Natural Science Forum / Earth Science / Oceanography / May 2007

 Truckloads of methods, so much so that to be serious about it,
you'll also have to devote significant effort to figuring out
what you mean by 'best results' before going much in to them.

 Meteorology also uses (usually before oceanography) such methods, so
there's another search term.  Some families of methods, loosely in
order of complexity:

Cressman Filtering
Splining
Kriging
2d Var (2 dimensional variational analysis)
Optimal Interpolation
3d Var
4d Var
Kalman Filtering

 The variational analysis methods and Kalman filter require that
you have a model of the evolution of the field you're working on.
The requirement may only be for a 'first guess' in the case of the
variational analysis, but you can do it for full-scale models as
well.

 Speaking of such, throw in adjoint models to the list.