Test of numerical prediction of sea water temperature in the Taiwan Strait

Acta Ceanologica Sinica, Vol. 20, No. 4, pp. 473~481, 2001.ISSN 0253-505XPublished by China Ocean Press.Test of numerical prediction of sea water temperaturein the Taiwan Strait *Wang Xiuqin! , Huang Huowang，Dong Jian2，Qian Chengchun'(Reeived July 25, 2001; accepted Axugust 8, 2001)Abstract- A dynamic numerical prediction model of sea water temperature for limited sea area isused to predict the sea water temperature at the sea area near Fujian. Essential adjustments have beenmade in accordance with the characteristics of this region. Two tests have been made. One is insummer (3 d) and the other is in winter (10 d). In the summer test，a typhoon is just passing byand the calculated current field well responds to typhoon. In the winter test，variation tendency ofthe predicted sea water temperature field agrees with that of the observation basically, the absolutemean error in the whole sea area is 0.6 C. The variation of the sea water temperature is mostly af-fected by entrainment and pumping, which is related to the topography of the strait.Key words The Taiwan Strait, numerical prediction of sea water temperatureINTRODUCTIONIn the ocean environment prediction numerical models in the China seas, the TaiwanStrait was usually considered as open boundary either in the study of the East China Sea or theSouth China Sea，so it is difficult to get a satisfactory predicted result in this strait area.Recently many studies on the Taiwan Strait have been made, and the prediction serviceprojects of the marine prediction stage are increasing and have achieved high-favorable economicbenefit. Nevertheless, the numerical sea water temperature prediction for a large area whichconsiders the region characteristics has never been carried out in China. And the sea area nearFujian is an important fishery of China. All kinds of fish strongly depend on the sea water tem-perature at their every growing stage and are seriously influenced by the ocean environment.The sea water temperature change will determine the aggregation and movement of fish and thetime and positions the fish groups formed. So, the study pf pon water temnorrature prediction中国煤化工* The project was financilly supported by the Natural Science FouJIYRCN M H Gxt No.099E02 andthe special fund for authors excellent Ph. D. dissertation of the whole country under contract No. 200021.1. Department of Oceanography，Ocean University of Qingdao, Qingdeo 266003, China2. Marine Prediction Stage of Xiamen, Xiamen 361002, China474Wang Xiuqin et al .especially for this area is very important.The study of numerical prediction of sea water temperature has been made for many yearsand many scholars had reported their study results. Recently some models adapted to differentareas were reported, such as the Ocean University of Qingdao' s model ( hereafter referencedas OUQ model)(Wang et al.，1998， 2000)， McCreary' s model (McCreary and Yu,1992)，Wang' s model (Wang and Chang, 1995) and Ravindran' s model (Ravindran et al.,1999). The OUQ model is manufactured according to the region characteristics of the Chinaseas. In the present paper, the OUQ model is adopted and necessary modification has beenmade according to the region feature.The detailed description of the prediction model is given in Section 2. The source of datarequired in models is briefly described in Section 3. The detailed descriptions of the test resultsare given in Section 4. And main conclusions are summarized in Section 5.PREDICTION MODELIn this paper，the OUQ model is adopted. The OUQ model is a two -dimensional dynamicnumerical prediction model of sea water temperature for the limited sea area, it was establishedin the study of the numerical prediction of abnormal sea water temperature in the Huanghai Seaand the East China Sea (Wang et al., 1998， 2000)， and can predict sea surface temperatureand current fields. The simplified equations of this model are as follows:Dynamic equations areaudd=-g(n +z外0ot)-(V.Wu+V.u) .3t“ax2poax.+fva+ Am Vug+1(tx-cb|V:|us),(1)Pohnud=-g(哥+ h )-(V.J+V..)-B\ay + 2po ay+ fus+Am V吃。+ (r, -co|Va|ra),(2)pohcontinuity equation isandd=-v.(V:h),(3)Otequation of state isPa=f(T,S)- Pe,(4)thermodynamic equation isaT__ 1at=cophQm +(vg+ V)VT+A中国煤化工(5)whereYHCNMHGQm=Q1-Qb-Qn-Q。(6)in Eqs (1)~(5), V。，v。 