Atmospheric Radar Research Center

In recent years, increasingly larger wind turbines are being constructed due in part to generator efficiencies and the desire to tap stronger, higher-level wind fields. Of course, the economic incentive is the main driving force and careful analysis has provided the impetus for larger wind turbine designs. A typical turbine structure consists of a tower, nacelle, rotor, and three blades.

Due to their large size, modern wind turbines cannot be operated under all wind conditions. Too light wind will not drive the large generators and too strong wind could cause damage to the turbine. Generally, wind farms will not operate unless the winds are within an appropriate range of approximately 3-25 m/s. While some models have a fixed rotor speed, others allow the rotation rate to vary with the wind speed and can be in the range of 10-20 rpm.
Wind turbines typically have two degrees of freedom to optimize power generation.

1. The ability to change their yaw or compass orientation by turning (using motors) the entire nacelle unit so the rotor is pointed directly into the wind.
   This process is controlled by wind direction information from nearby wind vanes which are located to minimize the effect due to wake turbulence from the wind turbines.
2. The pitch of the blades which can be changed to keep a near-constant rotation rate under varying wind speeds, where the rotation rate is chosen to optimize the power-generation efficiency of the turbine.
   Another purpose of both the blade pitch control and yaw mechanisms is to act as a brake under extremely strong wind conditions [General Electric Power Systems Web page Products, 2005].

Figure 2 shows this structure along with a photograph of a wind turbine located in Hagesholm, Denmark. The tower can have heights of 50-80 m with blades in the range of 30-80 m in length [American Wind Energy Association, 2001b]. The large turbine shown in the figure has a diameter of 72 m and a tower height of 68 m.

Wind farms will typically contain many wind turbines. For example, the Gray County Wind Farm near Dodge City, Kansas, has 170 such turbines [Aquila Inc., 2005]. Clusters of turbines are organized in such a way as to minimize the effects of wake turbulent, which can reduce efficiency. The general practice is to place turbines 2-3 rotor diameters apart [Butler and Johnson, 2003].

Location	Number of Towers	Height	Length of Blades	RPM	Tip Velocity
25 miles SW of Dodge City, Kansas	170	217 feet (66 m)	77 feet (23.5 m)	28.5 (fixed)	70 m/s

Ground clutter (GC) is a persistent problem for the WSR-88D network. Typically, GC signals exhibit large reflectivity, near-zero Doppler shift, small spectrum width, and are consistently localized. These features can be exploited for clutter mitigation using Level-I digital filters, such as an elliptic filter which is currently used on the WSR-88D radars [Heiss et al., 1990; Chrisman et al., 1995]. Of course, limitations exist especially for the case when the weather echo has a small Doppler shift. Improvements on the current GC filter design have been developed for applications such as non-uniformly sampled data [Torres and Zrnic, 1999]. Further, the Open RDA (ORDA) upgrade to the WSR-88D network will facilitate algorithms based on spectral processing [Siggia and Passarelli, 2005].
Compared to commonly occurring GC, interference caused by wind turbines is a much more difficult challenge. Direct reflections will be received from both the tower (stationary) and the blades (non-stationary). Like GC, the Wind Turbine Clutter (WTC) signal should still have a significantly large reflectivity, with a possible modulation due to blade rotation causing a systematic variation in radar cross-section.
The Doppler shift will be affected by several factors, including the blade rotation speed and rotor orientation with respect to the radar beam. Doppler velocities should be maximum when the rotor axis is oriented 90° from the radar line-of-sight and near zero when the rotor axis is along the line-of-sight. Since the resolution volume of the radar will likely encompass the entire wind turbine structure, it is expected that the spectrum width will be significantly enlarged.
This is due to the blade rotation away and toward the radar. Multiple turbines within one resolution volume would only exacerbate this effect.

In addition to WTC signals caused by reflections from the actual wind turbines, backscattered energy from turbulent eddies in the wake of the wind farm may be observed. It is expected that these echoes would exhibit characteristics similar to clear-air backscatter from discontinuities in the refractive index at the Bragg scale of the radar. These wake echoes would drift with the wind field and would likely have much lower reflectivity compared to the direct reflections from the turbines. Nevertheless, they could significantly enlarge the radar coverage area affected by WTC and thus exacerbate the problem.

Two distinct examples of Level-II data from the Dodge City, Kansas, WSR-88D radar (KDDC) are provided in Figure 3. As expected, the reflectivity shows large values near 45 dBZ with sporadically large spectrum widths of over 10 m/s. The relatively small region of high reflectivity to the south-west of the radar is clearly visible and matches the location of the Gray County Wind Farm approximately 45 km from KDDC.
Without prior knowledge, it would be extremely difficult to distinguish between the WTC and the thunderstorms. Since the blades rotate toward and away from the radar, one would expect a near-zero mean Doppler velocity. Of course, the large spectrum widths will reduce the accuracy of the Doppler velocity estimates as illustrated in the figure by small deviations from zero.
Given the complexity of the WTC signal, Level-I data are necessary to gain a proper understanding and for a complete analysis.