b PER Spectra Velocity Results

[James Foster]

This is a continuation of the topic covered in the Spectroscopic forum in AASVO under the b PER campaign; see:
https://www.aavso.org/february-2018-b-p ... ment-58763

Thanks to Robin and Peter's input, I was able to process my nie (9) days of b PER data with heliocentric corrections applied to my Ha, Hb, & CaK data btw 4 & 15Feb18.
Before I went into Isis to calculate/apply the corrections, I made a small spreadsheet that listed each spectra set taken per day, with its mid-point
of exposure (from the velocity calculation screen of BaSS V1.98) and converted that number into decimal days as seen here:

Once this was done, I went into Isis, Misc tab, heliocentric Velocity sub-tab, and input these data with the spectra filename that the correction needed to be applied to.
After correcting these 20 spectra for heliocentric velocity correction I went back into Bass and looked at seven (7) H-Alpha spectra to compare what is happening.
Because I was mostly interested in the very bottom (cusp) of the H-Alpha absorption, I cropped the spectra to only show 6558.5-6567A for the dates btw
4 and 14 February 2018 as seen here:


I will also do this for my H-Beta and CaK data. Unfortunately, I only took one spectra set for the He5885 so I will have to get another spectra set to process this.
Now I need to find the deltas of the H-Alpha/H-Beta/CaK "dips" with the standard H-Alpha (6562.85A) wavelength to see how the C-Star affected the main A-star's
velocity as it neared periastron around 12Feb18......

James

[Peter Somogyi]

Hello James,

Your coverage is highly valuable, I am sure your rare spectra will be hunted by more of us

Your graph basically showing the known RV shifts, like I presented here already (80 km/s change within 1.5 day):
http://www.spectro-aras.com/forum/viewt ... 952#p10225

The final sum is less meaningful because the absorptions may shift even by 12 km/s within an hour at certain periods, and that is resolved by the LHires. (You didn't seem to mention total exposure times.)
In case of ISIS (I don't know BaSS), you can check a flag to keep the temporary files - then it generates @pro1, @pro2, ... I'd be more interested in these files, if you allow me to look.

To extract the eclipse and non-eclipse differences, need to align subexposures (out of eclipse time) by RVs (RV precision and phase is irrelevant for this operation), make a median of them, align it back to each exposures (all days), and subtract from every days single exposures. (That's what I did with my spectra, all out of eclipse and no change above noise...) So my principle was, to subtract the known profiles (with out of eclipse shape) with their known RV changes, uncovering the eclipse-related changes.

Cheers,
Peter

[James Foster]

Here are my H-Beta results presented in the same way as H-Alpha:

Peter, thanks for the great feedback and offer to better make sense of these data. As requested here are my exposure times for Ha & Hb:

04Feb18, Ha: 8x200sec, Hb: 6x300sec
07Feb18, Ha: 7x200sec
08Feb18, Ha: 7x200sec, Hb: 5x300sec
09Feb18, Ha: 7x300sec
10Feb18,Ha: 7x300sec, Hb: 6x420sec
12Feb18,Ha: 9x200sec, Hb: 5x300sec
14Feb18,Ha10x180sec, Hb:8x200sec
15Feb18,Hb: 7x240sec

Please note that exposures could not be constant do to inclement weather and clouds. I need to read your topic article a few dozen times before 15-50% of it sinks into my
non-analytical brain! I'll post tha CaK and email you the spectra latter today......need to sleep for an upcoming photo session I'm participating in.

James

[James Foster]

Here is a close-up around the CaH & H-Eps line in my UV spectra for b PER taken btw 07 & 15 Feb 2018:


The CaK spectral data for 13Feb18 is very jagged due to poor skies and passing clouds. Will work with References and Peter to get RV orbital velocities from these data.

James

[Peter Somogyi]

Hello James,

The CaII 3968 line is mixed together with H7. Hence I doubt anything extractable here (other than for checking the other H8+CaII 3933 lines, or increasing measurement accuracy).
You will always want to present the lower CaII 3933 feature first (and the nearby H8).
In this special case, even the Balmer is a sum of the 3 stars (and much wider than CaII), hence I'm rather curious how the professionals decomposing it.

Another - yet unseen for b Per - graph could be, to show a velocity plot of different features that you took on the same night, centered on 0-velocity (grouped and shifted by day, to allow human eye comparing velocities of different features). PlotSpectra is capable of doing that...

