Ocen Dynmics (214) 64:557 571 DOI 1.17/s1236-14-73-z On the dynmics of low ltitude, wide nd shllow costl system: numericl simultions of the Upper Gulf of Thilnd Suriyn Srmul & Tl Ezer Received: 9 August 213 /Accepted: 29 Jnury 214 /Pulished online: 25 Ferury 214 # Springer-Verlg Berlin Heidelerg 214 Astrct A high-resolution ( 1 km horizontl grid nd 21 verticl lyers) numericl model sed on the Princeton Ocen Model (POM) hs een used to study the 3D dynmics of the Upper Gulf of Thilnd (UGOT). While influenced y tides nd rivers like other esturine systems, the UGOT is unique ecuse it is wide ( 1 km 1 km), it is shllow (verge depth of only 15 m), it is locted in low ltitudes ( 12.5N 13.5N), nd it is influenced y the sesonl monsoon. Sensitivity studies were thus conducted to evlute the impct tht surfce het fluxes, monsoonl winds, river runoffs, nd the low ltitude my hve on the dynmics; the ltter hs een evluted y modifying the Coriolis prmeter nd compring simultions representing low nd mid ltitudes. The circultion in the UGOT chnges sesonlly from counter-clockwise during the northest monsoon (dry seson) to clockwise during the southwest monsoon (wet seson). River dischrges generte costl jets, wheres river plumes tend to e more symmetric ner the river mouth nd remin closer to the cost in low ltitudes, compred with mid-ltitude simultions. River plumes re lso dispersed long the cost in different directions during different stges of the monsoonl winds. The model results re compred fvorly with simple winddriven nlyticl esturine model. Comprisons etween n El Niño yer (1998) nd L Niñ yer (2) suggest tht wter Responsile Editor: Jrle Berntsen This rticle is prt of the Topicl Collection on the 5th Interntionl Workshop on Modelling the Ocen (IWMO) in Bergen, Norwy 17-2 June 213 S. Srmul Deprtment of Mrine Science, Chullongkorn University, Phythi Rod, Pthumwn, Bngkok 133, Thilnd T. Ezer (*) Center for Costl Physicl Ocenogrphy, Old Dominion University, 4111 Monrch Wy, Norfolk, VA 2358, USA e-mil: tezer@odu.edu tempertures, wrmer y s much s 2 C in 1998 reltive to 2, re lrgely driven y decrese cloudiness during the El Niño yer. The developed model of the UGOT could e used in the future to ddress vrious environmentl prolems ffecting the region. Keywords Numericl model. Monsoon. Tides. Rivers. Upper Gulf of Thilnd 1 Introduction The dynmics of semi-enclosed sins nd costl regions re often fll into one of two ctegories, either dominntly winddriven system, or n esturine-like system driven y tides nd river dischrges. The motivtion for this study of the Upper Gulf of Thilnd (UGOT) comes from the fct tht this re is unique in its topogrphy nd loction nd thus its dynmics my involve chrcteristics of oth shllow wind-driven system ssocited with the sesonl monsoonl winds nd n esturine-like system ssocited with tides nd uoyncydriven river flows. This region is especilly vulnerle to impcts of climte chnge, since sesonl monsoonl floods my ecome more severe due to fst se level rise (SLR) in this region. In some loctions in the northern UGOT, the reltive SLR cn e s much s ten times fster thn the glol men SLR (Srmul 213). Lnd susidence on Thilnd s costs is especilly lrge due to erthqukes nd groundwter extrctions (Nicholls 211), which results in lrge sptil vritions in reltive SLR tht significntly exceed those t other plces with lnd susidence (e.g., the U.S. Est cost; Ezer 213). Other environmentl prolems, such s pollutnts crried y rivers nd costl erosion, will require etter knowledge of the circultion nd forcing mechnisms in the UGOT. In region with limited ocenic oservtions, these prolems cn e ddressed with the help of 3D numericl
558 Ocen Dynmics (214) 64:557 571 costl ocen model s done here. This UGOT re hs not een modeled or studied s much s mny other semienclosed ses t higher ltitude such s the North Se (e.g., Alretsen nd Røed 21). In costl systems tht re long nd nrrow, stedy windinduced circultion cn often e descried with reltively simple rotropic models (Winnt 24; Sny nd Vllelevinson 25), wheres only ottom topogrphy nd wind re considered (the Coriolis prmeter, f, cn e neglected if the sin or estury is nrrow enough reltive to the Rossy Rdius of deformtion). In such cses, the dynmics is generlly descried y longshore wind-induced trnsport in the direction of the wind in shllow res nd trnsports ginst the wind direction in deeper regions. For exmple, Csndy (1973) used n nlyticl model of Lke Ontrio ( long nd nrrow lke) to show flow pttern similr to the type descried ove. However, in rotting system, or when strtifiction is not neglected, the trnsport pttern cn e more complex. A simple model, like those mentioned ove, will e compred here with more relistic 3D roclinic numericl model to evlute the extent of which the sic wind-driven shllow dynmics cn descrie the circultion in the UGOT. On the other hnd, in esturine-like costl systems, river flow nd tidl mixing ply the dominnt roles, ut wind is sometimes neglected. River plumes of freshwter crete uoyncy flows nd cn contriute to the dynmic of the costl circultion (Horner-Devine 29). Due to the Coriolis effect, river plumes re deflected to the right/left in the Northern/Southern hemisphere nd the costl current will generlly crry the plume wy from the river mouth (Horner-Devine 29). This dynmics ssumes tht there is reltively lrge river dischrge nd tht the Coriolis term is lrge enough (i.e., nonequtoril regions). Therefore, nother purpose of the study is to evlute the dynmics of river plumes in low ltitudes, nd hence, etter understnd how properties such s pollutnts, slinity, nd temperture in the costl region re ffected y the dvection nd mixing of river plumes in the UGOT. The UGOT is shllow (verge depth of 15 m), semienclosed sin, with lmost squre ( 1 1) shpe; it is situted in low ltitudes ( 12.5N 13.5N), t the hed of the Gulf of Thilnd, which is lso reltively shllow compred to the South Chin Se nery (Fig. 1). It is locted in region tht is strongly ffected y monsoonl winds nd where four mjor rivers