6 : 797, 3 B9SI,,, [9 ] (CSI) 2. 1 WSR288D Build 7. 0, ( B7SI) B7SI 1997,Johnson [10 ] 1 7, ( T Z ), T Z ;, ( T DZ ), WSR288D Build : 1 ( L S

Size: px

Start display at page:

Download "6 : 797, 3 B9SI,,, [9 ] (CSI) 2. 1 WSR288D Build 7. 0, ( B7SI) B7SI 1997,Johnson [10 ] 1 7, ( T Z ), T Z ;, ( T DZ ), WSR288D Build : 1 ( L S"

爷骅牛
5 years ago
Views:

1 ACTA METEOROLO GICA SIN ICA Vol. 64, No. 6 December 2006 Ξ 1, , , , : (1) WSR288D Build 7. 0, (2) WSR288D Biuld 9. 0 (B9SI) 7, 2D B9SI,, (3) CSI B9SI B9SI, CSI ;,, ; : ; ;,, :, Austin [4 ], 20 70, Rosenfeld [5 ], Michael [6 ] : Rinehart [1 ] C ij, ; Rinehent [2 ], [3 ; ] N EXRAD (WSR288D Build 7. 0) ;, ;,,,Mansur [7 ],Arthur [8 ] Ξ : ; : : (2004B ) ( ) :,,1974 com. cn

2 6 : 797, 3 B9SI,,, [9 ] (CSI) 2. 1 WSR288D Build 7. 0, ( B7SI) B7SI 1997,Johnson [10 ] 1 7, ( T Z ), T Z ;, ( T DZ ), WSR288D Build : 1 ( L S ), ( 1 A ), WSR288D Build 7. 0 ( ( 2) : B7SI),, ( T AZI ) ( T BIN ) ( 1 B ) ; 2 WSR288D Build 9. 0 ( 1 C ) (B9SI) B7SI, ( N STORM ) ( N DZ ),, N DZ T DZ, 1 B7SI Table 1 Operational parameters of storm segment in the B7SI algorithm ( ), 1 ( N DZ ) 2 ( ) 2 ( T DZ ) (dbz) 25 3 ( T Z ) (dbz) 30 4 ( L S ) (km) B7SI Table 2 Operational parameters of storm centroid in the B7SI algorithm ( ) 1 ( T AZI ) ( ) 1 2 ( T BIN ) ( ) 2 3 ( N STORM ) ( ) 20

3 (A ;B, ;C ) Fig. 1 A general schematic for storm identification ( Part A, identifying storm segments ; Part B, defining a storm component ; Part C, constituting a storm cell) 2. 2 WSR288D Build B9SI ( B9SI) Table 3 Operational parameters added in the B9SI B7SI,,B9SI : 2D ( 3) (1) :B9SI 7 ( Z 1-7 ),B7SI,B9SI,, 7 ( ) 1 Z (dbz)30,35,40,45,50,55,60 2 R LONG (km) R DEL ETE (km) D DEL ETE (km) 4. 0, 2 30,35,40,45,50 dbz ( 2) (2) : Fig. 2 Storm segments of different lengths identified using thresholds of 30, 35, 40, 45, and 50 dbz,, respectively,along a radial (3) 2D : 2D 2. 3 ( CSI) ( R LON G ) 2D B9SI B7SI ;,, ;, : ( R DEL ETE ), ( D DEL ETE ), B9SI

4 6 : 799, ( I C ), ;, : : B9SI 7 ( [0,1 ] ) ( 25, 30, 35, 40, 45, 50, 55 dbz), ;, I C ( 3) CSI (, ) Fig. 3 Flow chart for the CSI algorithm (Steps of processing include : inputting the radar base data and the output of the B9SI, generating the features derived from the base data fields, using a fuzzy logic engine to determine the initial interest output, and finally outputting the convection index),2002 Kessinger,, [14 ] WSR288D, 1995 Steiner [11 ] BET [15 40 dbz ], [16 ] 2000 Biggerstaff [12 ] [17 ] 3 : ( Z Texture ) 3 km, ( Z SIGN ), ( 9 H ) ( Q V IL ) 2. 0 dbz/ km ; BBF ( V ),, ;, ( 4) 4, 2002 Table 4 Characteristic quantities and their weight coefficients Seo [13 ] WSR288D CSSA ( Convective/ used in the fuzzy logic engine Z Stratiform Separation Algorithm) Texture Z SIGN Q 9 VIL V H

