Eddy covariance measurements on mountain slopes:: The advantage of surface-normal sensor orientation over a vertical set-up

The measurement of scalar fluxes employing the eddy covariance method is a widely used experimental approach, for which the flow distortion due to obstacles (e.g., sensor mounts and mast) is a well-known but not fully solved problem. In order to reduce flow distortion we installed a sonic anemometer in a surface-normal orientation relative to the terrain slope, and a second instrument in a vertical position at a horizontal distance of 1.54 m from the first instrument We found a significant reduction in the rotation angle necessary for the coordinate rotation procedure in the x-z plane when computing 30-minute flux averages with the surface-normal orientation. In 91% of all cases this rotation angle remained within the angle of incidence of +/- 10 degrees recommended by the manufacturer. In contrast, only 24% of the measurements taken with the vertically mounted anemometer were obtained at an angle of incidence within +/- 10 degrees, and 3% were outside the +/- 30 degrees range specified for an acceptable operation. A data quality test based on the variance of vertical wind speed normalized with friction velocity (sigma(w)/u(*)) revealed problems for application under stable conditions due to large uncertainties in the determination of the Monin-Obukhov stability parameter z/L. An alternative test using the bulk drag coefficient C-D revealed other problems related to the dependence of C-D on z/z(0), the measuring height normalized by the roughness length, which do not appear to be constant in complex terrain. With both tests, a tendency for a slightly improved data quality was found for the surface normal set-up, which, however, proved statistically insignificant. It is concluded that the surface-normal set-up of a sonic anemometer significantly reduces flow distortion by the sensor head. Although the surface-normal mounting position therefore appears to be the preferred one, with decreased flow distortion and a slightly improved data quality, no significant differences in turbulent quantities were found between the two set-up positions. Hence, the consequences for short-term measurements of mass and energy fluxes with a surface-normal set-up in complex terrain appear to be relevant only if single flux events are to be inspected, while for long-term measurements of integrated fluxes both the surface-normal and vertical installation of the sonic anemometer are adequate, indicating that eddy covariance measurements in complex terrain are less delicate than expected.