dischrge (Fig. 2). A stellite imge tken during flood in 211 shows very xisymmetric river plume (Fig. 2), with no pprent deflection to the right of the river flow (i.e., westwrd), s expected in the northern hemisphere; numericl model simultions will e used to test if this oservtion is the result of the low ltitude of the UGOT. The sesonl monsoon pttern includes wet seson when winds previl from the southwest direction, dry seson when winds previl from the northest direction, nd trnsitionl periods in etween (Fig. 3e). Previous numericl model studies suggest tht the southwest/northest monsoonl winds drive clockwise/counter-clockwise circultion in the UGOT (Burnprtheprt et l. 28, 29). However, the previous studies used only monthly men winds, while here more relistic semi-dily wind forcing, s well s dily het fluxes will e used (Fig. 3). Multi-yer simultions will llow us to look t internnul vritions, nd test the influence of lrge-scle climte vritions ssocited with El Niño nd L Niñ. In prticulr, the El Nino-Southern Oscilltion (ENSO) hs lrge influence on precipittion nd cloudiness over the region (Chndr et l. 1998), which could ffect ir-se het exchnge nd wter tempertures. Studies show the influence of ENSO on the circultion in the South Chin Se (Cho et l. 1996), so it is likely tht ENSO influences the UGOT s well. Unlike midnd high- ltitudes, surfce het fluxes in the UGOT (Fig. 3, ) do not show significnt sesonl cycles tht cn explin the sesonl vritions in surfce temperture (Fig. 3c). However, lrge sesonl vritions re seen in cloud coverge (Fig. 3d) nd winds (Fig. 3e), s result of the sesonl monsoon cycle. The southwest monsoonl winds (out dys 9 3) ring wrm nd moist ir from the southwest nd the sky is lrgely covered with clouds. Note tht in 1997 1998 there ws n El Niño event, with wrmer tempertures nd longer period of cler sky, while in 1999 2 there ws L Niñ event, with colder tempertures nd more cloud cover. Therefore, the influence of ENSO will e ddressed in this study. The study follows previous 3D esturine circultion modeling of the impct of winds, tides, nd freshwter dischrges in plces such s the Chespeke By (Guo nd Vlle-Levinson 27, 28), the Chrlotte Hror estury (Zheng nd Weiserg 24) nd even in the UGOT (Burnprtheprt et l. 28, 29), ut will focus on studying the prticulr chrcteristics of the UGOT, including its topogrphy, forcing mechnisms, nd loction in low ltitudes. The pper is orgnized s follows: in Section 2, the numericl model nd the experiments re descried; in Section 3, the sesonl wind- nd river-driven circultions re descried; in Section 4, the numericl model is compred with simple nlyticl model; in Section 5, internnul vritions nd ENSO impct re descried; nd finlly, Section 6 offers discussions nd conclusions. 2 Model description, forcing dt, nd experimentl design The model used in this study is sed on the Princeton Ocen Model (POM; Blumerg nd Mellor 1987; Mellor 24). The POM is 3D, free surfce, primitive eqution ocen model tht includes complete thermodynmics nd the level 2.5 Mellor Ymd turulence closure scheme (Mellor nd Ymd 1982) to provide verticl mixing. Bottom friction is
Ocen Dynmics (214) 64:557 571 559 Fig. 1 Left The topogrphy of the region, including the South Chin Se nd the Gulf of Thilnd; the study re is indicted y the red ox. Right Focus on the topogrphy of the study re in the Upper Gulf of Thilnd. Color represents ottom depth in m. The southern oundry of the numericl model is the B-B solid line where stronomicl tides were imposed, using the se level dt from the two tidl sttions ner B nd B. Dshed line A-A is the cross-section shown in Fig. 6. The loctions of four mjor rivers t the hed of the model domin re indicted, s well s the tide guge sttion (lck squre) nd the uoy sttion (lck circle) used to evlute the model in Fig. 4 sed on lw of the wll formultion for oundry lyers (see Mellor 24) with mximum drg coefficient of.25. The model domin nd ottom topogrphy of the UGOT re shown in Fig. 1. The verge depth of the UGOT is only 15 m, which includes smooth ottom in most of the domin, except some deeper chnnels nd more complex topogrphy round islnds long the estern oundry. The horizontl grid is curviliner, ut is lmost rectngulr with grid cells Δx Δy 1 km. In the verticl direction, z, terrinfollowing, sigm-coordinte grid, σ, is used with n=21 lyers, i.e., σi =(zi η)/(h+η); wheres <σi < nd H<zi <η, for i=1,,n, where H(x,y) nd η(x,y,t) re the ottom depth nd surfce elevtion, respectively. The minimum depth in the model is set to 3 m, which is sufficient for simultions of tides in the UGOT without inundtion. Simultions with the wetting nd drying option in POM (e.g., Srmul nd Ezer 21), which could simulte much shllower regions of few centimeters deep, will e considered in future studies. At the southern open oundry (B-B line in Fig. 1), rdition oundry condition is pplied to llow wves to propgte cross the oundry without reflection (Mellor 24). The open oundry conditions for temperture nd slinity llow properties set on the oundry from oservtions (see elow) to e dvected into the model domin when flow is northwrd. Tidl forcing includes 8 tidl constituents, Q1, O1, P1, K1, N2, M2, S2, nd K2, otined from hrmonic nlysis of oserved surfce elevtions t 2 tide guge sttions locted ner the two edges of the southern oundry (Fig. 1). Liner interpolted etween these 2 sttions is used nd pplied to the oundry s comintion of tidl elevtion nd rotropic flow oundry conditions (Mellor 24). Initil conditions of slinity nd temperture re sed on oservtions (Burnprtheprt et l. 28). Slinity is initilly ssigned s constnt (32.15; dimensionless units of Prcticl Slinity Scle, PSS) for the entire model domin, while wter temperture is nerly constnt nd it is vried from 3.63 C t the surfce to 3.2 C t the ottom, sed on typicl oserved profiles. After spin-up (which is very short, order of weeks for such shllow sin) the model rn for the period 1997 2. All the sensitivity experiments discussed in Section 3 re sed on simultions with different forcing fields nd different prmeters representing the yer 2, nd only in Section 5 internnul vritions re discussed. Surfce forcing fields re shown in Fig. 3. These dt re dily nd sptilly uniform. Relile dt with high resolution sptil distriutions re not ville for this region. However, since this study ims to understnd processes with sensitivity experiments using different forcing nd is not imed t developing relistic ocen forecst system, the ssumption of
