An observational campaign was conducted at Nova Roma do Sul, in the montainous region of southern Brazil (Serra Geral), from November to December, 2001. The main purpose was to provide an estimation of the climatic impact of the flooding of the Antas river that is supposed to happen after the construction of a Power Plant dam. The very sharp nature of the terrain made it impossible to determine the local turbulent fluxes at a horizontally homogeneous location. Therefore, a tower was placed by the edge of a cliff, with a 450-m vertical relief.

Footprint considerations are essential to the interpretation of the observations. The mean winds rotated along the day, following a typical mountain-valley circulation. Thus, nighttime and morning footprints were entirely located at the drier highlands, while afternoon observations are related to the air that comes from the valley, where the river runs. Eddy correlations of scalar and vertical wind perturbations are not necessarily reflecting the turbulent vertical fluxes, specially during the afternoon. A consequence of this is the fact that the daily cycle of these correlations is not in phase with net radiation, peaking in average three hours after local noon. The “Bowen Ratio” determined from the correlations exhibits a daily cycle, being smaller at the morning, and increasing at those periods when the wind blew from the bottom of the valley.

The observations were used as inputs to numerical models of the climatic impact of the flooding. Two models were used. The first is a very simple one, based on energy partition and considering that the local variation of temperature and humidity is entirely dependent on the local fluxes. The second is a mesoscale, finite-differences model (RAMS), using detailed description of the surface before and after the flooding. Both models indicate a very small increase of relative humidity and decrease of temperature due to the flooding, a consequence of the small increase of the water surface and moderate winds that are able to distribute the input differences over a large area.