Abstract
Methods of eliminating or reducing three types of errors found in the Gill UVW anemometer have been investigated by utilizing field experiments comparing this sensor with a three-component sonic anemometer. The non-cosine response of each of the three orthogonal propellers to a wind which is not parallel to the propeller axis was adequately corrected during computer processing of the data, using the manufacturer's wind-tunnel-measured calibrations. The accepted theory describing a propeller as a first-order system with a time constant τ = L/ū (where L is a distance constant characterizing the propeller inertia and ū is the mean wind) was found to be only a fair description of the frequency response, probably due to dependence of L on properties of the flow, but was used to qualitatively delineate proper applications for this sensor. The threshold response was improved for the U and V components by orienting the anemometer so that the mean wind direction bisects the angle between the horizontal axis propellers. Improvement was also achieved for the vertical component of the wind by rotating the formerly vertical W propeller 45° into the horizontal, in the plane bisecting the horizontal propeller axes. The orthogonal components of the wind must then be calculated during computer processing of the data. Since for many applications the finite, but small, response threshold of the vertical component was not found to be a serious problem, the additional complication of this modification may be unnecessary.