Wind Turbine Radar Interference Research
The rapid increase in the number of wind farms has created serious concerns about wind turbine interference on surveillance, navigation, and Doppler weather radars. This concern has resulted, in particular, in halts or delays in a number of proposed installations in Europe and the US.
A wind turbine is a large structure that includes a stationary mast and rotating blades. To radar, the turbine can appear as a larger target than a B747 aircraft because of the amount of signal that is scattered back to the radar. This unwanted interference is termed radar clutter. Wind turbine clutter (WTC) includes a strong interference around zero Doppler frequency due to the scattering from the stationary mast, as well as signals with large Doppler frequency spread due to different parts of the rotating blades. This spread in velocity and its resulting Doppler frequency spectrum may, for example, be misinterpreted by a weather radar processor as a storm.
A number of interference mitigation techniques have been under consideration. These include material and designs for turbines to affect stealth characteristics and systems, and signal processing approaches to reduce or eliminate the effects of the interference. The focus of our research is on signal processing mitigation approaches. Many algorithms exist for filtering out stationary clutter returns. However, these algorithms are ineffective in removing WTC because, unlike stationary clutter, WTC is usually wideband in the Doppler spectral domain and highly variable.
Our current research is built on NEXRAD Level 1 scanning and spotlight Doppler weather radar data. In azimuth-Doppler plots of the scanning data, in which we plot the frequency information of a sliding window in azimuth at a fixed range, WTC is present as high-power wideband spectral interference over certain intervals in azimuth (Fig. 1). A similar phenomenon also occurs in plots of range-Doppler, wherein azimuth angle is fixed and the frequency information of each range cell is plotted. An algorithm is currently being developed to notch the spectral coefficients in the corrupted azimuth intervals and to perform bilinear interpolation using pure weather coefficients. The interpolation is currently being conducted and compared across azimuth-Doppler, range-Doppler, and range-azimuth (Fig. 2) coefficients. The measurements of the algorithm’s performance include the recovery of both the reflectivity and the Doppler spectrum of each cell under interpolation.
In addition to the interpolation algorithm, the statistical information of the wind turbines’ Doppler components is also under investigation. If these components are viewed as random variables, statistical information such as the probability distribution is of great value for subsequent filter development and evaluation. Currently, a composite Gaussian model is being established by utilizing the data from NEXRAD.
Finally, we are investigating approahes for adaptively determining the wind turbine clutter map for a given site, using current conditions in the mitigation process. An example of such an approach using spectral and certain cross-spectral information is given in Figure 3.
Faculty: M. Kaveh
Students: B. Perfetti and J. Zheng