r = V max t 2. 2 DITREC, [16 ] [16 ] TREC, [16, ] t i f ( x, y, t i ) [16 ] D ITREC(Difference Image2based Tracking Radar E2 t j f ( x, y, t j

64 5 2 0 0 6 1 0 ACTA METEOROLO GICA SIN ICA Vol. 64, No. 5 October 2006 Ξ 1,2,3 1 4 1 5 1,, 100081 2,, 210044 3,, 401147 4,, 210008 5,, 430074 TREC ( Tracking Radar Echo by Correla2 tion), ( ) (Difference Image2 based Tracking Radar Echo by Correlations) DITREC TREC,DITREC TREC DITREC TREC TREC,DITREC TREC DITREC 700 hpa CINRAD/ SA 1 h,ditrec 700 hpa Z2R :, 1 [1 ], ( [629 ], Rinehart ) [10 ] [2 ] TREC ( Tracking Radar Echo by Correlation),Smythe [11 ] Tuttle [12 ] [3 ], TREC [5,13214 ] Li [15 ] TREC TREC CO TREC, (Continuity of TREC vectors) TREC (QPF) [4 ] [5 ], 2 Ξ :2005 4 26 ; :2006 3 3 : (40375007,40575009,40475017) (BS2003054) :,1969, E2mail :ZYP @121. cq. cn

632 64 r = V max t 2. 2 DITREC, [16 ] [16 ] TREC, [16, ] t i f ( x, y, t i ) [16 ] D ITREC(Difference Image2based Tracking Radar E2 t j f ( x, y, t j ) cho by Correlations) TREC f ( x, y, t i ) - f ( x, y, t j ) DITREC, DITREC D ij ( x, y) = f ( x, y, t i ) - f ( x, y, t j ) > T 700 hpa CIN2 0 RAD/ SA 1 h (2) T Z2R 1 h, D ij ( x, y ) DITREC 700 hpa f ( x, y, t i ) - f ( x, y, t j ) T 1 R = 6 X 1( i) X 2 ( i) - n - 1 6 X 1( i) 6 X 2( i) [ 6 X2 1( i) - nx 2 1) ( 6 X2 2( i) - nx 2 2) ] 1/ 2 (1) X 1 X 2, n DITREC V max 2, T = 1 db 1 t 1 Z( x, y, t 1 ) 2003 7 8 10 :09 ( 2. 1 TREC ) (32. 194 N, 118. 401 E) CIN2 TREC RAD/ SA 1. 5 PPI t ( PPI), t 1 t ( 5 min) v ( 19 m/ s ) Z ( x, y, t 2 ) t 2 t Z ( x, y, t 3 ), 1d 1e ( 2) D 12 D 23 D 12 v t 1 t D 23, t 2 TREC t 1, (DITREC) D 12 t TREC, D 23 ( TREC ) t 1 t 2 TREC D 12 t 1 t 2 t DITREC ( DITREC ) 0 t 1 t 3 TREC

5 : 633 DITREC 15 km 44 m/ s ( ),,,,, [12 ] 3 3. 1 2003 7 8 (31. 866 N, 117. 257 E) CINRAD/ SA 2005 CINRAD/ SA 7 19 (27. 995 N, 120. 733 E) CINRAD/ SA 366) 1 km 1, km 39 km 39 km, 1521, 6 km, x ( Rinehart [19 ] TREC ) y ( ) 5 min, [17 ] 2 2003 7 8 1. 5 PPI 4 TREC ( 1) 10 dbz 4a (30 km, - 10 km) (65 km, - 50 km) TREC T 1 db ( 2a 4b 2c) [18 ] 7 8 500, 700, 850 hpa 4d (50 km,100 km) ( ) [18 ],, 7 8 12, 3, 4e (50 km,60 km) 50 km 50 km 500 hpa 5d 24 32 m/ s, 700 hpa 18 20 m/ s, 850 hpa 12 16 m/ s [18 ] 3 2005 7 19 1. 5 PPI 2005 0505 4c 4f,DITREC 19 09, Tuttle [12 ] Li [15 ] Tut2 19 18 tle [20 ] TREC 5 3 39 km 39 km ( 2) 3. 2, ( 1 km, TREC (chaotic) 43 m/ s (25 km,10 km) Li [15 ], CO TREC 5a 2003 7 8 15 :19 TREC DITREC DITREC ( 4c) 5b 1 26 2005 7 19 17 :13 m/ s 10 km DITREC ( 4f) 5 29 m/ s 2, 4 TREC 38 m/ s DITREC

