Modeling of storm surge in the coastal waters of Yangtze Estuary and Hangzhou Bay, China

Transcription

1 Jurnal f Castal Research SI 5 pg - pg ICS7 (Prceedings) Australia ISSN Mdeling f strm surge in the castal waters f Yangtze Estuary and Hangzhu Bay, China K. Hu, P. Ding and J. Ge State Key Labratry f Estuarine and Castal Research, East China Nrmal University, Shanghai, P.R. China, klhu@sklec.ecnu.edu.cn ABSTRACT Hu, K., Ding, P. and Ge, J., 7. Mdeling f strm surge in the castal waters f Yangtze Estuary and Hangzhu Bay, China. Jurnal f Castal Research, SI 5 (Prceedings f the 9th Internatinal Castal Sympsium), pg pg. Gld Cast, Australia, ISBN The regin f Yangtze Estuary and Hangzhu Bay (YE-HB) is subjected t trpical cyclne (TC) impacts during summer almst every year, resulting in extensive lss f life and prperty. An integrated mdel system, which includes a strm-induced wind mdel, wave mdel and hydrdynamic mdel, was develped fr strm surge simulatin in this area. Nesting cmputatin was adpted fr cnsidering the effect f a remte wind field. Wave effects were included by cnsidering radiatin stress and enhanced bttm shear stress. Taking Typhns Agnes () and Matsa (59) as examples, the mdel system was applied t the waters f the YE-HB. The calculated results f winds, waves and strm surges agree well with the bserved data, which means there is gd applicability f the mdel system fr the study area. The distributins f the surge setup were analyzed. Finally, the effects f waves and remte winds n strm surge were prbed by numerical experiments. ADITIONAL INDEX WORDS: Strm-induced wind, Wind-waves, Nesting, Tide simulatin, Surge setup. INTRODUCTION Predictins f strm surge in castal areas are f great imprtance fr castal structure design, castal management and the safety f human life and prperty. Many studies have been cnducted n simulating strm surges (MADSEN and JAKOBSEN, ; SHEN et al., ). Fr instance, the famus SLOSH mdel (JELESNIANSKI et al., 99) was widely used fr tw-dimensinal (-D) applicatin. With the fast develpments f cmputer science and numerical technlgy, sme three-dimensinal (3-D) mdels were develped (PENG et al., ) that can cnsider the hydrdynamic prcesses mre realistically and precisely. Furthermre, wave effects were intrduced by including radiatin stress and enhanced bttm shear stress (JONES and DAVIES, 99), which can imprve the results nearshre. Cnsidering strm surge is a -D variable and -D calculatin has an advantage fr cmputatinal speed r time, -D mdels are still irreplaceable, especially in sme cases, where a rapid predictin is required. Whether the mdels are -D r 3-D, the accuracy f the input wind field during a trpical cyclne (TC) is crucial fr the results f mdeling strm surge. Sme classic parametric pressure r wind mdels (FUJITA, 95; HOLLAND, 9) are frequently used t cnveniently generate symmetric wind fields. These mdels have similar frmulae, and the accurate chice f TC parameters is the key pint. Bth lcal wind and remte wind can cause strm surge, as studied by WONG and MOSES-HALL (99) with a cnceptual methd. Their studies shw that the cmputatinal regin shuld be large enugh t include the effects f remte winds, r the nesting technique shuld be applied. The regin f Yangtze Estuary and Hangzhu Bay (YE-HB, see Figure ), lcated n the east cast f China facing t the pen sea, the East China Sea (ECS), is subjected t TC impacts almst every year during summer. These TCs are mainly generated at the Latitude (N) Jiangyin m m 3m m 5m Matsa (59) YE-HB Agnes () Ganpu 3 5 Tajialu Waigaqia Shanghai Hangzhu Bay Yangtze Estuary 3 Lngitude (E) East China Sea Figure. Regin f the YE-HB, survey statins (black pints) and typhn paths (sub-figure). Jurnal f Castal Research, Special Issue 5, 7

