Modeling Stars


    The modeling of stars in general is very complex.

    In this insert, we outline the simplest approach to star modeling in order to understand how it is done and how models are fitted to observations. The fit verifies the models insofar as they are realistic: i.e., contain the right physical structures and the right physics theories.

    Verified models then allow us to know things about star which CANNOT be known by direct NOR indirect observations (which are both discussed in IAL 19: Star Basics I: Introduction).

    As preview/review of some of the topics of to come in IALs on stars, see the Pleiades open star cluster in the figure below/above (local link / general link: pleiades.html).

    In the sections following this one, we explicate further the ingredients of star modeling: i.e., sections Star Shape and Structure, Stellar Composition, Luminosity, Flux, Photometry, Distance and Stellar Parallax, Luminosity Determination and Spectroscopic Parallax.

    In the subsections below, we cover star modeling itself:


    1. The Very Basic Qualitative Model of a Star:

      The very basic qualitative model of a star is a large self-gravitating sphere of hot gas.

      Going beyond this to quantitative star modeling necessarily includes the relevant physics theories: hydrostatics, hydrodynamics, nuclear physics, quantum mechanics, radiative transfer, statistical mechanics, thermodynamics, and whatever other physics theory is needed.

    2. The Stellar Structure Model and the Atmosphere Model:

      Going beyond the very basic qualitative star model actually requires two quantitative models: the (interior) stellar structure model and the stellar atmosphere model.

      The stellar structure model is entirely about what we CANNOT observe since it is all beneath the opaque photosphere.

        There is an important exception in the case of the Sun. We can observe solar neutrinos that are produced by the nuclear burning in the Sun's core. Neutrinos are so unreactive that the Sun is nearly transparent to them. Our detectors are nearly transparent to them too, but we detect enough that give important information about the Sun and neutrino physics. However, we CANNOT neutrinos from any other star than the Sun. We will NOT further expand on neutrinos here.

      The stellar atmosphere model (which is from just below the photosphere outward) is about lots that we do NOT see too, but we do get direct observations of photometry and spectroscopy. For the Sun, we get a lot more information about solar atmosphere, than for any other star because we are so close. As always the we are so close to the Sun is special case because of its closeness.

      The reason for needing two models is that the scales of stellar structure modeling and stellar atmosphere modeling are so different that doing them in one model is vastly impractical. The two kinds of models can be fitted together: outer boundary conditions of the stellar structure model are the inner boundary conditions of the stellar atmosphere model and vice versa. The two kinds of models are connected as we discuss below subsection Stellar Atmosphere Models.???? (Well we will one day when I write that up explicitly.???)

      Since we are discussing very simple modeling, we ignore the complications of stellar rotation, stellar magnetic fields, and star-star interactions which occur close binaries. Close binaries show stellar evolution NEVER seen for single stars.

    3. Stellar Structure Models:

      One can develop a simple stellar structure model for a main-sequence star with only mass and composition as free parameters.

      The stellar mass range is ∼ 0.08--300 M_☉ (i.e., solar masses).

      The composition is usually the solar composition with metallicity (Z) as a free parameter for zero-age main sequence (ZAMS) (i.e., the time right after star formation). As a star ages, hydrogen burning in the stellar core converts hydrogen (H) to helium-4 (He-4).

      In the post-main-sequence star phase, there is complex layered nuclear burning including hydrogen burning, helium burning, and nuclear burning of heavier elements. The post-main-sequence star phase is discussed in IAL 23: The Post-Main-Sequence Life of Stars

      Note that the hypothetical Population III stars (see below section Population I, II, and III Stars) that formed at cosmic time ∼< 1 Gyr may have had stellar masses much larger than 300 M_☉ and have the primordial cosmic composition (fiducial values by mass fraction: 0.75 H, 0.25 He-4, 0.001 D, 0.0001 He-3, 10**(-9) Li-7).

    4. An Example Stellar Structure Model:

      What a stellar structure model gives you is "runs" of quantities (i.e., their distribution with radius coordinate).

      The figure below (local link / general link: sun_model_interior.html) is an example of the "runs" for a stellar structure model of the Sun with a brief discussion of how stellar structure models are calculated.

      We discuss stellar structure models further in IAL 22: The Main Sequence Life of Stars: Stellar Structure and Stellar Modeling.


