Mo delling Variability in Hot-Star Winds
1
Stanley P.Owocki
Bartol Research Institute University of Delaware, Newark, DE 19716 USA
Abstract. I review 2-D hydro dynamical simulations of rotating hot-star winds with
azimuthal structure induced by mo dulation of the radiative driving force near the wind
base. As a rst step toward examining more realistic p erturbation mechanisms (e.g.,
nonradial pulsations, or magnetic elds), the driving mo dulation here is taken to arise
from bright and dark sp ots in the stellar photophere. These sp ots induce decreases or
increases in wind ow sp eed, and as the star rotates, spiral \Co-Rotating Interaction
Regions" (CIRs) form, much as in the solar wind, from from interaction b etween fast
and slow ow streams. A new feature unique to line-driven owisavelo city-gradient
kink that propagates inward from interaction fronts at a fast radiative-acoustic mo de
sp eed. The slowly evolving velo city plateaus that form b ehind suchkinksgiveriseto
absorption features with a slow apparent acceleration, muchlike the Discrete Absorp-
tion Comp onents (DACs) often observed in UV wind lines from hot-stars. In simulation
mo dels with base driving sinusoidal ly mo dulated b etween increases and decreases, there
arise alternating spiral streams of enhanced or decreased density, asso ciated resp ec-
tively with decreased or increase ow sp eeds. These sp eed variations have substantial
impact on the line pro le, and so these dynamical simulations are not as successful as
analogous kinematic mo dels of corotating density streams in repro ducing the \phase-
bowing" of p erio dic absorption mo dulations observed in the recent IUE `Mega' pro ject.
1 Intro duction
Several pap ers at this meeting (e.g. by Prinja, Kap er, Kaufer, Massa, and Wolf )
have summarized the extensive observational evidence for explicit variability,
cyclical or otherwise, in the winds from hot, luminous, early typ e (OB) stars. A
general challenge for theory is to understand the nature of the physical changes
and/or structures asso ciated with this variability.To b e visible as direct varia-
tions in line-pro les formed from globally integrated radiative ux, the asso ciated
ow structure must b e on a relatively large scale, on order the stellar radius. This
makes it unlikely that suchvariations could stem from pro cesses entirely intrin-
sic to the wind out ow, such as the inherent instability in the line-driving of the
wind to small-scale p erturbations (Rybicki 1987; Owocki, Castor, and Rybicki
1988; Owocki1994b). The dynamical evolution of such large-scale structure can
be simulated using the lo cal, computationally ecient, CAK/Sob olev expres-
sions for the line force (Castor, Abb ott, and Klein 1975; Sob olev 1960), making
it feasible to carry out multidimensional simulations of the wind structure re-
sulting from large-scale p erturbations from the underlying, rotating star.
In this review, I will fo cus on recent e orts to develop initial dynamical
mo dels for two distinct classes of suchvariability, namely the `classical' Discrete
2 Stanley P.Owocki
Absorption Components (DACs) and the more recently identi ed Periodic Ab-
sorption Modulations (PAMs) discovered in the IUE `Mega' pro ject (Massa et al.
1995). Unlike the quasi-episo dic, slowly evolving, net absorption enhancements
that characterize most DACs, the PAMs recur regularly at p erio ds a loworder
fraction of the rotation p erio d, include b oth reductions and enhancements of
absorption, and evolve relatively quickly over the line pro le. Indeed, in contrast
to the slow blueward evolution of DACs, the PAMs in one case (BO I star HD
64760) show a \phase-b owing" that re ects apparentredward as well as blue-
ward propagation (Owocki, Cranmer, & Fullerton 1995; ; Fullerton et al. 1997).
These distinct observational characteristics likely re ect di erences in the under-
lying p erturbation mechanisms, p erhaps, for example, with the more sto chastic
DACs b eing induced by magnetic activity, and the regular PAMs b eing initiated
by Non-Radial Pulsations (NRPs). But given the present uncertainty, the initial
simulations here simply induce wind variations rather arti cially, through direct
mo di cation of the radiative driving in the inner wind, much as might o ccur
from \sp ots" on the underlying star (Cranmer and Owocki 1996, hereafter CO).
I rst (x 2) describ e the e ect of isolated sp ots, b oth brighter and darker than
the ambiant photophere. As the star rotates, Co-rotating Interaction Regions
(CIRs) form along spiral patterns by collision b etween faster and slower wind
streams originating from di erent longitudes relative to the sp ot. These simula-
tions thus represent the rst dynamical test of the original prop osal by Mullan
(1984a,b; 1986) that the wind density enhancements in such CIRs could cause
the DACs. A central goal here is to determine whether key characteristics of ob-
served DACs, particularly their apparent slow acceleration, can b e repro duced
in synthetic line-pro les generated from dynamical mo dels with CIRs.
I also examine (x 3) the e ect of a sinusoidal modulation of the radiativedriv-
ing near the wind base, assuming a xed number (m = 4) no des around the star.
This is intended as a dynamical version of the simple kinematic picture prop osed
to explain the \phase b owing" of the PAMs in HD 64760 (Owocki, Cranmer, &
Fullerton 1995.) In this picture the PAMs arise from corotating streams of al-
ternating increased or decreased density, within the key simpli cation that the
velo city is xed to a sp eci c law, una ected by the density p erturbations. The
dynamical simulations here self-consistently include suchvelo cityvariations, and
so allow us to examine their e ect in the line-formation.