56 Ocen Dynmics (214) 64:557 571 Stellite imge, Novemer 29, 211 Monthly men river flow Fig. 2 A flse color imge from the THEOS stellite tken in Novemer 29, 211, during mjor flood. The edge of turidity plume from the Cho Phry River is indicted y the dshed lck line. White nd red res re clouds nd lnd vegettions, respectively. This imge is modified (with permission) from the originl imge otined from www. gistd.or.th/gistd_n/gllery/img/flood211. Monthly men river dischrges of the four mjor rivers shown in nd used y the model uniform forcing fields seems justified. Note, however, tht despite the sptil uniform surfce forcing, the model developed quite relistic sptil ptterns of surfce tempertures due to the ocen dynmics. Surfce wind stress for the sensitivity experiments during 2 is sed on the 1-m wind vectors retrieved semi-dily from Jet Propulsion Lortory (poet.jpl.ns.gov); for the internnul simultions (1997 2) dily fields re otined from the Europen Center for Medium rnge Wether Forecsting (ECMWF) renlysis dt (dt-portl.ecmwf.int/dt/d/er4_dily). Surfce het fluxes in the model re implemented sed on the pproch descried in Ezer (2) with some modifiction needed for this region. The het fluxes, Q, in the model is clculted y Q ¼ Q o þ Q ðt o T m ÞþQ T sw F s ð1þ where Q o is the oserved net flux (Q net ) without shortwve rdition (Q sw ). The second term is feedck term, where T o nd T m re surfce temperture otined from oserved nd model dt, respectively, nd the coupling coefficient, Q/ T, issetto5wm 2 K (Ezer 2). The third term is shortwve rdition, which llows solr rdition to penetrte into the upper few meters of the ocen (Mellor 24). In this region, with lrge sesonl vritions in cloud cover (Fig. 3d) due to the dry/wet monsoon periods, it ecme pprent tht cloud cover must e tken into ccount s it strongly ffect the mount of short wve rdition reching the surfce of the wter. F s is the cler sky fctor clculted s.6+.4 (% Cler sky/1), sed on tril nd error tests. The cler sky fctor is estimted from stellite-derived se surfce temperture retrieved from AVHRR Pthfinder version 5 (http://poet. jpl.ns.gov). Q net, Q sw,ndt o re otined from renlyzed dt from the work of Yu et l. (28), nd retrieved from http://rd.ucr.edu. The implementtion of rivers in the model is sed on the work of Oey (1996), wheres river dischrge Q riv is considered s freshwter verticl flux upstrem the river mouth. The monthly wter dischrges in four rivers re shown in Fig. 2. Mximum river flow is in Septemer
Ocen Dynmics (214) 64:557 571 561 Fig. 3 Surfce forcing dt used in the model (Eq. 1): net het fluxes, Q net ; shortwve rdition, Q sw ; c surfce temperture, T o ; d % cler sky; nd e wind vectors Q net (W m 2 ) 4 2 2 1997 1998 1999 2 4 4 Q sw (W m 2 ) 3 2 1 c d Temperture ( C) 32 3 28 26 24 1 e % Cler Sky Wind vector (m s ) 5 1 5 5 5 m s 2 4 6 8 1 12 14 Julin dy from 1997 Octoer, during the trnsition from the southwest to northest monsoons. Note tht internnul vritions in river flows re neglected here. Tle 1 summrizes the six experiments nlyzed here to investigte the seprte impct of winds (experiments Wnd_1 nd Wnd_2), river dischrges (Riv_1 nd Riv_2), nd the comined impct of winds nd rivers (WRiv_1 nd WRiv_2). The impct of the Coriolis prmeter, f, is evluted y compring experiments denoted exp_1 (rel UGOT loction in low ltitude) nd exp_2 (mid-ltitudes). The Coriolis prmeter in the model vries in spce (function of ltitude), ut its sptil vrition in ech experiment is not significnt in such smll domin. Surfce slinity, surfce temperture, nd current velocity otined from the model results re sved hourly nd the monthly verged vlues of these prmeters re discussed in the following sections. Tle 1 Model experiments Experiment Forcing Model domin center Wnd_1 Surfce het fluxes, tides, wind 13N Wnd_2 Surfce het fluxes, tides, wind 45N Riv_1 Surfce het fluxes, tides, rivers 13N Riv_2 Surfce het fluxes, tides, rivers 45N WRiv_1 Surfce het fluxes, tides, wind, rivers 13N WRiv_2 Surfce het fluxes, tides, wind, rivers 45N
562 Ocen Dynmics (214) 64:557 571 Becuse of the importnce of tidl mixing, ll the experiments in Tle 1 include tides. The performnce of the tidl model is evluted ginst oservtions nd exmples of model-dt comprisons for surfce elevtion nd currents re shown in Fig. 4. Tidl velocity in the UGOT is minly in the north south direction, so only v-velocity is shown. Both surfce elevtion nd v-velocity otined from the tidl model re compred well with oserved dt; the Root Men Squre Error is 16.9 cm for elevtion nd 3.75 cm s for velocity, nd the dt-model correltion coefficients re lrger thn.9 for oth fields. More detiled nlysis of model nd oserved tides, s well s se level dt, cn e found in Srmul (213). 