Historically, turbulence measurements for wind energy purposes have been based on variances of horizontal wind speed observed with cup anemometers (Brower 2012; Emeis 2012), usually based on an average over 10-minute intervals, as documented in IEC standards. In contrast, atmospheric scientists, micrometeorologists, and boundary-layer meteorologists usually consider “turbulence” to be based on variation in three directions, including vertically (Panofsky and Dutton 1984; Stull 1988; Arya 2001; Wallace and Hobbs 2006; Foken 2008), and quantify turbulence using an eddy-correlation approach. The eddy-correlation approach usually relies on the use of a sonic anemometer recording fluctuations of the three components of the flow measured at 10 Hz or faster. A Reynolds decomposition is applied to a time series to separate the fluctuating or turbulent component from a slowly varying or stationary mean component considered to be representative of mesoscale variability. The time interval over which this Reynolds decomposition is applied varies, but is usually on the order of 2030 minutes (Lenschow et al. 1994; de Franceschi et al. 2009). Further, other parameters defining turbulence, such as the rate of turbulence destruction or the turbulence dissipation rate, can also be employed to define turbulence.
The quantification of turbulence is closely tied to the instruments used to estimate it and therefore the effective measurement volume of the instruments. When remote-sensing devices, such as lidars, are used, variances of wind speed can be calculated. However, these variances are subject to spatial averaging from the pulse length of the laser as well as the sampling frequency of the return signal, and so turbulence statistics measured from lidar are distinct from those that would have been measured from cup or sonic anemometers and should not be considered interchangeable (Sathe et al. 2011). Some researchers refer specifically to “lidar turbulence” (e.g., Rhodes and Lundquist 2013). Recent approaches have suggested methods that might adjust lidar measurements to more closely approximate what may have been measured by a co-located sonic anemometer, but these developments are ongoing (Brugger et al. 2016; Newman and Clifton 2017). Information on turbulence, such as turbulence dissipation rate or structure functions, may also be extracted from analysis of the Doppler spectra from scanning lidars (Frehlich et al. 1998; Smalikho et al. 2005; Smalikho and Banakh 2013; Smalikho et al. 2013) and profiling lidars (O’Connor et al. 2010; Bodini et al. 2018).
Comprehensive reviews on approaches for measuring turbulence with lidars are presented in Engelbart et al. (2007), Sathe et al. (2011), and Sathe and Mann (2013). Recent work has suggested that a six-beam approach (Sathe et al. 2015) may be useful. Numerous intercomparisons of measurement approaches (Pauscher et al. 2016; Lundquist et al. 2017) have also been addressed recently or are ongoing.