Wind vector retrieval algorithm for a coherent Doppler lidar based on KNN-COOKS

The wind vector retrieval from a coherent Doppler lidar system in the plan position indicator (PPI) scanning mode often suffers from high inversion errors in horizontal wind speed and direction at far-range gates due to "erroneous" radial wind speed. To address this, we propose a weighted sine wave fitting algorithm that combines K-nearest neighbors and Cook's distance (KNN-COOKS). Numerical simulation experiments show KNN-COOKS achieves higher accuracy than direct sine wave fitting (DSWF) and adaptive iterative reweighted sine wave fitting (AIR) and performs comparably to filtered sinusoidal wave fitting (FSWF). Validation with real-world data shows KNN-COOKS increases valid data by 22.5% and 12.5% over DSWF and AIR, respectively, while reducing computation time by 62% compared to FSWF and 38% compared to AIR.