems where this might be useful, typically close in planets (51 Eri b or HR 8799 e in the M-band, for example).

I might look into implementing this on my fork...

[tomasstolker]

Hi William, thanks for bringing up this question!

The way I did that previously (PDS 70 b, for which I included the ALMA rms as upper limit) was by using the flux_density parameter of add_object and setting the flux to zero and the uncertainty to the 1σ upper limit. That could be done for any upper limit that is given in number of sigma, so indeed the underlying assumption is Gaussian noise.

The units parameter can be used for setting the units of the flux_density if it is not W/m2/um.

Feel free to leave this issue open since it might be useful for others as well since it is not mentioned anywhere in the documentation...