634 64 2 3 [21222, TREC ] [21222 ] 4, DITREC 3 150 m TREC, 1 km T DITREC 700 hpa, DITREC 1 h, TREC 6 TREC DITREC 1 h DITREC 1 700 hpa 7 8 13 :47 15 :57 24 TREC DITREC 340 s ( 4. 2 6a 6d) 2 8a 2003 7 8 13 :57 ( 6e,6f) 8b 13 :48,13 :53 13 :58 1. 5 6a 6d, PPI DITREC TREC DITREC 8b 14 :44,14 :49 14 :56 1. 5 PPI, DITREC ( ) 13 :57 ( 8a) 14 :55 ( ) [20 ], 5 min [20 ] TREC DITREC Z2R ( Z, ( 6e,6f) 6 TREC mm 6 / m 3, R mm/ h), TREC 1 h 700 hpa ( 6a,6c 6e) DITREC ( 7 8 12 ( 6b,6d 6f) TREC DITREC 143 km 83 ) 235 TREC DITREC 18 m/ s ; ( 150 km 188 ) ( 7a 1 7b 2), DITREC TREC 179 km 311 ) 220 8 m/ s DITREC TREC TREC 225 15 m/ s DITREC Z2R 1 h 4 4. 1 Z2R 1 h 1 h 225, 19 m/ s ; ( 3 700 hpa Z2R ( ) 150 km [21,23 ] WSR288D Z = 300 R 1. 4 (3) Z = 200 R 1. 6 (4)

5 : 635 6 TREC DITREC (a. 4a 4b TREC 5a, ; b. 4c DITREC 5a ; c,d a,b, ; e. 4d 4e TREC 5b ; f. 4f DITREC 5b ; 1 :1 ( ) R N E RMS ) Fig. 6 Scatterplots of TREC and DITREC radial components versus averaged Doppler radial velocities (a. TREC radial components shown in Fig. 4a and Fig. 4b versus averaged Doppler velocities in Fig. 5a, but only the data north of Hefei radar were used ; b. DITREC radial components in Fig. 4c versus averaged Doppler velocities in Fig 5a, but only the data north of Hefei radar were used ; c,d. same as (a) and (b),respectively, except the data south of Hefei radar were used ; e. TREC radial components in Fig. 4d and Fig. 4e versus averaged Doppler velocities in Fig. 5b ; f. DITREC radial components in Fig. 4f versus averaged Doppler radial velocities in Fig 5b. The 1 :1 line, regression line, correlation coefficient R between compared quantities, and the number, N, of retrieved vectors, and the Root Mean Squared Error ( E RMS ) are also given in each plot)