5 ( Z Texture ) 4 km 4, 0. 5 PPI ( N beams N gates ) : Z Texture = N N beams gates 6 6 j = 1 i = 2 ( Z i, j - Z i - 1, j ) 2 ( N gates - 1) N beams (1) Z SIGN 3, : 1 : (1) B9SI ( Z i Z i ) > 0, Z SIGN i = + 1 ; ( Z i Z i ) = 0, Z SIGN i = 0 ; ( Z i Z i ) < 0, Z SIGN i = - 1 ; Z SIGN = 6 N B9SI 394 Z SIGN i N (2) 125, V = i = 1 N N beams gates 6 6 j = 1 i = i ( V i, j - V ) 2 N gates N beams,, 2 cm 25, m/ s 1 h 20 mm ; ( 9 H ), 60 dbz 40 kg/ m 2 3 km (2) P 0, 4 H 0 ; P 0 1 (3), 2 P 1 P 2 H 1,, H 2 ;, H 2 - H 0 3 km, 9 H = - ( Z 2 - Z 0 ) / ( H 2 - H 1 ), 0 9 H = - ( Z 1 - Z 0 ) / ( H 1 - H 0 ) 1, 0 ( ( Q V IL ) 4. 5), 0 0 Z2I,, Z SIGN, Q V IL Z SIGN ( V ) : : Z SIGN 0. 15, Z SIGN % 394, Z SIGN (3), Z SIGN V i, j, V, Z SIGN, : Z SIGN Z Texture, Z SIGN V Z SIGN 1 ; PPI Z SIGN 0. 15,

6 6 : ; (92. 8 %) Z SIGN ( 5) 4 (a. Z Texture,b. Z SIGN,c. 9 H,d. Q VIL,e. V ) Fig. 4 (a) Z Texture, (b) Z SIGN, (c), (d) Q VIL, and (e) V patterns for all storms (solid line) and vigorously convective storms (dotted line)

802 64 5 (a. Z Texture,b. Z SIGN,c. 9 H,d. Q VIL,e. V ) Fig. 5 Membership functions (a. Z Texture,b. Z SIGN,c. 9 H,d. Q VIL,e. V ) F( Z Texture ) = F( Z SIGN ) = Z Texture / 7. 5 0 Z Texture 7.

7 (a. Z Texture,b. Z SIGN,c. 9 H,d. Q VIL,e. V ) Fig. 5 Membership functions (a. Z Texture,b. Z SIGN,c. 9 H,d. Q VIL,e. V ) F( Z Texture ) = F( Z SIGN ) = Z Texture / Z Texture ( Z Texture ) / < Z Texture 50 0 Z Texture > Z SIGN ( Z SIGN ) < Z SIGN Z SIGN 0. 4 < Z SIGN < Z SIGN Z 9H 0. 5 F( 9Z 9H ) = 1 - ( 9Z 9Z ) / < 9H 9H 4. 5 (6) 9Z 0 9H < 0 9Z 9H > Q V IL 20 F( Q V IL ) = Q V IL / < Q V IL 40 (7) 1 Q V IL > 40 F( V ) = 2 V 0 V ( V ) / < V 2. 7 (8) 0 V > 2. 7 B9SI,,B7SI B9SI ( 6),, 4, ( ), ,, 7, 14 :51 B7SI B9SI 3 ( ) 3. 1 B7SI B9SI B9SI B7SI :15,, B9SI (4) (5)

6期胡胜等 : 风暴的多普勒雷达自动识别 1994-2007 China Academic Journal

9 804 64

10 6 : 805, : 3 ( ),B9SI B9SI 2, B7SI 4, 25 dbz, CSI 8 B7SI B9SI 25 dbz, 3 :B9SI 4 5 CSI, B7SI B9SI, 2 ;,B9SI 4 B9SI ( ) 4 1, CSI B7SI, 0. 5 ( 6. 0 PPI PPI 45 dbz) Q V IL, PPI B7SI 2 4,B9SI,, B7SI, B9SI CSI 16 :09,B7SI 16 :09,,, 9, B7SI,B9SI CSI ( 1 B9SI CSI CSI ) 3, 5 B7SI B9SI 2,,, 6 ; 1 5,6,7 ( ) 0. 5 CSI, ( 3 4) B9SI Table 5 The features and the convection indices for four storms in the yellow rectangle Z Texture Z SIGN Q 9 VIL V I C H Table 6 The features and the convection indices for six storms in the blue rectangle Z Texture Z SIGN Q 9 VIL V I C H (1) D (2) (3) : (1) V ; (2) 2D, 9 H ; (3),

