Enhancement of Doppler weather radar signal processing

Advanced signal processing techniques applicable to the mean Doppler velocity estimation of weather radars are reviewed and in particular linear prediction (LP) based frequency estimators are studied. In order to reduce the computational burden of the LP based estimators, the Quotient-Difference based algorithm for fast frequency estimation is developed. The relationships between the classical and advance methods are established. A unified approach to three eigendecomposition-based methods for frequency estimation at low Signal-to-Noise Ratio (SNR) is also presented.; The two most troublesome problems in Doppler weather radar signal processing are range-velocity ambiguities and ground clutter. By suitable combination of two or more different Pulse Repetition Times (PRTs) it is possible to increase the unambiguous velocity effectively. Two approaches for mean Doppler frequency estimation in the presence of clutter are presented.; The filtering-estimation approach uses a high-pass filter to remove clutter. The pulse pair (PP) estimator is then applied to estimate the mean Doppler frequency of the weather signal. It is shown that these filters can be designed to work successfully with staggered PRT. The developed filter is time-varying, with periodically changing coefficients. However, the nonlinearities of the phase characteristic are concentrated in a small part of the velocity interval which requires that a special decision logic be applied for velocity estimation.; The modeling-estimation approach uses methods that allow parametric modeling of more than one signal. Therefore, it is possible to model not only weather signals but also ground clutter. It is shown that the LP based estimators have the ability to model and resolve weather signals from ground clutter in unfavorable conditions, i.e., when clutter is stronger than the signal and when only a small number of samples is available for processing. Also, using a priori information about the ground clutter the LP based estimator is simplified and a generalized PP estimator is derived. The eigendecomposition-based minimum-norm method is applied to resolve weather signals from ground clutter at SNRs as low as 10dB.