are current vectors; the suffix d and c respectively represent thephysicalariables corresponding monthly drift and residual current; η is the surface elevationTest of numerical prediction of sea water temperature in Taiwan Strait475relative to the undisturbed water level; f the Coriolis parameter; g the gravitational accelera-tion; h the depth of UML (upper mixed leyer); Am the coefficient of horizontal kinematic vis-cosity; Cb the stress coefficient of the mixed layer in the bottom; Tx and Ty are the componentsof the surface stress; T and S respectively the monthly temperature and sality on the surfaceand in this paper are considered as constants; Qm denotes the heat flux on the surface; Ak isthe coefficient of heat viscous diffusivity; We and Wp are the speed of entrainment and pump-aTing, respectively; OT is the temperature difference across thermocline; 2 denotes the varia-tional ratio of the surface temperature and is determined by the heat flux at the surface，ther-mal advection，diffusion, entrainment/detrainment and pumping caused by the surface wind.The variations of the heat flux at the surface Qm are represented as follows:solar radiation Q1 isQ1= Qg(1-0.61n)(1-r)(1-βe-*)，(7)effective radiation through sea surface Qb isQb=eσT3[Ta(0.39 - 0.05Ve2)(1 - kn2)+4(T- T.)],(8)evaporation latent heat Qh isQn= p.ChLV。(qw- - qo),(9)sensible heat Qs isQ.= p.ChcpV。(T- T.),(10)where Q1o is the surface astro radiation; n the cloudage; r the reflectance of sea surface; β thetransmission ratio of solar ratiation on surface; Y the attenuation of short wave radiation in seawater; ε the grey scale of sea surface; σ the constant of Stefan - Boltzmann; cp the specificheat at constant pressure; L the potential energy; Ch the exchanged coefficient of the heat andvapour between the sea and air; k the obstruction coefficient of cloud and its empirical formulaisk=0.59+0.005(φ- 20) with φ the latitude; Ts， Va， Pa， ea， 9a are the air tempera-ture, wind speed, density of air, vapour pressure and specific humidity respectively at 10 mabove sea surface; Ta= T。+ 237 is absolute temperature; qw the saturation specific humidityat surface which is the function of T only.In the present paper, entrainment/detrainment and pumping are especially ilustrated be-cause they have important effect on the change of the sea water temperature in this region.(1) The entrainment/detrainment speed of the UML, We, by means of the balance equa-tion of the turbulent energy, is parameterized as (Nille and Kraus, 1977)1W.=h{2m?: +气(1+)|Bo|-(1-1)|Bo|1+Jo(h -到,(11)where ux = V τ1ρ denotes the friction velocity, τ中国煤化I surface buoyancyflux; Jo the buoyancy flux of the transmission radiat:THCNMH Gf buoyancy acrossthe thermocline; m=1.5 and l=0.5 are two experiment parameters related with diffusion.Under the control of the surface wind, the turbulent energy flux becomes surplus with theintense wind and cooling, the ULM will deepen，and at the same time, the cold water will be476Wang Xiuqin et al.entrained into the UML and causes the decrease of SST. It is more reasonable to consider theUML changing with the surface wind than to consider it as rigid bottom boundary. The varia-tion of UML depth due to We is given as follows:ah[0， w.≤0 (detrainment),(12)AtWe, w.>0 (entrainment).(2) The pumping speed Wp is expressed aswp=1cr(系), .(13)where τ is wind stress; f Coriolis parameter. The pumping is induced by strong cyclone/anti-cyclone circulation. Under the force of cyclone, Ekman pumping with the horizontal divergenceof the UML and the upwelling of the deep water will decrease the SST of the UML. After en-trainment or pumping，the difference of temperature across thermocline OT will be changed,and obviously, it will restrain the progress of the entrainment.When t = 0 the initial conditions isv=Va.0，na= 70.0,(14)T= To.(15)The lateral boundary conditions: Sliding condition is adopted. At the lateral boundary,the normal component of current is zero, and the tangent component of current is equal to theinternal current，and there is no heat exchange for temperature:avAn0， v=0,(16)aT=0,(17)where the suffix l and n represent the components in the tangent and normal directions of theshore, respectively.The open boundary conditions: At the open boundary, the drift current adopts the valuethat extends from the internal current along the streamline s. And the temperature is stationaryas flowing in, and no gradient as flowing out:aVd(18)asa[=0, (v,≤0);m=0, (v.≥0).