Development of an oil spill forecast system for offshore China*

Transcription

1 Chinese Jornal of Oceanology and Limnology Vol. 34 o. 4, P , 6 Development of an oil spill forecast system for offshore China* WAG Yonggang ( 王永刚 ),, WEI Zexn ( 魏泽勋 ),, * *, A Wei ( 安伟 ) 3 First Institte of Oceanography, State Oceanic Administration (SOA), Qingdao 666, China Laboratory for Regional Oceanography and merical Modeling, Qingdao ational Laboratory for Marine Science and Technology, Qingdao 666, China 3 China Offshore Environmental Services Ltd., Tianjin Tangg 345, China Received Jan. 9, 5; accepted in principle Apr. 3, 5; accepted for pblication Jn. 8, 5 Chinese Society for Oceanology and Limnology, Science Press, and Springer-Verlag Berlin Heidelberg 6 Abstract An oil spill forecast system for offshore China was developed based on Visal C++. The oil spill forecast system incldes an ocean environmental forecast model and an oil spill model. The ocean environmental forecast model was designed to inclde timesaving methods, and comprised a parametrical wind wave forecast model and a sea srface crrent forecast model. The oil spill model was based on the particle method and flfills the prediction of oil particle behavior by considering the drifting, evaporation and emlsification processes. A specific database was embedded into the oil spill forecast system, which contained fndamental information, sch as the properties of oil, reserve of emergency eqipment and distribtion of marine petrolem platform. The oil spill forecast system was sccessflly applied as part of an oil spill emergency exercise, and provides an operational service in the Research and Development Center for Offshore Oil Safety and Environmental Technology. Keyword : oil spill; China offshore; particle method; emergency service ITRODUCTIO With the rising demand for petrolem, there has been an increase in the freqency of oil spill accidents dring marine oil exploration, transportation, loading and nloading processes. In the past few years, several serios accidents have occrred that have had a considerable impact on the marine environment, sch as the Deepwater Horizon oil spill in the Glf of Mexico, the Dalian Port oil spill and the Penglai 9-3 oil field in the Bohai Sea. Oil spills have become a serios problem, casing offshore marine environmental polltion and ecological damage. Therefore, it is crcial to develop an oil spill forecast system, which can provide accrate predictions of oil spill movement for polltion control and cleanp (Li et al., a; M et al., a; Cai et al., 3). The Deepwater Horizon oil spill, which contined for 3 months, was the largest and the most costly oil spill in US history (Li et al., a). In rapid response to this oil spill, a system for tracking the oil was developed by combining satellite-monitoring information with an ensemble of six nmerical ocean circlation models (Li et al., b, c). Faced with the potential risk of an oil spill offshore of China, mch needs to be done to nderstand the behavior and movement of oil spills sing experiments, and theoretical and nmerical models. To that end, researchers have developed several real-time forecast systems for oil spills. An et al. () developed a system for predicting oil spill movements and corresponding operations offshore of China, composed of a 3-D (three-dimensional) hydrodynamic model, an oil weathering model, environmental sensitive area maps, and a decision spport model. M et al. (a, b) developed an oil spill emergency warning and predicting system for the Bohai Sea, which inclded an ocean environmental forecast model and an oil spill drift model. Li et al. () set p a nmerical model of marine oil spills based on a -D (two-dimensional) nested regional hydrodynamic model for the Bohai Sea, and stdied the effect of * Spported by the ational High Technology Research and Development Program of China (863 Program) (o. 3AA9A56) and SFC- Shandong Joint Fnd for Marine Science Research Centers (o. U4644) ** Corresponding athor: weizx@fio.org.cn