2 Mdeling f strm surge cean surface either east f the Philippine islands r near Guam. The TCs that affect this regin can be divided int tw main grups accrding t the different tracks f their lw-pressure center (see sub-figure f Figure ), namely the turning type and the landfalling type. The latter type can smetimes turn and mve int the sea again. In this study, an integrated mdel system is develped fr strm surge simulatin. Nesting cmputatin is adpted fr cnsidering the effect f remte wind fields. The shallw depth in the YE-HB makes wave effects n strm surge imprtant. Wave effects are included thrugh radiatin stress and bttm shear stress in the hydrdynamic mdel. Taking tw strng TCs, in the turning and landfalling categries, respectively, the mdel system is applied t the YE-HB fr the simulatin f strm surge. The characteristics and mechanics f strm surge in this regin are analyzed. The effects f waves and remte winds are als discussed. MODEL SYSTEM DESCRIPTION The numerical mdel system cnsists f a wind mdel, a nearshre wave mdel and a hydrdynamic mdel. The wind mdel prvides strm-induced wind fields fr bth the wave mdel and the hydrdynamic mdel. Wave elements prvided by wave mdel can be input int the hydrdynamic mdel fr calculating radiatin stress and bttm shear stress under waves and currents. The hydrdynamic mdel includes the dynamic factrs f river runff, astrnmical tides, and winds and waves. Mdel Grid Settings Yangtze Estuary (YE) and Hangzhu Bay (HB) are cnsidered as ne research system because f the substantial exchange f water and sediment between them. In rder t reslve the cmplicated tpgraphies and castal shapes in the YE-HB, it is necessary t use curvilinear grids with high reslutin in the area f interest. Figure shws the calculatin grids fr the YE-HB. The grids are 7 3; the average grid spacing is abut km; the maximum is abut km ffshre; the minimum is abut 3 m inside the Estuary. The nesting methd is used fr cnsidering the remte wind effect during strm events. The regin f ECS is chsen fr the larger calculatin dmain, in which the wind- r strm-induced water level change is cmputed. This level change can be added int the bundaries f the small dmain f YE-HB by interplatin. The grids fr the ECS are generated by de-refining the grids fr the YE-HB and expanding t the whle dmain (subfigure f Figure ). The grids are 5 7; the average grid spacing is abut km. Wind and Wave Mdels A symmetrical strm-induced wind mdel with a backgrund wind is adpted in this study. The -hurly backgrund wind data at m abve sea level n a T Gaussian grid (apprximately ) are btained frm NCEP/NCAR Reanalysis. The parametric wind mdel uses Fujita s frmula (FUJITA, 95) fr the atmspheric pressure field due t strms (Refer t HU et al., 5, fr details). The SWAN wave mdel develped by TU Delft was intrduced fr simulatin f strm-induced wind-waves. The spectral wave mdel based n the wave actin balance equatin is made suitable fr nearshre applicatins by incrprating sme specific treatments f nearshre wave prcesses. Detailed descriptins can be fund in the SWAN user manual (BOOIJ et al., ). Driven by the wind frm the wind mdel intrduced abve, the SWAN mdel was applied t the YE-HB fr simulatin f strm-induced waves with reasnable results btained by HU et al. (5), using Latitude (N) the methd n which this study is based. Hydrdynamic Mdel A -D hydrdynamic mdel fr strm surge calculatin was develped frm the mdel by SHI et al. (997) using the wet-dry grid pint methd. This mdel was applied t the YE-HB fr simulatin f astrnmical tides and currents (HU et al., ). In this study, the riginal mdel was revised by adding the water level change due t the atmspheric pressure field and cnsidering the effect f waves. The gverning equatins in Cartesian crdinates are as fllws: H Hu Hv + + = () t x y u u u ζ P + u + v f v = g t x y x ρ x YE-HB Lngitude (E) Figure. Curvilinear mdel grids fr the regins f the YE-HB and ECS (sub-figure). S S xx xy ( τ ) + ax τbx x y v v v ζ P + u + v + f u = g t x y y ρ y S S yx yy ( ) τ ay τby + x y in which Eq. is the cntinuity equatin, Eqs. and 3 are mmentum equatins in x- and y-directin; (u, v) are the cmpnents f depth-averaged velcity U v ; g is the gravity acceleratin; H=h+ζis the ttal depth, h is the depth under the mean water level, ζ is the water elevatin; f is the Crilis parameter;ρis the water density; P atm is the atmspheric pressure; (S xx, S xy, S yx, S yy ) are the cmpnents f radiatin stress induced by waves. atm atm () (3) Jurnal f Castal Research, Special Issue 5, 7