3 Model results 3.1 Sesonl wind-driven circultion nd dynmics First, simultions of the sesonl wind-driven su-tidl circultion sed on experiment Wnd_1 (i.e., wind forcing, ut no rivers) re descried. Exmples of monthly men surfce elevtions nomly nd depth-verged velocities re shown in Fig. 5, representing four sesons, first inter-monsoon (Mrch), southwest monsoon (My), second inter-monsoon (Octoer), nd northest monsoon (Novemer). The most pprent pttern is tht during the southwest monsoon (My Septemer), the previled winds cuse the wter in the UGOT to pileup on the northest corner of the domin nd the circultion is mostly clockwise (Fig. 5), while during the northest monsoon (Novemer Ferury), counterclockwise circultion is developed, with lower wter level in the northest (Fig. 5d). During the first nd second intermonsoons the surfce elevtion hs north south (Fig. 5) nd est west (Fig. 5c) ptterns, respectively. While the flow long the northern nd western costl oundries re reltively smooth, the lrger vritions in ottom topogrphy nd the existence of islnds long the estern oundry (Fig. 1) result in more complex circultion ptterns nd the developments of smll ( 1 2 km in size) gyres (e.g., t 1.8E, 13.3N, there is gyre with n nti-clockwise circultion in Mrch nd clockwise circultion in Novemer). Note tht the circultion is fr from geostrophic lnce since the flow is mostly cross lines of constnt elevtion nd not prllel to them (e.g., in Fig. 5c, the flow long the northern cost is estwrd, while the surfce elevtion contours re in the north south direction). The flow seems to resemle nonrotting wind-driven circultion in shllow sins (Csndy 1973; Sny nd Vlle-levinson 25), wheres the flow tends to e in the direction of the wind long the shllow res ner the cost. To evlute the 3D verticl distriution of the wind-driven flow, the velocity distriution long n est west cross section (line A-A long 13.35N in Fig. 1) isshowninfig.6 for the sme months s those shown for the horizontl circultion in Fig. 5. Lrge sesonl vritions in the verticl structure re noticele, with periods of more rotropic flow nd periods of more roclinic flow (i.e., depth-dependent). During the Surfce elevtion (cm) v velocity (cm s ) 2 1 1 2 5 25 25 Th Chin Si Chng Buoy R 2 =.94; Oserved RMSE=16.9 (cm) Modeled Oserved Modeled 5 6 65 7 75 8 85 9 Fig. 4 Exmples of model-dt comprisons during Mrch 2 for surfce elevtion t Th Chin tide guge sttion nd v-velocity t Si Chng uoy sttion (see Fig. 1 for loctions). Oserved points re mrked Julin dy R 2 =.9; RMSE=3.75 (cm s ) with strs, nd model output with solid lines. Correltion coefficients etween the dt nd the model, R 2, nd root men squred errors (RMSE)redepictedonthetop right corner of ech pnel
Ocen Dynmics (214) 64:557 571 563 Fig. 5 Monthly men nd depthverged current velocity (lck rrows, incms )ndmonthly men surfce elevtion nomly (color, in cm) otined from experiment Wnd_1 for four different monsoon sesons during 2: first inter-monsoon (Mrch), southwest monsoon (My), c second inter-monsoon (Octoer), nd d northest monsoon (Novemer). Surfce elevtion nomly is reltive to the re verged vlue of ech month. Blue rrows t the top left of ech pnel show the men monthly wind; for scling reference, the wind speed in My is 5 ms Ltitude 12.8 N 13. N 13.2 N 13.4 N Ltitude 12.8 N 13. N 13.2 N 13.4 N Mrch 2 Octoer 2 c My 2 Novemer 2 d 1. E 1.2 E 1.4 E 1.6 E 1.8 E 1 cm s 1. E 1.2 E 1.4 E 1.6 E 1.8 E 5 5 southerly wind (Mrch), the u-velocity (long the section) is lmost uniformly estwrd throughout the entire wter column (Fig. 6), while the v-velocity (cross the section) is 2- lyer flow with northwrd flow in the upper 5 1 m nd southwrd flow in the deep lyer. The estwrd u-velocity flow intensified ner the surfce in My nd Octoer (Fig. 6c, e), reflecting the costl jet shown in Fig. 5, ut the flow is completely reversed in Novemer (Fig. 6g) during the northestern monsoon. Note tht in the deeper chnnel (round 1.7E), there is convergence in Mrch, My, nd Octoer, ut divergence in Novemer (Fig. 6g). During Novemer, the cross-section flow (v-component) shows southwrd upper lyer flow, in the direction of the wind, nd northwrd susurfce jet of return flow in the center of the chnnel (Fig. 6h). More detils of the Novemer circultion will e discussed lter with the help of n nlyticl model. The verticl structure of flows my depend on three prmeters (Ksi et l. 2; Vlle-Levinson et l. 23): the verticl mixing coefficient, K M, (derived in the model from the Mellor-Ymd turulence scheme), the Ekmn numer E k = K M /fh 2 (which represents the importnce of viscosity versus geostrophy; H o is depth scle, tken here s the mximum pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ottom depth), nd the Ekmn lyer depth D E = 2K M =f ¼ pffiffiffiffiffiffiffi 2E kh (which represents the influence of the wind). Estimtion of the verge vlue of these prmeters cross the section in Fig. 6 is shown in Tle 2. The Ekmn numer thus is proportionl to the rtio etween the Ekmn lyer depth nd ottom depth, showing wht portion of the wter column is ffected y wind-driven Ekmn trnsport. For lrge Ekmn numer (i.e., smll f or lrge K M ), E k =1 nd f cn e neglected (Ksi et l. 2), which leds to D E =1.4H. Therefore, in such cse, the Ekmn lyer cn occupy the entire wter column nd the pttern of flow cross will e in form of n inflow in one direction in the deeper prt nd return flow in opposite direction on the shols (Wong 1994). If on the other hnd, the Ekmn numer is modertely smll, sy E k =.1, then D E =.45H. This mens tht the upper prt of the wter column is controlled y f, while the lower hlf my e ffected y seprte Ekmn ottom lyer. In the cse of very smll Ekmn numer, sy E k =.1 or D E =.14H, the verticl eddy viscosity is wek, the flow is nerly geostrophic nd the Ekmn lyer is pressed ner the ottom (i.e., two-lyer flow is expected). In the simultions shown in Fig. 6, the Ekmn numer rnge (Tle 2) is etween.3 (Mrch) nd 1.2 (Octoer), thus the flow is clerly not geostrophic. Octoer is the only month with E k >1, nd thus, the long-chnnel flow
