Since midocean eddies migrate westward, they eventually reach the west
ern boundaries. It is, therefore, of interest to find out what happens
after the eddies collide with the walls. An isopycnic, two-layer, pri
mitive equation model on a beta plane and a simple analytical model on
an f plane are constructed to investigate the meridional migration of
an oceanic eddy along a western wall. On a beta plane, three factors
determine the eddy's migration along a western meridional wall. First,
the image effect pushes an anticyclonic (cyclonic) eddy northward (so
uthward). Second, the beta force (resulting from the larger Coriolis f
orce on the northern side of the eddy) pulls an anticyclonic (cyclonic
) eddy southward (northward). Third, after an anticyclonic (cyclonic)
eddy collides with the wall, parts of the anticyclonic eddy's interior
fluid leak out southward (northward) along the wall forming a thin je
t. In an analogy to a rocket, this jet pushes the eddy northward (sout
hward). Our aim is to investigate in which direction the eddy ultimate
ly migrates along the wall (i.e., to determine which of the above thre
e processes dominates). The combined effect of the three processes is
a rather complicated process and the results are counterintuitive. For
instance, imagine a lenslike anticyclonic eddy situated on a sloping
bottom (analogous to beta). This highly nonlinear eddy migrates with s
hallow water on its right (''westward'') and encounters a meridional w
all. Intuitively, it is expected that, once the ''westward'' migration
is arrested by the wall, gravity will pull the eddy downhill (southwa
rd) so that the eddy will migrate toward deep water (i.e., toward the
equator). Surprisingly, however, the authors' numerical computations s
how that the eddy migrates uphill. This bizarre behavior results from
the leakage along the wall that, in terms of the eddy energy, compensa
tes for the uphill drift. Namely, the leakage plays a crucial role in
the eddy-wall interaction process because it allows the uphill migrati
on. Eventually, it causes a destruction of the lens by completely drai
ning its fluid. The above highly nonlinear experiments are supplemente
d by quasigeostrophic analytical solutions and isopycnic numerical exp
eriments of cyclones and anticyclones. It is found that, in contrast t
o the situation with the lens, the leakage does not play a crucial rol
e in quasigeostrophic eddies. However, all of these experiments show t
hat the image effect is the most dominant process. It turns out that,
as the eddy responds to the presence of the wall, it is transformed in
to a half-circular shape that is very different from its original prei
nteraction circular shape. This results from the fact that, even thoug
h the westward beta-induced speed (forcing the eddy into the wall) is
small, it is active over an extended period of time so that its final
effect is relatively large. The final half-circular eddy that migrates
along the wall is nearly independent of beta as long as the eddy is n
ot extremely far from its original latitude. This is demonstrated by b
oth our numerical solution (of the primitive equations) as well as our
quasigeostrophic analytical solution. The authors term this final mig
rating eddy a wodon as it represents a combination of a wall and a mod
on. Possible applications of these models to various oceanic situation
s are discussed.