806 64 10 E : V Z Texture, Z SIGN, 1,2,3,4,6 8 ( 9 I C 9 H Q V IL I C 5 6 ) 5, 7 A 6, A : Z Texture Z SIGN, I C, Z Texture V I C ; 9 H, Q V IL B : Z SIGN Z Texture, 9 H, Q V IL V I C ; C : 9 H Z

11 E : V Z Texture, Z SIGN, 1,2,3,4,6 8 ( 9 I C 9 H Q V IL I C 5 6 ) 5, 7 A 6, A : Z Texture Z SIGN, I C, Z Texture V I C ; 9 H, Q V IL B : Z SIGN Z Texture, 9 H, Q V IL V I C ; C : 9 H Z Texture, Z SIGN, Q V IL V I C ; D : Q V IL Z Texture, Z SIGN, V I C ; I C 9 H B C I C, Z SIGN 9 H,, E I C,, 7 Table 7 The convection indices of the several sensitive experiments A B C D E ,, 15 :21,, (, ) ( 9 1) (14 :15 16 :27 6 min 10) 16 :09, 14 :15 14 :45 15 :09 15 :27 10 Fig. 10 Temporal variations of the convection index of a supercell occurred in Guangzhou on August 11th, 2004, during its evolution

12 6 : B7SI, ;B9SI, ; CSI B9SI,,,B7SI B9SI,B9SI, B7SI,,B9SI ; B7SI B9SI 2D CSI, ; I C,, I C ; 15 :21 I C ; I C, 16 :09 I C,,CSI [13 ] Seo D J, Ding F, Fulton R. Final report interagency MOU a2 mong the NEXRAD program. The WSR288D Radar Operation, V Center and the NWS Office of Hydrologic Development, Hydrol2 9 H, ogy Laboratory, Office of Hydrologic Development, National Weather Service, Silver Spring, MD, 2002 [ 1 ] Rinehart R E. A pattern2recognition technique for use with con2 vention weather radar to determine internal storm motion, recent progress in radar meteorology. Atmospheric Tech, 1981, 13 : [ 2 ] Rinehart R E, Garvery E T. Three2dimensional storm motion detection by conventional weather radar. Nature, 1978, 273 : [ 3 ]..,1991,10 (3) : Liu Liping, Xu Baoxiang, Wang Zhijun. A preliminary study of tracking and warning severe storms by the method of tracking e2 cho centroids. Plateau Meteorology(in Chinese), 1991, 10 (3) : [4 ] Austin G L, Bellon. Very short2range forecast of precipitation by the objective extrapolation of radar and satellite data. In : Broning K, Ed. Nowcasting. Academic Press, [ 5 ] Rosefeld D. Object method for analysis and tracking of convec2 tive cells as seen by radar. J Atmos Oceanic Tech, 1987, 4 : [ 6 ] Michael Dixon, Gerry Wiener. Thunderstorm identification, tracking, analysis and nowcasting. J Atmos Oceanic Tech, 1993, 10 : [ 7 ] Vance Mansur M. Examples of the strength and weakness of the WSR288D storm tracking products. Preprints, 26th National Conventional Conference on Radar Meteorology, [ 8 ] Arthur Witt, Johnso J T. An enhanced storm cell identification and tracking algorithm. preprints, 26th National Conventional Conference on Radar Meteorology, [ 9 ]..,1998,21 (2) : Xiao Yanjiao, Tang Dazhang, Li Zhonghua, et al. Storm auto2 matic identification, tracking and forecasting. J Nanjing Institute of Meteor (in Chinese). 1998, 21 (2) : [ 10 ] Johnson J T, Pamela L, Mackeen. The storm cell identification and tracking algorithm : an enhanced WSR288D algorithm. Wea Forec, 1997, 13 : [ 11 ] Steiner M, Houze R A, Yuter S E. Climatological characteriza2 tion of three dimensional storm structure from operational radar and rain gauge data. J Appl Meteor, 1995, 36 : [ 12 ] Biggerstaff M I, Listenmaa A. An improved scheme for Convec2 tice/ Stratiform echo classification using radar reflectivity. J Appl Meteor, 2000, 39 : [14 ] Kessinger C. The radar echo classifier : a fuzzy logic algorithm for the WSR288D. Preprints, 3rd Conf on Artificial Intelligence Applications to the Environmental Science, Amer Meteor Soc, 2003 [ 15 ],..,2004,62 (3) : Zheng Yuanyuan, Yu Xiaoding,Fang Chong,et al. Analysis of a strong classic supercell storm with Doppler weather radar data. Acta Meteor Sinica (in Chinese),2004,62 (3) :