3 Hu et al. Statin Wind speed (m/s) Statin Wind speed (m/s) 3 3 Typhn Agnes () Date (Aug.~ Sep., 9) Wind directin (deg) Wind directin (deg) Wind speed (m/s) Wind speed (m/s) 3 3 Typhn Matsa (59) 5 7 Date (Aug., 5) Wind directin (deg) Wind directin (deg) Statin Statin Hs (m) & T (s) Hs (m) & T (s) Typhn Agnes () Date (Aug.~ Sep., 9) Wave directin (deg) Wave directin (deg) Hs (m) & T (s) Hs (m) & T (s) Typhn Matsa (59) 5 7 Date (Aug., 5) Wave directin (deg) Wave directin (deg) Figure 3. Cmparisns f wind speed (slid line fr the mdeled; cicles fr the bserved), wind directin (dashed line fr the mdeled; triangles fr the bserved), significant wave height (slid line fr the mdeled; circles fr the bserved), wave directin (dashed line fr the mdeled; triangles fr the bserved) and average wave perid (dash-dtted line fr the mdeled; squares fr the bserved) at tw statins ( and ) during Typhns Agnes and Matsa. ( τ, ax τ ) are the cmpnents f surface wind stresses v τ ay a, parameterized by the quadratic law, that is: v v v τ a = ρ a C D W W () where ρ is the air density, W v is the vectr f wind speed, C a D is the drag cefficient based n the fllwing equatin (SMITH, 9): v 3 C D = ( W ) (5) ( τ, bx τ ) are the cmpnents f bed shear stresses (bttm by frictin) v τ. A wave-current bundary layer sub-mdel, Grantb Madsen (GM) mdel (GRANT and MADSEN, 9) is adpted fr calculating bed shear stress due t waves and currents. Fr pure astrnmical tide simulatin with n wave effects, v τ is b represented by the Manning s frictin relatinship, v v v / 3 τ b = ρgn H U U () where n is the Manning parameter. TYPHOONS AND OBSERVED DATA A typhn is a super strng TC that riginates n the Nrth- West Pacific Ocean. Tw typhns (see sub-figure f Figure ), Agnes, a very strng ne histrically, and Matsa, a recent ne, which greatly affected the YE-HB, were selected fr the simulatin f strm surge. TC Agnes (N. ) frmed near Guam n August 5, 9. On August 9, it became a typhn and then entered the ECS. On August 3, Agnes develped int its strngest stage with 99 hpa center air pressure and 5 m/s near-center maximum wind velcity at : UTC. After 3 hurs f slw nrthward mvement, it mved nrth-east. Agnes impacted greatly n the YE-HB because it was very clse t the area and lasted fr a lng perid. TC Matsa (N. 59) riginated at 9 km east f the Philippines, n July 3, 5. Increasing gradually in intensity, it became a typhn n August 3. During the time frm August t August 5, Matsa reached its strngest stage with 95 hpa center air pressure and 5 m/s near-center maximum wind velcity. At 9: UTC n August 5, it made landfall t the suth f the YE-HB. At : UTC n August, it weakened t a TC. After days, it mved int the sea again and finally diminished. Bth typhns cincided with spring tides, which caused extremely high water levels during their passage. During typhn events, wind and wave data were cllected at tw statins, and. Wave data were rare ( times a day) cmpared t wind data (every hur). Unfrtunately, the data fr wind directin and waves during Typhn Matsa are nt available. Water levels were bserved at three statins,, and Waigaqia. Fr verificatin f astrnmical tides, bserved harmnic cnstants are available at mre than statins (Refer t HU et al., fr details) thrughut the regin f YE-HB, and the predicted tide data can be btained frm the tide tables. WIND AND WAVE SIMULATIONS Winds and waves due t typhns were mdeled and cmpared with bservatins befre simulatin f strm surge, as shwn in Figure 3. Fr Agnes, the agreement f wind directin was very gd, while fr wind speed, the result was underestimated a little at bth statins n September. The reasn fr the discrepancy may be that the symmetrical wind mdel cannt accunt fr the asymmetry f winds, caused by, fr instance, the effect f land and the reslutin f backgrund wind is relatively lw. Fr Matsa, the agreement fr wind speed was gd. The result fr wind directin differs frm that f Agnes because f their different TC categries. In general, the mdeled wind reprduced the bserved winds fairly well and can be used in the wave mdel and hydrdynamic mdel. Jurnal f Castal Research, Special Issue 5, 7