564 Ocen Dynmics (214) 64:557 571 Wter depth (m) Wter depth (m) Wter depth (m) Wter depth (m) 5 5 8 4 4 8 u velocity (Mrch) 5 5 u velocity (My) 5 5 u velocity (Octoer) 5 5 c e g u velocity (Novemer) 2 1. E 1.2 E 1.4 E 1.6 E 1.8 E v velocity (Mrch) d v velocity (My) f v velocity (Octoer) h v velocity (Novemer) 1. E 1.2 E 1.4 E 1.6 E 1.8 E 11. E Fig. 6 Cross section of monthly men (sme months s Fig. 5) velocity distriution (in cm s ) of long-section, u-velocity (left pnels) nd crosssection, v-velocity (right pnels) long ltitude 13.35N (line A-A in Fig. 1). White solid lines seprte etween positive nd negtive velocities is quite rotropic (Fig. 6f) with southwrd flow on the estern side of the chnnel throughout the wter column nd northwrd westwrd of the deepest point. During the other months, two-lyer system is developed (with Ekmn lyer occupying only portion of the wter column) s expected in cses of modertely smll Ekmn numer. Note tht the verticl structure depends on the Coriolis prmeter, f, which is smll, ut not totlly negligile, in the low ltitudes of the UGOT; in the next section, the impct of Coriolis will e further evluted. Tle 2 Verticl eddy viscosity (K M ), Ekmn numer (E k ), nd Ekmn depth (D E ) sptilly verged long ltitude 13.35N (line A-A in Fig. 1) nd monthly verged for Mrch, My, Octoer, nd Novemer 2 (experiment Wnd_1) Month K M (m 2 s ) E k D E (m) Mrch.33.3 14.2 My.77.6 21.8 Octoer.164 1.2 31.7 Novemer.65.5 2. Figure 7 summrizes the sesonl north/south trnsport cross the section shown in Fig. 6. During the period of southerly or southwesterly wind (Mrch nd My), northwrd trnsport occupies the shllow regions in the west while southwrd trnsport exists in the deeper chnnel in the est. During the northesterly wind, the flow pttern reverses, so the southwrd nd northwrd trnsports re locted t the shllow nd deeper regions, respectively. Novemer seems to hve especilly lrge trnsport, centered t the deeper chnnel, ecuse the wind is strongest. Even though ottom friction nd Coriolis effects re included in the numericl experiment Wnd_1, the results seem to gree to some degree with the nlyticl results for the non-rotting frictionless costl system shown in Csndy (1973). 3.2 River plume dynmics nd the impct of Coriolis To evlute the impct of winds nd Coriolis on the dynmics of river plumes, experiments with rivers, ut without winds (Riv_1 nd Riv_2) re compred with experiments tht
Ocen Dynmics (214) 64:557 571 565 Fig. 7 Top North south trnsports (m 3 s ) cross ltitude 13.35N (cross-section A-A ) for sme months s Figs. 5 nd 6: Mrch (red), My (green), Octoer (lue), nd Novemer (mgent). Bottom Topogrphy nd monthly men wind velocity (in m s ) for ech shown month include wind forcing (WRiv_1 nd WRiv_2). To evlute the impct of the Coriolis prmeter, low-ltitude experiments ( exp_1 ) re compred with mid-ltitude experiments ( exp_2 ). The simulted monthly men surfce slinity nd surfce temperture for those experiments re shown in Figs. 8 nd 9, respectively. Slinity distriution is shown for Septemer nd Novemer to represent months with reltively lrge mount of wter dischrge (Fig. 2), ut with different wind directions, while temperture distriution is shown for June (summer monsoon) nd Novemer to represent reltively wrmer nd colder months with different wind directions. Cho Phry River, the lrgest river in the Gulf of Thilnd, produces the lrgest freshwter plume, though the river plume is quickly mixed y the tidl currents even without winds (Fig. 8, ). The cse with relistic low-ltitude Coriolis prmeter shows n xisymmetric river plume (Fig. 8) tht is very similr in shpe to the oserved river plume in Fig. 2. Symmetric plumes re lso found t the other three rivers, ut the extent of these plumes is much smller thn tht of the Cho Phry River. The mount of freshwter nd intensity of mixing re the key fctors tht control how fr from the river mouth the plume cn penetrte. After initil mixing ner the river mouth, tongue of low slinity plume is trnsported s uoyncy-driven current long the oundry. The impct of lrger Coriolis prmeter is to push the low slinity plume 3 km frther long the west cost (Fig. 8 in mid ltitude reltive to Fig. 8 in low ltitude). Additionl impct of lrger Coriolis effect is less xisymmetric plume ner the river mouth nd the development of clockwise circulting gyre offshore the river mouth (1.7E, 42.3N; Fig. 8). The clssicl tendency of the river plume to turn to the right in the northern hemisphere is not seen here, proly due to the strong tidl mixing ner the river mouth. The impct of the wind on the river plume is to push the low slinity plume estwrd during the southwest monsoon (Fig. 8c, d) nd westwrd during the northest monsoon (Fig. 8e, f). The comintion of northest wind nd lrger Coriolis prmeter is especilly effective in producing costl uoyncy plume tht propgte frther long the west cost (lmost to the southern edge of the model; Fig. 8f) thnny other experiment; the contriution of the wind y itself (without rivers) to this jet cn e seen in Fig. 5d. The impcts of rivers, winds, nd Coriolis on surfce tempertures (Fig. 9) re quite similr to the impct on slinity. It should e noted first tht despite the uniform surfce het fluxes forcing, lrge sptil vritions re seen in the model simultions due to the ocen dynmics; similr sptil vritions cn sometimes e seen in stellite imges (Srmul 213). These temperture ptterns re the result of the fct tht the solr rdition will crete wrmer tempertures ner the cost, especilly over the shllow regions in the north nd west (Fig. 1) where the entire wter column (of only few meters) is wrmed up; while in deeper regions, the ction of tides cools the surfce wters y mixing them with colder wter elow. The uoynt river plume nd winds trnsport the wrm wters long the cost. During the southwest monsoon, the wrmest wters re ccumulted t the northestern corner,
566 Ocen Dynmics (214) 64:557 571 44.8N 45.N 45.2 N 45.4N 12.8 N 13. N 13.2 N 13.4 N Septemer/Mid lt/without wind Ltitude Septemer/Low lt/without wind Septemer/Mid lt/with wind Ltitude d 44.8N 45.N 45.2 N 45.4 N c 12.8 N 13. N 13.2 N 13.4 N Septemer/Low lt/with wind Novemer/Mid lt/with wind 44.8N 45.N 45.2N 45.4N e Ltitude 12.8 N 13. N 13.2 N 13.4 N Novemer/Low lt/with wind 1. E 1.2 E1.4 E1.6 E1.8 E f 1. E1.2 E1.4 E1.6 E1.8 E 3 cm s Fig. 8 Model simultions of river plumes showing monthly men surfce current velocity (lck vectors in cm s) nd monthly men surfce slinity (in color). Left pnels re simultions with relistic low ltitude loction, nd right pnels re simultions t mid-ltitude loction. From top to ottom re simultions without wind (Riv_1 nd Riv_2) nd with 1 2 3 33 winds (WRiv_1 nd WRiv_2) for Septemer (south-westerly monsoon with men wind speed of 2.7 m s) nd Novemer (north-esterly monsoon with men wind speed of 5.7 m s). The wind direction in c f is shown y the lue rrows. The chosen months re those with significnt river flow, ut different wind direction
