Kinematics of ob stars in the nearby galactic disk

The goal of this thesis is it to learn something about the kinematics and structure
of the Milky Way. By analysing a sample of young stars in the nearby Galactic disk
we can get to know more about the global large-scale structure of our host Galaxy.
If we see resonance phenomena in a velocity field, then we are able to back-reference
from this effect to a possible asymmetric features in the potential.
The reality of science is of course not as simple as that. There are many assumptions
to make and weighted for being able to work. But step by step information can be
filtered out of the entire ensemble.
In the first part of this work we relate the velocities of a sample of OB stars, to
the average velocity field of the nearby areas of the Galaxy. Our analysed region
covers a �2:2 kpc square around the Sun. A star with an orbit passing near the Sun
would spend more than 12% of its time period1, in this analysed field.
Our sample of young OB stars has complete phase space information. For kinematic
analysis and to investigate the characteristics of this sample, such as its
completeness-, spectral type dependence, and errors- tests are made. The completeness
of our sample is particularly discussed in the view of possible kinematic
biases. Different spectral type limited sub-samples are analysed, and error bars are
estimated with Monte Carlo simulations.
Our aim is to understand the large-scale velocity field, thus we remove runaway
stars and members of prominent OB associations out of the Gould belt, and we
implement a random procedure to construct a spatially nearly homogeneous sample
of OB star tracers for the young star velocity field.
The individual velocity vectors of these stars are randomly distributed with respect
to the mean field in the solar neighbourhood, perhaps due to a variety of dynamical
processes occurring at or after the birth of these stars. Our goal is to recover the
mean velocity field by fitting a smooth velocity field to the in-plane velocities of all
our sample stars. For this task we use the non-parametric smoothing algorithm and
software of Wahba and Wendelberger (1980, hereafter WW-algorithm), originally
developed for analyzing meteorological data. We fit a two-dimensional surface to
the data points for each of the in-plane velocity components, as a function of posi-
tion relative to the Sun, and thus derive a smoothed velocity field. For testing the
WW-algorithm and adjusting a smoothing parameter, and for understanding the
results of applying the algorithm to the OB star data, we analyse simulated data
sets. The general idea is to draw Monte Carlo realisations from a known velocity
field, so that the resulting artificial data sets closely resemble the OB star samples
under investigation.
We made a number of tests to ensure that our fitted velocity field is independent of
the (sub)sample used in the analysis, of the assumptions made for the Galactic parameters,
and of the technical details of the fitting procedure. In particular we tested
if the fitted velocity field does not depend on: (1) the size of the region around the
Sun used for the fitting, (2) the rotation of the region and of the coordinate system
in which the fit is made, (3) whether we use the nearly complete sample of earlytype
OB stars, or the full sample which has the advantage of the best space coverage
but with a bad sample completeness, (4) by changing subsamples through different
modulo functions, (5) by restricting the sample to lower heights (jzj < 100 kpc vs.
jzj < 200 kpc), (6) by using only the half of the sample with the better distance
estimates, (7) by changing the assumed rotation velocity of the LSR, and (8) by
changing the assumed galactocentric solar radius.
To make these differences between our fitted velocity field and a circular field more
intuitive, we convert the observed velocity field to the Galactic Center reference
frame, by subtracting the solar motion in the LSR and adding the LSR rotation for
an assumed position of the Sun at R0 = 8 kpc. We then determine streamlines by
integrating through the converted velocity field.
Deviations of the fitted field to a circular one are clearly visible. They are in the
sense that the streamlines derived from the OB star velocities are more elongated
than those expected from circular orbits, especially for radii R < R[sun]. These elongated
ow-lines reach their minimum galactocentric radii at points that are located
approximately on the line that connects the Sun with the Galactic Center. At
R > R[sun] the streamlines seem to be slightly turned forward.
This result is very robust appropriate to all the listed tests. And it inspired us to
think at periodic orbit families near resonances.
Using linear perturbation theory for near-circular orbits, one finds that closed orbits
in a barred potential are elongated either parallel or perpendicular to the bar.
The orientation changes at each of the fundamental resonances. The situation at
the OLR shows two closed orbits (in a frame of reference co-rotating with the bar)
just inside and outside the Outer Lindblad Resonance (OLR). The innermost is
antialigned and the outermost is aligned with the bar. Because of their ellipticity
they can reach the other side of the OLR. Clearly, if all disk stars moved on closed
orbits, the stellar kinematics would deviate from that of a nonbarred galaxy only
at positions very close to ROLR, where the closed orbits are significantly noncircular.
By using 4 different potentials we simulate the corresponding orbit families in a
rotating frame with the corresponding pattern speed of the implemented bar. The
main differences between the models are the positioning of the OLR and the incorporation
of an asymmetric potential represented by 4 equal spiral arms. Each model
includes a symmetric potential and an asymmetric bar potential, for two models we
included a halo component. In all the cases we worked with only one pattern speed.
As a first step, using the well established potential from Bissantz et al. (2003),
we found orbits having the characteristic of our non-circular velocity field, but not
at an expected solar region. By shifting the OLR we could provoke an intersection
between the wanted orbit families and the analysed OB star field. The ellipticity
of these orbits is strong, for a better result we have to shift the OLR even further
out, what we plan to do. Our modeling with the arm potential included, showed
that the inclination of the orbit families, as we see it in the observed field, can be
reproduced by these technique.