The polar wind is an ambipolar plasma outflow from the terrestrial ionosphere at high latitudes. As the ions drift upward along geomagnetic flux tubes, they move from collision-dominated (ion barosphere) to collisionless (ion exosphere) regions. A transition layer is embedded between these two regions where the ion characteristics change rapidly. A Monte Carlo simulation was used to study the steady-state flow of H+ ions through a background of O+ ions. The simulation domain covered the collision-dominated, transition, and collisionless regions. The model properly accounted for the divergence of magnetic field lines, the gravitational force, the electrostatic field, and H+--O+ collisions. The H+ velocity distribution, f(H+), was found to be very close to Maxwellian at low altitudes (deep in the barosphere). As the ions drifted to higher altitudes, f(H+) formed an upward tail. In the transition layer, the upward tail evolved into a second peak with a kidney bean shape, and hence, f(H+) developed a double-humped shape. The second peak grew with altitude and eventually became dominant as the ions reached the exosphere.

This behavior is due to the interplay between the electrostatic force and the velocity-dependent Coulomb collisions. Moreover, the H+ heat flux, q(H+), was found to change rapidly with altitude in the transition layer from a positive maximum to a negative minimum. This remarkable feature of q(H+) is closely related to the coincident formation of the double-humped structure of f(H+). The double-hump distribution might destabilize the plasma or, at least, cause enhanced thermal fluctuations. The double-hump f(H+), and the associated wave turbulence, have several consequences with regard to our understanding of the polar wind and similar space physics problems. The plasma turbulence can significantly alter the behavior of the plasma in and above the transition region and, therefore, should be considered in future polar wind models. The wave turbulence can serve as a signature for the formation of the double-hump f(H+). Also, more sophisticated (than the existing bi-Maxwellian 16-moment) generalized transport equations might be needed to properly handle problems such as the one considered here. ¿ American Geophysical Union 1995