    5. Stellar Structure Models and Observables:

      The only SYNTHETIC direct observable calculable from simple stellar structure models is luminosity (i.e., energy output per unit time: i.e., power)---which is typically given in units of solar luminosities L_☉ = 3.828*10**26 W.

      SYNTHETIC luminosities from stellar structure models can be compared to (observed) luminosities as test of the accuracy of the stellar structure models.

      But luminosity is only a direct observable when you know the distance to the star (see sections Distance and Stellar Parallax and Luminosity Determination and Spectroscopic Parallax), can effectively integrate observed flux over all wavelengths (see section Luminosity, Flux, Photometry), and can account for extinction.

        Extinction is the amount of absorption and scattering of EMR by the interstellar medium (ISM) along the line of sight from the star to the observer.

        Determining and correcting for extinction is a major problem in astronomy, but it's too intricate a subject for IAL---and so we will skirt it.

      In fact, in the modern age, distance (from the via stellar parallax Gaia spacecraft (mission 2013--2025?)) and extinction are much less of problems than in the past.

      However, comparisons of SYNTHETIC luminosities and observed luminosities are only a very limited test.

      Stars of the same luminosities may in general have very different phases (main sequence, post-main-sequence, masses, and stellar structure.

      Neither of the prime free parameters of stellar structure models (i.e., stellar mass and stellar composition) are direct observables in general.

      So they CANNOT be set by direct observation.

      In fact, the only SYNTHETIC direct observable calculable from simple stellar structure models is luminosity (i.e., energy output per unit time: i.e., power)---which is typically given in units of solar luminosities L_☉ = 3.828*10**26 W.

        Stellar atmosphere models allow you to calculate SYNTHETIC spectra and photometry

      But luminosity is only a direct observable when you know the distance to the star (see sections Distance and Stellar Parallax and Luminosity Determination and Spectroscopic Parallax), can effectively integrate observed flux over all wavelengths (see section Luminosity, Flux, Photometry), and can account for extinction.

        Extinction is the amount of absorption and scattering of EMR by the interstellar medium (ISM) along the line of sight from the star to the observer.

        Determining and correcting for extinction is a major problem in astronomy, but it's too intricate a subject for IAL---and so we will skirt it.

      Now observed luminosity and stellar mass (if we had them) are enough (if we can assume composition) to constrain simple stellar structure models, and so tell us what stars of thoses masses and luminosities are like.

      But in general we do NOT have stellar mass and often NOT luminosity.

        Actually, stellar mass is a direct observable for some binary star systems. Such systems are important tests of stellar structure modeling.

      The upshot of the above is that in order verify our understanding of stars based on stellar structure models we need another kind of model from which we can calculate sufficient sythetic observables to fit to actual observables to verify and constrain our stellar structure models. That model is the stellar atmosphere model which we discuss below in the subsections Stellar Atmosphere Models and Spectral Types and Stellar Models.

    6. Stellar Atmosphere Models:

      What we directly observe for stars are photometry (broad wavelength band measurements of flux: see section Luminosity, Flux, Photometry below), spectroscopy (narrow wavelength band measurements of flux), and, for sufficiently near stars, distance.

      Spectroscopy gives more detailed information than photometry, but is harder to obtain to the same level of accuracy and for distant stars NOT obtainable at all.

      With sufficient photometry and spectroscopy we can model the stellar atmosphere: i.e., create a model of the stellar atmosphere.

      Adjusting the free parameters of the model to fit the photometry and spectroscopy gives us values for those free parameters. The values are as good as the photometry, spectroscopy, and modeling allow.

        Like Stellar structure models, stellar atmosphere models must be calculated using numerical methods on the computer. There are NO analytic solutions, except for highly simplified cases like the grey atmosphere. Such simplified cases are very useful in understanding stellar atmospheres and in testing computer codes, but do NOT have realistic behaviors, except in a very approximate way sometimes.

