During the MAP SOP several field experiments focusing on the atmospheric boundary-layer took place. Out of these experiments, the most comprehensive turbulence measurement campaign was held in the Riviera valley, a typical Alpine valley in the southern part of Switzerland. The Riviera project's observational phase, which lasted from August to October 1999, produced an highly detailed picture of the thermal, dynamical and turbulent structure of the atmosphere within and above the Riviera valley. Near-surface eddy correlation data were continuously collected in a cross-section of the valley using several micrometeorological towers. In addition, during selected periods, a small aircraft, radio soundings, sodars, thethered baloon and a passive microwave profiler gave precious data describing the atmosphere above.

In the search for a general description of the boundary-layer in highly complex terrain, the associated turbulent fluxes and the interaction of the boundary-layer with mesoscale winds, we present results from the investigation of the near-surface boundary-layer turbulence strucure associated with well developed valley- and slope-wind system circulations (thermal forcing) or with foehn events (mechanical forcing). The choice of days with this type of well defined circulations enables us to define a mean diurnal cycle of the local flow, with highly regular features, thus simplifying the description of the turbulence observations. Criteria for the selection are mainly based upon observations of the surface pressure gradient along the valley and the superimposed pressure gradient across the Alps. During the entire field phase, 20% of the total days are valley wind days, another 30% are south- or north-foehn events.

Turbulent momentum flux observations at the valley floor and from a tower on a steep slope covered with forest are first compared with theoretical expectations from Monin-Obukhov similarity theory. To do this a set of procedures is selected to correctly process the eddy correlation data collected over very steep and inhomogeneous terrain. After a simple quality control, a coordinate tranformation into the mean flow system is applied using the 'planar fit' method. This method is compared to the more standard double rotation method. Finally the common flux corrections are applied to the computed covariances. Results show a complex interaction between valley-wind and slope-wind, leading to a vertical directionl shear at the slope site. The presence, in the late afternoon of this directional shear, increases the vertical turbulent fluxes significantly. Local scaling is shown to bee applicable also in this kind of situation. On the valley floor the Reynolds stress is found to exhibit a strong decrease with height, related to the presence of the along-valley pressure gradient. This renders the definition of a characteristic velocity scale (other than the local velocity scale) under different forcing conditions a subtle task.