636 64 144, 14 :00 N OBS 15 :00 8c, 6 ( P n - G n ) 2 n = 1 E 150 km RMS = (9) N OBS, R H ( Hit Rate), 9, R FA ( False Alarm Rate), Y Y,, P NA ( No Alarm Probability), Y N,, I CS (Critical Success Index), E N Y, RMS ( Root Mean Squared Error), (9) [24 ] : R H = Y Y / ( Y Y + N Y ) (5) R FA = Y N / ( Y Y + Y N ) (6) P NA = N Y / ( Y Y + N Y ) (7) I CS = Y Y / ( Y Y + N Y + Y N ) (8) N OBS, P n, G n 1 N = Y Y + Y N + N Y 1 Table 1 Evaluation of the extrapolative hourly precipitation and the accumulated hourly precipitation retrieved from the radar data Z2R (mm) / N Y Y Y N N Y R H ( %) R FA ( %) P NA ( %) I CS ( %) E RMS (mm) 700 hpa 151 82 31 38 68 27 32 54 7. 09 2 DITREC 149 86 29 34 72 25 28 58 5. 86 Radar 147 103 27 17 86 21 14 70 4. 61 700 hpa 85 23 16 46 33 41 67 27 8. 76 5 DITREC 87 26 18 43 38 41 62 30 7. 49 Z = 300 R 1. 4 Radar 80 43 11 26 62 20 38 54 5. 86 700 hpa 37 4 11 22 15 73 85 11 12. 51 10 DITREC 33 9 5 19 32 36 68 27 10. 88 Radar 31 11 5 15 42 31 58 35 7. 85 700 hpa 16 2 6 8 20 75 80 13 17. 23 15 DITREC 12 2 2 8 20 50 80 17 14. 29 Radar 13 3 3 7 30 50 70 23 10. 61 700 hpa 161 90 41 30 75 31 25 56 6. 79 2 DITREC 153 92 32 29 76 26 24 60 6. 34 Radar 151 106 31 14 88 23 12 70 4. 63 700 hpa 85 23 16 46 33 41 67 27 8. 44 5 DITREC 87 26 18 43 38 41 62 30 7. 44 Z = 200 R 1. 6 Radar 80 43 11 26 62 20 38 54 5. 92 700 hpa 37 4 11 22 15 73 85 11 12. 31 10 DITREC 31 8 5 18 31 38 69 26 10. 76 Radar 31 11 5 15 42 31 58 35 8. 34 700 hpa 12 0 6 6 0 100 100 0 22. 14 15 DITREC 13 1 2 10 9 67 91 8 15. 18 Radar 13 3 3 7 30 50 70 23 11. 62

5 : 637

638 64

5 : 639

640 64

5 : 641

642 64

5 : 643 14 : 00 15 : 00 15 : 00 16 : 00, 700 hpa 700 hpa,ditrec DITREC 9 m/ s (90 km, - 60 km),radar 20 m/ s, 1 : Z2R (3), 700 hpa (3) (4) ;, 700 hpa, ; DITREC 700 hpa 2 Table 2 The comparing the extrapolative hourly precipitation with 8d Z2R (3) the accumulated hourly precipitation retrieved from the radar data (14 :00 15 : 00) 8e Z2R (3) DITREC (14 : 00 15 :00) 8f Z2R (3) 700 hpa (14 :00 15 :00) 8c 8d,14 :00 15 :00 (0 km, - 40 km), 8c (0 km, - 80 km),, 5 TREC (1) 700 hpa DITREC DITREC DITREC 1 h, 8, : 301 km 301 km (1) DITREC TREC 9 2 DITREC TREC DITREC TREC N d, N e N f 8d 8e 8f, R de (2) DITREC ( 8e) TREC DITREC ( 8d), R df, 700 hpa ( 8f) ( TREC DITREC 8d) 2,, (3) 3 5 min, DITREC, T DITREC DITREC TREC 700 hpa (4) DITREC 700 hpa 15 :19 DITREC CIDRAD/ SA ( 4c) ( - 100 km,100 km) DITREC 4 m/ s (mm) N d N e N f R de ( %) R df ( %) 2 41300 44604 40973 74 65 5 18704 19541 17716 69 60 10 6605 7339 6830 65 54 15 2667 3090 2596 63 52 20 2013 2117 1801 60 49 25 976 911 701 45 29 30 563 529 281 35 20 35 298 210 90 24 12 40 142 44 1 15 0 1 h

