MOTC-IOT-9-H-BB0 ( )

Size: px

Start display at page:

Download "MOTC-IOT-9-H-BB0 ( )"

倍熙蒙
7 years ago
Views:

1 MOTC-IOT-9-H-BB0 ( )

2 MOTC-IOT-9-H-BB0 ( )

3 ( ) GPN: 定價元

4 40 (0) (0) (0) (0) B (0) (04) (04) (07)33490 GPN

5 H-BB #54933 Holland 93 3 I

6 PUBLICATION ABSTRACTS OF RESEARCH PROJECTS INSTITUTE OF TRANSPORTATION MINISTRY OF TRANSPORTATION AND COMMUNICATIONS TITLE: Typhoon Wave Calculation Model around Harbors of Taiwan (II) ISBN(OR ISSN) GOVERNMENT PUBLICATIONS NUMBER DIVISION: Harbor & Marine Technology Center DIVISION DIRECTOR: Chiu Yung-Fang PRINCIPAL INVESTIGATOR: Ho Liang-Sheng PROJECT STAFF: Lin Shou-Shiun PHONE: (04) 6583 FAX: (04) RESEARCH AGENCY: National Chiao Tung University PRINCIPAL INVESTIGATOR: Chang Hsien-Kuo PROJECT STAFF: Chien Wei-An, Chen Wei-Wei, Huang Zen-Yu ADDRESS: 00 Ta Hsueh Road, Hsinchu, Taiwan 300, ROC PHONE: #54933 IOT SERIAL NUMBER PROJECT NUMBER 9-H-BB0 PROJECT PERIOD FROM 0/003 TO /003 KEY WORDS: Typhoon Waves Neural Network Wave Prediction ABSTRACT: The project presents a typhoon wave model using neural network and fuzzy theory. Input winds obtained by popular Holland typhoon wind model are used in the present wave model. Fast and accurate calculation on typhoon waves by the present model provides good information for a coastal warning system to prevent or decrease from possible life loss and property damage. We assemble wave data at Hwa-lien Harbor, Su-Au Harbor and Kee-lung Harbor. The wind data are classified into some groups by paths, scales, and land effect. An auto-adapting connection between typhoon waves and corresponding wind data is established by using neural network. The proposed model is useful in calculating typhoon waves in both learning stage and verification stage. The outcomes indicate that the wave model can be applied to future practical calculation. We extend the proposed model to a neighboring point and find that the model also provides good calculation on waves at that point, even used the ready wave model established for other point. The present model can be applied to the eastern waters of Taiwan. DATE OF PUBLICATION March 004 NUMBER OF PAGES PRICE CLASSIFICATION SECRET CONFIDENTIAL UNCLASSIFIED The views expressed in this publication are not necessarily those of the Ministry of Transportation and Communications. II

12 .. 3. Bretschneider (976) Young(988b) SWAMP (985) WAMDI(988) Young(988a) SWAN(996) Tolman et al.(997) -

15 3 I (00) blend 4 I (00) Holland.4 Holland -4

16 . ( ) (hpa)

17 - -

18 viscosity coefficient (drag coefficient) (Coriolis parameter) (atmospheric models) RUC, MM-5, AVN (Aviation), MRF (Medium Range Forecast), NGP (NoGaps)

19 - KMA 95 hpa. Holland(980) (Holland model) p p n pc p c B R max = exp (-) r R max r p r p p c n f (Coriolis parameter) r V B = ( p p ) n ρ c R r max B R exp r max B + r f rf (-) -4

20 V max B = ( p p ) n ρ e c 0.5 (-3) B Harper and Holland(999) p 900 c B = for.0 < B <. 5 (-4) 60 V = K V (-5) 0 m Harper and Holland(999) model o K V m Jelesnianski(966) f R r = (-6) max U ( r) V f Rmax + r -5

21 3 3. WAN WW3 N t, φ, θ, λ N t + cos φ φ ( ) ( φ& N cos θ ) + ( λ& N ) + ( θ& N ) λ θ = S σ 3- θ φ λ & φ λ & θ & S S = S + S + S + S 3- in nl ds bot S S S in nl S bot ds Rankine vortex model SLOSH model Holland model Bretshneider(976) H = K R p s, max max c (3-3) 3-

