[A-AS08] General Meteorology

[AAS08-08] Wind Turbine Clutter Mitigation Using Sparse Modeling for Weather Radar

*Hiroki Matsumoto¹, Daichi Kitahara¹, Yuuki Wada¹, Tomoo Ushio¹ (1.Osaka University)

Keywords:weather radar, wind turbine clutter, sparse modeling, range-frequency domain, convex optimization

Renewable energies, such as wind power generation, are attracting increasing attention from the perspective of environmental protection and energy self-sufficiency, but these wind power facilities (i.e., wind turbines) can cause false echoes and have a significant impact on radar weather observations. False echoes, called clutters, appear when radio waves reflected by objects other than raindrops (such as the ground, sea surface, and buildings) are received by the radar. Pulse Doppler radar detects the Doppler velocity of the target by transmitting radio waves several times in the same direction and analyzing the correlation of each received signal. While precipitation echo shows non-zero Doppler velocity due to wind and the falling motion of raindrops, ground clutter shows zero Doppler velocity. Therefore, the conventional Moving Target Indicator (MTI) method separates the precipitation echo and the ground clutter by removing the frequency component (DC component) that corresponds to the zero Doppler velocity. However, the MTI method cannot remove clutters from rotating wind turbines because they also show non-zero Doppler velocity. In this research, we propose a wind turbine clutter mitigation method based on the sparse modeling. From the fact that raindrops are distributed targets and the variations in wind speed and the falling velocity of each raindrop, the Doppler velocity distribution of precipitation echo has a spread in the range and velocity (frequency) directions. On the other hand, since each wind turbine can be regarded as a point target rotating at a constant speed, the Doppler velocity distribution of wind turbine echo is concentrated at a single point. Therefore, we assume that the wind turbine echo is sparse in the range-frequency domain. Then, we construct a convex optimization problem based on the idea of the sparse modeling and solve this problem by an iterative algorithm to separate the precipitation echo and the wind turbine echo. In simulations, we use received signals in which artificial wind turbine echoes are added to actual precipitation echoes and show that the proposed method can adequately separate the wind turbine echoes from the received signals.