644 64 2 min, 20 min Z2R Z2R DITREC ; DITREC, 700 hpa DITREC 700 hpa Z2R Z2R Z2R (5) Li [13 ], Z2R 5 min DITREC Z2R 700 hpa, 3, 5 min DITREC 2 1 [12,15,20 ] TREC Smith [25 ],, [26 ] 700 hpa :, (pattern matching) TREC (segmentations) TREC [1 ],,.. :, 2004. 254pp 39 km 39 km TREC, DITREC TREC 2004. 254pp TREC DITREC ( ) Li [15 ] 2 min CO TREC, Z2R,DITREC Cheng Minghu, Liu Liping, Zhang Peiyuan, et al. The Re2 trieval Theories and Techniques of Observed Doppler Radar Data for Heavy Rainfall Events. Beijing : China Meteorological Press, [ 2 ],,. 6. 30. :. (7). :, 2003. 423pp Yang Hongping, Wan Rong, Shi Yan, et al. The precipitation estimation of 6. 30 heavy rainfall process in Huaihe basin using tri2weather radar. In : CMS Commission for Hydrometeorology, eds. The Science and Technology Innovations and the Develop2 ment in Atmosphere Science (7). Beijing : China Meteorological Press, 2003. 423pp [3 ],. 2003., 2004, 62 (6) : 8032813

5 : 645 Yao Xuexiang, Xu Jing. A study of volumetric precipitation dur2 ing the Huaihe river basin floods in 2003. Acta Meteor Sinica (in Chinese), 2004, 62 (6) : 8032813 [ 4 ] Wilson J W, Crook N A, Muller C K, et al. Nowcasting thun2 derstorms: a status report. Bull Amer Meteor Soc, 1998, 79 (10) : 207922099 [ 5 ] Berenguer M, Davila J, Corral C, et al. Hydrological evaluation of a nowcasting technique applied to flood forecasting. Preprints, 31th Conf on Radar Meteorology, Seattle, Washington, Amer Meteor Soc, 2003. 7082709 [6 ] Hilst G R, Russo J A J r. An objective extrapolation technique for semi2conservative fields with an application to radar patterns. Tech Memo No 3, Travelers Weather Research Center, Har2 ford, CT, 1960. 34 pp [7 ] Kessler E, Russo J A. Statistical properties of weather echoes. Preprints 10th Weather Radar Conf, Washington, D C, Amer Meteor Soc, 1963. 25233 [ 8 ] Crane R K. Automatic cell detection and tracking. IEEE Trans. Geosci Electron, 1979, GE217 : 2502262 [9 ] Bjerkaas C L, Forsyth D E. Operational test of a three2dimen2 sional echo tracking program. Preprints 19th Conf Radar Meteo2 rology, Miami Beach, Amer Meteor Soc, 1980. 2442247 [ 10 ] Rinehart R E, Garvey E T. Three2dimensional storm motion de2 tection by conventional weather radar. Nature, 1978, 273 : 2872289 [11 ] Smythe G R, Zrni Η D S. Correlation analysis of Doppler radar data and retrieval of the horizontal wind. J Climate Appl Mete2 or, 1983, 22 : 2972311 [12 ] Tuttle J D, Foot G B. Determination of the boundary layer air2 flow from a single Doppler radar. J Atmos Ocean Tech, 1990, 7 : 2182232 [13 ] Li P W, Wong W K, Chan K Y, et al. SWIRLs2an evolving nowcasting system. Technical Note, 100, Hong Kong Observa2 tory, 2000. 28pp [14 ] Mueller C, Saxen T, Roberts R, et al. NCAR Auto2Nowcast System. Wea Foreca, 2003, 18 (4) : 5452561 [ 15 ] Li L, Schmid W, Joss J. Nowcasting of motion and growth of precipitation with radar over a complex orography. J Appl Mete2 or, 1995, 34 : 128621300 [16 ] Jain R, Kasturi R, Schunck B. Machine vision. New York : McGraw2Hill, 1995. 568pp [17 ] Dixon M, Wiener G. TITAN : Thunderstorm Identification, Tracking, Analysis, and Nowcasting : A radar2based methodolo2 gy. J Atmos Oceanic Tech, 1993,10 (6) : 7852797 [18 ],,. 2003 7 8., 2004, 30 (1) : 38240 Zheng Yuanyuan, Yu Xiaoding, Fang Chong, et al. Analysis of a series of tornado events during 8 J uly 2003 in Anhui province with new generation weather radar data. Meteor Mon (in Chi2 nese), 2004, 30 (1) : 38240 [ 19 ] Rinehart R E. A pattern recognition technique for use with con2 ventional weather radar to determine internal storm motions. At2 mos Tech, 1981, 13 : 1192134 [20 ] Tuttle J, Gall R. A single2radar technique for estimating the winds in tropical cyclones. Bull Amer Meteor Soc, 1999, 80 (4) : 6532668 [21 ] Fulton R A, Breidenbach J P, Seo D, et al. The WSR288D rainfall algotithm. Wea Forecasting, 1998, 13 : 3772395 [22 ],,.., 2002, 25 (5) : 6402647 Zhang Yaping, Liu J un, Xia Wenmei, et al. The calculation of beam blockage coefficients in estimaing regional precipitation with radar. J Nanjing Inst Meteor (in Chinese), 2002, 25 (5) : 6402647 [23 ] Marshall J S, Palmer W M. The distribution of raindrops with size. J Meteor, 1948, 5 : 1652166 [ 24 ] Colle B, Westrick K, Mass C F. Evaluation of MM5 and Eta210 precipitation forecasts over the Pacific Northwest during the cool season. Wea Foreca, 1999, 14 : 1372154 [25 ] Smith D L. The application of manually digitized radar data to short2range precipitation forecasting. Preprints 16th Conf on Radar Meteorology, Houston, Texas, Amer Meteor Soc, 1975. 3472352 [ 26 ] Tatehira R, Sato H, Makino Y. Short2term forecasting of digi2 tised echo pattern. Kisho2cho Kenkyu Jiho, 1976, 26 : 1882189