And not to forget, many weak features (SiII, Ti II) exist in this region with large velocities (already presented in the thread with my link above), their shift is very similar to CaII (and I'm sure professionals already using them with the eShel's wide coverage, for correlating the RVs - especially when they don't use Balmer, for the already mentioned contamination trouble).

Cheers,
Peter

[James Foster]

I've re-visited this object with re-fined H-Alpha b PER data. I did the following:
1.) Re-Aligned my 4-14Feb18 spectra, in Isis, using the H2O line near 6516.50A.
2. ) Applied Heliocentric V-C corrected all spectra.
3.) Determined new b PER Ha values based on the "dips" in the b PER spectra as shown here



Finally, made a RV spreadsheet based on (HaAPPARENT-Ha/Ha)*300,000=RV (Km/sec):


Let me know if this seems logical or if i need another more accurate RV formula.

James

[Peter Somogyi]

Hello James,

"Let me know if this seems logical or if i need another more accurate RV formula."

This is a nice attempt!
Currently there are 2 basic things to consider:
- we should never connect the measurement points, unless the curve is assumed to be continuous (indeed, seeing such mistakes in past professional articles, too - giving absolutely wrong suggestions). Not a hard rule, but here applies.
- in case of b Per, the RVs change on a timescale of minutes(!), so you mustn't sum up the exposures. For that, you need to estimate RVs of 1 single (and noisy...) exposure. This is exactly what I did here:

http://spectro-aras.com/forum/viewtopic ... 952#p10225

You may still do these measurements manually on still the noisy frames, without sophisticated math tools: remove tellurics (tons of them around H=alpha), perhaps using them to check your calib. After removed tellurics (ISIS => Tellurics tab as an example) on each single (noisy) exposures, you can use ISIS -> Compare tab, checking the Ratio box, and select 'km/s' as shift, then try manually guessing on 2 spectra at which shift you see the graph most symmetric - so the best aligned (many guess needed - based on what your eyes tell). This way you might better cope with noise (and my usual technique for checking the result of any Vrad-shift measurement).

If you think your calib was moved in time (this is why I did take calibrations per every exposure), you might try using tellurics to fix the shift (perhaps use multiple shift estimations to average out a linear shift).

If you are lazy, I personally only interested a difference-graph (again: of single exposures!), where you can apply an arbitrary constant as a shift for both Vrad and time start point (just respect the total period - see my link -, and a reference exposure frame as a Vrad fix point).

Peter

[James Foster]

To: Peter,

RE:"- we should never connect the measurement points, unless the curve is assumed to be continuous (indeed, seeing such mistakes in past professional articles, too - giving absolutely wrong suggestions). Not a hard rule, but here applies."
Yes I agree, however I'm using MS Excel 2001 and it has very limited graph functions, but I'll see if I can interpolate the values to make the graph more "curvy."

RE:"- in case of b Per, the RVs change on a timescale of minutes(!), so you mustn't sum up the exposures. For that, you need to estimate RVs of 1 single (and noisy...) exposure. This is exactly what I did here:"
Yes i see that thread and will do individual days with as many individual Ha spectra as I get reasonable Snr. Is a better formula for RV calculations?:

Vitesse ejecta = (((L1 - L0) / L0) x c) - V helio - V (The naked V is Red shift velocity in Km/Sec...so I ignore this for stellar objects)

I had this example from another Aras forum member(saved in Note without his name):
NGC6217 Vitesse ejecta = (((6401.5 - 6562.8) / 6562.8) x 299792.458) - 3.382 - 1363

The V helio would apply to the spectra Ha shift before I make the VC helio corrections in Bass.

James

[Peter Somogyi]

Hello James,

I don't know these formulas how to use (and wouldn't play with them without practical verification - the interpretation is rather nontrivial, even though correct).
2 easiest methods:
- ISIS automatic heliocentric correction ("2. General" whilst processing)
- ISIS -> "Misc" -> Heliocentric velocity
I do not know BASS how to do heliocentrics.
But, I don't expect your result would fall in the error range of heliocentric skew within that a few days, compared to the error of the (noisy) single raw frames + 80 km/s total shift within a few hours.

Peter

[Peter Somogyi]

"Vitesse ejecta = (((L1 - L0) / L0) x c) - V helio - V (The naked V is Red shift velocity in Km/Sec...so I ignore this for stellar objects)"

BTW speaking of 2 separate things:
- calculating Vhelio
- above forumula is the shift conversion from km/s to wavelength

For doing the shift, you can either use ISIS -> 5. Profile -> Shift (switch to km/s), or Heliocentric tab also easy to do.
The manual calculations can be fully skipped in the process.

- Peter