1
Reproducing the upper ocean response of the Meso-American Barrier
Reef System to Hurricane Mitch and coastal freshwater inputs: an
approach using SeaWiFS remote sensing data and a nested-grid ocean
circulation model
Liang Wang1, Jinyu Sheng2, Serge Andréfouët3, Chuanmin Hu1, Bruce G. Hatcher4, and Frank E.
Muller-Karger1
1
Institute for Marine Remote Sensing, College of Marine Science, University of South Florida, St.
Petersburg, FL 33701, USA
2
Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, B3H 4J1, Canada
3
UR 128 Coréus, Institut de Recherche pour le Développement, BP A5-98848, Noumea Cedex,
New-Caledonia
4
Marine Ecosystem Research, Cape Breton University, Sydney, Nova Scotia, B1P 6L2, Canada
(To be submitted to JGR-Oceans in August, 2006)
2
Abstract
The growing body of work on coral reef connectivity calls for a better representation of the circulation
between reefs in normal but also during catastrophic storm conditions. Previous work has identified
the passage of the category-5 Hurricane Mitch (October 1998) on the Meso-American Barrier Reef
System (MBRS) as a potentially important event in terms of unusual reef connectivity. The impact of
this hurricane on the upper ocean offshore the MBRS is investigated using a triply nested-grid
modeling system and SeaWiFS ocean color data. The nested system is forced by 6-hourly
NCEP/NCAR winds for the first 294 days, and then by the combination of the NCEP/NCAR wind
forcing and an idealized vortex associated with Hurricane Mitch in the following 20 days. The nested
system is also forced by the monthly mean sea surface heat and freshwater fluxes, and buoyancy
forcing associated with river runoff at the coast. The low salinity and extent of estuarine waters along
the northern coast of Honduras are specified based on SeaWiFS data. As expected during hurricane
conditions, the nested system generates strong divergent surface currents under the storm, intense
inertial currents and sea surface temperature cooling behind the storm, both of which are more
energetic on the right of the storm track. The model also reproduces well the strong estuarine
low-salinity plume extending from the coast to the deep water region off Honduras, which is
consistent with SeaWiFS-derived salinity observations shortly after the passage of Hurricane Mitch.
As a perspective, the reliability of model predictions offers fresh perspectives for modeling the reef
connectivity within MBRS and the Caribbean region.
3
1. Introduction
The Meso-American Barrier Reef System (MBRS), which contains the largest coral reef system in
the Caribbean Sea, extends from the Bay Islands of Honduras, north through the coastal and
continental shelves of Guatemala and Belize, to the northeast tip of Yucatan Peninsula of Mexico
(Figure 1). Several million people living in the adjacent coastal areas of the MBRS benefit from the
natural and commercial resources that this system offer, including stabilization and protection of
coasts, fishery resources, tourism industry, and aesthetic value. Coral reefs in the MBRS have been
affected by various natural and human disturbances and stresses including hurricanes, coral bleaching
and subsequent mortalities, disease outbreaks, overfishing, and contamination through land-based
sources of pollution over the last 30 years [Kramer and Kramer, 2002]. Therefore, this region is the
focus of a large number of conservation programs.
Conservation programs typically intend to promote the creation of marine protected areas (MPA)
in order to decrease the stress that reef ecosystem endures. One of the criteria to define MPA, or
network of MPA, is the level of connectivity that exists between the different reefs. Reefs act as
source and sinks of larvae. Especially in the context of fishery management, MPA should be
conceived to maximize the rates of exchanges between protected sources towards non-protected sinks.
Clarifying and quantifying the temporal and spatial scales of physical connectivity between reefs is
thus a new growing challenge for physical oceanographers. Numerical models have been applied for
this task for about 20 years, but requirement for scientifically designed MPA network has speed up the
developments worldwide. The Caribbean Sea and MBRS is no exception [Tang et al., 2006; Cowen et
al., 2006].
Spatially, the needs to represent the connectivity of dense matrices of reefs call for high-resolution
ocean circulation models. Options are: (1) a very high-resolution finite-difference model throughout
the entire domain, and thus with prohibitive processing time, (2) a finite-element model with variable
mesh size but the design of the grid may not be not trivial [Legrand et al., 2006], or (3) a
high-resolution finite-difference model nested within a coarser regional model. Temporally,
connectivity patterns for a given area may evolve in a matter of hours, days, months, years depending
on a variety of factors such as tidal regime, wind climate or global change. Climatological data can be
used to construct the time-mean connectivity patterns at the adequate time-scale, but short-term events
may generate unusual short-term connectivity patterns that may have significant impacts on
population transfer if they occur concurrently to fish or coral spawning periods for instance. In
October 1998, the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) ocean color sensor captured
such short-term patterns of connectivity right after Hurricane Mitch landed in Honduras [Andréfouët et
4
al., 2002]. The patterns were evidenced by large river plumes originating from the Honduras coastline
and reaching in a matter of hours the Mexican and Belizean reefs (Figure 3). Since numerical models
have been used to study connectivity under normal (or climatologically monthly mean) condition
[Tang et al., 2006], we propose here to verify that numerical models can also reproduce short-term
ocean surface circulation events with the simple and effective model parameterizations. For this goal,
Hurricane Mitch provides an ideal case study, with its specific challenges.
Following this introduction, section 2 presents the current knowledge on the general mean
circulation over the MBRS, the observations collected at the time of Hurricane Mitch and a brief
review on numerical modeling of hurricanes. Section 3 describes the triply nested-grid modeling
system for the MBRS based on the smoothed semi-prognostic method [Sheng et al., 2005], the
parameterization of the model to account for the Hurricane, and the use of SeaWiFS data to
parameterize the coastal boundary of the model. Section 3 presents the model results, including
near-surface and sub-surface currents, SST cooling, and river plumes simulations. Section 5 provides a
brief summary and discussion.
2 Background: ocean circulation over the Meso-American Barrier Reef System
and the Hurricane Mitch
2.1 Observed and Simulated Ocean Circulation over the MBRS in Normal Conditions
Very limited in-situ oceanographic measurements were made on the MBRS in the past. Historical
observations compiled by Craig [1966] suggest three distinct features of the general mean circulation
in the upper ocean of the region [see also Figure 2 of Ezer et al., 2005]: an intense northwestward
offshore flow as part of the Caribbean Current in the deep water off continental shelves of Honduras
and Belize; an equator-ward coastal current that flows first along the east coast of Belize and then
eastward along the north coasts of Guatemala and Honduras; and a cyclonic (counter-clockwise)
circulation in the Gulf of Honduras (GOH). As discussed in Ezer et al. [2005], two sub-surface drifters
were deployed in April 2000 at 15 m depth to the south and west of Glover Reef respectively
[Fratantoni, 2001]. The first drifter revealed a south moving, then eastward trajectory following a
cyclonic gyre in the GOH, which is consistent with the general mean circulation pattern in the region
suggested by Craig. The second drifter moved northward about 200 km in 20 days, indicating a
northward flow from Glover’s Reef and through the passage between Turneffe Islands and Lighthouse
Reef Atolls. This northward current was in the opposite direction in comparison with the general mean
circulation pattern suggested by Craig. Ezer et al. [2005] attributed the above contradiction of the flow
direction from Glover’s Reef (south side) through the passage between Turneffe Islands and
5
Lighthouse Reef Atolls to the mesoscale variability of the near-surface circulation in the region.
