lundi 6 novembre 2017

BeSS 10th anniversary symposium

In 2003 took place in Oléron a professional/amateur astronomy school organized by astronomer Jean-Pierre Rozelot, the CNRS – french research organization – and AUDE association promoting the use of electronic detectors in amateur astronomy for several decades now. At that important meeting, amateurs astronomers showed what they were doing around pro/am collaboration and professional astronomers showed what their research was about and how amateur could contribute. Several project kicked off then, such as the Spectro-L yahoo discussion group and the ARAS web front-end for pro/am spectroscopy.
Coralie Neiner, astronomer from Paris Meudon observatory, explained to us what Be stars were and how important it would be to do a long term Halpha high resolution spectroscopy survey, continuing and expanding the work Christian Buil had been doing for years. This kicked off the need for a high resolution spectrograph (the one available at that time from SBIG had a limited resolving power R<3000) and the development within AUDE of the Lhires (Littrow High Resolution Spectrograph).
Later on, a Be Star Spectra database was created and hosted by Paris Meudon observatory. A FITS file format was described so amateur software could produce spectra with the right information in their header – a long way from the DAT or SPC file format that were used so far. This BeSS database is now operational for ten years and it was time, under the lead of Coralie Neiner, to celebrate the 10th anniversary and to build for new adventures – this was the purpose of the 10th BeSS symposium that took place in Meudon on october 23rd to 27th, 2017.


It would be impossible to review everything that was said during this intensive week but I’ll try to give some summary. The sumposium slides are available on-line :

A dozen people gathered for this event, couple professional astronomers (big thanks to Thomas Rivinius for joining) and some amateurs well involved in the BeSS project. Around 75 % of the spectra in BeSS were acquired or archived by one of the person in the room ! Top « score » amateurs like Christian Buil, Olivier Garde, Thierry Lemoult and myself were there. It was also international with Ken Hudson from the USA, but also Jose Ribeiro from Portugal, Hugo Van den Broeck from Belgium and Arnold de Bruin from the Netherlands. Jean-Jacques Broussat was reprensenting the beginners and gave us a nice talk on his experience with BeSS and Be stars spectra in general. Last but not least, key amateur pilars and founders of the BeSS & ARASBeAm tools were there with François Cochard and Valérie Desnoux.



The notes that I am providing here are my own and of course subject to caution as they are my interpretation of what was said there. Basically, any mistakes would be mine! :-)


The first day was a review of the Be stars and the BeSS database. Coralie Neiner gave a talk on the Be stars & the Be phenomenon. To summarize, Be stars are non supergiant B-type (actually O7 to A2) stars that show or have shown at least once emission in their spectra. Emission is usually first visible on Halpha, thus the focus on this particular spectral line in BeSS.
Around 20 % of the B stars are Be stars and at any given time, around 15 % of the Be stars are in emission phase. The emission come from a cool circum stellar disk and hot polar winds. They are fast rotators, which certainly help for the creation of the disk which is not from the star formation (ie : it is a deccretion disk, not an accretion one like for exemple for the young Herbig Ae/Be stars). The emission profile shape depends on the viewing angle as well as inhomogeneities in the disk.
Be stars have variation of several time scale which are visible in photometry and in spectroscopye (Halpha but not only). They can go from absorption to emission and back to absorption state within months or years. They also show short term variations.
One of the main question regarding Be stars is how the disk is formed, how material is ejected from the B star. In ten years, several progresses have been made but questions remains unanswered. The rapid rotation helps. Some Be stars have a very flat shape and are close to the critical velocity but still at 90 % of it so it is missing some.
Magnetism is one aspect of the ejection mechanism. Around 10 % of all hot stars are magnetic, with a field ‘Bpol’ above 50 Gauss. The magnetism seem fossil, created while the original material was collapsing into what became the star. It seems simple structured and particular strong inside the star. The fossil field remains active through the (short) life of the hot star. For Be stars, they are then considered as « Magnetic Be » stars as opposed to « Classical Be » stars.
The magnetic field is oblique and does not coincide with the stellar rotational axis. This leads to a rotational modulation of the longitudinal field toward the line of sight, photospheric variation if spots are present on the star, and the UV wind lines.
Wind going through the magnetic field eject material which get confined on the equatorial plane of the star. It gives rotational modulation of Xray emissions, photometric light curve and Halpha emission itself.
The main problem is that for such hot stars, the fiels is weak, the fast rotation broaden the line profile making the detection through spectropolarimetry difficult, there are few spectral lines visible from thosen hot photospheres and emission lines disturb the detection !
Omega Orionis is a peculiar Be stars for which magnetism was detected (Neiner, 2003) but then not anymore. Material concentrated at the junction of the star rotation equatorial & magnetic field equatorial planes could be hidden by the emission but wiped out when an outburst occures (Neiner, 2012).
Non Radial Pulsations (NRP) is another explanation. Be stars are slightly above the main sequence B type stars in the HR diagram, similar to beta Cep and Slow Pulsating B (SPB) stars, so it was logical to look for pulsations on those stars. Added to fast rotation, NRP could bring additional momentum for material ejection. Mu Centauri was studied and outburst & NRP correlation found (Rivinius, 1998). A model with quiescence phase followed by a precursor, the outburst itself and a relaxation phase was described.
COROT satellite was launched to measure stellar pulsations through very precise photometry. Exoplanet were also searched by the transit method. Several Be stars were observed (Neiner, 2012). Compared to models, observations differed and some additional mixing of the star was required to match both. In general, the convective cores of Be stars are 20 % larger and 25 % heavier than the normal B stars.
As a summary, all Be stars show some level of pulsations, excited by the kappa-machanism (depending on metallicity) but also stochastic modes as shown by g-modes visible in the photometry curve study (gravity g-mode are good to tranport angular mode and those stochastic mode are visible at time of outburst while we mainly see pressure p-mode otherwise).
Slow rotators seem to never go to Be phasis, so the rotation is a key factor. Non Radial Pulsations seem the key required help, sometimes with magnetism added. And for several Be stars, binarity does provide complexity.
A quick note to say that VV Cep is not a Be star as the disk is an accretion disk from the supergiant main star – this is not the Be phenomenon we know of.
Also, beta Cep is a special case with the bet Cep pulsating star and a Be star companion (linked but further away from interactions).