13 [16 ]. [ 17 ]..,2004,62 (6) : Li Yongping, Zhu Guofu, Xue Jishan. Microphysical retrieval from Doppler radar reflectivity using variational data assimila2 tion. Acta Meteor Sinice (in Chinese), 2004, 62 (6) : ,2000,58 (2) : Qiu Chongjian, Yu Jinxiang. Use of Doppler radar in improving short2term prediction of mesoscale weather. Acta Meteor Sinice (in Chinese), 2000,58 (2) : AUTOMATIC ID ENTIFICATION OF STORM CELLS USING DOPPL ER RADARS Hu Sheng 1,2 Gu Songshan 1 Zhuang Xudong 2 Luo Hui 3 1 N anjing U niversity of Inf orm ation Science & Technology, N anjing Guangz hou Cent ral Meteorological Observatory, Guangz hou S hanxi Provincial Meteorological B ureau, Xi an Abstract Three storm automatic identification algorit hms for Doppler radar are discussed. The WSR288D Build 7. 0 (B7SI) test s t he intensity and continuity of t he objective echoes by multiple2prescribed t hresholds to build t hree2 dimensional storms, and when storms are merging, splitting, or clustered closely, t he detection errors become larger. The B9SI algorithm is part of the Build 9. 0 Radar Products Generator of the WSR288D system. It uses multiple t hresholds of reflectivity, newly designs t he techniques of cell nucleus ext raction and close2storms pro2 cessing, and t herefore is capable of identifying embedded cells in multi2cellular storms. The st rong area compo2 nents at a long distance are saved as 2D storms. But, the B9SI can t give information on the convection strength of storm, because text ure and gradient of reflectivity are not calculated and radial velocity data are not used. To overcome this limitation, the CSI algorithm is designed in this paper. By using the fuzzy logic technique, and under the condition that the levels of the seven reflectivity thresholds of B9SI are lowered, the CSI processes the radar base data and the output of B9SI to obtain the convection index of storm. Finally, the convection index is obtained from the weighted average of all the likelihood values. The CSI is verified with the case of a supercell occurred in Guangzhou on 11 August, The computational and analysis results show that the two rises of convection index match well with a merging growth and strong convergent growth of the supercell, and the in2 dex is when the supercell is strongest, and then decreases. Correspondingly, the height of the maximum reflectivity, detected by the radar also reduceds, and heavy rain also occurred in a large scale area. Key words : Storm identification, Nucleus ext raction, Fuzzy logic technique, Convection index.

200 31 5 SWAN 0. 01 0. 01 6 min TITAN Thunder Identification Tracking Analysis SWAN TITAN and Nowcasting 19 TREC Tracking Radar Echo by Correlaction T

200 31 5 SWAN 0. 01 0. 01 6 min TITAN Thunder Identification Tracking Analysis SWAN TITAN and Nowcasting 19 TREC Tracking Radar Echo by Correlaction T 31 1 2013 3 Journal of Arid Meteorology Vol. 31 No. 1 March 2013. SWAN J. 2013 31 1 199-205. doi 10. 11755 /j. issn. 1006-7639 2013-01 - 0199 SWAN 730020 2011 7 7 SWAN SWAN 2012 2 SWAN 1006-7639 2013-01