Estimation of Doppler Spectrum Parameters Comparison between FFT-based processing and Adaptive Filtering Processing J. Figueras i Ventura 1, M. Pinsky 2, A. Sterking 3, A. Khain 2, H.W.J. Russchenberg 1 1 IRCTR-TU Delft 2 Institute of Earth Sciences-The Hebrew University of Jerusalem 3 Weizmann Institute of Science
ERAD 2006: Sept Contents Signal and noise models of the adaptive processing Adaptive filter estimation Implementation of the two algorithms Comparison of the two algorithms
ERAD 2006: Sept Model of the atmospheric radar signal For each range cell: Complex autoregressive series of first order with unknown coefficient Spectrum symmetric relative to average Doppler frequency
ERAD 2006: Sept Calculation of Doppler spectrum parameters Doppler spectrum parameters Mean Doppler frequency: Doppler spectrum width:
ERAD 2006: Sept Simulated Signal Simulated signal: Noise: uncorrelated additive normal complex white noise
ERAD 2006: Sept Adaptive filter estimation Objective: Suppress noise Accurately estimate a
ERAD 2006: Sept Adaptive filter estimation Recurrent estimation of a
ERAD 2006: Sept Implementation of the algorithms
ERAD 2006: Sept Comparison: Processing speed Number of operations: FFT: 6N log 2 N Adaptive: 41 N Advantage in real time operations !
ERAD 2006: Sept Convergence of spectral moments Overestimation of Doppler spectrum width Mean Frequency: 0.5 Doppler width: 0.4
ERAD 2006: Sept Estimation of weak signals ParameterGivenAdaptiveFFT 6 dB Clipping Signal Power NaN Mean Velocity 11.03NaN Doppler width NaN SNR=0.09
ERAD 2006: Sept Comparison: Real Data Clipping:
ERAD 2006: Sept Conclusion Adaptive Doppler processing: Good estimation of the Reflectivity and mean Doppler velocity Overestimation of Doppler spectrum width Faster signal processing than FFT processing Estimation of weaker signals Simple signal model: Bi-modal atmospheric signal?