Three-dimensional ocean circulation models have increasingly been used for simulating the
large-scale circulation in the Caribbean Sea [Sou et al., 1996; Murphy et al., 1999; Candela et al.,
2003; Ezar et al., 2003; Sheng and Tang, 2003 and 2004; Ezar et al., 2005, Oey et al., 2005; Tang et
al., 2006]. Among them, numerical studies made by Sheng and Tang [2003 and 2004], Ezer et al.
[2005] and Tang et al. [2006] focus specifically on the western Caribbean Sea and the MBRS. Sheng
and Tang [2004] studied the monthly mean circulation over the MBRS using a two-level nested-grid
z-level model with a fine-resolution (about 6 km) inner model imbedded inside a coarse-resolution,
20-km model for the western Caribbean Sea developed earlier by Sheng and Tang [2003]. Tang et al.
[2006] used a triply nested-grid z-level modeling system with three different horizontal resolutions of
20 km, 6 km and 2 km to study the upper ocean circulation and hydrodynamic connectivity associated
with the reef atolls on the Belize shelf. By using the Princeton Ocean Model (POM) with a variable
horizontal resolution ranging from 3 km along the MBRS to 8 km on the open boundary, Ezer et al.
[2005] examined the influence of topography, circulation, wind, density and eddies on the 3D
circulation on the MBRS. Despite the existing numerical ocean circulation models reproduce the
general mean circulation in the MBRS, more work is needed in order to have better knowledge of the
detailed MBRS circulations, particularly during sporadic and extreme events.
2.2 Brief Review on Hurricane-Induced Circulation Simulations
Despite the pioneering numerical simulations reported in section 2.1, responses of the upper ocean
circulation of the MBRS to major storms or hurricanes have not well been studied. Our goal in this
study is to examine the storm-induced circulations as well as temperature and salinity variations in the
upper ocean of the MBRS during Hurricane Mitch in 1998. Various numerical studies have been
conducted in the past to examine the storm-induced circulation in a variety of coastal and open oceans
[e.g., Geisler, 1970, Chang and Anthes, 1978; Price, 1981; Greatbatch, 1983; Slodal et al., 1994;
Sheng et al., 2006; Oey et al., 2006]. Geisler [1970] investigated the linear dynamics of the ocean
response to a moving storm without considering mixed layer effects. Price [1981] suggested a simple
and effective parameterization for estimating the vertical eddy viscosity and diffusivity coefficients in
the upper ocean in terms of the mean velocity difference across the base of the mixed layer. With
Price’s parameterization, Sheng et al. [2006] simulated reasonably well the storm-induced currents
associated with a moving storm known as Hurricane Juan in 2003 using a nested-grid ocean model.
Oey et al. [2006] studied the response of the Caribbean Sea and Gulf of Mexico to Hurricane Wilma in
2005 using the Princeton Regional Ocean Forecast System (PROFS). All these studies demonstrate
that the upper ocean response to a moving storm can be characterized as the intense inertial
6
oscillations and sea surface temperature cooling in the wake of storm biased to the right of the storm
track, strongly depending on the hurricane translation speed. Hereafter, we use a triply nested-grid
modeling system to study the upper ocean circulation over the western Caribbean Sea with special
emphasis of the storm-induced currents and sea surface temperature changes over the MBRS during
Hurricane Mitch in 1998.
2.3 In situ and Remote Sensing Observations during Hurricane Mitch
Hurricane Mitch is one of the most disastrous storms in history that struck the central American
countries of Nicaragua, Honduras, El Salvador and Guatemala and resulted in a region-wide death toll
of more than 9000. Mitch originated from a tropical wave over west Africa on October 8, 1998 and
progressed across the tropical Atlantic for the next seven days [http/://www.nhr.noaa.gov]. Mitch
moved through the eastern Caribbean Sea on October 18 and 19 and upgraded from a tropic
depression into a hurricane storm in the southwestern Caribbean Sea (76.1oW, 11.6oN) on October 22
(Figure 1), with a minimum surface pressure of 1002 mb and a maximum wind speed of about 55 km
h-1. Thereafter, the storm strengthened to a Saffir-Simpson category-5 hurricane storm by October 26
with maximum sustained wind speeds of about 285 km h-1 and minimum surface pressure of 905 mb
over the southwest Nicaragua Rise (83.1oW, 16.9oN). From October 27, Mitch started to skirt the
north coast of Honduras and generated anomalous precipitation, coastal flooding and landsliding. The
storm made landfall over Honduras during the morning of October 29 with the maximum wind speeds
of about 160 km h-1 and a minimum central pressure of 970 mb. After landing Mitch was downgraded
to a tropical depression and moved slowly from southward to westward and entered the Gulf of
Mexico on November 2, 1998 (Figure 1).
Large-scale ocean observations by means of remote sensing provide critical information for the
initialization and verification of atmosphere and ocean numerical models [Ishizaka, 1990]. In this
study, we use remote sensing data to parameterize the low salinity waters near the coast to provide
proper coastal boundary conditions and validate the performance of the circulation model in
simulating the upper ocean salinity fields.
Because of intensive vertical mixing, significant drops in sea surface temperature (SST) behind a
moving storm, typically from 1 to 6oC, have been reported [Jordan, 1964; Wright, 1969; Fedorov et
al., 1979; Pudov et al., 1979; Smith, 1982; Cornillon et al., 1987; Nelson, 1996]. We examined SST
imagery from the Advanced Very High Resolution Radiometer (AVHRR) and TRMM Microwave
Imager (TMI) during Hurricane Mitch for the study region. Unfortunately, due to the significant cloud
cover and precipitation, there are no valid satellite remote sensing SST data for the study region. We
therefore favor the use of ocean color data from the Sea-viewing Wide Field-of-view Sensor
7
(SeaWiFS) in this study.
SeaWiFS high-resolution (1.1 km/pixel at nadir) data were captured and processed using the
software package SeaDAS4.4 at the University of South Florida. After several rounds of reprocessing
to incorporate calibration and algorithm updates, the data products are considered of science quality
[McClain et al., 2004]. Images from an earlier processing were used to demonstrate the water
circulation patterns around MBRS [Andréfouët et al., 2002]. As an example, Figure 2 shows the
contrast between chlorophyll-a concentration images before (Fig. 2a) and after (Figs. 2b-2d) Hurricane
Mitch. On October 24 when Mitch was still far away the turbid water was restricted to the Honduras
coast and Belize shelf. After Mitch, the turbid plume extended from the northeast coast of Honduras to
the deep ocean, the Bay Islands (150 km, eastward, Figure 2b), and further north to the Belize shelf on
3 November (Figure 2c).
3. Methods: The Ocean Circulation Model Setup and External Forcing
3.1 The Triply Nested-Grid Ocean Circulation Modeling System
The triply nested-grid ocean circulation modeling system was constructed from a primitive-equation
z-level model known as CANDIE, which stands for the Canadian version of Diecast, [Sheng et al.,
1998]. CANDIE has been successfully applied to address various modeling problems on continental
shelf seas, including wind-driven circulations over an idealized coastal canyon [Sheng et al., 1998], a
density-driven coastal current [Sheng, 2001], tidal circulation in the Gulf of St. Lawrence [Lu et al.,
2001], and seasonal circulation in the northwestern Atlantic Ocean [Sheng et al., 2001]. Most recently
CANDIE has been applied to the western Caribbean Sea by Sheng and Tang [2003 and 2004] and
Tang et al. [2006], Lunenburg Bay of Nova Scotia by Sheng and Wang [2004] and Wang et al. [2006],
and Lake Huron and Georgian Bay by Sheng and Rao [2006].