4 Mdeling f strm surge The nn-statinary mde in spherical crdinates was used in SWAN. The nesting cmputatin was set up between the YE-HB dmain and the ECS dmain. Twenty-tw expnentially spaced frequencies frm.5 Hz t Hz with evenly spaced directins (f reslutin) and a time step f 5 minutes were adpted. Other parameters tk the default values given in the SWAN mdel. Wave directin here is defined as the directin where waves cme frm, which, fr easy cmparisn, is similar t wind directin. During Agnes, wave directin and perid agreed well with the bservatins except that the mdeled wave height did nt reach the huge measurements bserved (abut 5.5 m) at, and at the same time, it verestimated the wave height a little at. There are many islands lcated in the suth part f HB, the effect f which cannt be fully cnsidered by a wave spectral mdel. Overall, the wave results are reasnable and suitable fr cnsideratin f the wave effect in the hydrdynamic mdel. STORM SURGE SIMULATION Astrnmical Tide Simulatin The astrnmical tides in the YE-HB shuld be first verified befre the calculatin f strm surge. This wrk has been dne by HU et al. () using a similar hydrdynamic mdel. As shwn in Figure, the bundary f the YE is at Jiangyin, the upper limit f the tidal current, where the river runff enters. The bundary f the HB is at the line between Ganpu and Tajialu. There are three ffshre bundaries (shwn in Figure ). Water level cnditins were used t frce the mdel. Seven main cnstituents were cnsidered at these bundaries, namely M, S, K, O, N, K and P. Harmnic cnstants f M, S, K and O were derived by interplatin frm bserved distributins in the ECS. Harmnic cnstants f N, K and P were derived frm the values f M, S and K, respectively, by the statistical relatinship resulting frm the bserved data. The time step is s and the Manning parameter is abut.3. The mdel runs days fr spinning-up, and then starts t utput results fr verificatin. The mdeled harmnic cnstants gained by using 3 days f calculated water levels fr harmnic analysis were cmpared with the bserved cnstants. At mst statins, the relative errr f the lcal tidal amplitude and the abslute errr f the lcal phase were less than % and, respectively, which means the astrnmical tide simulatin in this mdel is successful. Furthermre, the astrnmical tides during Typhns Agnes and Matsa were calculated with the mnth-average river runff, that is, 5 m 3 /s and m 3 /s, respectively, and n wind input. The calculated results f water level agreed quite well with the predicted data f tide tables at three tidal statins,, and Waigaqia (Figures were mitted here due t limitatins n space). Strm Tide Simulatin Strm surge during tw TCs was simulated by using the nesting methd. There are tw steps in the calculatin. Firstly, driven by the mdeled strm-induced winds, the water level due t TCs in the larger dmain, the ECS, was calculated and interplated (every hur) t the pen bundaries f the small dmain, the YE-HB. Secndly, the strm surge in the YE-HB was simulated. Water level bundaries were the cmbinatin f the astrnmical tide frm harmnic cnstants and the water level change frm the larger dmain, thrugh which the effect f remte winds can be cnsidered. The mnthly-average river runff was added. Figure shws the cmparisns f the mdeled water levels against the bserved data. It can be seen that the agreements between the mdeled and the bserved data were fairly gd. The effects f strms can be clearly shwn by the time series f surge setup. The mdeled surge setup is the difference between the results in the strm tide simulatin and thse in the pure astrnmical tide simulatin. The bserved setup is the difference between the bserved water level and the predicted level frm tide tables. During Agnes, the surge setup was bvius during tw days, frm August 3 t September, due t strng