2 86 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol.34 wind and crrents on the path of spilled oil. Li and Xie (3) proposed an oil spill trajectory predictive model based on an EFDC (Environmental Flid Dynamic Code) model, and oil particle and weathering models. They conclded that an oil particle model is more realistic and accrate in simlating spill trajectory and morphology, as compared with traditional spill models based on advection and diffsion formlation. Based on a -D mathematical tidal crrent model, G and Yao (3) developed an oil spill model sing the oil particle method and stdied the inflence of tidal crrents and wind on a predicted oil spill, in Yeqing Bay. Li et al. (3) condcted a case stdy to analyze error sorces in a 3-D oil spill model. They developed an inverse model to estimate the temporal variability of emission intensity at the oil spill sorce. Li et al. (4) applied OI (optimal interpolation) assimilation technology in the oil spill emergency forecasting system in the Bohai Sea, to improve the accracy of oil spill nmerical forecasting reslts. Althogh some oil spill systems have been developed, these forecast systems either focs on the Bohai Sea or focs on a small region. Frther stdy is needed to develop an oil spill forecast system that covers the entire Chinese offshore region. In the present stdy, an oil spill forecast system was developed for the entire Chinese offshore region. Section describes the oil spill nmerical model, which consists of an ocean environmental forecast model and an oil spill model. Section 3 presents the strctre of the oil spill forecast system and its application in predicting spilled oil in the East China Seas (ECS). Section 4 contains discssion and conclsions. OIL SPILL FORECAST SYSTEM Sea srface wind, crrents, waves and water trblence are the main inflences on the drifting and spreading of spilled oil on the sea srface. In other words, the horizontal movement of oil is determined by wind, wind-indced crrent, tidal crrent, tidal residal crrent, wave-indced residal crrent, density-indced crrent and trblent diffsion. The sea srface wind and the properties of oil determine the evaporation and emlsification, and frther affect the density and viscosity of the oil. Srface waves affect the oil in two ways. First, they bring some of the oil into the water colmn becase of the mixing effect and wave breaking. Second, the waves affect the trajectories of the oil spill via wave-indced Stokes drift and wave-indced Coriolis-Stokes force (Röhrs et al., ). In Chinese offshore areas, the wind, wind-indced crrent, tidal crrent, and ocean circlation are the most important factors determining the trajectory of oil. To better forecast the trajectory and fate of oil spills, we designed an oil spill forecast system for offshore China. This forecast system was composed of two sb-models: an ocean environmental forecast model and an oil spill model.. Ocean environmental forecast model The ocean environmental forecast model is responsible for forecasting waves and srface crrents. Two kinds of interface were designed in this oil spill forecast system. The first interface is the operational ocean environmental forecast system, which can spply real time forecast prodcts. It was designed based on the forecast prodct of an operational ocean environmental forecast system. This forecast system consisted of a high resoltion wave-crrent copled model with a data assimilation modle, which covers the Soth China Sea and its adjacent seas (99 3 E, 3 S 4 ), and provides a 7 hor forecasting prodct of the marine environment (Wei et al., 5). The second interface was based on some timesaving methods, inclding a parametrical wind wave model and a qick forecast sea srface crrent model, to meet the demand of a qick response to the emergency event of an oil spill... Parametrical wind wave model A parametrical wind wave model (Wen et al., 989) can be sed to qickly predict significant wave height, wave period and wave nmber. Given wind speed and fetch x, a significant wave height and period were fond, as follows: For shallow water, gd.8 ( ) gh s -3 gx ( ) th[3 ]. gx.35 ( ) () gd.8 ( ) gt gx.33 /3 5.5( ) (th[3 ]). gx.35 ( ) () For deep water, gh s -3 gx ( ), (3)