The nested-grid system has three subcomponents (Figure 3): a coarse-resolution outer model (~19
km) covering the west Caribbean Sea (WCS) (72˚W-90˚W, 8˚N-24˚N), an intermediate-resolution
middle model (~6 km) covering the MBRS and Yucatan Strait (84˚W-89˚W, 15.5˚N-20˚N), and a
fine-resolution inner model (~2 km) covering the north coast of Honduras and the Bay Islands
(85˚W-88˚W, 15.6˚N-17˚N).
The time steps are set to about 14.4 minutes in the outer model, 5.5
minutes in the middle model and about 2.2 minutes in the inner model, respectively. The nested
system uses the digital bathymetry database of 2-minute resolution (DBDB2) developed by the Ocean
Dynamics and Prediction Branch, the Naval Research Laboratory of the United States
[ http://www.7320.nrlssc.navy.mil/DBDB2_WWW].
The boundary definitions of the middle and
inner domains are justified here by the geographic location of the MBRS in the
8
Mexico-Belize-Honduras coast and the significant impact of Hurricane Mitch in 1998 on the Honduras
coast where most of the SeaWiFS detected plume originated from. The three subcomponents of the
nested system have the same 28 unevenly spaced z-levels with the fine-resolution z-level of 2 m for
top ten sea levels and relatively coarse-resolution z-levels of about 500 m near the bottom. The centers
of z-levels are located respectively at 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 25, 40, 75, 140, 230, 340, 450,
575, 725, 900, 1250, 1750, 2250, 2750, 3250, 3750, 4250, and 4750 m. It should be noted that the
nested-grid system used in this study is very similar to the one used by Tang et al. [2006], except that
(1) the inner model domain in this study covers the coastal region of Honduras, the Bay Islands and
GOH; (2) the vertical resolution of the system is finer and 2 m in the top 20 m than about 10 m in
Tang et al., and (3) a simple and effective vertical mixing scheme (to be discussed next) is used here.
The nested-grid system uses the sub-grid scale mixing parameterization suggested by Price [1981]
for the vertical eddy viscosity and diffusivity coefficients Km and Kh. In this scheme a scale velocity
(V) defined as the magnitude of the mean velocity difference across the base of the upper ocean
mixed layer is used to parameterize the vertical mixing coefficients. As shown in Sheng et al. [2006],
application of Price’s vertical mixing scheme led to realistic simulations of the bias to the right of the
storm track of the sea surface temperature response. This is mainly because the storm-induced currents
and therefore vertical mxing are usually stronger to the right of the storm track than to the left [Price,
1981, Sheng et al., 2006]. The horizontal mixing scheme of Smagorinsky [1963] is used to
parameterize the horizontal eddy viscosity and diffusivity coefficients (Am, Ah), which are related to
the model grid spacing (x, y), and velocity shear and strain in the horizontal direction. Since the
Smagorinsky scheme is resolution dependent, the parameterization of horizontal mixing in three
subcomponents of the nested system is different. The nested system also uses the fourth-order
numerical [Dietrich, 1997] and Thuburn’s flux limiter to discretize the nonlinear advection terms
[Thuburn, 1996].
To exchange information between the three subcomponents of the nested-grid system, we follow
Tang et al. [2006] and use the two-way nesting technique based on the smoothed semi-prognostic
method developed by Sheng et al. [2005]. A free-slip boundary condition is used at lateral solid (or
closed) boundaries in the three subcomponents of the nested system. Along the open boundaries of
each subcomponent, the normal flow, temperature and salinity fields are calculated using adaptive
open boundary conditions [Marchesiello et al., 2001]. The depth-mean normal flows across the outer
model open boundaries are set to be the monthly mean results produced by a (1/3)˚ Atlantic model
based on FLAME. [Carsten Eden, personal communication, 2003]. The outer (middle) model results
are used to specify the boundary conditions along the open boundaries of the middle (inner) models.
9
3.2 Model Forcing and Parameterization of a Moving Storm
We initialize the nested-grid ocean circulation system with the monthly mean climatology of
temperature and salinity (TS) in January constructed from in-situ hydrographic observations at the
standard z-levels extracted from the World Ocean Database 1998 compiled by the US National
Oceanographic Data Center (NODC) using the objective analysis technique know as Barnes’
algorithm [Barnes, 1964]. Readers are referred to Geshelin et al. [1999] for a more discussion on the
methodology used in constructing the three-dimensional monthly mean hydrographic climatology for
the study region. The nested-grid system is forced by the wind stresses converted from 12 hourly wind
speeds extracted from the National Centers for Environmental Prediction of the Nation Center for
Atmospheric Research (NCEP/NCAR) 40 year reanalysis dataset [Kalnay et al., 1996], and monthly
mean heat flux constructed by da Silva et al. [1994] in the first 294 days (i.e., from January 1 to
October 21) of 1998. The conventional bulk formula of Large and Pond [1981] is used to convert
NCEP/NCAR wind speeds to wind stresses, except that the drag coefficient is set to a constant value
of 2.2x10-3 if the maximum sustained wind speeds are greater than 33 m s-1 [Powell et al., 2003].
The net heat flux through the sea surface Qnet is expressed as [Bariner et al., 1995]:
Q = Q +  (SST + SST )
c
net
where
c
(1)
m
net
Q is the monthly mean net heat flux taken from da Silva et al. [1994], SSTc is the monthly
c
net
mean sea surface temperature climatology, SSTm is the model calculated sea surface temperature, and γ
is the coupling coefficient defined as z1ρocp/τQ, where z1 is the thickness of the top z-level, cp is the
specific head, and τQ is the restoring time scale which is set to 10 days except for a special area in
Campeche near the Yucatan Strait where the restoring time scale is set to 5 days. We also restore the
model sea surface salinity to the monthly mean climatology at a time scale of 10 days.
A simple vortex is used to represent the wind stresses associated with a moving storm during
Hurricane Mitch [Chris Fogarty, personal communication, 2006]:
±
max
r
f ( )
r
r r
1 1
(
) f ( )
r r
r
r
r<r
(2)
min
min
 ( r , ) = ± 
r ,
max
max
min
r
min
≤r ≤r
max
max
max
0
min
r>r
max
where τr,θ(r,θ) is the radial (r) or tangential (θ) components of wind stress as a function of radius r and
azimuth θ measured counterclockwise from the right hand side of a perpendicular line through the
storm motion vector, τmax is the amplitude of the maximum wind stress located at rmin, rmax is the outer
radius where τr,θ vanishes, and f(θ) is the asymmetry factor set to unit in this study. We set rmin to 20
10
km and rmax to 300 km based on the satellite images during Hurricane Mitch, and τmax to be the
observed maximum sustained wind speeds provided by the Southeast Regional Climate Center
(SERCC) of the United States. The realistic storm track provided by SERCC is also used (Figure 1),
with the instantaneous translational speeds of Hurricane Mitch calculated from the 6-hourly SERCC
storm track data.