BeSS database was launched ten years ago. Professional have lot of archives and some of them have been uploaded into the database. Amateurs are mainly observing and putting their spectra in BeSS as they acquire them.
Echelle spectra are stored in BeSS with separated orders. Merged spectra are usually not recommended for line profile analysis so professional astronomers prefere to work on orders individually.
There are more than 140000 spectra/orders in BeSS, or around 21000 ‘observations’. 9000 observations were made with R~20000 spectrographs (ie : Littrow type) and 7000 with R~10000 (ex : echelle) spectrographs.
91 amateur astronomers contributed to BeSS (most of them from Europe, 50 % from France). Six observers represent 50 % of the observations, 16 for 80 % and 28 for 90 %. Most observers have 0,1-0,5m telescopes.
Two stars have more than 1000 observations : gamma Cas (this may be too much, isn’t it?) and delta Sco (due to specific monitoring during periastron). 38 stars have more than 100 observations. 81 stars with more than 50 observations.
In high resolution (R>5000), 92 % of the up to mag 6 stars have a spectrum within the expected observing period. 88 % up to mag 7 and 74 % up to mag 8. Considering some targets are not visible as too close to the Sun and that there are not enough observers in the southern hemisphere, this is not a bad score !
There were some discussions regarding the statistics. Halpha monitoring, which is the primary focus of the long term Be stars spectra monitoring, should be placed first. A 12 month roling statistics will be added to promote new observers which need time to catch up with the « all time top scores ». Also, a new statistics with integration time will be added in order to promote high SNR and long exposure observations.




Coralie Neiner presented the BeSS status and some future work. The database is first a catalog of 2256 classical Be stars, 64 Herbig Ae/Be stars and 9 other stars.
Herbig Ae/Be were added to the program several years after it started. They are young objects (like T Tauri for cooler stars) with an accretion disk. There are 16 Herbig Ae/Be of magnitude 4-8 which are not very active but 14 of mag 8-9 which would require some more active monitoring. We should take SNR=30-100 of those stars and look for an absorption on the right (red) side of the Halpha emission line ; this would be the signature of an accretion disk beeing formed. If detected, an alert through Spectro-L discussion group should then be given for frequent monitoring.
Coralie suggested to add some new categories into BeSS database and catalog :
  • around 60 magnetic Be stars
  • few wind driven Oe stars
  • around 10-20 B[e] stars (program led by Michaella Kraus)
For the later one, the B with forbidden emission lines, a group of few stars with actually different categories, SNR~30 spectra would be required. Some are mag 10 or so.
There are around 50000 spectra taken with MUSICOS spectrograph pending validation (a huge task) as well as spectra from NARVAL and ESO archives waiting to be uploaded. More validators are required – Olivier Garde, Thierry Lemoult and myself volunteered to be validator immediately (we had a training session with Valérie on this) ; Ken Hudson volunteered to join the team after he gets more experience.
In summary, while we acquired ten years of data more is required and Halpha is still the primary goal. Thomas Rivinius highlighted for exemple the work on omega CMa which was based on 40 years of photometry.
A suggestion was to define a set of 10-15 bright Be stars for which a very active monitoring (one spectrum per night, target SNR around 100) would be required. Without taking too much observing time (bright targets), this active monitoring could help in better understanding the outburst process.
SNR information will certainly be added in BeSS database for each spectrum, and then displayed on ARASBeAm.