22 H s, max sinφ R V max K m K Ω = ( X ) ( X ) 0. 53( X ) 3 (3-4) ΩRsinφ X = (3-5) V max V (m/s) f H s,max = H s,max V f cos( θ + β ) + V max (3-6) θ β 3- T H s,max s, max H s T = 0.734V tanh.07 tanh,max max (3-7) s Vmax R r/ r/r 3- (Shore Protection Manual,984) 3-

23 (984) R p (Energy Index) H s,max Rmax p = e c + 0.9V V max f (3-8) T s,max Rmax p = e c V V max f (3-9) m max (3-8) (3-9) H s, max V 0 m / s V max V p max = 0.387(4.5 c 0.6RΩsinφ ) V f (3-0) (Shore Protection Manual,984) Hs (3-) f T = (3-) s H s Young(988b) V (m/s) V (m/s) F (m) R max R F = av + bv + cv + dv + ev + max max f max f max f (3-) a = b = c =.3 0 d =.90 0 e = f = R max R R (3-3) 3 3 =.5 0 log R max max JOSNWAP 3-3

24 0.5 gh s 0 gf.006 = (3-4) Vmax Vmax (00) (Shore Protection Manual) Young(988) 3- Saffir Simpson (SS) 3-3- (Shore Protection Manual, 00) 3- Saffir Simpson (SPM, 00) SS p c (mb) H(m) T(sec) > <

25 Global Guide to Tropical Cyclone Forecasting(Bureau of Meteorology Research Centre) H = 0. 0 ( p p ) (3-5) s, max n c p p ( hpa ) n c Model (3-) (3-5) 0 V 0 3-5

26 ( p, V, V, V, R, φ, λ, S, S, S, S t) H s ( t) = f h c max f 0 max in nl ds bot, (3-6) p c max 3-6 V V f 0 V R max S in 6 V p c f V max R max V 0 S in S ( S in ) ( S nl ) ( S ds ( S bot ) S. ( S in ) Holland (980) 0 V 0(m/s) V 0 Sin V0 r θ t. ( S nl ) S ) ( nl 3-6

27 3. ( S ds ) ( S ds ) 4. ( S bot ) ( S bot ) 0 5. t t t V f V f 3-7

28 D S r θ φ λ r θ V V max f r θ θ 3-3 r θ V f θ

29 θ V0 t j S V, r, θ θ S j ( t j i 0 i= 0 0, ) = a f ( V, r, θ, θ ; t ) (3-7) i (3) a i f t j t j S s j = j i= 0 i [ S( t ] H ( t ) b f (3-8) i ) (3-7) (3-8) V, θ 0 r, θ, H θ ; t ) (3-9) = j s ( t j ) ci f ( V0, r,, θ i= j m i (3-9) m t j m m (3-9) (3-9) Model O f ( W I b ) P = S R R + S (3-0) 3-9

30 O P P f I R W R S S R b (3-0) S S-P (3-8) (3-9) V V max f V 0 r θ Model O f W f ( W [ I ] + b ) b ) (3-) = ( S S S ( R * n ) S + R n S O f W f ( W [ I ] + b ) b ) (3-) = ( S S S ( R * n ) S + R n S H s O = Ts O = H T T = [ O, r, θ, θ n I ] 3.. Model s s [ ] T I = V0, r, θ (membership function) MF(Hualien) MF(Keelung) (3-9) 3-0

31 Model Model [ I ] + b ) b ) = f R n S ( W S S f ( W S ( R * n ) S + O (3-3) [ I ] + b ) b ) O = f ( W f ( W S S (3-4) S S S ( R* n ) + R n [ I ] + b ) b ) O = f 3 ( W S S f ( W S ( R * n ) 3 S + R n S (3-5) H s H s H s O = O = T O 3 = s T s Ts [ V, r, θ ] T I = T 0 I = [ O, r, θ, θ ] = T 3 [ O, MF n I ] 3..3 Model 3 Model (membership function) MF(Hualien) MF(Keelung) (membership function) MF(Hualien) MF(Keelung) (3-3) (3-5) Model Model 3 H s H O = O = T s T [ V, r, θ ] T I = T 0 I = [ O, r, θ, θ ] n s s O 3 = H T I ] s s T 3 = [ O, MF, MF 3-

