During the CLU/ESCOMPTE experimental campaign (June-July 2001) two optical (0.93 m) large aperture scintillometers (LAS) were installed over the city center of Marseille (France) along perpendicular path lengths of approximately 2 kms. The two trajectories chosen present distinct topographical differences, thus providing an addition influence on the observations: one situates above a valley, the other includes a hill in the middle. Reference eddy correlation measurements were made at 35m on an instrumented mast in the valley. The aim of the experiment was twofold: (i) to validate the LAS method over an urban area and (ii) to characterize the spatial variability of sensible heat flux over the city. Results of the validation exercise are presented: they show a good agreement between LAS derived flux and eddy correlation measurements. The influence of topography is then analyzed using numerical simulations based on different case studies of relief (triangular valley, triangular hill, and step of different heights), and the differences between the two paths are finally discussed.