The evolution of near-inertial frequency currents is often thought to
be controlled by the linear, inviscid equations of motion. This hypoth
esis is tested by simulating the near-inertial currents described in P
art I using a two-dimensional, nearly inviscid, nonlinear layer model
with realistic wind forcing and stratification. The beta effect and mi
xing of momentum below the mixed layer during the storm are crucial to
realistic modeling, whereas the nonlinear terms have only a minor eff
ect. The model fails to simulate the observations in several ways. Fir
st, the mixed layer inertial currents decay more rapidly than predicte
d and propagate into the thermocline with a different pattern. Second,
the shear at the base of the mixed layer decays much more rapidly tha
n predicted. Third, mesoscale eddies modulate the evolution of the ine
rtial currents much less than predicted. These differences are much la
rger than the errors in the observations and cannot be removed by reas
onable variations of the forcing or stratification. The nearly linear
and inviscid internal wave equations thus cannot accurately predict th
e observed evolution of the near-inertial currents; additional physica
l processes, perhaps nonlinear interactions with smaller-scale interna
l waves and/or fronts, are required in the equations.