35 3..4 Joint Typhoon Warning Center (JTWC) Unisys Weather Holland Wind Model

36 JTWC Unisys Weather Holland Wind Model 3-3-6

37 JTWC UNISYS WEATHER JTWC 6 Datawell (directional waverider) 0.8 Hz ( N , E ) 5 ( N , E 45 ) 43 4-

38 Levi 05/5(800)~05/30(0600) 05/5(800)~05/30(0600) 997 Opal 06/5(0600)~06/(0600) 06/5(0600)~06/(0600) Peter 06/3(0600)~06/9(0000) 06/3(0600)~06/9(0000) 998 Otto 08/0(0000)~08/05(0600) 08/0(0000)~08/05(0600) 999 Sam 08/8(0600)~08/3(0000) 08/8(0600)~08/3(0000) Jelawat 08/0(0000)~08/(0000) 08/03(0800)~08/(0000) 000 Bilis 08/8(0600)~08/4(0000) 08/8(0600)~08/(00) Bopha 09/05(800)~09/(0600) 09/08(0400)~09/(0600) Yagi 0/(00)~0/7(0600) 0/(00)~0/7(0600) 4-4-

39 Tim (994) Fred (994) Kent (995) Haiyan (00) Maggie (999) Haiyan (00) Tim 07/05~07/ 07/09~07/0 994 Fred 08/9~09/0 08/9~09/0 995 Kent 08/4~09/0 08/4~08/ Maggie 06/0~06/06 06/0~06/06 00 Haiyan 0/~0/7 0/~0/

40 4. ( 7.m/s) m/s

41 0 8 Hs(m) D (m) Ts(sec) D (m)

42 4. Windows XP Matlab 5.3 Neural Toolbox (3-) (3-5 ) Model Model Model (3-) Model V 0 r ϑ θ 4 4 V 0 r θ V 0 r θ t t- t-4 t-4 V0 r θ O H T ( ) S S O r θ θ t t- t-4 t- O r θ θ H T S S ( ) (3-5) Model Model Model O O MF O MF t t- t-4 t- O MF T ( ) S Model Model 3 Model 3 Model O O MF MF H S 4-6

43 O MF MF t t- t-4 t- O MF MF H T ( ) S S Model 9 R R Model 9 R 0.87 Model 9 R Model 3 9 R 0.89 Model 3 9 R

44 8 R = H s (m) model H s (m) Observed data 4-6 Model 4 R = T s (sec) model T s (sec) Observed data 4-7 Model 4-8

45 0 R = H s (m) model H s (m) Observed data 4-8 Model 4 R= T s (sec) model T s (sec) Observed data 4-9 Model 4-9

46 0 R= H s (m) model H s (m) Observed data 4-0 Model 3 4 R= T s (sec) model T s (sec) Observed data 4- Model 3 4-0

48 Hs(m) Model Model Model 3 Data Time (hr) 4-3 Model Model Model 3 Tim 4-4 Fred 4-5 Model Model Model 3 Model Model.5 Model Model 3. Model Model 3 Model Model 3 Model Model R Model Model R 0.77 Model Model Model 4-

49 Model Model Fred Hs(m) Model Model Model 3 Data Time (hr) 4-5 Model Model Model 3 Fred 4-3

50 4-6 Kent(995) Kent Kent Model Model Model 3 Model Model Model Haiyan Model Model Model 3 Model Model Model 3 Model 4-9 Model Model Model RMS = N N ( H s, p H s, m ) i= (4-) H = MAX H ) MAX ( H ) (4-) S, p ( S, p S, m (4-3) t p = t p, p t p, m 4-4

