Sonic anemometers in aeolian sediment transport research

Fast-response wind and turbulence instruments, including sonic anemometers, are used more and more in aeolian sediment transport research. These instruments give information on mean wind, but also on fluctuations and turbulent statistics, such as the uw covariance, which is a direct measure of Reynolds' stress (RS) and friction velocity. This paper discusses the interpretation of sonic anemometer data, the transformations needed to get proper results and turbulence spectra, and how they are influenced by instrument size, sampling frequency, and measurement height. Turbulence spectra characterize how much the different frequencies in the turbulent signals contribute to the variance of wind speed, or to the covariance of horizontal and vertical wind speed. They are important in determining the measurement strategy when working with fast-response instruments, such as sonic anemometers, and are useful for interpreting the measurement results. Choices on the type of sonic anemometer, observation height, sampling period, sampling frequency, and filtering can be made on the basis of expected high and low-frequency losses in turbulent signals, which are affected by those variables, as well as wind speed and atmospheric stability. Friction velocity and RS, important variables in aeolian sediment transport research, are very sensitive to tilt or slope errors. During a field experiment, the slope sensitivity of the RS was established as 9% per degree of slope, which is 1.5 times the value reported in literature on the basis of theoretical considerations. An important reason for the difference probably is the large influence of streamline curvature on turbulence statistics and thereby on the slope sensitivity of the RS. An error of 9% per degree of slope in the RS will translate into an error of approximately 4% per degree of slope in the calculated friction velocity. Space-time correlation of the horizontal wind speed is much larger than that of the vertical wind speed and the instantaneous RS. This largely explains why, in previous studies, a poor correlation was found between instantaneous RS measured at 3 m height and saltation flux near the surface, whereas the correlation between wind speed at some height and saltation flux was much better. Therefore, the poor correlation between RS away from the surface and saltation flux does not contradict that saltation flux is caused by RS.