Laboratory experiments were conducted to investigate the generation of anticyclonic gyres by separation of a surface current from a coast in a rotating, two layer system. The experiments were motivated by the hypothesis that the flow of coastal currents around capes can generate oceanic eddies, as well as by the observation of gyres at the mouths of various straits. In the experiments the gyre is formed when the current, which flows with the coast to its right if one is oriented in the downstream direction, encounters a sharp convex corner. The current overshoots the corner, loops to the right, and reattaches to the coast downstream of the corner. Between the current loop and the coast is an anticyclone whose width grows with time. If a countercurrent flows under the surface current, a similar separation in the lower layer results in the generation of a cyclone as well; under some circumstances the cyclone and anticyclone advect each other away from the coast as a heton. Previous studies on related systems found that the corner must be sufficiently sharp for a gyre to form. I show that for a very sharp corner the angle made by the corner must be above a critical value of between 40¿ and 45¿ for a gyre to form. This is in contrast to nonrotating flows of comparable Rayleigh number, which will separate from a sharp corner at virtually any angle. For angles below the critical value, the current profile downstream of the corner changes as a function of corner angle, indicating that it is the stagnation of the flow nearest the wall which causes the anticyclone to form.

This stagnation is reminiscent of the two-dimensional, nonrotating picture of viscous boundary layer dynamics forcing separation of a boundary current. However, the gyre grows more slowly when the lower layer is much thicker than the upper layer, indicating that baroclinic processes are at least quantitatively important in the generation of the gyre. By varying the initial condition of the current, it is shown that the gyre formation is not a product of the interaction of the nose of the current with the corner. In conclusion, the experiments indicate that the basic mechanism of gyre formation may be viscous boundary effects as in nonrotating systems, but that rotation tends to inhibit eddy generation while baroclinic effects tend to enhance it. ¿ American Geophysical Union 1994