Hello,
I've finished reprocessing my spectra absolutizing them relative to '7 cep', disabling normalisation in ISIS.
Besides relativisation to '7 cep', I am now able to provide an absolute flux estimation, using a simple constant multiplier relying on David Boyd's spectrum from the ARAS Database "asdb_vvcep_20160911_860_flux.fit" being closest to my "asdb_vvcep_20160911_007.fits", by a common flux in the 3904-4004A region.
Here is how it works:
- 01_fluxes_lores_2016.png (14.34 KiB) Viewed 10753 times
It would be nice to have a few more checkpoints, but I don't have any in 2017 at this moment, and the final aim is only an attempt presenting my results in absolute flux roughly (all relative to 7 cep indeed), rather than providing precise flux measurements.
In general, shot 2 regions: 3706-3865A, and 3865-4010A.
Picking now 2 regions to represent the fluxes of continuum for them: [3754.313, 3756.745] for the lower, and [3867.818,3870.765].
In these graphs in general, the lower continuum flux from [3867-70] is multiplied by 1.5x to get to the same level visually.
Additionally, I've found a method to determine M ratio based on absorptional match, using a trivial 2-dimensional search for (shift, multiplier) in sane limits, searching for the minimum of the error. Error function: median(abs(difference spectrum)). Please be aware that the usual stddev (or least square method) won't work well, really need median (or any kind of clipping) to ignore errors.
Of course, rectified all spectra before by an order=3 polynom, subtracting to 0-level (hence the algorithm was unaware of any total flux).
In the lower UV region, restricted wavelength range for that search, while at CaII I've given the whole.
Here is the continuum fluxes result, estimating M continuum flux:
- 02_fluxes_cmp_2017.png (14.13 KiB) Viewed 10753 times
By chance, estimation on the lower UV region (black dots) is almost matching with the higher CaII's W-shape fitting.
And, here is a very important finding - M ratio before eclipse was always 1:4 in the CaII region:
With an stddev=10%, the ratio is always 1:4, even at the blue drop in 2016 ! (That's coming from the spectrum, no flux was needed.)
In other words: B and M component faded at the same time, the same ratio was kept!
That diagram contains also my CuII 3806 and NiII 3993 flux measurements, the latter one seems NOT to follow the flux drop. This is interesting, as I was expecting to follow it.
After subtracting M components (using ratio estimation from the algorithm above), continuum got straight enough to measure Balmer features in both UV regions (H8, H11), even after eclipse (where 1:1 subtraction would give senseless continuum, till deep eclipse as my last 2 observation where 100% is really the M).
So the Balmer fluxes:
As you see, Balmer was clearly reacting on the B and M flux changes linearily, hence I think in this special case, my earlier report of continuum relative EWs is making more sense before the eclipse.
Conclusions:
- '7 cep' looks stable and usable for an estimation (yet no other option for narrowband UV)
- the M ratio in 2016, 2017 has a 1:4 ratio till eclipse
- right after eclipse, the total M-ish flux has a jump (that shouldn't be from M)
- CuII 3706, NiII 3993 weak features seems to have a stable baseline, with larger errors
- both B, M may have fades at the same time, hence simple EWs for Balmer looks to be more meaningful (see my earlier report)
- Balmer and continuum changes are not strictly bound
- Peter