3 o.4 WAG et al.: Oil spill forecast system for China offshore 86 H s (m) T s (s) 8 4 /5/87 /6/87 /7/87 /8/87 /9/ /5/87 /6/87 /7/87 /8/87 /9/87 gt a b Fig. Comparison between model otpts (red line) and observation (black line with asterisk) a. significant wave height; b. significant wave period. gx ( ), (4) where, H s, T, g, d denote significant wave height, period, gravitational acceleration and water depth respectively. Fetch lengths were prepared in advance in eight directions as a backgrond data set. For both deep and shallow water, the following wave heightperiod relationship exists: /3 H T 8.5 s. (5) /3 After wave height and period were compted, wave freqency and wave nmber were calclated by f, (6) T =gkth( Kd). (7) The forecasted H s, T, K were sed in the oil spill forecast, which will be described in the oil spill model E (Section.). Figre shows a comparison between the parametrical wind wave model otpts and observations. It can be seen that the parametrical wind wave model can basically captre the variations of ocean waves. The root mean sqared error is. m for significant wave height and.4 s for significant wave period. These accracies can preliminarily meet the demands of an oil spill model to calclate the vertical eddy diffsivity and mixing of the oil spill into the water volme. The timesaving parametrical wind wave model takes less than one minte to predict 3 days, whereas in contrast, the nmerical wave forecast model takes several hors. Faced with the demand for a qick response to an emergency oil spill, the parametrical wind wave model is acceptable and serviceable... Sea srface crrent forecast model The sea srface crrent forecast model consists of a tidal crrent model, a circlation model and a wind drift model. This oil spill forecast system was developed to focs on oil slicks at the sea srface of Chinese offshore waters, where the tidal crrent model plays an essential role in coastal dynamic processes. Ths, for short term forecasting (several hors), the tidal crrent forecast model is key in oil spill drifting, however, for longer-term forecasting (several days), residal crrents and wind drifting velocity are more important. Figre shows the percentage of the tidal crrent that acconts for the sea srface crrent analyzed based on the operational P 8% 6%<P 8% 4%<P 6% P % Fig. The ratio of tidal crrent to sea srface crrent %<P 4%

4 86 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol.34 tidal crrent forecast system (Fang et al., 8) and an ocean general circlation model (Wang et al., 6) by Eq.8. P U tide, (8) U U tide crrent where U tide is the maximm possible tidal speed, U crrent is the annal mean speed of circlation. The tidal crrent acconted for over 8% of the sea srface crrents in most parts of northern 3 and the Beib Glf. On the coasts of Fjian Province, this percentage is sally between 6% and 8%, whereas it is arond 5% in the coastal region of Gangdong Province. To correspond with the emergency services, the qick forecast techniqe was selected to forecast sea srface crrents based on residal crrents and tidal crrent harmonic constants, sing Eq.9a and 9b. I Ut () U f() tucos t () t, (9a) i i i i, i i i I Vt () V f() tvcos t () t, (9b) i i i i, i i i w here U ( t ) and V ( t ) denote the forecasted eastward and northward crrent. U and V are eastward and northward residal crrent, the sbscript i represents the tidal constitents, U, ξ and V, η are the amplitde and phase lag for eastward and northward of tidal crrent, ω is the anglar speed, v is the initial phase of the eqilibrim tide. f is the nodal fact, and is the nodal adjstment angle of the initial phase. For the residal crrent ( U and V ), one sorce was derived from the forecast prodct of an operational ocean environmental forecast system (Wei et al., 5), and the other sorce was derived from the climatological monthly mean srface crrent, which originated from an ocean general circlation model. At present, the second interface is sed as the defalt option to flfil the qick forecast demand. Figre 3 shows the srface residal crrent based on the variable grid global ocean model (Wang et al., 6; Fang et al., 9). It can be seen that the sea srface crrents of offshore China are a typical, monsooncontrolled circlation. In winter, the sea srface crrents are prevailingly sothwest in the coastal region, except for in the Taiwan Strait. In the Bohai Sea, the sea srface crrents exhibit weak circlation and the residal speeds are less than. m/s. At the west and north coast of the Yellow Sea, the sea srface crrents also show weak circlation. For the rest region of the Yellow Sea, the sea srface crrents are less than. m/s. In the west part of the East China Sea, the sea srface crrents are fairly strong, with speeds greater than. m/s. In the northern Soth China Sea (except for the Beib Glf), the sea srface crrents are stronger than in other coastal region. In the smmer, the sea srface crrents are basically weaker than in the winter, and are in the reverse direction. The sea srface crrents in most of the coastal region are less than. m/s except for offshore of Zhejiang and Fjian Provinces. The sea srface crrents are in transition conditions in the spring and atmn. As the sea srface crrents of offshore China (especially for the Bohai Sea, Yellow Sea and East China Sea) are weak and dominated by monsoon, the residal crrents, which are interpolated from climatological monthly mean crrent, can be sed for the oil spill forecast system. The tidal crrents forecast model was based on the nmerical simlation reslts for China s adjacent seas (Fang et al., 8). The tidal ellipses for the principal tides of offshore China are shown in Fig.4, with different scales for each tidal constitent. It can be seen that the semi-dirnal tidal crrent dominated the srface tidal crrents of the East China Seas, except for in a small region near the Bohai Strait. On the contrary, the tidal crrent in the Beib Glf was controlled by the dirnal tidal crrent. To evalate the tidal crrent constitents, over 35 stations of tidal crrent constitents (Fig.5) were selected and compared with model reslts. To demonstrate the comparisons, the tidal crrent constitents were separated into a, a, a 3 and a 4 sing Eq. for the eastern ( U, ξ ) and northern ( V, η ) tidal crrent constitents: a = U cos ξ, a = U sin ξ, a 3 = V cos η, a 4 = V sin η, (a) (b) (c) (d) Figre 6 is the scattered diagram of the separated vales ( a, a, a 3 and a 4 ) for the principal semidirnal tidal crrent M and principal dirnal tidal crrent O. Abscissa represents observed vales and ordinate represents model reslts. It can be seen that most of the scattered points are near the diagonal line (the red solid line), indicating the reliable model reslts. To evalate the validation of the crrent forecast model, for sea crrent observation stations (Fig.7) were sed for time series comparison of crrent speed and direction (Fig.8). The defalt interface was