In summary, to simulate the Mitch-induced circulation in the west Caribbean Sea we first integrate
the nested-grid ocean circulation modeling system for 294 days from January 1 to October 21 in 1998
with the monthly-mean hydrographic climatology, monthly mean surface heat and freshwater fluxes,
and 12-hourly NCEP/NCAR wind stresses. We then add a parameterized vortex associated with a
moving storm to the NCEP/NCAR wind forcing for next 20 days from October 22 to November 10.
3.3 Deriving Sea Surface Salinity Field from SeaWiFS Ocean Color Data
Since the nested-grid ocean circulation modeling system does not simulate chlorophyll
concentrations, we use the surface salinity distribution maps derived from the satellite remote sensing
ocean color data to tune and validate the model performance. Several studies have documented the
inverse relationship between sea surface salinity and absorption coefficient of colored dissolved
organic matter (aCDOM, m-1), and the latter can often be derived from satellite ocean color data [e.g.,
Ferrari and Dowell, 1998; D’Sa et al., 2002; Hu et al., 2003; Hu et al., 2004]. Although this technique
is still in the early stages of the development, and in particular there is no in situ data around MBRS to
calibrate the relationship between the sea surface salinity and aCDOM, as a first attempt we derived the
sea surface salinity from SeaWiFS data in the following way.
SeaWiFS spectral remote sensing reflectance data in the visible bands (412, 443, 490, 510, 555, and
670 nm, respectively) were first used to derive the backscattering coefficient (bbp) of detritus (a sum of
all non-living particles) and the total absorption coefficient of CDOM+detritus (i.e., aCDM = aCDOM +
aD) using a quasi-analytical algorithm [Lee et al., 2002]. Then, to derive aCDOM from aCDM, an
empirical equation was used to estimate aD as: aD(440) = 2.075x(bbp(555))1.02 (n=110, r=0.89, 0.001 <
aD(440) < 0.12). The empirical relationship was derived from field data collected from 8
oceanographic cruises on the west Florida shelf in 2000 and 2001 [Jennifer Cannizzaro, University of
South Florida, unpublished data]. After the removal of aD(440) from aCDM(440), aCDOM(440) was
converted to salinity as: salinity = 36.1 - 10xaCDOM(440) (0 < aCDOM(440) < 3.61 m-1). The conversion
was based on limited field data collected in SW Florida coastal waters and is subject to uncertainties
off the Honduras coast and Belize shelf because there is no in situ data to tune the empirical
relationship between salinity and aCDOM(440). However, we emphasize that our focus is to see if the
model can reproduce the general salinity patterns (river plumes) rather than the absolute salinity
11
values derived from SeaWiFS.
3.4 Numerical Experiments
In addition to using the SeaWiFS derived sea surface salinity fields to validate the model
performance, we examine the influence of river runoffs in the generation of river plumes by specifying
different salinity values along the coastlines in three numerical experiments.
In the control run
(Exp-Control, Table 1), we specify the salinities at 12 river mouths along the coasts of Honduras and
Belize shelf to minimum values of 25 psu (practical salinity unit) from day 294 (October 22) to day
301 (October 29) during Hurricane Mitch and gradually restore them to the climatological values after
day 301 with an e-folding time scale of 12 days. For the purpose of sensitivity studies, we change the
minimum salinities at the river months from 25 psu to 20 psu in the second experiment (Exp-Low) and
28 psu in the third experiment (Exp-High), respectively (Table 1). Other model parameterizations in
the three experiments are the same. In addition, we conduct the fourth numerical experiment
(Exp-Norm, Table 1) by forcing the nested-grid system with the monthly mean heat and freshwater
fluxes and NCEP/NCAR wind, but without the parameterized vortex associated with Mitch and
without specification of buoyancy forcing associated with freshwater inputs at the river mouths. Since
the horizontal resolution of the NCEP/NCAR reanalysis data is about 200 km in the western
Caribbean Sea, which is too coarse to resolve Hurricane Mitch. Therefore, the model results in
Exp-Norm are therefore used to represent the ocean circulation under the normal conditions in the
study region.
4 Results: The Simulated Upper Ocean Response to Hurricane Mitch
4.1 Simulated Ocean Currents
Mitch upgraded from a tropical depression to a hurricane in the southern Colombian Basin on
October 22, 1998, with maximum sustained wind speeds of about 85 km h-1. Figures 4a and b present
the simulated ocean currents at depths of 1 m and 75 m respectively at 12:00 UTC (all times in this
study are in Universal Coordinated Time, UTC) October 23 (day 295.5) in the three sub-domains of
the nested-grid system in the control run. The outer model results (Figure 4a) demonstrate that Mitch
at day 295.5 affects mainly the near-surface circulation in the southern Caribbean Sea, with strong
divergent near-surface currents of about 1 m s-1 forced by the local wind forcing over an area around
the storm center with a radius of influence to be about 100 km. Outside this area of influence, the
near-surface circulation produced by the outer model is very similar to the near-surface circulation
under the normal condition (without the storm), and characterized as a relatively broad and
approximately westward flow associated the Caribbean Current in the northern and central Colombian
12
Basin and a bifurcation of the flow before reaching the Nicaragua Rise, with the main branch turning
northwestward and flowing along the outer flank of Nicaragua Rise to form a narrow offshore flow
running onto the southern MBRS, and a weak branch veering southwestward to form the cyclonic
Panama-Colombia Gyre over the southwestern Colombian Basin [Mooers and Maul, 1998; Sheng and
Tang, 2003; 2004]. Over the northern Caribbean Sea and MBRS, the near-surface circulation
produced by the outer and middle models at this time (Figure 4a) has not been affected directly by
Mitch and is characterized by the typical Caribbean Current that flows northwestward from Nicaragua
Rise to the continental shelf off southeastern Mexico and then turns northeastward along the east coast
of Yucatan Peninsula [Ezer et al., 2005; Tang et al., 2006]. The nested-grid middle and inner models
also produce a cyclonic recirculation gyre in the Gulf of Honduras (GOH) and several smaller-scale
gyres over the coastal region between the Bay Islands and the north coast of Honduras (Figure 4a).
The model-calculated sub-surface (75 m) circulation at day 295.5 is not significantly affected by
Mitch (Figure 4b) since the storm was still at the early stage of a hurricane and less storm-induced
energy penetrated into deep layers. The simulated sub-surface circulation at this time has a broad
westward flow over the northern Colombian Basin, with a large-scale cyclonic recirculation over the
southwestern Colombian Basin and several small-scale recirculation gyres near the coastal waters off
Panama and Colombia (Figure 4b). Part of the westward flow over the northern Colombian Basin
flows onto the central Cayman Basin through the outer flank of Honduras Rise, which turns gradually
westward on to the central MBRS, and then veers anticyclonically to form a narrow intense coastal jet
running northward along the east coast of Yucatan Peninsula. The sub-surface currents produced by
the middle and inner models at day 295.5 also have a cyclonic recirculation over the GOH and
approximately eastward currents along the outer flank of the Bay Islands. The large-scale sub-surface
circulation shown in Figure 4b is very similar to the general sub-surface circulation in the region under
the normal condition in the western Caribbean Sea produced by Sheng and Tang [2003; 2004] and
Ezer et al. [2005].
