Research Blog
Welcome to my Research Blog.
This is mostly meant to document what I am working on for myself, and to communicate with my colleagues. It is likely filled with errors!
This project is maintained by ndrakos
In the previous post, I went through some checks on the image SNR for our DREaM catalogs, and concluded that the background noise was about 5x too low.
I think I figured out why!
I begin with AB magnitudes returned from FSPS, which are calculated based on the Roman filter transmission curves. I then convert these values to fluxes in the usual way. To put this into galsim, I then convert to photons/cm^2/s. Bruno uses this to assign a total brightness to each galaxy. The final image is converted back into nJy using the same conversion.
My method of converting galaxy fluxes to photon density did not properly take into account the filter information. It should be:
\(\int \frac{T(\nu)f_\nu}{h \nu } d\nu = \frac{f_\nu}{h} \int \frac{T(\nu)}{\nu } d\nu\).
Therefore I was missing a factor of \(\int \frac{T(\nu)}{\nu } d\nu\) in the assigned galaxy fluxes. This is filter dependent, but corresponds to a factor of ~0.15. Assigning this correctly will increase the background when we convert back to nJy by a factor of 6 or so!
In galsim, the conversion can be calculated as:
lam = filter.wave_list
T = filter.func(lam)
correction = np.trapz(-T*lam,1/lam)