On the appropriateness of the baroclinic mode or the SQG dynamics to describe the surface mesoscale motions