Extended Abstract (436K)P5.7
A new multiple-PRT scheme for Météo-France's Doppler radar network to improve spectral moment estimation of weather radar signal and ground clutter filtering
Mohammed Tahanout, Météo France, Trappes, France; and J. Parent du Châtelet and C. Augros
This work describes a new scheme of staggered pulse repetition time (PRT) design to enhance the spectrum processing of received weather radar signal, to de-aliase Doppler velocities and to mitigate the ground clutter in Météo-France Doppler radar network “ARAMIS”, which is currently working with a triple-PRTs scheme and [– 60 to + 60 ] m.s-1 extended Nyquist interval. The problem of power spectrum recovering, and ground clutter filtering, with dual PRT scheme has been studied by Sachidananda and Zrnic in 2000. They have developed an algorithm based on magnitude deconvolution procedure using a matrix inversion. This algorithm was tested successfully on the research and development radar WSR-88D. Trying to apply this algorithm to our C-band triple PRT scheme (T1 = 2.64 ms, T2=3.08 ms and T3= 3.3 ms), we experienced problems as far as the spectral width σv is bigger than approximately 1 m.s-1.
To overcome these difficulties and make a good estimation of the weather radar spectrum, particularly in the ARAMIS context, we have developed a new method based on an optimization of the radar transmission scheme to reach a large number of points of the autocorrelation function R(τ). Basically, a dual PRT scheme gives 3 points of R(τ) for time lags 0,T1 and T2 ; a triple PRT scheme gives 4 points of R(τ) for time lags 0,T1, T2 and T3. A multiple nPRTs scheme gives (n+1) points of R(τ) for time lags 0,T1, T2 … Tn. Additional points can also be obtained for combined lags (T1+T2, T2+T3,…).
Choosing Ti+1 = (Ti + δT), the function R(τ) is available over the interval [T1 , Tn] with a uniform sampling time δT. R(τ) can be forced to zero for the other time lags and the power spectrum simply recovered using classical FFT algorithm within an extended Nyquist interval of λ/(2 δT). This power spectrum is convolved by the sinc function which is the power spectrum of the time window (Tn – T1).
After theoretical developments, we give in this paper simulation results showing the advantage of the new method with respect to the Sachidananda and Zrnic method in the Aramis network context. We then present an experimental validation campaign with an operational radar of the network, showing multiple PRTs spectra from ground echoes, and rain echoes. We finally conclude by operational recomandations.
Poster Session 5, Advanced Radar Technologies and Signal Processing I
Tuesday, 6 October 2009, 1:30 PM-3:30 PM, President's Ballroom
Previous paper Next paper