(19)SOURCE OF DATAThe nunerical temperature prediction needs man中国煤化主e data are veryfew especially in the Taiwan Strait. So， for some datd parameteriza-TYHCNMHGtion have been done.The predicted meteorologic fields ( solar radiation, wind velocity, air temperature，dew-point and cloudage) are obtained from the Chinese Center Observatory.驶pth of mixed layer, ocean current field and monthly salinity fieldl are obtained fromTest of numerical prediction of sea water temperature in Taiwan Strait477Ocean Atlas pressed by the State Oceanic Administration.The initial drift current adopts the calculated current that is driven by wind force on undis-turbed surface for 12 h before prediction, and then the predicted current is adopted as the nextinitial current field.The initial sea surface temperature field will seriously influence the precision of tempera-ture prediction, and it is derived from NOAA satellite and the statistic modification for the in-fluence of the cloudage has been done. At the first day，the initial temperature field adopts theobservation data, and then the predicted temperature field acts as the initial field of the next day.TEST RESULTSIn this paper, two tests have been done. One is in summer and the other is in winter.The computational domain in which there are 87 mesh points is shown in Fig.1. In the sum-mer test, a typhoon is just passing by and the calculated current field well responds to the ty-phoon. In the winter test, variation tendency of the prediction sea water temperature agreeswith that in reality basically.117° 118° 119° 120° 121°E270N26°25°24° pe23°22°Fig. 1. The computational field.Test in summerThe period of summer test is from August 27 to 30，1997， which is just during the Ty-phoon Amber passing this region and the necessary data needed in the model have been got.The Typhoon Amber (No. 9714) originated on August 21，1997, and made landfall nearHualian and Xingang of Taiwan, China at 03:00 ~中国煤化工a central pressureof 960 hPa and a maximum wind speed of 35 mdHCNMHGmadelandfallatFuqing of Fujian with a central pressure of 985 hPa and a maximum wind speed of 20 m/s.The path of Amber (see Fig.2) and its wind field (see Fig.3) show that Amber reached the478Wang Xiuqin et al.southeast of this region at 08:00 on August 29，and moved northwestwards. It arrived at themorthwest part of this region at 14:00 on August 30, and then went to northwards.9714 AmberAugust 21~30120125Dongshi Is027Fig. 2. The path of Typhoon Amber.117°118°119°120°121°E1189119° 12026°- a26bN」25° t25°24°23° .ol\t23022022Fig. 3. The wind fields. a. At 08:00 on August 29，1997 and b. at 14:00 on August 30，1997.The parameters of Amber are listed in Table 1 and the computational results are shown inFig. 4. Figure 4a is the initial temperature field and ocean current field. Figure 4b is the pre-dicted temperature field and current field after 48 h and it can be found that the calculated cur-中国煤化工rent well corresponds with the typhoon.To find out the important influencing factors (.MYHc N M H Gater temperature,two points have been selected. One is in the centre of the region (25°N，120°E)， and theother is near the offshore (25°N, 119°E). The results are listed in Table 2.Table 2 shows that pumping and entrainment are the most important factors not only inTest of numerical prediction of sea water temperature in Taiwan Strait479Table 1.The parameters of Typhoon AmberCentral pressureMaximum windDate .Northern latitudeEastem longitude(hPa)speed(m/s)At 08:00 on Aug.2820.7*123.7*95045At 08:00 on Aug.2924.2"120.9*970At 08:00 on Aug.3028.1"116.7*1 0002.117° 118° 119° 120°121°E117° 1189 119° 120° 1219E人26.026.5b叮26° -N.25°25° 127.028.07, 28.0ror24° 123°。27.5"23° t27.0r*G22025.0 25526.0↑/27.Fig. 4. The predicted sea water temperature and current field (in summer). a. The initial temperaturefield and ocean current field and b. predicted temperature field and current field after 48 h.Table 2.The influencing factors contributing to the sea water temperature (C) (predicted 3 d)PositionSolarEffective radiationEvaporationSensible heatEntrainmentPumnping Totalradiationthrough sea surfacelatent heatCentre0.80- 0.09-0.590.16- 3.94-0.41 -4.07Offshore1.40-0.18-0.980.18- 4.29-1.11-4.98the centre of the sea but also near the offshore and especially for the offshore area, and the con-tribution due to entrainment surpasses pumping. This is because the topography of the strait isnarrow and there are some islands among there. This will induce the water mixed enough un-der the wind force. The influencing factors have more contribution to the offshore area than tothe central area. The effect of solar radiation and evaporation latent heat on the temperature inthis area cannot be ignored.Test in winter中国煤化工.