Ocen Dynmics (214) 64:557 571 567 44.8N 45.N 45.2N 45.4N 12.8 N 13. N 13.2 N 13.4 N June/Mid lt/without wind Ltitude June/Low lt/without wind June/Mid lt/with wind 44.8N 45.N 45.2N 45.4N c Ltitude 12.8 N 13. N 13.2 N 13.4 N June/Low lt/with wind d Novemer/Mid lt/with wind 44.8N 45.N 45.2N 45.4N e Ltitude 12.8 N 13. N 13.2 N 13.4 N Novemer/Low lt/with wind 1.E1.2E1.4E1.6E1.8E f 1.E1.2E1.4E1.6E1.8E 3 cm s 25 26 27 28 29 3 Fig. 9 Sme s Fig. 8, ut for surfce tempertures (color in C) during June nd Novemer 2. These months were chosen to represent reltively wrmer nd colder thn verge months with different wind direction. Wind speeds re 5 nd 5.7 m s for June nd Novemer, respectively with very little impct from the Coriolis prmeter (Fig. 9c, d). During the northest monsoon, the impct of Coriolis is somewht lrger (Fig. 9e, f) due to stronger costl jet long the western oundry. This my e the result of more
568 Ocen Dynmics (214) 64:557 571 pile up of wter on the western cost y the Ekmn trnsport nd incresed geostrophic long-cost flow in mid-ltitudes compred with low ltitudes. 4 Comprison etween the 3D numericl model nd n nlyticl solution Winnt (24, 21) descried n nlyticl solution for the wind-driven viscous flow in closed shllow sin; such model hs een pplied for exmple to the Nnsemond River flow, nd the results were consistent with oservtions (Nrv'ez nd Vlle-Levinson 28). The simple model ssumes tht the length, L* of the sin (in the y direction) is lrger thn the width, W* (in the x direction), the wter depth, H*, is smll nd the wind is lowing long the min xis of the sin (y-direction). Becuse of the nrrow width the Coriolis prmeter, ƒ, cn e neglected. In comprison, the UGOT is shllow nd wide, so L* W*. However, ecuse of its loction in low ltitudes (i.e., smll ƒ),thesimplenonrotting solution my still e prtly vlid for the UGOT. Here it is tested whether or not the simple nlyticl solution of Winnt (21) cn explin the flow derived from the more complex 3D numericl model. The momentum eqution in the y direction is ssumed to e, 2 v z 2 η y ¼ ; ð2þ where v nd η re non-dimensionl velocity nd surfce elevtion, respectively. The reltion of the dimensionless vriles to dimensionl vriles (mrked with strs) re v ¼ v ρk M τh ; η ¼ ηρgh τl ; x ¼ x z L ; nd z ¼ H ;,whereρ is density of se wter nd τ is surfce wind stress. After pplying oundry conditions t the ottom, z= h, nd surfce z=η, the solution is v ¼ η z 2 h 2 y 2 ηðz þ hþ þ z þ h ð3þ where η y ¼ 3 2 Z 1 Z 1 h 2 dx ¼ 3 h 3 2 dx h 2 h 3 : ð4þ So for given topogrphy h(x), the velocity v(x,z) driven y wind stress τ cneesilyclcultedfrom(3) nd(4). The nlyticl solution ws pplied to the cross-section long ltitude 13.35N (line A-A in Fig. 1) nd compred with the 3D numericl model solution during Novemer (Fig. 6h) when the wind ws roughly lowing from the north nd the se level grdient is roughly in the north south direction (Fig. 5d), similr to the dη/dy ssumed in Eq. 2. The comprison revels similr flow pttern etween results otined from the complex 3D numericl model (Fig. 1) nd the simple nlyticl solution (Fig. 1). The flow is northwrd, ginst the wind direction, in the deeper chnnel, nd southwrd, in the direction of the wind, on the shols on oth sides of the chnnel, s expected from the theory discussed efore. Compred with the smoother nlyticl solution, the 3D model results show lrger velocities nd more sptil vritions ssocited with gyres nd circultion ptterns. Given the fct tht the 3D solution is monthly verged flow of time-dependent simultion with vrile wind nd tides, it is somewht surprising how well it produces simple stedy stte nlyticl solution. The conclusion is tht the time-men flow of this shllow low ltitude sin still prtly oeys clssic non-rotting wind-driven dynmics, though some impct of rottion still exists, s indicted y the Ekmn numer (Tle 2). 5 Internnul vritions nd ENSO It is cler from the dt in Fig. 3 tht in ddition to the sesonl vritions ssocited with the monsoonl winds, discussed efore, there re lso internnul vritions, so here results from the multiyer simultions (1997 2) re nlyzed. In those low ltitudes, oth the net surfce het flux nd shortwve rditions hve little sesonl or long-term pttern tht cn explin the vritions in surfce tempertures (Fig. 3c). Therefore, the min cuse of the sesonl nd internnul vritions in the model tempertures is the comintion of the feedck term nd cloud cover in Eq. 1. To evlute how well the model cn simulte internnul vritions, comprison is mde etween the se surfce temperture during 1998, which ws n El Niño yer, nd 2, which ws L Niñ yer (Fig. 11). Se surfce tempertures otined from the model simultions were sptilly verged in spce nd dily verged in time in order to remove tidl nd other high-frequency vritions. The model results re compred with tempertures otined from two sources, the renlysis dt (which ws used in the feedck term in Eq. 1) nd oserved tempertures retrieved from AVHRR stellite dt for oth dy time nd night time. The ltter stellite dt were not used in the model nd they re only sprsely ville during the summer due to incresed cloud cover during the wet seson. The sesonl pttern of tempertures nd the internnul vritions re quite well reproduced y the model. However, the model s tempertures re persistently wrmer thn the renlysis dt; the verge error is.8 C±.29 during 1998 nd.97 C±.29 during 2. Compring the model results with the stellite dt shows tht the men dily temperture in the model is