53 4-3 Model Model Model 3 Model Model Model 3 RMS H R sp t p RMS H sp t p RMS H sp t p R R (m) (m) (hr) (m) (m) (hr) (m) (m) (hr) Tim* Fred Kent Haiyan Maggie Maggie (999) 4-0 Model Model Model 3 Maggie 4-4- Model Model 3 Model Model R 0.95 Model Fred Maggie Fred Maggie Model Model 3 Model R

55 4.4 Model R 0.89 R Utor 07/0(0600)~07/06(800) 07/0(0600)~07/06(800) 00 Rammasun 06/8(0600)~07/06(0000) 06/8(0600)~07/06(0000) Sinlaku 08/8(800)~09/08(0000) 08/9(000)~09/08(0000) R= Hs(m) Model Hs(m) Observed Data 4-4-9

56 0 R= Hs(m) Model Hs(m) Observed Data Haiyan 4-6 Haiyan

57 Model Data 8 Hs(m) Time (hr) 4-4 Haiyan 4.6 Model 3 HAIYAN MAGGIE HAIYAN MAGGIE

58 0 8 成功港花蓮港 H s(m) Time (hr) 4-5 HAIYAN 0 8 成功港蘇澳港 H s(m ) Time (hr) 4-6 MAGGIE 4-

59 5 Model Model Model3 Model Model Model Model 3 Model 3 Model 3 Holland model Model Model

60 5-

61 6. (00) (00) 3. (000) - 4. (999) (988) ( ) (998) (986) 7 8. (987) Booij, N.,L.H. Holthuijsen and R.C. Ris(996) The SWAN wave model for shallow water, Proceedings of the 4th Conference International Conference on Coastal Engineering, ASCE, Orlando, Vol., pp

62 0. Bretschneider, C. L. and E. E. Tamaye (976) Hurricane wind and wave forecasting techniques, Proceedings of the Fifteenth Conference on Coastal Engineering, Hawaii, Vol., pp Chen, Y. H. and H. Wang (983) Numerical model for non-stationary shallow water wave spectral transformation, Journal of Geophysical Research, Vol. 88, pp Deo, M.C., and C.S. Naidu, (00) Real time wave forecasting using neural networks, Ocean Engineering, Vol. 6, Eberhart, R.C., and Dobbins, R. W., Neural Network PC Tools A Practical Guide, Academic Press, Ins, pp (990). 4. Harper, B.A., Holland, G.J., (999). An updated parametric model of the tropical cyclone, Proceedings of the 3rd Conference of Hurricane and Tropical Meteorology, Dallas, Texas, pp Hajime, M.,M. Sakamoto, and T. Sakai (995). Neural network for stability analysis of rubble-mound breakwaters, Journal of Waterway Port, Coastal and Ocean Engineering, ASCE, Vol. 97, Hiraoka, T., H. Maeda and N. Ikoma (999) Two-stage prediction method of typhoon position by fuzzy modeling-fusion of outline prediction and detailed prediction Systems, Proceedings of IEEE SMC '99 Conference on Man, and Cybernetics, Vol. 6, pp Holland, G. J. (980) An analytical model of the wind and pressure profiles in hurricanes, Monthly Weather Review, Vol. 08, pp Holland, G.J. (993): WMO/TC-No. 560, Report No. TCP-3, World Meteorological Organization; Geneva, Switzerland 9. Jelesnianski, C.P., (966) Numerical computations of storm surges without bottom stress, Monthly Weather Review, Vol. 94, No. 6, Lee, R.S.T. and J.N.K. Liu (999) An Elastic Graph Dynamic Link Model for Tropical Cyclone Pattern Recognition, Proceedings of the 6th International Conference on Neural Information, ICONIP 99, Vol., pp Niwa, Y. and T. Hibiya (998) Nonlinear processes of energy transfer from traveling hurricanes to the deep ocean internal wave field, Oceangraphic Literature Review, Vol. 45, pp Shore Protection Manual (984) U.S. Army Coastal Engineering Research Center, Fort Belvoir, Vol. I-III. 6-