[H.CNMHGIn the srmmer test，the predicted temperature field was not examnea because of lack ofthe observed temperature field on the 3rd day. In order to check the precision of the predictionmodel, another test in winter has been done. In this test，the predicted period extends to 10 d480Wang Xiuqin et al .and the predicted meteorologic fields increase for 10 d. The observed temperature field on the10th day was derived from NOAA satellite.During the winter test, there were two strong cold airs occurred, one was from January27~31，1999, the other is from February 1~6，1999. The two cold airs all moved south-eastwards, and the intensity of the second cold air was stronger than the first one.During 10 d, the tendencies of temperature variation are not uniform in this region. Fromthe observed temperature field (Table 3)，it can be found that in the north and offshore area,the temperature increased, yet in the centre and south area it decreased, and only in a smallpart in southeast area the increasement was very little.On the 10th day, the predicted result showed that the tendency of the temperature varia-tion was in agreement with the observation well (Table 4). That conclusion can also be clearlyfound from the figures of the predicted and the observed temperature field on the 10th day (seeFig.5). .Table 3. The difference between the observed temperature field on the 10th day and theinitial temperature field ( the tendency of temperature variation during 10 d)unit:CNorthermEastern longitudelatitude116.5 117.0° 117.5° 118.0° 118.5° 119.0° 119.5" 120.0° 120.5" 121.0° 121.5*27.0°0.02.4 .1.926.5"1.21.11.326.0°1.71.50.0.125.5*0.33.00.825.0* .1.9 -0.1 -1.6 -1.1.02.91.024.5*0.20.5 -3.2 -2.02.13.524.0*0.9-1.00.4 -1.2 -3.5 - 1.22.423.5*0.5-0.2-0.90.6- 2.1. -2.7- 0.623.0*0.4 -0.6 - 1.6-0.1-0.422.5*-0.50.8 -0.6 -0.8 -0.6. - 1.30.0 -0.6 -1.322.0*0.6-0.4 -0.7-1.0 - - 1.60.1 - 0.121.5*0.70.4Table4. The dfference between the predicted tenperature field on the 10th day and the initial tenperature fieldunit:C .Northem116.5* 117.0° 117.5* 118.0° 118.5° 119.0° 119.5° 120.0° 120.5° 121.0 121.5 .27.02.52.326.5*2.2I .626.0*1.6.32.725.0°- 2.5- 1.51.8中国煤化工3.623.5°0.0 -0.3 - 0.5YHCNMHG23.0"0.0 -0.5 -2.4.0.6-0.8- 0.2 - 0.7- 2.122.5"0.0 -0.4 -0.9-0.5 -2.3 -3.40.0 -0.2-0.7 -1.1 - 1.0 -0.5 -1.2 - 1.3 - 0.1- 0.5-0.60.0 ..Test of numerical prediction of sea water temperature in Taiwan Strait4811170118° 119°120°1210E1170 118° 1 19°120° 121°E27°27Na26°26b25024°23°230-18 一I8、22°一9.21°210Fig. 5. The predicted (a) and observed (b) ternperature (C) fields on the 10th dey (in winter).According to Tables 3 and 4，the absolute mean error between the predicted and observedtemperature on the 10th day is0.6 C in the whole region, and the points with errors less than土1.0C are 80% of all, those with errors less than +0.6 C are 63%. This results show thatit is feasible to predict temperature in this area during this season by using the OUQ model.CONCLUSIONSThe test results show that using the OUQ model to predict temperature field in the Tai-wan Strait is practicable and effective. The model can well simulate to the weather system ei-ther typhoon or normal weather, and can be adopted for predicting the upper sea water tem-perature field in3~ 10 d.REFERENCESMcCrearyJ.P., Z. Yu (1992) Equatorial dynamics ina 1. 5-layer model. Prog. Oceanogr., 29, 61~ 32.Niller P.P.，E.B.Kraus (1997) One-dimensional models of the upper ocean. In: Modelling and Prediction of the UpperLayers of the Oeean, Clarendon Press, Oxford, pp.143 ~ 172.Ravindran P., D.G. Wright, T. Platt, S. Sathyendranath (1999) A generalized depth integrated model of the oceanic mixedlayer. J. Phys. Oceanogr., 29, 791~806.Wang B.. T.M.Li, P.Chang (1995) An intermediate model of the tropical PacificOcean. J . Phys. OCeanogr., 25, 1599~1 616.Wang Sizhen, Li Xuhua, Qi Jianhua, Su Yusong (1998) Analyses of factnrs firmino nffshre SST anomaly (OSSTA) inChina seas. Acta Oceanologica Sinica, 17(4), 447~457.中国煤化工Wang Sizhen, Li Xuhua，Qi Jianhua, Su Yusong (2000) A numericalTYHCN M H Gr anomaly. Acta O-ceanologica Sinica, 19(1), 25 ~34. .