Study of in situ calibration performance of co-located multi-sensor hot-film and sonic anemometers using a 'virtual probe' algorithm

The performance of an in situ calibration technique, implementing neural network (NN) algorithms for co-located multi-wire hot-film and sonic anemometers, is studied. The NN-based calibration technique, proposed by Kit etal (2010 J. Atmos. Ocean. Technol. 27 23-41), allows performing direct measurement of fine scales in turbulent air flow, and offers a robust tool for measurements of micro-scale properties in atmospheric flows. Accuracy of the suggested calibration technique is examined in view of isotropic and anisotropic flow fields of various turbulence intensity (TI). Anisotropic velocity datasets of various TIs were generated using a 'virtual probe' simulating hot-film anemometer response to the sensed flow, while a kinematic model of homogeneous isotropic turbulent flow was implemented to generate isotropic flow datasets. NN calibration performance is examined by quantitative comparison between the original and reconstructed velocity components, using a specially constructed norm and by visual comparison of original and reconstructed time series. The examined NN calibration technique performance is shown to be of reasonable accuracy for all TI velocity fields examined, while a notable drop in accuracy is detected with the increase in TI. The reconstruction of isotropic velocity fields, using the NN algorithm for calibration, is apparently of slightly lower accuracy than in the case of anisotropic flows. The results are discussed in view of possible implementation of the suggested technique in direct field measurements of atmospheric turbulence fine scales.