At 12:00 October 26 (day 298.5), Mitch reached the northern flank of Honduras Rise and fully
developed into a Category-5 hurricane with maximum sustained wind speeds of about 290 km h-1
(Figure 5). The near-surface currents produced by the nested-grid outer model in the western
Caribbean Sea have been affected significantly by Mitch by this time (Figure 5a). There are intense
divergent near-surface currents forced by the local wind under the storm over the Cayman Basin and
strong inertial currents in the wake of the storm over the northern Colombian Basin (see more
discussion on the inertial currents later), which is consistent with previous studies of storm-induced
circulation by Chang and Anthes [1978], Price [1981], Greatbatch [1983] and Sheng et al. [2006]. The
middle and inner model results at day 298.5 (Figure 5a) demonstrate that the storm also affects
13
significantly the near-surface circulation on the MBRS, with a broad and approximately westward
flow of greater than 50 cm s-1 over the central MBRS. Most of this westward flow turns northward to
flow along the east coast of Mexico, and the rest veers cyclonically to form a cyclonic gyre over the
GOH. In addition, there are
relatively strong southward coastal currents over the inner Belize shelf
and convergent near-surface currents over the coastal waters between the Bay Islands and the north
coast of Honduras (see the inner model results in Figure 5a).
The sub-surface (75 m) circulation at day 298.5 produced by the outer model shows the impact of
the storm, particularly over the northwestern Colombian Basin and southern Cayman Basin, but to less
degree in comparison with the near-surface currents at the same time (Figure 5). The maximum
sub-surface currents at 75 m produced by the outer model are about 2 m s-1 over the northwestern
flank of Nicaragua Rise (Figure 5b). There is also a strong northwestward sub-surface flow of about
50 cm s-1 through the deep water region between Honduras Rise and Jamaica, which differs from the
subs-surface flow under the normal condition over this area. Over the central and southern MBRS, the
sub-surface circulation at this time is very similar to that at day 295.5, indicating that the
storm-generated energy does not yet reach the sub-surface circulation over this region.
Hurricane Mitch approached to the north coast of central Honduras and made landfall in the early
morning of October 29 (day 301) with the maximum sustained wind speeds of about 160 km h-1. The
nested-grid model results at day 301.0 (00:00 October 29) a few hours before landfall demonstrate that
the near-surface circulation in the western Caribbean Sea is significantly affected by Mitch (Figure 6a),
with intense near-surface currents of about 5 m s-1 over the coastal waters between the Bay Islands and
the north coast of Honduras, strong northward currents over the western Yucatan Basins and intense
northward through-flow over the western part of Yucatan Strait. The middle and inner models
generate much stronger divergent surface currents over the southern MBRS than the outer model, as
expected (Figure 6a). The middle model also produces strong westward near-surface currents of about
5 m s-1 over the central MBRS and strong southwestward jet over the Belize shelf. The nested-grid
model results also demonstrate the significant influence of Mitch on the sub-surface circulation at
depth of 75 m at 301.0 (Figure 6b), indicating the storm-generated energy penetrates to the deep ocean
and disturbs the sub-surface circulation behind the storm track over the southern MBRS and coastal
region off Yucatan Peninsula by this time.
The middle and inner models generate strong southward
sub-surface coastal currents over the Belize shelf and complicated sub-surface circulation features
over the coastal waters around the Bay Island and (Figure 6b).
After landfall Mitch went through the inland of Honduras and Guatemala and degraded to a tropical
depression on November 1 (day 304). The near-surface and sub-surface circulations produced by the
outer model at day 304.5 (12:00 November 1) still have strong inertial currents along the storm track
14
and adjacent areas, particularly over the right side of the track (Figure 7a). The middle and inner
models generate broad and approximately northwestward currents over the central MBRS, strong
eastward coastal currents around the Bay Islands and along the north coast of Honduras, and
stronger-than-usual currents flowing into Gulf of Mexico through the western part of Yucatan Strait
(Figure 7).
On of the important characteristics of the storm-induce circulation is inertial oscillations excited by
the storm behind the storm, which are most energetic to the right of the storm track [Greatbatch, 1983;
Sheng et al., 2006]. Figure 8 shows the time series of the eastward component of the surface currents
produced by the outer model along the storm track. (more discussion will be added later by Jinyu).
4.2 Simulated Sea Surface Temperature and a Cold Water Wake behind Mitch
The other important characteristic of the upper ocean response to a hurricane is the generation of a
cold water wake behind the storm. Previous studies of the storm-induced circulation in a flat bottom
case [Chang and Anthes, 1978; Price, 1981; Greatbatch, 1985] suggest that the cold water wake is
biased to the right of the storm track and strongly dependent on the hurricane translation speed, with
greater cooling for a slower moving storm. Figure 8 presents the near surface temperatures produced
by the nested-grid outer model in the control run from October 23 (day 295) to November 1 (day 301).
The near-surface temperature at 12:00 October 23 (day 295.5) produced by the outer model is spatially
uniform and about 28ºC over the western Caribbean Sea, except for three pools of cold surface waters
(Figure 8a). One cold water pool is located behind Mitch over the southern Colombian Basin and the
other two pools are located respectively over the Campeche Bank off northern Yucatan Peninsula and
coastal waters off northern Colombia. It should be noted that the near-surface temperature in the deep
water region of the western Caribbean Sea is about 28ºC in late October and early November under
the normal condition (without large disturbances due to the extreme and sporadic events) [Sheng and
Tang, 2003]. The cold water pool behind Mitch and the rightward bias of the SST cooling to the storm
track are due mainly to the intense vertical mixing associated with Mitch, of which the translational
speed was slow and about 8 km h-1 on average from the noon of October 22 to the evening of October
24. As demonstrated in Sheng et al. [2006], vertical mixing plays a dominant role in the storm-induced
SST changes including the rightward bias behind a storm, while (horizontal and vertical) advection
terms only play a very minor role. In comparison, the cold water pools over the Campeche Bank and
coastal waters off northern Colombia are mainly associated with the intense coastal upwelling [Sheng
and Tang, 2003].
Mitch moved rapidly northward and then northwestward on October 25 and reached Honduras Rise
on October 26 with a translational speed of about 15 km h-1. The storm intensity increased from a
15
category-3 to category-34 hurricane during this 2-day period. The most interesting feature of the
near-surface temperature at day 298.5 (12:00 October 26) produced by the outer model is a narrow
strip of SST (or near-surface) cooling which is more intense to the right of the storm track behind the
storm on the Colombian Basin and northern flank of Honduras Rise. The cold water wake behind the
storm also has significant spatial variability, with relatively stronger SST cooling over the southern
Colombian Basin and relatively weaker SST cooling over the northern Colombian and outer flank of
Honduras Rise, which is resulted mainly from the highly variable translational speed of the storm.
Mitch slowed down from October 28 to October 30, with a slow translational speed of less than 5
km h-1. In addition to the strip of SST cooling over the Colombian Basin and Honduras Rise, the
nested-grid outer model generates a new pool of intense SST cooling in the southern MBSR at 12:00
October 29 (day 301.5), with the coldest near-surface waters of about 20oC located over the coastal
water between the Bay Islands and north coast of Honduras (Figure 8c).
