All Figures

$\begin{figure} \par\includegraphics[width=8.8cm,clip]{2778f1.eps} \end{figure}$ Figure 1: Grayscale representation of the density contrast (normalized to the initial relaxed CAK model obtained with identical wind and stellar parameters) for a sequence of models with distinct viscosity amplitude $\alpha _{\rm vis}$ (see Sect. 1 and Appendix), increasing clockwise over the range 0.01, 0.1 and 1. For readability, each computed wedge is duplicated and stacked four times in azimuth, using the periodic boundary conditions employed in that direction.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=16.8cm]{2778f2_1.eps}\par\includegraph... ...8cm]{2778f2_2.eps}\par\includegraphics[width=16.8cm]{2778f2_3.eps} \end{figure}$ Figure 2: Top row: radial variation of time- and angle-averages that characterize the nature of flow structure for the case with $\alpha _{\rm vis}=0.01$ . a) Clumping factor $f_{\rm cl}$ ; b) radial; and c) azimuthal velocity dispersion; and d) velocity-density correlation coefficient (Runacres & Owocki 2002). Middle row: same as top for a model with $\alpha _{\rm vis}=0.1$ . Bottom row: same as top for a model with $\alpha _{\rm vis}= 1$ . For ease of comparison, corresponding quantities are plotted over the same ordinate range.
Open with DEXTER

In the text

$\begin{figure} \par\mbox{\includegraphics[width=5.3cm]{2778f3_1.eps} \includegra... ....3cm]{2778f3_2.eps} \includegraphics[width=5.3cm]{2778f3_3.eps} } \end{figure}$ Figure 3: Left: model with $\alpha _{\rm vis}=0.01$ . Azimuthal variation of the normalized viscous diffuse force $g_{\rm diff,\phi }(r,\phi ) / g_{\ast }(r)$ , at $r=6 R_{\ast }$ , and corresponding to the time-snapshot of Fig. 1. The lateral velocity, normalized to the thermal speed, is over-plotted (dotted line). Middle: same as left for a model with $\alpha _{\rm vis}=0.1$ . Right: same as left for a model with $\alpha _{\rm vis}= 1$ .
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=8.8cm,clip]{2778f4.eps} \end{figure}$ Figure 4: Grayscale images of the density contrast normalized to an relaxed CAK model with identical wind and stellar parameters. At the top ( bottom), models are shown only out to 10 (3) $R_{\ast }$ . Each panel is composed of five different models (duplicated and stacked two times laterally to improve the visibility), ordered clockwise with increasing $\sigma _{\phi }$ , i.e. 0, 0.1, 0.5, 1 and $1.5^{\circ }$ .
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=8.5cm,clip]{2778f5.eps} \end{figure}$ Figure 5: Illustration of 3-rays with $p/R_{\ast} = 0, ~ \pm \sqrt{0.9}$ spaced azimuthally every $\delta \phi =0.9^{\rm o}$ , together with the radial grid spacing (circular arcs) that aligns the rays to intersect multiple $(r,\phi )$ grid nodes; this allows very efficient evaluation of the ray escape integrals needed for computation of the SSF line-force in 2D wind models. For clarity, only every ninth azimuthal or radial zone is shown.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=8cm,clip]{2778f6.eps} \end{figure}$ Figure 6: Projected diffuse acceleration in the lateral direction from the two lateral rays (solid line) for a long- ( top) and short- ( bottom) wavelength velocity perturbation (dotted line). For both cases, we use a relaxed CAK radial velocity.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=12.5cm,clip]{2778f7.eps} \end{figure}$ Figure 7: Grayscale representation of the density contrast for relaxed non-Sobolev models computed with the 3-ray ( left column) and 1-ray ( right column) approaches. All models were run up until a total real time of 120 000 s. Panels in the top ( bottom) row show models only out to 10 (3) $R_{\ast }$ . Each panel is sub-divided into four wedges corresponding to models A, B, C, and D, stacked clockwise from the vertical in order of decreasing radial resolution (Table 1). As before, each computed angular wedge is duplicated a number of times so that the total angular extent of each wedge shown equals $22.5^{\circ }$ , i.e. a quarter of the quadrant in each of the four panels. The key result shown here is the presence of laterally coherent wind structures close to the photosphere in the high resolution 3-ray models (A and B), while with the equivalent 1-ray model, one observes instead wind structures with no noticeable lateral coherence, i.e. of the order of the lateral numerical grid resolution.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=8cm,clip]{2778f6.eps} \end{figure}$	Figure 6: Projected diffuse acceleration in the lateral direction from the two lateral rays (solid line) for a long- ( top) and short- ( bottom) wavelength velocity perturbation (dotted line). For both cases, we use a relaxed CAK radial velocity.
Open with DEXTER