typhn intensity and the shrt distance f the lw-pressure center frm the regin. The mdeled maximum setup accrded well with the bserved ne, abut m bth at and, and.5 m at Waigaqia. During Matsa, the pattern f the surge prcess was similar, but with lwer value (mainly due t different wind directins between the tw categries), abut.5 m at and, and. m at Waigaqia. There was sme discrepancy during the last hurs f August, which may have been caused by sme ther weather prcesses that were nt cnsidered here. Overall, the gd agreements between the mdeled and the bserved data shwed that the simulatin f strm surge was successful. The distributins f the surge setup during tw typhns at the time when maximum setup ccurred are shwn in Figure 5. At 3: UTC n August 3, the lw-pressure center (pentagram in sub-figure) f Typhn Agnes was very clse t the YE-HB. The wind directin was cunter-clckwise near the regin, that is, easterly in the nrth and westerly in the suth, which caused surge set up in the nrth (.3 m) and set dwn in the suth (-. m). The maximum setup appeared at tw places. One was the muth bar regin f YE, where strm surge ffshre was cnverged by the gradually shallw depth under the east wind. The ther place is the suth part f HB, where strm surge induced by suth-west winds was blcked by the islands. Bth setups reached. m. At : UTC, August, the center f Matsa was very clse t the suth cast f HB. Winds were easterly and sutheasterly thrughut the regin, which caused a high surge setup in the nrth part f HB and the upstream f YE. These setups reached. m and. m, respectively. In additin, it shuld be pinted ut that winds in the nrth part f the larger dmain (ECS) were mainly caused by backgrund winds, nt by typhn-induced winds, because the center f Matsa was far away. DISCUSSION There are several factrs that affect the accuracy f strm surge calculatins, which include strm-induced wind, wind-waves, astrnmical tide and the setting f the cmputatinal dmain. In this study, the effects f waves and remte winds were prbed by cnducting numerical experiments. Remte wind was cnsidered by a nesting cmputatin. The nrmal simulatin with waves and nesting is defined as Case ; the test simulatin with n wave effects (with nesting) is defined as Case ; the test simulatin with n nesting (with waves) is defined as Case. Fr the legibility f figures, the results f Cases and are nt directly added t Figure b. The surge differences between Cases, and are shwn in Figure. Their differences can als be viewed as waveinduced setup and remte-wind-induced setup, respectively. The slid line in Figure represents the surge difference due t waves. During Agnes, the biggest difference appeared frm August 3 t September, when there were huge waves. This difference was abut. m, which was abut % f the ttal surge setup. During Matsa, the wave-induced setup was less than thse during Agnes, abut. m. It shuld be pinted ut that sme statins, like and, are nt lcated in the surf zne, where wave effects will be mre imprtant. S fr extreme strms Jurnal f Castal Research, Special Issue 5, 7

5 Hu et al. Typhn Agnes () Typhn Matsa (59) Water level (m) Waigaqia Date (Aug.~ Sep., 9) 5 7 Date (Aug., 5) Typhn Agnes () Typhn Matsa (59) Surge setup (m) Waigaqia Date (Aug.~ Sep., 9) Date (Aug., 5) Figure. Cmparisns f water level and surge setup at three statins (, and Waigaqia) during Typhns Agnes and Matsa (slid line fr the mdeled; dashed line fr the bserved). -.m m.m.m.m.m Latitude (N) 3 3 m/s m/s (b) Typhn Matsa (59) : UTC, Aug. 3 Lngitude (E) Figure 5. Distributins f the surge setup in the YE-HB and wind field in the ECS (sub-figure) during Typhns Agnes and Matsa. Jurnal f Castal Research, Special Issue 5, 7

