A new similarity approach is applied to the thermocline equations in o
rder to examine contrasting frontal and ventilated models of the main
thermocline. The method of solution involves reducing the number of in
dependent variables of the controlling partial differential equation,
leading to a particular form for the solutions which satisfy appropria
te boundary conditions. A frontal model of the thermocline is obtained
following the study of Salmon and Hollerbach (1991). When the vertica
l diffusivity becomes vanishingly small, an interior front in the subt
ropical gyre appears at the depth where the vertical velocity changes
sign. The front separates downwelling warm water in the subtropical gy
re from the underlying upwelling of cold, deep water. These solutions
appear to be robust to changes in the vertical diffusivity profile, as
long as there is a small, nonzero value in the interior. However, whe
n there is uniform diffusivity, there is no implied surface heat flux
and surface isotherms are coincident with streamlines. As the diffusiv
ity increases toward the surface, the surface heat input increases in
magnitude and the temperature field becomes more plausible. A ventilat
ed model of the thermocline is formed using the similarity approach wi
th a diffusive surface boundary-layer overlying an adiabatic interior.
In this case, the temperature and velocity fields are solved for in t
he limit of uniform potential vorticity. There is now a more plausible
cross-isothermal flow in the surface layer with a polewards decrease
in temperature, and the implied surface heat input increases equatorwa
rds. Fluid is subducted from the surface boundary layer into the adiab
atic interior and forms a continuous thermocline. In conclusion, the c
ontrasting frontal and ventilated solutions arise from modeling differ
ent aspects of the circulation, rather than depending on the type of m
odel employed. The ventilated solutions form a thermocline by advectin
g the surface temperature field into the interior of a subtropical gyr
e, whereas the frontal solutions create a thermocline from the interac
tion of the wind-driven gyre and the underlying thermohaline circulati
on. These thermocline solutions might occur separately or together in
the real ocean, although both solutions might be modified by higher-or
der processes or more complicated forcing.