5 o.4 WAG et al.: Oil spill forecast system for China offshore Janary nits: m/s April nits: m/s a b E E 4 Jly nits: m/s ovember nits: m/s c d E E Fig.3 Seasonal variations in sea srface crrents (a) winter, (b) spring, (c) smmer, (d) atmn selected for the residal crrent ( U and V in Eq.9), which was derived from climatological monthly mean srface crrents. Table shows the mean absolte errors and root mean sqared (RMS) errors of speed and direction for sea crrent forecast. The mean absolte error of crrent speed was 4.6 cm/s and the mean absolte error of crrent direction was.. The RMS of crrent speed was 5.7 cm/s and RMS of crrent direction was 6.9. It can be sggested that the second interface of the residal crrent is sitable for most parts of offshore China. However, for the offshore region of the northern Soth China Sea, where circlation is vigoros and complicated, the first interface was more favorable.

6 864 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol.34 4 a 4 b 3. m/s 3. m/s 5 5 E E 4 c 4 4 d 3. m/s 3. m/s E E Fig.4 Srface tidal crrent ellipses, (a) M, (b) S, (c) K, (d) O. Oil spill model The oil spill model was based on the particle method, which represents the oil spill as a large nmber of particles. The drifting velocity, evaporation and emlsification processes control the evoltion of each oil particle... Horizontal and vertical movement of the oil spill The drifting velocity of oil particle U p ( t ) and V p ( t ) can be denoted as: U () t U() t U () t U' () t p d V () t V() t V () t V' (), t () p d

7 o.4 WAG et al.: Oil spill forecast system for China offshore a E 4 3 b Fig.5 Stations of tidal crrent constitents 4 B A Fig.6 Comparison between model reslts and observations, (a) M and (b) O 3 Table Validation of speed and direction for sea crrent forecast C Stations Time series (5 days) Speed (cm/s) Direction ( ) Mean error RMS Mean error RMS 8 A D B C D Mean E Fig.7 Stations sed for sea crrent observations where U ( t ) and V ( t ) are the backgrond crrents provided by the sea srface crrent forecast model

8 866 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol Speed (cm/s) 6 4 Speed (cm/s) Direction ( C) Direction ( C) a. Station A b. Station B 6 5 Speed (cm/s) Speed (cm/s) Direction ( C) Direction ( C) c. Station C Fig.8 Comparisons of the sea crrents d. Station D (Eq.9), U d ( t ) and V d ( t ) are wind drifting velocities, and U ( t ) and V ( t ) are random trblent velocities of the oil particles. The horizontal movements of the oil particles within a sb-grid can be calclated throgh the following eqations: n n n p H x x U t 6 K t o( t ) n n n p H y y V t 6 K t o( t ). () We adopt central difference in time t to ensre the above difference eqations to be of a second order accracy. ξ, K H in the Eq. denotes the niformly distribted random nmber in [-,] and the horizontal eddy viscosity, respectively. As the sea srface sally becomes smoother when covered by the spilled oil, the vale of wave indced flow will be mch smaller than that of the backgrond crrent. Waves mainly affect the oil spill via the mixing effect and wave breaking, which indce some of the oil spill mixing into the water. The volme of the oil particle mixed into water is: Ve C th / L S e V, (3)