      For simple stellar atmosphere modeling, the free parameters (which are determined by fits to observed photometry and spectroscopy) are usually:
      1. Composition. Assuming the cosmic composition (which is NOT always valid), composition effectively means metallicity Z.
      2. Gravitational field
          g =  GM/R**2  , 
        where gravitational constant G = 6.67430(15)*10**(-11) (MKS units), M is stellar mass and R is stellar radius (i.e., photospheric radius). But note the modeling only gives g, NOT M and R separately.
      3. Effective temperature
          T_eff = (F/σ)**(1/4) = [(L/(4πR**2))/σ]**(1/4) , 
        F is flux, σ is the Stefan-Boltzmann constant σ = (5.670374 19 ...)*10**(-8) W/*m**2/K**4 (exact) (see NIST: Fundamental Physical Constants --- Complete Listing 2018 CODATA adjustment)), L is luminosity, and R is stellar radius (i.e., photospheric radius). Note effective temperature is the temperature the star would have if it radiated like an exact blackbody radiator of radius R. Stars do NOT radiate like exact blackbody radiators. Nevertheless, effective temperature is a good characteristic or sort-of average temperature for their photosphere.

      From the above, we can obtain three fitted free parameters: stellar atmosphere (or nearly equivalently metallicity Z), gravitational field g, and effective temperature.

      Alas, the there are 3 unknowns M, R, and L for 2 equations: the ones for gravitational field g, and effective temperature.

      So we cannot solve for M, R, and L separately without more information.

      If we had any 2 of M, R, and L, and some estimate of core composition, then stellar structure model could be fitted and we would understand the star insofar as simple modeling allows.

      However, as discussed in section Stellar Structure Models and Observables we usually do NOT have stellar mass or luminosity. And stellar radius is known to accuracy/precision only for the Sun. A few red supergiants can be barely resolved and have their radii determined: e.g., Betelgeuse with R = 887(203) R_☉ ≅ 4 AU (see solar radius R_☉ = 6.957*10**5 km = 109.1 R_eq_⊕ = 4.650*10**(-3) AU; Dolan et al. 2016).

      So are we stuck?

      No. Distances (and therefore luminosities) can be obtained for relatively nearby stars by stellar parallax (see section Distance and Stellar Parallax below). Distances by stellar parallax were originally only obtainable to very nearby stars within a few parsecs in the 19th century, (see Wikipedia: Stellar parallax: 19th and 20th centuries), but with advancing technology, smaller stellar parallaxes, and so greater distances have been obtained progressively.

      As of 2018, the Gaia spacecraft (mission 2013--2025?) has provided us with accurate/precise stellar parallax to distances up to 8 kpc (see Wikipedia: Gaia spacecraft: Objectives)) which is about the distance to center of the Milky Way.????

      Also as mentioned above in section Stellar Structure Models and Observables, stellar mass is a direct observable for some binary star systems.

      The upshot is that we can nowadays fit stellar models (both stellar structure models and stellar atmosphere models) to vastly many stars and thereby understand them and know their parameters insofar as our modeling is sufficiently realistic.

      We can go well beyond the simple modeling described here and include stellar rotation, stellar magnetic fields, and star-star interactions which occur close binaries.

    7. Spectral Types and Stellar Models:

      As previewed in IAL 8: The Sun and as discussed at greater length in IAL 20: Star Basics II, stars can be classified by spectral type.

      The full spectral type classification (which includes its luminosity class) very full characterizes a star. The spectral type classification is just empirical: it is a direct observable.

      Now all stars of the same full spectral type classification are very much alike and so all have the same stellar model (both stellar structure model stellar atmosphere model) insofar as they are alike.

      So we do NOT have to model every star we want to understand. We just have to model stars of all spectral types.

      And this has been done.

      Now there remain imperfections in spectral type classification and modeling of stars, but there is continually work to reduce those imperfections.

    8. Beyond the Main Sequence:

      The current status is that we understand main-sequence stars in their bulk properties and evolution very well.

      Pre-main-sequence stars and post-main-sequence stars are more difficult to model, and so are less well understood, especially quantitatively.

      A general reason is that pre-main-sequence stars and post-main-sequence stars evolve more rapidly than main-sequence stars, and that just makes them harder to model.

      Also in the case of pre-main-sequence stars, they are embedded in star forming regions which are opaque in the visible band (fiducial range 0.4--0.7 μm = 4000--7000 Å) and this makes them harder to undersand observationally.

      And in the case of post-main-sequence stars, they are subject to explosive events (core helium flashes, thermal pulses (AKA helium shell flashes), and for stars > ∼ 8 M_☉, supernova explosions) which are hard to model because they are so complex.

      See the figure below (local link / general link: stellar_evolution_overview.html) for an overview of the stellar evolution of a star of less than ∼ 8 M_☉.


    File: Star file: star_modeling.html.