Transcript Observing Convection in Stellar Atmospheres

Observing Convection in
Stellar Atmospheres
John Landstreet
London, Canada
August 22, 2006
IAU Symposium 239
Introduction
• Convection reaches photosphere in most stars of
•
•
•
Te < 104 K, perhaps also in hotter stars
Directly visible in Sun as granulation
Detected in stars as microturbulence,
macroturbulence, bisector curvature, etc
Comparison of convection models with observed
spectra provides interpretation of
observations and tests of models
August 22, 2006
IAU Symposium 239
Solar granulation
• Appearance of sun with
•
•
•
good seeing reveals
granulation
Sequences of images
suggest coherent
overturning flow
Granulation ~ visible
convection cells
=> Study convection
observationally
August 22, 2006
IAU Symposium 239
Indirect detection of velocity fields
• Granulation not directly visible on (unresolved)
•
•
•
stellar surfaces
But velocity fields in photosphere affect
spectral line profiles & energy distribution, so
we may still study convection observationally
Simplest example of velocity field: stellar
rotation
Small for “cool” stars, large for “hot” stars
August 22, 2006
IAU Symposium 239
Microturbulence
• Abundance analysis allows indirect detection of small•
•
•
•
scale velocity field (excess line broadening over thermal),
required to fit weak and strong lines
Microturbulence parameter x characterizes velocity
Required for most stars with Te < 10000 K, corresponds to
convective instability
=> Microturbulence ~ convection, at least in cooler stars
Detectable even in broad-line stars – much data
August 22, 2006
IAU Symposium 239
Convection effects on line profiles
• In Sun-like flow, expect rising and descending
•
•
•
gas to have different velocities along line of sight
Different areal coverage (filling factors) and
brightness lead to different contributions to total
flux
Result: spectral lines are shifted and asymmetric
Importance of these effects depends on where in
atmosphere the lines are formed – weak lines
will be different from strong lines
August 22, 2006
IAU Symposium 239
Macroturbulence
• Most main sequence stellar line profiles can be
•
•
•
roughly modelled with Voigt profile + rotation
Line profiles of giants & supergiants more
“pointed”, with broad shallow wings
Successful model: radial-tangential
macroturbulence. On half of surface, lines
have Gaussian spread radially, on other half lines
have Gaussian spread tangentially.
One parameter: macroturbulence zRT (velocity)
August 22, 2006
IAU Symposium 239
Macroturbulence - 2
• If zRT > 0, we conclude that large-scale velocity
•
•
•
•
field exists within stellar atmosphere
=> Velocity field may be studied by modelling
spectral line shapes
Values vary systematically over cooler part of
HR diagram
Large values of macroturbulence found among
(low v sin i) main sequence A stars near Te ~
8000K
Macroturbulence drops to zero above A0V
August 22, 2006
IAU Symposium 239
Macroturbulence - 3
• Among hotter stars (Te > 10000) situation is
•
•
•
quite confusing
For hot main sequence and giant stars,
microturbulence takes various values between 0
and several km/s, but not systematically – are
these values really > 0 (e.g. Lyubimkov et al
2004)?
B and A supergiants have microturbulence of
several km/s, and macroturbulence of 15 – 20
km/s (e.g. Przybilla et al 2006)!
Is supergiant macroturbulence due to winds,
non-radial pulsations, convection, or…?
August 22, 2006
IAU Symposium 239
Radial velocities
• In convecting stars (x > 0), radial velocity of
•
•
•
•
lines observed to vary with line strength
Reflects typical velocity (average over flows) at
depth where line is formed
Difficult to study: requires very accurate lab
wavelengths, sharp lines
Not yet studied over full HR diagram
Examples: Sun, Procyon (Allende Prieto et al
2002)
August 22, 2006
IAU Symposium 239
Bisector curvature (asymmetry)
• Line asymmetry (bisector curvature) reveals
•
•
•
•
asymmetric flows
Should provide a direct means to observe
convective velocity field in photosphere
Cool stars bisectors resemble solar bisector, but
with considerable variations
Gray & Nagel (1989) found bisectors reversed in
hotter stars: a “granulation boundary”
Two “different” types of convection??
August 22, 2006
IAU Symposium 239
Bisector curvature (asymmetry) - 2
• On MS, reversed bisectors also found
among A stars (Te <10500 K)
• Late B stars show no bisector curvature,
and have x < 1 km/s
• Bisector curvature not studied for hotter
stars, mainly because so few have v sin i
< 5 km/s
August 22, 2006
IAU Symposium 239
Multi-parameter models of flow
• Modelling of cool stars by Dravins (1990) with
•
•
four-component flow (2 hot upflows, 1 neutral, 1
cool downflow) reproduces line profiles
reasonably and supports general picture of flow
behaviour
Frutiger et al (2000, 2005) have used multiparameter models to derive temperature and
velocity structure of simple geometrical flow
models for Sun, a Cen A & B
Useful for searches of parameter space
August 22, 2006
IAU Symposium 239
3D hydrodynamic models
• Physically realistic modelling requires 3D
hydrodynamic models (e.g. Nordlund &
Dravins) but such models are very costly
• 3D models of low-metal stars with
convection reveal that temperature
stratification is changed significantly,
perhaps also changing derived Li
abundance (Asplund & Garcia Perez 2001)
August 22, 2006
IAU Symposium 239
3D models - 2
• Detailed model of Procyon allows comparison of
•
micro- macro-turbulence fits to fits of 3D line
profiles (Allende Prieto et al 2002)
Without free model parameters (except
fundamental parameters of star), 3D model lines
provide excellent fit to observations
August 22, 2006
IAU Symposium 239
3D models - 3
• CO5BOLD code used to compute coarse model of
•
•
entire M2 I star; find giant convection cells as
suggested by images (Freytag et al 2002)
Same code computed convective models of A
star, but found no reversed bisectors (Steffen et
al 2005)
Limitation of 3D codes – if one disagrees with
observation, testing changes is very costly
August 22, 2006
IAU Symposium 239
MLT and other convection models
• MLT, FST and non-local convection models
•
•
provide alternative description
Comparisons of predictions of such models with
Balmer lines, uvby colours of star (Smalley &
Kupka 1997; Gardiner et al 1999) show that
observational tests of models are possible
Kupka & Montgomery (2002) seem to predict
correct sense of A star bisectors from non-local
convection model
August 22, 2006
IAU Symposium 239
Conclusions
• Stellar atmospheric velocity fields clearly
detectable in spectrum: microturbulence,
macroturbulence, bisector curvature,
energy distribution,….
• Behaviour over HR diagram quite varied;
largest velocities in supergiants
• Modelling making progress at connecting
convection theory with observations
August 22, 2006
IAU Symposium 239