Ocen Dynmics (214) 64:557 571 569 Fig. 1 The v-velocity distriution cross 13.35N during the northest monsoon (Novemer, s ottom pnel of Fig. 6). From the 3D numericl model simultion nd from the nlyticl 2D model. A white solid line represents zero contour line nd the units re nondimensionl. For scling purposes, the rnge of the velocity color r is equivlent to out 6 to 6 cm s Wter depth (m) 5 5 2 v numericl model.5.5 Wter depth (m) 5 5 v nlyticl solution 2 1 E 1.2 E 1.4 E 1.6 E 1.8 E 11 E 32 Yer 1998 Temperture ( C) 3 28 26 Temp os Temp mod Temp m Temp pm 32 Yer 2 Temperture ( C) 3 28 26 5 1 15 2 25 3 35 Yer dy Fig. 11 Are verged se surfce tempertures (SST) during El Niño yer (1998) nd L Niñ yer (2). SST from model simultions (green solid line) is compred with the oserved temperture from the renlysis dt (red solid line) nd with stellite dt otined during the dy (lue strs) nd night (mgent strs)
57 Ocen Dynmics (214) 64:557 571 within the dily rnge, except t the eginning of 2. In generl, the model is doing etter jo during the summer monsoon thn during the winter seson. Note the lrge dily vritions etween dy nd night tempertures, up to 2 3C, which cnnot e simulted without much higher temporl forcing dt. The model lso does not tke into ccount potentil lrge-scle chnges in the Gulf of Thilnd south of the model domin; to ccount for such chnges, much lrger model domin will e needed. In 1998 (n El Niño yer), the surfce temperture ws generlly wrmer y 2 C thn tht in 2 ( L Niñ yer). This difference is ttriuted to the reduced cloud cover during El Niño events (Fig. 3d; dys 3 5 versus dys 1,1 1,3) tht llows more shortwve rdition to rech the wter surfce. The El Niño of 1997 1998 ws one of the strongest events of the century, which resulted in reduced clouds nd dry climte in the western Pcific nd the Indonesi region (Chndr et l. 1998), in greement with our nlysis of cloud cover ner Thilnd. The 1999 2 L Niñ ws more moderte event. 6 Summry nd conclusions A low ltitude esturine-like shllow costl system, the UGOT, hs een modeled to investigte the effects of wind stresses nd river dischrges on the 3D distriution of slinity, temperture nd velocity. While influenced y rivers nd tides like other esturine systems, the UGOT is not typicl esturine system ecuse of its topogrphy, forcing nd loction. The UGOT is very shllow nd its width equls its length; it is lso locted in low ltitudes where Coriolis effects re smll nd sesonl monsoonl winds dominte. The UGOT model is sed on 3D primitive eqution Princeton Ocen Model. The primry forces re the tides t the southern open oundry, dily surfce winds nd surfce het fluxes. Monthly men dischrges of four rivers hve een included in the simultions; it will e of interest to include internnul vritions in river flows in future studies. The simultions show tht the circultion in the UGOT is strongly influenced y the sesonl monsoonl winds, which result in clockwise circultion during the southwest monsoon nd counter-clockwise circultions during the northest monsoon. The southwest nd northest monsoons pile up wter in the direction of the winds, ut the se level grdients do not result in geostrophic flows in those low ltitudes. Insted, the men su-tidl trnsport resemles wind-driven non-rotting costl system such s those descried y Csndy (1973) nd others. To show this, the time-men 3D model flow cross n est west section ws compred with n nlyticl solution of simple wind-driven non-rotting stedy stte model (Winnt 21). This comprison confirms tht the clssic pttern of trnsport in the direction of the wind in shllow regions nd trnsport ginst the wind in regions deeper thn the verge depth is still lrgely vlid in the UGOT. However, some rottionl influence on the roclinic flow pttern is not negligile, even in those low ltitudes, s seen from nlysis of the Ekmn numer. Freshwter plumes from rivers ply importnt roles in the UGOT, so sensitivity experiments with the model were conducted to ssess the impct of uoyncy-driven versus winddriven dynmics, nd the impct of Coriolis. At low ltitudes nd without winds, the simulted river plume in the vicinity of the river mouth ws found to e xisymmetric (Fig. 8), result confirmed y stellite dt (Fig. 2). Frther wy from the river mouth low-slinity plume is spred s costl jet long the oundries. The clssic picture of river plumes turning to the right of the outflow when exiting the river mouth (in the northern hemisphere) is not seen here, even when the Coriolis prmeter is incresed to represent midltitudes. The reson is likely the strong tidl mixing which dilutes the river plume when it is still close to the river mouth. However, in the mid-ltitude experiments, the plume is less xisymmetric ner the river mouth nd the diluted low slinity plume spreds frther long the cost thn tht in low ltitudes. The low ltitudes lso limit the development of smllscle gyres seen offshore the river mouth in mid-ltitude simultions. When monsoonl winds re included in the model, the winds dominte the spreding of the plume long the cost in the direction of the wind. Internnul vritions in surfce tempertures, s much s 2 C difference etween one yer nd nother, re due to lrge-scle tmospheric nd climte vritions. A comprison etween n El Niño yer (1998) nd L Niñ yer (2) demonstrtes the ility of the model to cpture these chnges. However, unlike higher ltitudes where sesonl nd internnul vritions in surfce wter temperture re the result of vritions in solr rdition nd het fluxes, in the UGOT the most importnt fctor seems to e the chnge in cloud cover ssocited with the monsoon (wet nd dry sesons) or with the El Niño nd L Niñ ptterns (e.g., the dry climte in the region following the 1997 1998 El Niño; Chndr et l. 1998). Therefore, the cloud cover hd to e specificlly ccounted for in the model fluxes to produce relistic simultions of wter tempertures. The model still hs some deficiencies, in prticulr, it does not include the influence from chnges in the Gulf of Thilnd south of the UGOT; this spect cn e improved y future extension of the model domin to the entire Gulf of Thilnd. In summry, the UGOT model ws used here s test ed to study the dynmics in unique environment of shllow low-ltitude costl system with strong influences from river dischrges nd monsoonl winds. Better understndings of the dynmics nd forcing mechnisms in the UGOT cn help in the development of future relistic costl ocen forecst system to ddress vrious environmentl prolems fcing this region.
