Proceedings with Extended Abstracts (single PDF file) - Radio ...

WIND MEASUREMENTS BY THE CHUNG-LI RADARIN THE PRESENCE OF STRONG CLUTTER AND HARD TARGETSEleanor Praskovskaya 1 , Alexander Praskovsky 2 , Jenn-Shyong Chen 3 , and Yen-Hsyang Chu 41 Colorado Research Associates, 3380 Mitchell Lane, Boulder, CO 80301, USA2 National Center for Atmospheric Research, 3450 Mitchell Lane, Boulder, CO 80301, USA3 Chien-Kuo Institute of Technology, Chang-Hua, Taiwan 500, R.O.C. (Taiwan)4 Institute of Space Science, National Central Univ., Chung-Li, Taiwan 320, R.O.C. (Taiwan)The received signals from atmospheric spaced antenna (SA) radars might be contaminated byground clutter and, although less frequently, hard targets. The contamination adverselyaffects the performance of the correlation function (CF)-based data analysis techniques. Letus consider the standard complex signals from two receivers E1(t)and E2( t)where t is time.The second order cross CF is defined as follows (Doviak et al., 1996):**C ( τ ) = E ( t)E ( t + τ ) E ( t)E ( )(1)12 1 21 1tHereafter τ is the temporal separation, the brackets denote the ensemble averages, and thesuperscript * denotes complex conjugation. The auto CF C11(τ ) is a particular case of (1) atE 1(t ) = E ( t)2. The function C ( 12τ ) is not affected by white noise with the temporal scaleTn= 0, and C11(τ ) is affected by such noise only at τ = 0. For this reason, white noise in thereceived signals can be easily taken into account in the CF-based techniques such as FullCorrelation Analysis (FCA) and the Holloway – Doviak method; see Briggs (1984), Briggsand Vincent (1992); and Doviak et al. (1996), Holloway et al. (1997), respectively. However,no reliable approach to measuring characteristics of a scattering medium with CF-basedtechniques have been yet developed when noise with a large scale T ≥ τ , such as groundclutter and hard targets, is present; here τpis a separation where C12 ( τp) has the maximum.The major problem is that such noise affects CF at all separations τ and often cannot bedistinguished from a signal itself.A structure function (SF)-based method UCAR-STARS (University Corporation forAtmospheric Research - STructure function Analysis of Received Signals) has beendeveloped recently by Praskovsky and Praskovskaya (2003a, b). The STARS unique featureis a low sensitivity to noise with large temporal scaleT n. This feature was shown theoreticallyin Praskovsky and Praskovskaya (2003a), and demonstrated experimentally for the case ofstrong ground clutter in Praskovsky et al. (2003). The STARS low sensitivity to noise withlarge Tnhas simple physical explanation. The second order cross SF is defined as follows(Tatarskii, 1971, chap. 1A):2D ( τ ) = S ( t)− S ( t + τ ) S ( t)− S ( t(2)[ ] [ ] 212 1 21 1)where S ( t)= E(t)E * ( t)is the instantaneous signal power. The auto SF is a particular case of(2) at S1(t)= S2( t). The increment S1( t)− S2( t + τ ) is a band-pass filter that extractsfluctuations with temporal scales around τ . Only small temporal separations τ ≤ 3δtareused in STARS, where δtis the radar inter-sample time interval. It is clear that anycomponent in the received signal with a sufficiently large temporal scale T n>> 3δtcannotaffect SF because it is merely filtered by the increment. The objective of this paper is anp230