2 CIRs Induced by Isolated Sp ots
Let us rst review the CO simulations of wind structure induced by isolated
sp ots on a rotating hot-star. The aim is to mimicphysical pro cesses { e.g. mag-
netic eld, NRPs { that might increase or decrease the mass ux emerging from
some lo calized region of the star, and then study howsuchvariations are prop-
agated through the radiatively driven stellar wind. To reduce the complexity
and computational costs, the simulations are con ned to 2-D variations in ra-
dius and azimuth (r , ) within the equatorial plane. The line-driving force is computed using the standard CAK/Sob olev formalism (Castor, Abb ott, Klein
Mo dellin g Variability in Hot-Star Winds 3
Fig. 1. Grayscale p olar plots showing the spatial dep endence of the deviations from
a steady mo del for a. density, b. radial velo city,c.azimuthal velo city, and d. Sob olev
optical depth.
1975; Sob olev 1960), mo di ed by the nite-disk correction factor for a spherical
out ow(Friend and Abb ott 1986; Pauldrach, Puls, and Kudritzki 1986). This
ignores the azimuthal force that should arise from sideview p ersp ectives of the
sp ot, and simply mo di es the radial force by a xed enhancement/reduction
factor determined by the relativeproximity to the sp ot.
The sp ot parameterization allows sp eci cation of b oth a longitudinal width
and an amplitude A, with 1 0 for bright
sp ots. Of the mo dels listed in Table 1 of CO, I con ne attention here to the rst
o
two, for which =20 and A = 0:5, representing a standard bright and dark
sp ot. The stellar parameters represent those for the canonical O-typ e sup ergiant
Puppis, with a rotation sp eed V = 230 km/s, corresp onding to a rotation
rot
p erio d of P 4:2days. For convenience, let us assume two identical sp ots
p ositioned on opp osite hemispheres, allowing restriction of the computation to
o
an azimuth range of just 180 with p erio dic b oundary conditions.
For the bright sp ot case, gure 1 shows grayscale plots of the resulting density,
radial velo city, azimuthal velo city, and radial Sob olev optical depth, all relative
to values in the corresp onding steady, spherically symmetric, CAK wind mo del
without anyspot.Thespothereiscentered on the x-axis, at the origin the
spiral pattern of enhanced densitythatcharacterizes the resulting CIR. Note
that regions of enhanced density generally corresp ond to regions of lower radial
velo city. Because of the enhanced brightness over the sp ot, the wind driven from
there initially has a higher mass ux, and thus higher density. As the enhanced
brightness fades with increasing heightabove the sp ot, the higher densityofthis
4 Stanley P.Owocki
Fig. 2. left panels: Radial variation of velo city (upp er panel) and density (lower panel)
along selected, xed azimuthal angles in the bright sp ot mo del.
Fig. 3. right panels: Same as g. 2, but for the dark sp ot mo del.
material makes it harder to accelerate, re ecting the characteristic scaling of the
line-acceleration with the inverse of the density ,
1 @v
r
g (1)
lines
@r
where is the usual CAK exp onent. The lower velo city gradient @v =@ r asso-
r
ciated with the lower acceleration also contributes, through eq. (1), to a further
reduction in the acceleration. As the stellar rotation brings this higher-density,
lower-sp eed material into interaction with faster ambient wind originating away
from the sp ot, there results a sho ck compression of the gas into the dense spiral
stream that characterizes the CIR.
Though the general CIR formation here is analogous to that o ccuring in the
solar wind (Hundhausen 1972; Zirker 1977), there are imp ortant di erences. The
much higher density of hot-star winds makes radiative co oling very ecient, and
so the sho cks are e ectively isothermal. Unlike the nearly adiabatic sho cks in
the solar wind, whichhave a maximum compression ratio of four in the strong
sho ck limit, the density compression in such isothermal sho cks scales with the
square of the Machnumb er, and so can b e arbitarily large. Moreover, whereas
the nearly adiabatic sho cks of the solar wind propagate backward into the faster
wind at roughly a quarter of the sho ck sp eed, these isothermal sho cks remain in
close proximity to the compressed layer.
Mo dellin g Variability in Hot-Star Winds 5
Another signi cant di erence arises when we consider the nature of the pre-
shock ow. In ordinary gas-dynamics (or even MHD) sho cks, the presho ck ow
is assumed to b e completely una ected by the imp ending sho ck, simply b ecause
the sp eed of this material exceeds the sound sp eed (or in MHD, the fast-mo de
sp eed), preventing anyupwind information propagation that a disturbance lies
ahead. However, as rst shown by Abb ott (1980), the characteristic sp eeds c in
a line-driven ow are substantially mo di ed by the op eration of the line force,
given by
p
2 2
c = U=2 (U=2) + a ; (2)
where a is the ordinary (isothermal) sound sp eed, and U @g =@ (@v =@ r )
lines r
is a new characteristic sp eed, nowadays often called the \Abb ott sp eed". From
eqn. (1) we ndU g =(@v =@ r ), but from the radial equation of motion,
lines r
we exp ect v @v =@ r g , which suggests U v , i.e. this Abb ott wave sp eed is
r r r r
typically of order the radial ow sp eed. Since most the wind is sup ersonic, we
nd v U a, implying that inward wave propagation can now o ccur at a
r
sp eed c U v that is much faster than the propagation of an ordinary