Ocen Dynmics (214) 64:557 571 571 Acknowledgments This study is prt of the grdute studies of S. Srmul who received support from ODU s deprtment of Ocen, Erth nd Atmospheric Sciences (OEAS), including the Dorothy Brown Smith Scholrship, nd from the computtionl resources of the Center for Costl Physicl Ocenogrphy. Additionl support provided y the Thi Government Science nd Technology scholrship nd from Chullongkorn University, Thilnd. T. Ezer ws prtly supported y grnts from NOAA Climte Progrms nd the Keni Peninsul Borough, Alsk. References Alretsen J, Røed LP (21) Decdl long simultions of mesoscle structures in the northern North Se/Skgerrk using two ocen models. Ocen Dyn 6(4):933 955. doi:1.17/s1236-1- 296- Blumerg, AF, Mellor, GL (1987) A description of three-dimensionl costl ocen circultion model. In: N. Heps (ed.) Threedimensionl costl ocen models. Americn Geophysicl Union, 28pp Burnprtheprt A, Yngi T, Mtsumur S (28) Sesonl vrition in wter column conditions in the upper Gulf of Thilnd. Cont Shelf Res 28:259 2522 Burnprtheprt A, Yngi T, Niemnn KO, Mtsumur S, Sojisuporn P (28) Surfce chlorophyll- dynmics in the upper Gulf of Thilnd reveled y coupled hydrodynmics ecosystem model. J Ocenogr 64:639 656 Burnprtheprt A, Niemnn KO, Yngi T, Mtsumur S, Sojisuporn P (29) Sesonl vritions in wter column conditions in the upper Gulf of Thilnd. Burph Sci J 14(1):99 113 Chndr S, Ziemke JR, Min W, Red WG (1998) Effects of 1997 1998 El Niño on tropospheric ozone nd wter vpor. Geophys Res Lett 25(2):3867 387 Cho SY, Shw PT, Wu SY (1996) El Niño modultion of the South Chin Se circultion. Prog Ocenogr 38(1):51 93 Csndy GT (1973) Wind-induced rotropic motions in long lkes. J Phys Ocenogr 3:429 438 Ezer T (2) On the sesonl mixed lyer simulted y sin scle ocen model nd the Mellor Ymd turulence Scheme. J Geophys Res 15(C7):186 195 Ezer T (213) Se level rise, sptilly uneven nd temporlly unstedy: Why the U.S. Est Cost, the glol tide guge record nd the glol ltimeter dt show different trends. Geophys Res Lett 4(2): 5439 5444. doi:1.12/213gl57952 Guo X, Vlle-Levinson A (27) Tidl effects on esturine circultion nd outflow plume in the Chespeke By. Cont Shelf Res 27:2 42. doi:1.116/j.csr.26.8.9 Guo X, Vlle-Levinson A (28) Wind effects on the lterl structure of density driven circultion in the Chespeke By. Cont Shelf Res 28:245 2471. doi:1.116/j.csr.28.6.8 Horner-Devine AR (29) The ulge circultion in the Columi River plume. Cont Shelf Res 29:234 251. doi:1.116/j.csr.27.12.12 Ksi A, Hill AE, Fujiwr T, Simpson JH (2) Effect of the Erth s rottion on the circultion in regions of freshwter influence. J Geophys Res 15(C7):16,961 16,969 Mellor GL (24) Users guide for three-dimensionl, primitive eqution, numericl ocen model. 35 pp., Progrm in Atmospheric nd Ocenic Sciences. Princeton University, Princeton Mellor GL, Ymd T (1982) Development of turulence closure model for geophysicl fluid prolems. Rev Geophys 2(4):851 875 Nrv'ez DA, Vlle-Levinson A (28) Trnsverse structure of wind induced flow t the entrnce to n estury: Nnsemond River. J Geophys Res 113(C94), DOI 1.129/28JC477 Nicholls RJ (211) Plnning for the impct of se level rise. Ocenogrphy 24(2):142 155 Oey LY (1996) Simultion of mesoscle vriility in the Gulf of Mexico: Sensitivity studies, comprison with oservtions, nd trpped wve propgtion. J Phys Ocenogr 26(2):145 175 Sny R, Vlle-levinson A (25) Wind induced circultion in semienclosed homogeneous, rotting sins. J Phys Ocenogr 35: 252 2531 Srmul S (213) Oservtions nd modeling forcing mechnisms for the costl dynmics of the Upper Gulf of Thilnd. Ph.D. Disserttion, Old Dominion University, 154 pp Srmul S, Ezer T (21) Tidl-driven dynmics nd mixing processes in costl ocen model with wetting nd drying. Ocen Dyn 6(2): 461 478. doi:1.17/s1236-9-25-1 Vlle-Levinson A, Reyes C, Sny R (23) Effects of thymetry, friction, nd rottion on estury ocen exchnge. J Phys Ocenogr 33:2375 2393 Winnt CD (24) Three-dimensionl wind-driven flow in n elongted, rotting sin. J Phys Ocenogr 34:462 476 Winnt CD (21) Wind nd tidlly driven flows in semienclosed sin. In: Vlle-Levinson A (ed) Contemporry Issues in Esturine Physics, 1st edn. Cmridge University Press, New York, pp 125 144 Wong KC (1994) On the nture of trnsverse vriility in costl plin estury. J Geophys Res 99(C7):14,29 14,222 Yu L, Jin X, Weller RA (28) Multidecde glol flux dtsets from the Ojectively Anlyzed Air se Fluxes (OAFlux) Project: Ltent nd sensile het fluxes, ocen evportion, nd relted surfce meteorologicl vriles. Tech. Rep. OA-28-1, Woods Hole Ocenogrphic Institution, Woods Hole, Mss Zheng L, Weiserg RH (24) Tide, uoyncy, nd wind driven circultion of the Chrlotte Hror estury: A model study. J Geophys Res 19(C611), DOI 1.129/23JC1996