We now describe our attempts to understand the complex dynamics of the outflow in M82 with the aid of kinematic models. We have extracted one-dimensional velocity cuts and synthetic two-dimensional spectra from the Fabry-Perot data cubes. These cuts have been made relative to the outflow axis, at an estimated position angle of 150o (85o relative to the major axis of the galaxy; [McKeith et al. 1995]). While some authors have argued for a spherical wind in M82 (e.g., [Seaquist & Odegard 1991]), most recent models find evidence for a bipolar outflow, usually in the shape of cones with opening angles much less than 90o. This is borne out by our study: the line flux, splitting, and ratio maps all exhibit marked azimuthal variations indicative of an aspherical outflow morphology, strongly weighted toward the minor axis of the galaxy. Studies of the supernova distribution (e.g., [Kronberg, Biermann, & Schwab 1985]) and other disk observations (e.g., [Achtermann & Lacy 1995]) indicate that even the central injection zone is probably not spherical, but rather in the form of a flattened disk, with dimensions 600 pc wide and 200 pc thick.
The specific velocities predicted by [Chevalier & Clegg 1985] do not agree with our observations. An estimated supernova rate of 0.3 yr-1 ([Rieke et al. 1980]) yields an outflow velocity from their model of 2000-3000 km s-1 at the edge of the starburst injection zone. This value is several times larger than the deprojected velocities observed at optical wavelengths (525-655 km s-1; see below) on much larger scales. However, the extensive radio continuum halo ([Seaquist, Bell, & Bignell 1985]; [Seaquist & Odegard 1991]) could arise from synchrotron radiation due to a population of relativistic electrons, as they are transported outward in a wind at velocities in the range considered by Chevalier & Clegg. The relationship between such a wind and the slower, denser minor-axis outflow seen at optical and x-ray wavelengths remains unclear. We suspect that much of the H filamentation arises from large-scale shocks from a high-speed wind plowing into the gaseous halo and entrained disk gas. Only a small fraction of the total wind energy is encompassed by the radio halo (2%; [Seaquist & Odegard 1991]).
In modeling the bipolar outflow in M82, we model the individual line components rather than flux-weighted velocity profiles. In Figure 10, we show the velocities of line fits to the dual H components along the minor axis of the galaxy. While the northern outflow components show a relatively constant projected separation of 300 km s-1, the southern region of split lines reveals an intriguing variation. Within 200 pc south of the nucleus, where the favorable inclination of the disk allows us to measure line profiles closer to the starburst, the individual components of H are separated by a much smaller velocity, comparable to our resolution (50 km s-1). Between 200 and 500 pc from the nucleus, the components rapidly diverge, remaining at a constant separation of 300 km s-1 beyond 500 pc. Maps of the line component splitting reveal a separation of this order throughout the spatial extent of both lobes. Also plotted in Figure 10 are fits to the H profiles from the long-slit optical spectra of [McKeith et al. 1995], which clearly confirms our observed minor-axis trend in the line splitting. An identical radial trend is seen in the [NII] line components along the southern outflow axis, but is not shown in the figure.