At day 304.5 (12:00 November 1), which was more than 3 days after Mitch made landfall, the
model-calculated results still show significant SST cooling effects in the western Caribbean Sea, with
the strong SST cooling of about a few degrees over the central Colombian Basin and the southern
MBRS. The SST cooling over the Campeche Bank is again resulted from the local upwelling as
mentioned earlier.
To further demonstrate that upper ocean response to Hurricane Mitch, we calculate differences in
the near-surface temperature and currents between the model results in Exp-Control and Exp-Norm
and refer the differences as to the storm-induced SST cooling and currents (Figure 9). The
storm-induced near-surface currents at day 295.5 are horizontally divergent under the storm, with a
maximum speed of greater than 1 m s-1 (Figure 9a). Behind the storm, there is a cold water pool,
which is biased to the right of the storm track. Outside the area influenced by the storm, the
storm-induced currents and SST cooling are very small (Figure 9a). At day 298.5, the storm-induced
near-surface currents produced by the outer model are very strong with a maximum value of greater
than 5 m s-1 under the direct influence of the storm (Figure 9b). There are strong inertial oscillations
and a cold water wake in the vicinity of the storm track, both of which are more intense on the right
than on the left of the storm track. As discussed in the previous studies [Chang and Anthes, 1978;
Price, 1981; Greatbatch, 1985; Sheng et al. 2006], the rightward bias of the inertial currents and SST
cooling behind the storm can be explained largely by the fact that a more efficient energy transfer
from the storm to the ocean on the right side of the storm track than that on the left side of the storm
track. This is because the wind stress veers anticyclonically at a fixed point on the right side of the
storm track as the storm passes by, while the wind stress veers cyclonically on the left side of the
storm track. The Coriolis force (on the north hemisphere) turns the ocean currents in the same
16
direction as the wind stress on the right side of the storm track, leading to an efficient transfer of
energy from the storm to the ocean currents. By contrast, on the left side of the storm track, the ocean
currents are turned in the opposite direction to the wind stress, resulting in weak currents. In addition,
water parcels on the right side of the storm are accelerated by the wind forcing for a longer time than
those on the left side of the storm. The rightward bias of the intense inertial currents behind the storm
leads to stronger entrainment and mixing on the right side of the storm track, which, in turn, is mainly
responsible for the rightward bias of SST cooling.
At day 301.5, the outer model produces a large size pool of SST cooling in the southern MBRS,
with the maximum SST cooling of about 10oC over the coastal region around the Bay Islands, and
weaker SST cooling over the northern flank of Honduras Rise and central Colombian Basin. The
storm-induced near-surface inertial currents are relatively strong and widespread over the
northwestern Caribbean Sea and in the vicinity of the storm track over the central Colombian Basin. It
is interesting to note that part of the inertial energy excited over the northern flank of Honduras Rise
propagates southward along the east coast of Honduras and reaches the southwestern Colombian Basin
at day 301.5. At day 304.5, the near-surface cooling and inertial currents dissipate and spread to other
region of the western Caribbean Sea.
4.3. Simulated salinity and river plumes
We assess the model performance by comparing the simulated sea surface salinity (SSS) in the
control run with SSS derived from the SeaWiFS ocean color data.
The SeaWiFS remote sensing
images showed a large-scale river plume extending from the Honduras coast into the deep ocean
during Hurricane Mitch and indicated the important hydrodynamic connection between coral reefs and
land after Hurricane Mitch (Figure 2). Figure 10 shows the simulated river plume produced by the
nested-grid outer model in the control run, as compared with those from the SeaWiFS observations.
The low salinity waters of less than 35.5 psu along the coast are well captured by the model. Indeed,
the near-surface salinity measured by an oceanographic buoy in one of the Belize atolls dropped to 25
psu after Mitch (Bjorn Kjerfve, personal comm., 2006). On November 1 (day 304) the outer model
produces the low-salinity waters on the southern MBRS, from the Honduras coast of 85ºW to about
18.5ºN. In agreement with SeaWiFS images (Figure 10a), the model also generates eastward spreading
of the low salinity water along the Honduras coast from the Gulf of Honduras. On November 3 (day
306) the modeled plume expands further to 19ºN and turns westward with the Caribbean
through-flows (Figure 10d). After Mitch on 14 November (day 317, Fig. 10f), the model showed
another plume produced by the model. This new plume, characterized by salinity below 35.5 psu,
extends over 17ºN on the southern MBRS, and then moves northwestward with the Caribbean
17
through-flows to the Belize Shelf.
The simulated plume patterns defined by low salinity values are visually comparable with those
obtained from SeaWiFS observations. However, the coarse resolution bathymetry [2-minute grid], the
idealized parameterization of the sub-grid mixing, and the input salinity values could all affect the
model performance. Figure 11 shows the difference between Exp-Low and Exp-High, where different
coastal salinities were forced. Although the general patterns are similar, the plumes in Exp-Low grow
larger and further north, with lower salinity near the coast. The results further demonstrate that
although the general plume patterns can be modeled reasonably well under the current
parameterizations, the exact size of the plume as well as the absolute salinity values within the plume
strongly depend on the salinity input, whose accuracy can only be assured through systematic
measurements.
5. Summary and Discussion
The combination of the SeaWifS remote sensing data and a triply nested-grid ocean circulation
modeling system was used to study the upper ocean response of the Meso-American Barrier Reef
System to Hurricane Mitch in late October 1998. The nested-grid system has three subcomponents:
fine-resolution inner model (2 km), intermediate-resolution middle model (6 km), and
coarse-resolution outer model (19 km). The nested system used the two-way nesting technique based
on the smoothed semi-prognostic method and was forced by 6-hourly NCEP/NCAR wind, a
parameterized vortex associated with Hurricane Mitch, monthly mean heat and freshwater fluxes, and
buoyancy forcing associated with river runoff at the coast. The nested-grid system generated intense
divergent currents forced by the local wind under the storm, intense inertial currents and sea surface
temperature (SST) cooling behind the storm, and bias of the inertial currents and SST cooling to the
right of the storm track.
A novel attempt was made to simulate the SeaWiFS-observed low-salinity plume after Mitch.
Based on SeaWiFS imagery where the sea surface salinity (SSS) was derived empirically by assuming
the inverse relationship between SSS and colored dissolved organic matter, low salinity was specified
in the nested-grid system at twelve river mouths along the Honduras coast and Belize shelf and was
allowed to vary with time and depth in its value, corresponding to anomalous coastal precipitation and
flooding due to Mitch. The model produced comparable plume patterns to SeaWiFS observations in
both space and time. However, the simulated magnitude of the low salinity values is less satisfactory,
possibly due to the coarse-resolution bathymetry (2-minute resolution), idealized sub-grid scale
mixing parameterization and relatively arbitrary low-salinity input along the Honduras coast and
Belize Shelf. The sensitivity study demonstrated different plume scales with different model
18
parameterizations (Figure 11).