646 64 ESTIMATION OF WEATHER RADAR ECHO MOTION FIELD AND ITS APPL ICATION TO PRECIPITATION NOWCASTING Zhang Yaping 1,2,3 Cheng Minghu 1 Xia Wenmei 4 Cui Zhehu 1 Yang Hongping 5 1 Chinese Academy of Meteorological Sciences, Beijing 100081 2 N anjing U niversity of Inf orm ation Science & Technology, N anjing 210044 3 Chongqing M unicipal Meteorological B ureau, Chongqing 401147 4 Jiangsu Institute of Meteorolodical Sciences, N anjing 210008 5 Institute of Heavy Rain, CMA, W uhan 430074 Abstract An extension of TREC ( Tracking Radar Echo by Correlations ) technique, named Difference Image2based Tracking Radar Echo by Correlations (D ITREC) is presented and evaluated, in which individual feat ures in two difference images f rom t hree temporally successive radar reflectivity images are t racked, by finding t he maxima in t he cross2correlation f unction between t he two successive difference images. Examples are presented where the TREC and DITREC algorithm were applied to two cases : one is the squall line case on 8 J uly 2003 from Hefei CINRAD/ SA radar,and the other the tropical revolving storm case on 19 J uly 2005 from Whenzhou CINRAD/ SA radar. For the two cases, Doppler radar data were available and the radial components (with respect to the radar) computed by the TREC and DITREC could be compared. It is found that the DITREC eliminated those chaotic vectors shown in the TREC which are caused by the rapid changes of reflectivity within the radar echo patterns, and improved the temporal and spatial continuity of the velocity field of echo motion. In general, the DITREC radial componenrs were consistent with the TREC ones, and the TREC and DITREC appeared to work in the lower few kilometers of the tropical revolving storm where the vertical wind shear was relatively weak. On the other hand, the TREC and DITREC vectors determined in squall line situations tended to reflect the mean advection of the rain cell, which could be quite different from the Doppler radar observations. Advecting the hybrid scan reflectivity measured by HeFei Doppler radar with the field of the DITREC vec2 tors and t he 700 hpa wind fields, yields t he ext rapolative hourly precipitations, respectively. The evaluation of t he ext rapolated hourly precipitations based on t he gauge rain data shows t hat t he hourly precipitation ext rapolat2 ed using the DITREC vectors is closer to the gauge rain data than one by the 700 hpa wind fields, however its precision is also related to t he Z2R relation used. The growt h and decay of radar echoes are not considered when advecting t he hybrid scan reflectivity. The future work may be to investigate how to reduce the errors due to nonlinear temporal changes in the intensity and velocity of rainfall patterns. Key words : Weat her radar, Echo motion field, Precipitation, Nowcasting.