The Autocorrelation Spectral Density for Doppler-Weather-Radar Signal Analysis

Abstract

Time-domain autocovariance processing is widely accepted as a computationally efficient method to estimate the first three spectral moments of Doppler weather radar signals (i.e., mean signal power, mean Doppler velocity, and spectrum width). However, when signals with different frequency content (e.g., ground clutter) contaminate the weather signal, spectral processing using the periodogram estimator of the power spectral density (PSD) is the preferred tool of analysis. After spectral processing (i.e., filtering), a PSD-based autocorrelation estimator is typically employed to produce unbiased estimates of the weather-signal spectral moments. However, the PSD does not convey explicit phase information, which has the potential to aid in the spectral analysis of radar signals. In this paper, the autocorrelation spectral density (ASD) is introduced for spectral analysis of weather-radar signals as a generalization of the classical PSD, and an ASD-based autocorrelation estimator is proposed to produce unbiased estimates of the weather-signal spectral moments. A significant advantage of the ASD over the PSD is that it provides explicit phase information that can be exploited to identify and remove certain types of contaminant signals. Thus, the ASD provides an alternative means for spectral analysis, which can lead to improved quality of meteorological data from weather radars.