The results presented in this study demonstrate the advantage of an integrated approach (numerical
modeling and remote sensing) to assess environmental changes along a coastal system. From an
oceanography standpoint the approach is satisfactory since hurricane-induced patterns seem correctly
reproduced for temperature and currents. We lack here in situ measurements during hurricane
conditions, which are difficult to obtain without permanent buoys, but we compensated by using
SeaWiFS observations before and after Mitch to help parameterizing the coastal boundary (salinity
input at river mouths) and assess the realities of the model predictions (qualitative comparisons of
river plumes extent).
Our ultimate goals are to refine the predictions of connectivity patterns within the matrix of reefs
present along the Meso-American Reef System. The demonstration that the nested model is able to
reproduce the circulation patterns visible on remote sensing images is a strong motivation to continue
the modeling approach since it appears reliable. Modelling will help in defining connectivity in normal
conditions (studied with climatology data) but also during short term events, like Hurricane Mitch, and
this is a new finding. Like Cowen et al. [2005] pointed out “Ecological connections may be extended
in some situations by rare, extreme dispersal events where unusually large numbers of larvae are
exported to distant locations. When such events occur frequently enough (in terms of the demographic
longevity of a species), populations may be sustained (i.e., storage effect). Extending longevity in a
species has been suggested as a means of capturing dispersal related variability in flow events”. The
study we conducted here is the first step towards being able to include these extreme dispersal events
into the growing reef connectivity work.
Future work will now consist in computing connectivity matrices [Cowen et al., 2005] during
normal [Tang et al., 2006] and extreme situations and compare the patterns. Several improvements
need to be considered for this. First, better representation of shallow reefs topography and rugosity are
required for the reef themselves. Second, coastal salinity need to be calibrated with in situ
measurements along the Honduras abd Belizea coasts. Third, more scenario need to be identified from
remote sensing imagery, in order to identify coastal patterns that can be used to test the models output.
These patterns can be catastrophic or not. Such model-images comparisons are qualitative here, but
could be better quantified with some morphometric indicators. All these developments are in progress
and will be reported in a near future.
Acknowledgments
We wish to thank Chris Fogarty and Liqun Tang for their useful suggestions. This project is supported
by the National Aeronautics and Space Administration (NASA).
19
References
WILL NEED MORE CONNECTIVITY REFERENCES
Andréfouët, S., P. J. Mumby, M. McField, C. Hu and F. E. Muller-Karger, Revisiting coral reef
connectivity, Coral Reefs, 21, 43-48, 2002.
Barnier, B., L. Siefridt, and P. Marchesiello, Thermal forcing for a global ocean circulation model
using a three-year climatology for ECMWF analyses, J. Mar. Sys., 6, 363-380, 1995.
Chang, S. W., and R. A. Anthes, Numerical simulations of the ocean’s nonlinear baroclinic response
to translating hurricanes, J. Phys. Oceanogr., 8, 468-480, 1978.
Cornillon, P., L. Stramma, and J. F. Price, Satellite Measurement of Sea Surface Cooling during
Hurricane Gloria. Nature, 326 (611), 373-375, 1987.
Craig A. K., Geography of fishing in British Honduras and adjacent coastal waters. Louisiana State
University Press, 143 pp., 1966.
da Silva, A. M., Young, C. C., and Levitus, S., Atlas of surface marine data 1994, Volume 3,
Anomalies of heat and momentum fluxes. NOAA Atlas NESDIS 8, 413 pp. NOAA, Washington,
DC, 1994.
Dickey, T., and coauthors, Upper-ocean temperature response to hurricane Felix as measured by
Bermuda tested mooring, Monthly Weather Rev., 126, 1195-1201, 1992.
Dietrich, D. E., Application of a modified Arakawa ’a’ grid ocean model having reduced numerical
dispersion to the Gulf of Mexico circulation, Dyn. Atmos. and Oceans, 27, 201-217, 1997.
D’Sa, E. J., C. Hu, F. E. Muller-Karger, and K. L. Carder, Estimation of colored dissolved organic
matter and salinity fields in case 2 waters using SeaWiFS: Examples from Florida Bay and
Florida Shelf. In Proc. Indian Acad. Sci. (Earth Planet Sci.), 111, September 2002, pp. 197-207,
2002.
Ezer, T., L-Y Oey, and H-C Lee, The variability of currents in the Yucatan Channel: analysis of
results from a numerical model, J. Geophys. Res., 108, 3012, 2003.
Ezer, T., D. V. Thattai, B. Kjerfve, and W. D. Heyman, On the variability of the flow along the
Meso-American Barrier Reef System: a numerical model study of the influence of the Caribbean
current and eddies, Ocean Dyn., DOI 10.1007/s10236-005-0033-2, 2005.
Fedorov, K. N., A. A. Varfolomeev, A. I. Ginzburg, A. G. Zatsepin, A. Yu. Krasnopevtsev, A. G.
Ostrovsky, and V. E. Skylarov, Thermal reaction of the ocean on the passage of hurricane Ella,
Okeanologiya, 19, 992-1001, 1979.
Ferrari, G. M., and M. D. Dowell, CDOM absorption characteristics with relation to fluorescence
and salinity in coastal areas of the Southern Baltic Sea. Estuarine, Coastal and Shelf Science,
47, 91-105, 1998.
Fratantoni D.M., North Atlantic surface circulation during the 1990s observed with satellite-tracked
drifters, J. Geophys. Res., 106, 22067-22093, 2001.
Geisler, J. E., Linear theory of the response of a moving hurricane, Geophysical Fluid Dynamics, 1,
249-272, 1970.
20
Geshelin, Y., J. Sheng, and R. J. Greatbatch, Monthly mean climatologies of temperature and salinity
in the western North Atlantic, Can. Tech. Rep. Hydrogr. Ocean. Sci. 153, 62 pp., 1999.
Greatbatch, R. J., On the response of the ocean to a moving storm: The nonlinear dynamics, schemes,
J. Geophys. Res., 13, 357-367, 1983.
Hu et al., 2003.
Hu, C., E. T. Montgomery, R. W. Schmitt and F. E. Muller-Karger, The dispersal of the Amazon and
Orinoco River water in the tropic Atlantic and Caribbean Sea: Observation from space and
S-PALACE floats, Deep-Sea Res. II, 51, 1151-1171, 2004.
Hu, C., J. R. Nelson, E. Johns, Z. Chen, R. H. Weisberg and F. E. Muller-Karger, Mississippi River
water in the Florida Straits and in the Gulf Stream off Georgia in summer 2004, Geophys. Res.
Lett., 32, L14604, doi:10.1029/2005GL022942, 2005.
Ishizaka, J. Coupling of coastal zone color Scanner data to a physical-biological model of the
southeastern U.S. continental shelf ecosystem 1. CZCS data description and Lagrangian particle
tracing experiments, J. Geophys. Res., 95, 20167-20181, 1990.
Jordan, C.L., On the influence of tropical cyclones on the sea surface temperature. Proc. Symp.
Tropical Meteoro, Wellington, New Zealand, 614-622, 1964.
Kalnay, E., and 20 others, The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteor. Soc., 77,
437-472, 1996.
Kramer, P. A., and P. R. Kramer, Ecoregional conservation planning for the Mesoamerican
Caribbean Reef, 140 pp., 2002.
Large, W. G., and S. Pond, Open ocean momentum flux measurements in moderate to strong winds, J.
Phys. Oceanogr., 11 , 324-336, 1981.
Lee et al., 2002.
Legrand et al., 2006.
