Context. Detailed knowledge of surface dynamics is one of the key points in understanding magnetic solar activity. The motions of the solar surface, to which we have direct access via the observations, tell us about the interaction between the emerging magnetic field and the turbulent fields. Aims. The flows computed with the coherent structure tracking (CST) technique on the whole surface of the Sun allow for the texture of the velocity modulus to be analyzed and for one to locate the largest horizontal flows and determine their organization. Methods. The velocity modulus maps show structures more or less circular and closed which are visible at all latitudes; here they are referred to as donuts. They reflect the most active convective cells associated with supergranulation. These annular flows are not necessarily joined as would seem to indicate the divergence maps.Results. The donuts have identical properties (amplitude, shape, inclination, etc. ) regardless of their position on the Sun. The average donuts computed from all the donuts shows an asymmetry east-west of the amplitude which is related to previous works on the wave-like properties of supergranulation. A kinematic simulation of the donuts' outflow applied to passive scalar (corks) indicates the preponderant action of the selected donuts which are, from our analysis, one of the major actors for the magnetic field diffusion on the quiet Sun. Conclusions. The absence of donuts in the magnetized areas (plages) indicates the action of the magnetic field on the strongest supergranular flows and thus modifies the diffusion of the magnetic field in that location. The detection of the donuts is a way to locate -in the quiet Sun -the vortex and the link with the jet, blinkers, coronal bright points (campfires), or other physical structures. Likewise, the study of the influence of donuts on the evolution of active events, such as the destruction of sunspots, filament eruptions, and their influences on upper layers via spicules and jets, could be done more efficiently via the detection of that structures.