9 o.4 WAG et al.: Oil spill forecast system for China offshore 867 where V, t, H s, L are the initial volmes of the oil spill, time, significant wave height and wave length (provided by parametrical wind wave model) respectively, and C is a constant, / V.6. The oil spill mixed into water can be represented by oil drops, which are mch smaller than the oil particle. The vertical displacement of oil drops can be calclated by Eq.4, where W b is the vertical velocity of an oil drop, W L is the floating velocity cased by the boyancy, and K V is the vertical eddy viscosity. z WbWLt 6Kv t. (4) The vertical eddy viscosity is derived from the Reynolds stresses, de to srface waves (Ichiye, 967): H s -Kz Kv.8 e C, (5) T where H s, T, Z, K are significant wave heights, wave periods, water depth and wave nmber, respectively, and C is a constant. The float velocity was calclated on the basis of the diameter of each oil drop. Under the action of boyancy, assming the critical diameter of the oil drop is de, we have: /3 9.5v de, (6) /3 /3 g o/ W when di < de and Stokes law is applied, leading to: W g d o/ W/8 v, (7) L i when d i > d e 8 / W L g di o / W 3, (8) where g, d i, ν, ρ o, ρ W are gravity accelerators, the diameter of oil drops, kinetic viscosity, the density of oil and the density of water, respectively. Ths, the vertical coordinates of oil drops can be calclated throgh the following eqation: n n n/ Z Z Wb WL t 6 Kv t t. (9).. Evaporation and emlsification of the oil spill The evaporation and emlsification of the oil are related to the oil properties. The rate of evaporation can be written as follows (Stiver and Mackay, 984): T T A B G V ln B T T F e, () T BT G where A =6.3, B =.3, T is the temperatre of spilled oil, T G is the crve gradient of the oil boiling point, T o is the initial boiling point of the oil, and θ is the coefficient of evaporation, which was defined by the following eqation (Bchanan and Hrford, 988):.78 C W ta/ V, () o where C is a constant, W is wind speed, t is time, A is the area of the oil slick, and V o is the initial volme of the oil spill. The effect of emlsification is defined by the moistre content Y W, following Mackay et al. (98): -K K A B W t YW e K B, () where Y W is the moistre content in the emlsion, K A is taken as 4.5-6, K B is taken as / Y F W, in which Y F W is the final moistre content and is taken as.5. Ths the volme of oil particles on the sea srface shold be: V V F / Y. (3) i o Vi Wi 3 GRAPHICAL USER ITERFACE AD APPLICATIO A graphical ser interface (GUI) of the oil spill forecast system was designed sing Visal C++. Figre 9 shows the main interface of the system. The main men of the system incldes six parts. The first men is File. It presents the oil spill forecast service, which is the most fndamental fnction of the system. The second men can set the display settings, sch as zoom in, zoom ot and selective amplification. The third men controls the display information of the forecasted oil spill, sch as the oil spill trajectory, the shape of the oil slick and the scatter of the oil particles. The forth men gives some information on decision spport, inclding the handbook of emergency eqipment and some emergency preplans. The fifth men is the management of the database of the system. The oil spill forecast system embeds a database to manage the properties of oil (sefl for forecasting the rate of evaporation), a reserve of emergency eqipment and the distribtion of the marine petrolem platforms. The sixth men contains a ser gide for this system. Figre shows a demonstration of the oil spill forecast system. The system can qickly forecast the trajectory of spilled oil and provide the shape of the oil slick, the residal qantity of spilled oil, and affected areas. Other serviceable information, sch as positions of petrolem platforms, oil pipelines, and emergency

10 868 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol.34 Fig.9 Main interface of the oil spill forecast system Fig. Demonstration of an oil spill forecast