6 Mdeling f strm surge.. Typhn Agnes ().. Typhn Matsa (59) Waigaqia Surge difference (m) Date (Aug.~ Sep., 9) Date (Aug., 5) Figure. Surge difference with different mdel cnditins at three statins (, and Waigaqia) during Typhns Agnes and Matsa (slid line fr the difference between Case and Case ; dashed line fr the difference between Case and Case ). with huge waves, wave effect shuld be accunted fr t prduce an accurate simulatin f strm surge, especially in shallw places. The dashed line in Figure represents the difference in surge due t remte winds. It can be seen that the nesting cmputatin is essential fr bth typhns. The difference in surge was abut. m fr Agnes and. m fr Matsa. If the effect f remte winds were excluded, the final results wuld be a lt less accurate. CONCLUSIONS Taking Typhns Agnes and Matsa as examples, an integrated mdel system was applied t the waters f the YE-HB fr the simulatin f strm surge. The effects f waves and remte winds were cnsidered. The results f mdeling winds, waves, astrnmical tides and strm tides were cmpared with bserved data. Gd agreements between the bserved and mdeled data shwed that this mdel system is suitable fr strm surge calculatin in the YE-HB. Thrugh numerical experiments, it was fund that waves shuld be cnsidered in the case f strng strms, and nesting cmputatin shuld be adpted if the calculatin dmain is nt large enugh t include the effect f remte wind. Many factrs, such as winds, waves, astrnmical tides and the setting f the calculatin dmain, affect the final accuracy f strm surge mdeling. The wind mdel can be imprved by using an asymmetric pressure distributin and the methd f data assimilatin. The spectral wave mdel SWAN cannt fully accunt fr the effect f wave diffractin. Therefre, the cupling f SWAN with mmentum-based wave mdels, such as mildslpe r Bussinesq mdels, is desirable fr sme lcal places. 3- D simulatin can reflect mre realistically the prcesses in nature. These will be ur future wrk. ACKNOWLEDGEMENTS This study was funded by the Natinal Key Basic Research Prject (N. CB3) in China and the Research Prjects in Science and Technlgy Cmmissin f Shanghai Municipality, China (N. DZ5 & DZ). The referees cmments imprved the manuscript substantially and are kindly appreciated. LITERATURE CITED BOOIJ, N., HAAGSMA, IJ.G., HOLTHUIJSEN, L.H., KIEFTENBURG, A.T.M.M., RIS, R.C., van der WESTHUYSEN, A.J. and ZIJLEMA, M.,. SWAN User Manual (Cycle III versin.). Delft University f Technlgy, Delft, the Netherlands, 5p. FUJIATA, T., 95. Pressure distributin within a typhn. Gephysical Magazine, 3, GRANT, W.D. and MADSEN, O.S., 9. The cntinental shelf bttm bundary layer. Annual Review f Fluid Mechanics,, 5-. HOLLAND, G.J., 9. An analytic mdel f the wind and pressure prfiles in hurricanes. Mnthly Weather Review,, -. HU, K., DING, P. and CHEN, Q., 5. Mdeling f typhn windwaves arund the Yangtze Estuary, China. Prceedings f the 9th Internatinal Cnference n Castal Engineering (Lisbn, Prtugal, ASCE), Vlume, pp HU, K., DING, P., ZHU, S. and CAO, Z.,. -D current field numerical simulatin integrating Yangtze Estuary and Hangzhu Bay. China Ocean Engineering, (), 9- JELESNIANSKI, C.P., CHEN, J. and SHAFFER, W.A., 99. SLOSH: sea, lake, and verland surges frm hurricanes. NOAA Technical Reprt NWS, 77p. JONES, J.E. and DAVIES, A.M., 99. Strm surge cmputatins fr the Irish Sea using a three-dimensinal numerical mdel including wave-current interactin. Cntinental Shelf Research,, -5. MADSEN, H. and JAKOBSEN, F.,. Cyclne induced strm surge and fld frecasting in the nrthern Bay f Bengal. Castal Engineering, 5, PENG, M., XIE, L. and PIETRAFESA, L.J.,. A numerical study f strm and inundatin in the Cratan-Albemarle-Pamlic Estuary System. Estuarine, Castal and Shelf Research, 59, -37. SHEN, J., WANG, H., SISSON, M. and GONG, W.,. Strm tide simulatin in the Chesapeake Bay using an unstructured grid mdel. Estuarine, Castal and Shelf Research,, -. SHI, F., SUN, W. and WEI, G., 997. A WDM methd n a generalized curvilinear grid fr calculatin f strm surge flding. Applied Ocean Research, 9, 75-. SMITH, S.D., 9. Wind Stress and heat flux ver the cean in gale frce wind. Jurnal f Physical Oceangraphy,, WONG, K.-C. and MOSES-HALL, J.E., 99. On the relative imprtance f the remte and lcal wind effects t the subtidal variability in a castal plain estuary. Jurnal f Gephysical Research, 3(C9), Jurnal f Castal Research, Special Issue 5, 7