63 3. Shore Protection Manual (00) U.S. Army Coastal Engineering Research Center, Fort Belvoir, Vol. I-III. 4. SWAMP Group (4 Authors) (985) Ocean wave modeling, Plenum Press, New Pork, 56 pp. 5. Tolman, H.L. (997) User manual and system documentation of WAVEWATCH-III, version.5, NOAA / NWS /NCEP / OMB Technical Note 5, 97pp. 6. WAMDI group The WAM model (988) A third generation ocean wave prediction model, Journal of Physical Oceanography, Vol. 8, pp Young, I.R. (988a) Observations of the spectra of hurricane generated waves, Ocean Engineering, Vol. 5, pp Young, I.R. (988b). A parametric hurricane wave prediction model, Journal of Waterway Port, Coastal and Ocean Engineering, ASCE, Vol. 4, pp

64 A. () () (3) (4) Y i = f Wij X j θ i () j Yi f (transfer function) Wij X j θ i MATLAB (back-propagation neural network BPN) (gradient steepest descent method) W W θ θ X T (hyperbolic tangent function) Y A-

65 () Z Y net = W X i i θ () i net net e e Z = f ( net) = net net e + e (3) net = j j θ (4) j net net e e Y = f ( net ) = (5) net net e + e () δ δ ( Z ) ( W jδ j ) δ = Z (6) j δ = ( + Y )( Y )( Z Y ) (7) (3) W Tj E E E = ( T j Y j ) j E W = η (9) W Y j (8) E W ij n = δ A (0) j n i η (learning rate) n n δ W n A W j ij n i ij (4) W W θ θ W W + W = () A-

66 θ = () θ + θ W θ W + W = (3) θ + θ = (4) Root Mean Squared Error RMSE = n n i ( ) T i Y i (5) n A-3

67 B (membership function) (step function) 3 B-

68 .0 spring summer fall winter degree of membership degree of membership Ma 4 J 3 Septe Decem M -.0 spring summer fall winter M 9 J 8 Septe 7 Decem - 6 M B-

69 (a) (b) (c) (d) -3 4 AND OR NOT A B A and B A B A or B A not A AND OR NOT A B min(a,b) A B max(a,b) A -A AND OR NOT -4 B-3

70 5 Input Input 模糊化函數 Input Input Rule Rule Rule 3 Σ Output Output -7 B-4

71 8-8 B-5

72 C. 6 Data Model 4 Model Model 3 H s (m) Time(hr) 3- Model Model Model 3 Jelawat(000) T s (sec) Data Model Model Model Time(hr) 3- Model Model Model 3 Jelawat(000) C-

73 6 Data Model 4 Model Model 3 H s (m) Time(hr) 3-3 Model Model Model 3 Bopha(000) 0 8 T s (sec) 6 4 Data Model Model Model Time(hr) 3-4 Model Model Model 3 Bopha(000) C-

74 6 Data Model 4 Model Model 3 H s (m) Time(hr) 3-5 Model Model Model 3 Yagi(000) 0 8 T s (sec) 6 4 Data Model Model Model Time(hr) 3-6 Model Model Model 3 Yagi(000) C-3

75 6 Data Model 4 Model Model 3 H s (m) Time(hr) 3-7 Model Model Model 3 Sam(999) 8 T s (sec) 4 Data Model Model Model Time(hr) 3-8 Model Model Model 3 Sam(999) C-4

76 Data Model 8 Model Model 3 H s (m) Time (hr) 3-9 Model Model Model 3 Otto(998) 8 T s (sec) 4 data model model model Time(hr) 3-0 Model Model Model 3 Otto(998) C-5

77 6 Data Model 4 Model Model 3 H s (m) Time(hr) 3- Model Model Model 3 Levi(997) 8 T s (sec) 4 Data Model Model Model Time(hr) 3- Model Model Model 3 Levi(997) C-6

78 6 4 Data Model Model Model 3 H s (m) Time(hr) 3-3 Model Model Model 3 Opal(997) 8 T s (sec) 4 Data Model Model Model Time(hr) 3-4 Model Model Model 3 Opal(997) C-7

