It is currently debated if either thermal erosion of the lithosphere or dynamical support is the source of topography and geoid anomalies. The origin of this controversy lies probably in the difficulty to model simultaneously these two effects. For this purpose we have studied the time dependent behavior of two-dimensional convection with a temperature and pressure dependent viscosity. The use of a control volume method allows us to define a rigid zone simulating the mechanical lithosphere. The interface between the lithosphere and the convective mantle is determined by a viscosity cutoff. First, some experiments model the rise of a plume below the lithosphere in order to observe the evolution of the uplift and thus to appreciate the various processes involved in the swell formation. Before the plume reaches the base of the thermal lithosphere, an uplift a few hundred meters in amplitude develops which can only be ascribed to a pure dynamical support. The major uplift occurs when the ductile part of the lithosphere, the convective boundary layer, is squeezed by the plume. The reheating of the mechanical lithosphere takes place after this transient stage of dynamical erosion.

However, this late process is very slow but can magnify the amplitude of the swell if the lithosphere stays long enough above the plume. These results shed some light on the different mechanisms occurring during the swell formation, but the configuration modeled does not correspond to the one expected for actual hotspot swells. They feature plume rising up to the lithosphere while natural situations correponds to lithosphere drifting above prexisting plumes. An experiment with a moving lithosphere was run and shows that thermal erosion does not affect significantly a moving lithosphere even for relatively slow drifting velocities (few centimeters/year). Indeed, the thermal structure of the lithosphere is not modified above the 800 ¿C isotherm except for a motionless plate. In this case the resulting swell should be greater: this could explain why Azores, Crozet or Cap Verde swells are so high. On the other hand, the shape of a swell over a moving lithosphere is strikingly reminiscent of the Hawaiian swell. ¿American Geophysical Union 1993