In this letter, we study the scaling properties of multi-year observed and atmospheric model-generated wind time series. We have found that the extended self-similarity holds for the observed series, and remarkably, the scaling exponents corresponding to the mesoscale range closely match the well-accepted inertial-range turbulence values. However, the scaling results from the simulated time series are significantly different.

We have analyzed long-term wind speed time series from five field sites up to a height of 300 m from the ground. Structure function-based scaling analysis has revealed that the scaling exponents in the mesoscale regime systematically depend on height. This anomalous behavior is likely caused by the buoyancy effects. In the framework of the extended self-similarity, the relative scaling exponents portray quasiuniversal behavior.