79 6 4 Data Model Model Model 3 H s (m) Time(hr) 3-5 Model Model Model 3 Peter(997) 8 T s (sec) 4 Data Model Model Model Time(hr) 3-6 Model Model Model 3 Peter(997) C-8

80 Data Model 8 Model Model 3 H s (m) Time(hr) 3-7 Model Model Model 3 Bilis(000) 6 T s (sec) 8 4 Data Model Model Model Time(hr) 3-8 Model Model Model 3 Bilis(000) C-9

81 D Hmax 3. Hmax D-

82 4... Model Model. Model Model D-

83 D-3

84 E BPN normalization 4. P7 P0 3. Lippman (987) 5. Model fetch I V 0 r 4. 6 MF MF E-

85 .... Holland(980) Model S S in S nl S dl S bot. pp.7. 7.m/s E-

86 Model Model E-3

87 F. (II) F-

88 3 (I) 4 F-

89 5 6 F-3

90 r θ θ 7 V max V f r V 0 θ 8 F-4

91 F-5 9 Model [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O = s s T H O = s s T H O [ ] T r V I θ,, 0 = T r O I ],,, [ θ θ = 0 Model ( ) [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O [ ] ) ) ( ( 3 ) * ( = S S n R n R S S S b b I W f W f O = s s T H O = s s T H O = s s T H O 3 [ ] T r V I θ,, 0 = T r O I ],,, [ θ θ = T MF O I ] 3 = [, n

92 F-6 Model 3 ( ) [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O [ ] ) ) ( ( ) * ( + + = S S n R n R S S S b b I W f W f O [ ] ) ) ( ( 3 ) * ( = S S n R n R S S S b b I W f W f O = s s T H O = s s T H O = s s T H O 3 [ ] T r V I θ,, 0 = T r O I ],,, [ θ θ = T MF MF O I ], 3 = [, n (MF)

93 (MF) 3 (MF) 4 F-7

94 (MF) F-8

95 --- 7 TIM 0 H s (m ) Model Model Model 3 Data Time (hr) 8 F-9

96 FRED 0 H s (m ) Model Model Model 3 Data Time (hr) 9 KENT 0 H s (m ) Model Model Model 3 Data Time (hr) F-0

97 HAIYAN 0 H s (m ) Model Model Model 3 Data Time (hr) MAGGIE 0 H s (m ) Model Model Model 3 Data Time (hr) F-

98 HAIYAN 0 8 H s (m ) Time (hr) 3 MAGGIE 0 8 H s (m ) Time (hr) F-

99 HAIYAN Utor 00 Rammasun Sinlaku 00 8 Model Data H s (m ) Time (hr) 5 Model Model 3 Model Model 3 Model 3 Holland model 6 F-3

100 F-4

101 N RMS = ( H H ) H N i = sp, sm, = MAX( H S, p ) MAX( H S, S, peak m t p, delay = t p, p t p, m H S, p H S, m t p, p t p, m ) 9 Model Model RMS (m) R H S, peak (m) t p, delay (hr) RMS (m) R H S, peak (m) t p, delay (hr) Tim* Fred Kent Haiyan Maggie F-5

102 F-6 3 peak S H, delay p t, peak S H, delay p t, Maggie Haiyan Kent Fred Tim* (hr) (m) R RMS (m) (hr) (m) R RMS (m) Model 3 Model 3 Model 3) HAIYAN

103 33 F-7

104-012-7794 MOTC-IOT-103-H1DB001a 臺灣港務公司之監督與公司治理績效評估研究 (2/2) 著者 : 謝幼屏吳榮貴朱金元吳朝升孫儷芳王克尹林玲煥張淑滿陳銓楊世豪陳秋玲

104-012-7794 MOTC-IOT-103-H1DB001a 臺灣港務公司之監督與公司治理績效評估研究 (2/2) 交通部運輸研究所中華民國 104 年 3 月 104-012-7794 MOTC-IOT-103-H1DB001a 臺灣港務公司之監督與公司治理績效評估研究 (2/2) 著者 :