Peter looked at synthetic spectra in a range of 20 K (20,000 degrees Kelvin) to 30 K. P-Cygni profiles from expansion of supernovae. Mostly the lines in the optical are Helium, He-I, He-II, and Oxygen-3 and Carbon-3. In UV differences in temperature are more clear. As we move toward higher redshifts these will be easier to observe. So spectrograph should have higher resolution and go as far as possible in the blue end.
Peter has generated synthetic spectra for SNe II and then compared the expected Planckian blackbody and says that it looks pretty good. The P-Cygni profiles are observed to conserve flux.
Did a comparison between LTE and non-LTE simulations. LTE: local thermodynamic equilibrium. In LTE the temperature determines everything using Saha equations etc. In non-LTE incident radiation and other effects are also included (much more computationally expensive). Difference apparent in the features of the spectra.
Alex Kim points out that repeated observations of the same SNe would be great because that would have the same extinction characteristics. And that brings us to Dan ....
Statistical analysis for interstellar dust (very nearby SNe). There is a known Clayton-Cardellian-Mathus law shows how the dust extinguishes as a function of wavelength. The important parameter is R_V (varies in the range of 2-6, typically ~3.1).
Dan made some fake spectra and measure the effect of the total extinction using R_V ranging from 2.9 to 3.4. With a 50 angstrom resolution spectra at 1 percent photometry (IN blue) and a wavelength range from 3000 angstroms to 2.5 microns. B and U are the crucial bands. Peter says that the other bands are important for determine extinction because they shouldn't be extinguished and serve as a known calibration. Grey dust is also an issue. With one magnitude of extinction the dust becomes easier to see. Saul asks how good a measurement of extinction can one get integrating out the other parameters. Dan says "I don't know." With 100 angstrom resolution the error bars get "puffed up". But if we can still narrow the temperature from spectral constrains then we've drastically constrained the extinction coefficient. With 1000 angstrom resolution and a 1000 resolution on the temperature we can measure the extinction to within 0.4 in R_V.
Peter says that the mean extinction for Type II SNe is 0.5 with a sigma of 0.5. Carl argues that it must be asymmetric. Peter replies that this is just a rough measure. Clearly there is asymmetry because we don't see the 5 magnitude extinguished SNe or the -0.5 magnitude extinguished SNe.
Alex Kim asks why do this? Why not just take the residual from blackbody (A_lambda) for the SNe? Peter points out that that's crude. Dan says that if we don't know the temperature that well then you can't do that. Fitting temperature using spectra is a different approach. With a thousand degree swing a 100 angstrom feature shifted by 13 percent. If you don't know the epoch you just fit a model which fits the epoch. The floating parameter here is the size of the atmosphere.
For nearby galaxies we can check out their dust properties. We also would have a great way of removing dust effects from Type II observations. For SNAP this is a great way of measuring dust change as a function of redshift. Susana mentions recent work regarding galaxy dust models as a function of redshift (reference?). So this would allow us to remove a systematic for SNAP: the understanding of the dust.
Alex Kim asks how well do you have to know your blackbody. Peter again says that's just determining A_lambda. Alex doesn't want to use CCM as a benchmark but rather solve for more general dust models. Peter says that will be impossible. Saul says just use three ranges of lambda in large bins then let the theorists predict relationships between the A_lambda in this range versus that range. Then they have to match whatever data we get. Alex Kim says the question becomes how much unlike a CCM model.
Peter says that both SNAP and the SNfactory can do some of this work. Saul says that the observations to support doing this with SNfactory will be tough. Peter argues for HST UV time and IR spectroscopy with VLT. Between 1000 and 3000 angstroms would be the good range to measure to within 1000 maybe even 500 degrees. In just optical the overall spectrum will start departing from the tail, but there are new lines that allow you to see a whole series of new lines for cooler SNe. Peter says below 10,000 degrees K to see an H-alpha bump.
Looks promising ....