Adaptive Processing for Fully Digital Polarimetric Phased Array Weather Radar

Abstract

This dissertation investigates advanced adaptive signal processing techniques for fully digital polarimetric phased array weather radar (PAR) systems, with an emphasis on improving data quality, clutter mitigation, and operational feasibility. Conventional weather radars based on mechanically steered parabolic antennas are fundamentally limited in temporal resolution, restricting their ability to observe rapidly evolving atmospheric phenomena. Fully digital PAR systems overcome these limitations through electronic beam steering and digital beamforming (DBF), enabling rapid volumetric scanning and flexible beam synthesis. This work focuses on adaptive beamforming and space-time adaptive processing (STAP) as key enabling technologies for next-generation weather radar. While adaptive methods such as Capon beamforming offer improved angular resolution, sidelobe suppression, and interference mitigation compared to conventional Fourier-based techniques, their application in polarimetric weather radar introduces significant challenges, including calibration complexity, computational cost, and sensitivity to sample support. To address these challenges, a comprehensive framework for adaptive processing, specifically the Capon (also called minimum variance distortionless response) method, is developed, combining theoretical analysis, simulation studies, and validation using real data from the fully digital S-band Horus radar system. A novel calibration methodology is proposed to preserve dual-polarization consistency under dynamically varying adaptive beam patterns. The performance of adaptive beamforming is qualitatively evaluated against conventional methods in terms of clutter suppression, sidelobe reduction, and polarimetric variable estimation accuracy. It is also evaluated quantitatively with the impact of signal-to-noise ratio (SNR), diagonal loading, and number of pulses. In addition, space-time processing techniques are investigated for mitigating complex, non-stationary clutter sources such as wind turbines, demonstrating improved separation of meteorological signals in both spatial and Doppler domains. Practical considerations for operational deployment are also addressed. Computationally efficient implementations are developed through reduced-dimension covariance estimation, subarray-based processing, and conditional application of adaptive methods. The impact of array architecture, signal-to-noise ratio, and environmental conditions on adaptive processing performance is systematically analyzed. Results demonstrate that adaptive processing significantly enhances radar data quality, particularly in challenging environments with strong interference and spatial gradients, while improving polarimetric performance for several variables, including reflectivity, differential reflectivity, radial velocity, and differential phase, when properly calibrated. It shows that adaptive processing mitigates contamination from complex clutter that includes moving point targets and/or wind turbines, which are challenging to mitigate. However, the limitations are that (1) adaptive processing performance depends on SNR, which requires at least ~0 dB SNR from our study, (2) it requires a large number of samples, typically $2N$, twice the number of receivers, which can lead to longer dwell time, (3) there is a computational burden from adaptive processing; in one experiment, the Capon DBF took 28 times more time than the Fourier DBF to perform a full Plan Position Indicator (PPI) scan, and (4) there remains a challenge in $\rho_{\textrm{hv}}$ estimation, arising from different optimization in each H and V channel. This work establishes a pathway toward the practical integration of adaptive beamforming and space-time processing in next-generation weather radar systems.