On the second day, Thomas Rivinius presented his work on Be stars from space photometry. Based on several space missions (Kepler, COROT, SMEI, BRITE, MOST…), he focused on several Be stars and analysed large set of data. For exemple, the SMEI solar telescope scanned the sky with one strip in 90 minutes, thus for 9 years of data !
He explained that when the disk, in equilibrum, gets denser, emission from H (~1.6µm) and K (~2µm) bands come from further away region from the star than V band (0,5µm). He also highlighted that disks grow and decay both inside and out. Some significant part of ejected material fall back onto the star photosphere.
Thomas Rivinius recommended to monitor the BRITE wiki and observe the same bright targets in spectroscopy.
He also recommended to take very high SNR (>300) spectrum of Be stars when in quiescence in order to get a baseline spectrum.


We then had a serie of amateur talks. Jose Ribeiro presented the work on delta Sco 2011 campaign. I presented the status on V442 And outburst that was detected in august and closely followed up for more than two months. An advice to show V/R graph in log scale was given then.
Coralie Neiner and Thomas Rivinius told us that it is very important to catch Be star outburst as early as possible (high SNR spectra are best to ensure we capture faint emission when they begin) and that an almost 24h monitoring in high resolution is required for the first one to two weeks. Then observations can be done daily for few weeks then weekly for couple of months until outburst is done. To achieve this, we certainly need to expand the BeSS network in the US (Ken Hudson is running a practical workshop in New Mexico in february ; this could help) and in the southern hemisphere.
Valérie Desnoux made a talk on Be stars outburst. Out of the 814 classical Be stars of mag <9 in BeSS, 599 have at least two spectra. 210 show no change, 135 with some changes (decreasing emission, V/R variations…) and 254 have at least one outburst. She extracted a list of 62 « remarquable » stars to watch for.
Christian Buil presented his work on other wavelengths than Halpha. A Littrow (Lhires, L200) spectrograph with 600 gr/mm grating is very good for near IR, R~4000, work. It can also be used in the near UV (H & K calcium doublet) to study chromospheric activites but chromatism is strong ; UVES is a special spectrograph designed for near UV work.
The eShel can be modified with a Samyang 135mm F/2 lens, modified lens mechanical adaptation, UV optical fiber, high temperature (4700K) halogen lamp used for painting lightning for exemple, a cooled ASI 1600MM camera or equivalent which shows good performance fo such work. This looks like a kit to be proposed for eShel users ! Then focus can be achieved through a large spectral domain including calcium triplet and Paschen lines in the near IR and H & K in the near UV – all in one exposure ! This would be perfect with the use of mirror based telescope such as Newton or GSO 254mm F/8 aluminium telescope.
Near IR spectral domain seems interesting for Be stars (specially binaries for the calcium triplet) as well as magnetic Be stars (Paschen lines). The OI 8446 line comes from the same region than the Lyman beta line but is more transparent making it a good line to study.
H & K near UV lines are mainly interstellar on Be stars so they are not very interesting lines. Otherwise, the Balmer discontinuity (with a second discontinuity from the disk) gives information on the effective temperature of the star photoshere and the pressure/gravity [log(g)].


Vincent Robert from the IMCCE presented the NAROO scanning machine project to scan old photographic plates to get old astrometry and, thanks to GAIA improved astrometry of reference stars, improve planetary astrometry for exemple. The new machine beeing built at Paris observatory will be operated 24h/24h, 7 days a week. It will also be able to scan spectrographic plates such as the Be stars spectra recorded at OHP (4400 plates out of a total of 55-80000 plates).


On day three, Valérie Desnoux howed us how the validation process works. She developped some specific tools in VisualSpec (anyone can use them!) to check telluric lines around Halpha and upload last two spectra in BeSS for comparison.
She also gave some common mistakes :
  • constant shift in wavelength of 0.1-0.2A, due to delay with calibration acquisition
  • in general, we expect calibration error lower than 0.1A ; this could (should!) be checked by observer before submitting the spectrum in BeSS
  • date/time collision between different spectra, unobservable object
  • wrong latitude/longitude placing the new observing in the ocean :-)
  • wrong observing site (OHP typically weeks after OHP workshop!)
  • hot pixels not well removed, leading to spikes in the spectrum
  • ripples in the spectrum : can be accepted but warning given to the observer
  • intensity saturation, hard to catch but sometimes when comparing with other spectra
  • continuum level/shape : can also be accepted with some warning to the observer to improve
  • bumps in the continuum that can be real or not (real hard to catch)
  • object error (ex : filename different from object name) : where is the mistake ?
  • Signal to noise ratio too low are accepted but usually a warning and some advices to improve given to the observer
Valérie proposed a PDF to add on BeSS web site, giving advice on how to check your own spectra before submission.
This beeing said, Valérie mentionned that lot of spectra are ok and that the overall validation process ensure an excellent data consistency which is a plus for the database and its reputation among the professional astronomers community.
Valérie also showed the monthly reports which desserve to be more widely communicated. She is spending ot of time to edit those reports but they highlight the « hot stars » of the moment and are very useful tools for the observers.
Several feedback were given on BeSS as well as the ARASBeAm website (and the associated robot file). While a low resolution program was launched few years ago (mainly for faint targets and to automatically search for outbursts ; it didn’t kick off), high resolution Halpha is our top priority.


