A Signal Processing Technique to Mitigate Wind Turbine Clutter on the NEXRAD Network

In the US, the amount of electricity generated by wind continues to increase at a record pace with wind turbines growing both in number and size. It is well known that wind turbines in the radar’s line of sight can negatively impact weather radar data, even if not in the immediate vicinity of the radar. If not mitigated, radar returns from wind turbines, referred to as wind turbine clutter (WTC), can severely contaminate weather signals and introduce significant biases in all radar products. To address the coexistence of weather radar and wind turbines at the national level, the US Congress encouraged the National Oceanic and Atmospheric Administration (NOAA) to “identify, test, and validate proposed technical solutions to reduce potential wind turbine impacts to NEXRAD radar data”. Compared to stationary ground clutter, WTC contamination is more difficult to mitigate due to the motion of the wind-turbine blades, which produces non-stationary radar signals with varying spectral content. Additionally, the presence of multiple turbines in relatively small geographical areas (wind farms) can result in complex returns and may include multipath effects. Conventional ground clutter filters that are designed to remove contamination from stationary clutter are typically unable to fully mitigate WTC; thus, special techniques must be used that are tailored to the unique characteristics of WTC. Several WTC mitigation techniques have been proposed in the literature, but none have been shown to (1) perform well under a wide range of realistic contamination conditions, (2) be suitable for real-time implementation, and (3) be compatible with operational radar scans, which are all requirements for an eventual operational implementation. In this work, we present a novel signal-processing technique for WTC mitigation that is targeted for implementation on the NEXRAD network. It uses features in the short-time Fourier transform (STFT), which allows for the spectral analysis of localized segments of the radar signal as it changes over time. The STFT features are heuristically combined using a relatively simple but robust set of rules to determine spectral bins with and without WTC contamination. Finally, this information is used in a clutter-filtering-like scheme to remove as much of the WTC as possible while preserving the weather signal. The performance of the technique was evaluated using data with WTC collected with NEXRAD radars under a variety of realistic weather conditions. Our results confirm that the proposed technique can filter WTC contamination and recover the underlying weather signals in a wide range of conditions. To our knowledge, this is the first signal-processing technique for WTC mitigation that is compatible with NEXRAD operations and, because of it, has the potential to help alleviate growing and competing concerns about the coexistence of weather radar and wind turbines.