11 o.4 WAG et al.: Oil spill forecast system for China offshore 869 a b Observe Forcast b Fig. Oil spill emergency exercise (a) and validation of the oil spill forecast system (b) eqipment bases, can be displayed simltaneosly. An oil spill emergency exercise was held at 35.8 E, , in the East China Sea, on th Agst, 8. A tracer was released and tracked by the Hai Yang Shi Yo 689 ship (Fig.a). Prior to this emergency exercise, the trajectory of the oil spill was forecasted by or oil spill forecast system (the brown line in Fig.b). Dring the emergency exercise, we tracked the tracer and recorded the position horly (the white line in Fig.b). Based on this oil spill emergency exercise, the average Lagrangian separation distance was.56 km. To evalate the performance of the oil spill model, the skill score, as proposed in Li and Weisberg (), was sed. The skill score ss for the trajectory model is defined by Eq.4, where s is a non-dimensional index (defined as Eq.5), n is a positive nmber that defines the threshold of no skill, d i is the separation distance between the forecasted and observed track at time step i, and l oi is length of the observed trajectory. In this paper we chose n =, and a skill score of.5 for this oil spill emergency exercise. According to the application on this oil spill emergency exercise, the developed oil spill forecast system exhibited a good performance and was able to satisfy an oil spill emergency response. ss = ( s / n ), (4) d s l i i oi i, (5) 4 DISCUSSIO AD COCLUSIO A nmerical model for an oil spill was described, which inclded an ocean environmental forecast model and an oil spill model. The timesaving ocean environmental forecast model comprised a parametrical wind wave forecast model and a sea srface crrent forecast model, and considered the demand for a qick response to the emergency event of oil spill. The oil spill model was based on the particle method. The evoltion of each oil particle is controlled by drifting velocity, evaporation and emlsification processes. The GUI for an oil spill, based on these nmerical models, is also described. Significant information, sch as the properties of oil, emergency eqipment reserves and distribtion of marine petrolem platforms, were embedded into the system with a manageable database. The oil spill forecast system was sccessflly applied for an oil spill emergency exercise, and provides an operational service in the Research and Development Center for Offshore Oil Safety and Environmental Technology. As described in Section., two interfaces were designed for the residal crrent ( U and V ) in the sea srface crrent forecast model. The second interface, which was based on a climatological monthly mean srface crrent, was set to the defalt option. Althogh this defalt option shows some satisfactory reslts in the East China Seas (the Bohai Sea, Yellow Sea and East China Sea), the first interface, which was derived from the operational ocean environmental forecast system with mltiple model ensemble (Li et al., c), is more favorable. Althogh the oil spill forecast system was validated by an oil spill emergency exercise, more extensive validations shold be implemented sing the skill scores, as proposed in Li and Weisberg (). As re-initialized locations of oil spills, dly obtained from oil spill monitoring systems, are critical