The fourth day, we had several amateur talks. Olivier Garde presented his work on eta Ori ; Ken Hudson showed his installation in New Mexico, USA ; Jean-Jacques Broussat presented how he got into Be star spectroscopy and his progress over time ; Hugo Van den Broeck showed his access to several instruments in Belgium.
Thierry Lemoult presented his work to automate his observatory and how he searched (and found) Be candidates among a large group of B stars. Coralie Neiner indicated that you can plot the Hbeta versus Halpha EW as Be stars should show off the normal B stars. Also, she encourages us to publish the results and new Be stars can then be added in BeSS with the publication referenced.


Thomas Rivinius gave an excellent talk on spectral data analysis with lot of tips. He explained how to get effective temperature and log(g) pressure from Balmer discontinuity.
For Equivalent Width (EW), the spectrum has to be normalized by taking two sections of the continuum well outside the line profile and doing a local linear renormalization. Thomas is actually doing this hundred times with different random domains ; he then take the median value of the calculated EW and the RMS scatter as the 1 sigma error margin. He mentionned that telluric lines do not have a significant impact so removing them is not critical (this can be tested by taking a ‘dry’ and a ‘wet’ spectrum and compare the measured EW).
He mentionned VarTools as a good package for time series data analysis.
SNR (Signal to Noise Ratio) is usually easy to calculate on Be stars where there is lot of continuum. Just flatten a continuum domain and divide the average value by the RMS. Be careful with the electron scattering that could cover very large portions of the wings (ex : chi Oph).
Peak separation (Huang’s law, Huang 1972) relates to disk size. This can be done on OI 8446, FeII 5169 or SiII 6347 lines.
V/R ratio can be measured from zero level, continuum (=1) level, or a profile modeling. It really doesn’t matter but the method should be explained/defined in your publications. Zeta Tau shows cyclic (ie : not periodic) variations with four cycles over 1500 days (Carciofi 2009), indicating a spiral shape of the disk. Pleione is a binary with 219 days period ; you can easily use BeSS data to fold a dynamical spectrum and plot two periods for better read out.
You can also substract spectra taken during quiescence with one taken during an active phase (exemple with Achenar between 2000 and 2006, or 66 Oph). It does show disk loss. This is why he recommends to take very high SNR (taking one spectrum in one night for exemple!) of Be stars during quiescence time.
When Be stars are in absorption, look for short term changes in the line profile due to Non Radial Pulsations ; for exemple on 31 Peg, nu Cen, omega UMA…
Doppler tomography can be applied for exemple on Hbeta, HeI5876. See Maintz PhD thesis in 2003 and Jason Grunhut work. Good targets are phi Per, 59 Cyg, FY CMa, o Pup, HD161306. Remember that in tomography, the inside of the star is the border of the ‘image’ and that the center is actually the outter space !
Orbital velocity can show short term cyclicity such as omega Ori. Look for HeI 4471 line for exemple.
And as already mentionned, he encouraged us to look for the targets observed by satellites such as BRITE, K2, TESS and observe in spectroscopy the bright ones.


Valérie Desnoux showed some automatic analysis on BeSS data. After ten years of observations, there are some gold mine waiting to be digged out. Amateur should work on those and publish the data analysis, for exemple in IBVS.


Coralie Neiner then showed some tips for data analysis too :
  • Tlusty is a good source for O or B type stars models. There are pre-calculated grids of models (OSTAR2002 and BSTAR2006) that can be downloaded from the author's website. These grids should be more than enough for our needs. Start with the file BGvispec_v2.tar available at http://nova.astro.umd.edu/Tlusty2002/tlusty-frames-BS06.html
  • v.sin(i) can be measured as first approximation by the Fourrier Transform of a photospheric line without emission. The first value of the FT gives the v.sin(i) directly ; spectrum should be scaled in km/sec of course. Measure it on several spectra to average the values.
  • Plot spectra in log(wavelength) to ensure a fix scale in km/s.


The conclusion of this symposium is that the next one should be in few years only (4-5 years max). We got plenty of « to do » actions out of the symposium and a motivated team to go out and expand the BeSS observing community.

Spectroscopy rocks, Be Stars Spectroscopy even more !