12 87 CHI. J. OCEAOL. LIMOL., 34(4), 6 Vol.34 components of the oil spill forecast system (Li et al., a), combined with an operational oil spill monitoring system was planned to implement in following development. We fond that the oil spill model was sitable for forecasting the drifting and spreading of the spilled oil on the sea srface. However, we also fond that when the spilled oil was attached to the coastline, the oil spill model was not very good at forecasting, becase it is very difficlt to determine how mch oil will remain on the coastline, and how mch oil will move back to the ocean. Therefore, we are still developing the oil spill nmerical model, to improve its sability and accracy. References An W, Wang Y G, Wang X Y, i Z G, Zhao Y P.. An oil spill forecast and emergency decision spport system in China offshore. Marine Sciences, 34 (): (in Chinese with English abstract) Bchanan I, Hrford Methods for predicting the physical changes in oil spilt at sea. Oil and Chemical Polltion, 4(4): Cai Y, M L, Li H, Song J, Chi Y X, Gan C Y, Li C. 3. A review of nmerical modeling research on the marine oil spill. Marine Science Blletin, 5 (): Fang G H, Wang Y G, Wei Z X, Fang Y, Qiao F L, H X M. 9. Interocean circlation and heat and freshwater bdgets of the Soth China Sea based on a nmerical model. Dyn. Atmos. Oceans, 47 (-3): Fang G H, Wei Z X, Wang Y G. 8. Development of tide and tidal crrent regional prediction in China. Advance in Earth Science, 3 (4): (in Chinese with English abstract) G E H, Yao Y M. 3. Fractal simlation of oil spill trajectory in the northern port area of the Yeqing Bay. Marine Science Blletin, 3 (4): , 84/j.issn (in Chinese with English abstract) Ichiye T Upper ocean bondary-layer flow determined by dye diffsion. The Physics of Flids, (9): S7-S77. Li D M, Li J C, W D, Bai L.. Mathematical model of marine oil spill in Bohai. Jornal of Tianjin University, 45 (): (in Chinese with English abstract) Li T, Xie Z Y. 3. Stdy and application of oil spill trajectory model. Environmental Science and Management, 38 (7): (in Chinese with English abstract) Li Y, Zh J, Wang H, Kang X D. 3. The error sorce analysis of oil spill transport modeling: a case stdy. Acta Oceanol. Sin., 3 (): 4-47, s Li Y, Zh J, Wang H, Lin C Y. 4. The assimilation technology application in the oil spill emergency forecasting system of the Bohai Sea. Acta Oceanol. Sin., 36 (3): 3-, (in Chinese with English abstract) Li Y G, MacFadyen A, Ji Z G, Weisberg R H. a. Trajectory forecast as a rapid response to the Deepwater Horizon oil spill. In : Li Y G, Weisberg R H, H C M, Zheng L Y eds. Monitoring and Modeling the Deepwater Horizon Oil Spill: A Record-Breaking Enterprise. American Geophysical Union, 95 : 53-65, doi.org/.9/gm. Li Y G, MacFadyen A, Ji Z G, Weisberg R H. b. Monitoring and Modeling the Deepwater Horizon Oil Spill: A Record-Breaking Enterprise. AGU/Geopress, Washington D.C., Li Y G, Weisberg R H, H C M, Zheng L Y. c. Tracking the deepwater horizon oil spill: a modeling perspective. Eos Trans., AGU, 9 (6): 45-46, EO6. Li Y G, Weisberg R H.. Evalation of trajectory modeling in different dynamic regions sing normalized cmlative Lagrangian separation. J. Geophys. Res, 6 (C9): C93, Mackay D, Paterson S, Trdel K. 98. A Mathematical Model of Oil Spill Behavior. Environmental Protection Services. Fisheries and Environment, Canada. M L, W S Q, Song J, Li H, Li S H, Li Y, Gao J. a. merical model research on Emergency Warning and Predicting system of ocean oil spill: I. Research on predicting of ocean dynamical factors. Marine Science Blletin, 3 (5): (in Chinese with English abstract) M L, W S Q, Song J, Li H, Li S H, Li Y, Gao J. b. merical model research on Emergency Warning and Predicting system of ocean oil spill in Bohai Sea: II. The visalization and the research on application. Marine Science Blletin, 3 (6): (in Chinese with English abstract) Röhrs J, Chistensen K H, Hole L R, Broström G, Drivdal M, Sndby S.. Observation-based evalation of srface wave effects on crrents and trajectory forecasts. Ocean Dynamics, 6 (-): Stiver W, Mackay D Evaporation rate of spills of hydrocarbons and petrolem mixtres. Environmental Science & Technology, 8 (): Wang Y G, Fang G H, Wei Z X, Qiao F L, Chen H Y. 6. Interannal variation of the Soth China Sea circlation and its relation to El iño, as seen from a variable grid global ocean model. J. Geophys. Res., (C): CS4, Wei Z X, X T F, Wang Y G, Li H P, Yang X L, Feng W Z, Zhang Z Y. 5. Introdction to the fine-resoltion windwave-crrent copled forecasting and information service system for the Soth China Sea and its adjacent seas areas. Jornal of Ocean Technology, 34 (3): (in Chinese with English abstract) Wen S C, Zhang D C, Chen B H, Gao P F A hybrid model for nmerical wave forecasting and its implementation: Part I, the wind wave model. Acta Oceanol. Sin., 8 (): -4.