Lu, Y., K. R. Thompson, and D. G. Wright, Tidal currents and mixing in the Gulf of St. Lawrence: An
application of the incremental approach to data assimilation, Can. J. Fish. Aquat. Sci., 58,
723-735, 2001.
Marchesillo, P., J. C. McWilliams, and A. Shchepetkin, Open boundary conditions for long-term
integration of regional oceanic models, schemes, J. Ocean Modelling, 3, 1-20, 2001.
McClain, C. R., G. C. Feldman, and S. B. Hooker, An overview of the SeaWiFS project and strategies
for producing a climate research quality global ocean bio-optical time series. Deep-Sea Res. II,
51, 5-42, 2004.
Mooers, C. N. K., and G. A. Maul, Intra-Americas Sea circulation, coastal segment (3,W), The Sea, 11,
John Wiley and Sons, 183-208, 1998
Murphy, S. J., H. E. Hurburt, and J. J. O’Brien, The connectivity of eddy variability in the Caribbean
Sea, the Gulf of Mexico, and the Atlantic Ocean, J. Geophys. Res., 104, 1431-1453, 1999.
Nelson, N. B., The wake of hurricane Felix, Int. J. Remote Sensing, 17, 2893-2895, 1996.
21
Oey, L-Y, T. Ezer, and H-C Lee, Loop current, rings and related circulation in the Gulf of Mexico: a
review of numerical models and future challenges, Circulation in the Gulf of Mexico:
Observations and Models, edited by W. Sturges and A. L. Fernandez, American Geophysical
Union Washington, DC, 31-56, 2005.
Oey, L-Y, T. Ezer, D. Wang, and S. Fan, Hurricane-induced motion and interaction with ocean
currents, Cont. Shelf. Res., in press, 2006.
O’Reilly et al., ???
Powell, M. D., P. J. Vickery, and T. A. Reinhold, Reduced drag coefficient for high wind speeds in
tropical cyclones, schemes, Nature, 422, 279-283, 2003.
Price, J. F., Upper ocean response to a hurricane, schemes, J. Phys. Oceanogr., 11, 153-175, 1981.
Pudov, V. D., A. A. Varfolomeev, and K. N. Fedorov, Vertical structure of the wake of a typhoon in
the upper ocean. Okeanologiya, 21, 142-146, 1979.
Sheng, J., Dynamics of a buoyancy-driven coastal jet: The Gaspe Current, J. Phys. Oceanogr., 31,
3146-3163, 2001.
Sheng, J., and Y. R. Rao, Circulation and thermal structure in Lake Huron and Georgian Bay:
Application of a nested-grid hydrodynamic model, Cont. Shelf Res., in press, 2006.
Sheng, J., and L. Tang, A numerical study of circulation in the western Caribbean Sea, J. Phys.
Oceanogr., 31, 3146-3163, 2003.
Sheng, J., and L. Tang, A two-way nested-grid ocean-circulation model for the Meso-American
Barrier Reef System, Ocean Dyn., 54, 232-242, 2004.
Sheng, J., and L. Wang, Numerical study of tidal circulation and nonlinear dynamics in Lunenburg
Bay, Nova Scotia, J. Geophys. Res., 109, C10018, doi:10.1029/2004JC002404, 2004.
Sheng, J., R. J. Greatbatch, X. Zhai, and L. Tang, A new two-way nesting technique based on the
smoothed semi-prognostic method, Ocean Dyn., 55, doi:10.101007/sl0236-005-0005-6, 162-177,
2005.
Sheng, J., R. J. Greatbatch, and D. G. Wright, Improving the utility of ocean circulation models
through adjustment of the momentum balance, J. Geophys. Res., 106, 16,711-16728, 2001.
Sheng, J., D. G. Wright. R. J. Greatbatch, and D. E. Dietrich, CANDIE: A new version of the
DieCAST ocean circulation model, J. Atmos. and Ocean. tech., 15, 1414-1432, 1998.
Sheng, J., X. Zhai, and R. J. Greatbatch, Numerical study of the storm-induced circulation on the
Scotian Shelf during Hurricane Juan using a nested-grid ocean model, Prog. in Oceanogr. in
press, 2006.
Slordal, L. H., E. A. Martinsen, E. A., and A. F. Blumberg, Modelling the response of an idealized
coastal ocean to a traveling storm and to flow over bottom topography, J. Phys. Oceanogr., 24,
1689-1705, 1994.
Smagorinsky, J., General circulation experiments with the primitive equation. I. The basic experiment,
Mon. Wea. Rev., 21, 99-165, 1963..
22
Smith, N. P., Response of Florida Atlantic Shelf Waters to hurricane David, J. Geophys. Res., 87,
2007-2016, 1982.
Sou, T., G. Hollaway, and M. Eby, Effects of topographic stress on Caribbean Sea circulation, J.
Geophys. Res., 101, 16449- 16453, 1996.
Tang, L., J. Sheng, B. G. Hatcher, and P. F. Sale, Numerical study of circulation, dispersion and
hydrodynamic connectivity of surface waters on the Belize shelf, schemes, J. Geophys. Res., 111,
C01003, doi:10.1029/2005JC002930, 2006.
Thuburn, J., Multidimensional flux-limited advection schemes, J. Comput. Phys., 123, 74-83, 1996.
Wang, L., J. Sheng, A. E. Hay, and D. J. Schillinger, Storm-induced circulation in Lunenburg Bay of
Nova Scotia: Observations and numerical simulations, J. Phys. Oceanogr., in press, 2006.
Wright, R.,Temperature structure across the Kuroshio before and after typhoon Shirley. Tellus, 21,
409-413, 1969.
23
Table 1. List of four numerical experiments forced by the different combination of the NCEP/NCAR
wind stress (NECP), a parameterized vortex associated with Mitch (Vortex), monthly mean heat and
freshwater fluxes (MF), and specification of low salinity at 12 river months (River).
Name of run
External forcing
Salinity at river months (psu)
Exp-Control
NCEP+Vortex+MF+River
25
Exp-Low
NCEP+Vortex+MF+River
20
Exp-High
NCEP+Vortex+MF+River
28
Exp-Norm
NCEP+MF
NA
24
Figure 1. Topographic map of the Gulf of Mexico and Caribbean Sea based on the 2-miniute gridded
global relief data (ETOPO2), and the storm track (red line) of Hurricane Mitch from October 22 to
November 6, 1998. The storm symbol along the storm track denotes the beginning location of the
storm center on each day. Abbreviations are used for the Meso-American Barrier Reef System
(MBRS), Yucatan Strait (YS), Gulf of Honduras (GOH), Nicaragua Rise (NR), Republic Dominion
(RD) and Windward Passage (WP). The contours in the lower left panel are labeled in meters.
25
Figure 2: Spatial patterns of turbid coastal water plumes on the Meso-American Barrier Reef System
(MBRS) derived from SeaWiFS remote sensing data during and after Hurricane Mitch [Andréfouët et
al., 2002]. The images were processed for cloud masking and bio-optical products using the SeaWiFS
processing package distributed by NASA (SeaDAS V3.2), where chlorophyll-a concentration was
estimated using the OC2 algorithm of O'Reilly et al. Clouds and land are masked as purple and grey
colors respectively. (a) Typical dry season conditions showing clear ocean and narrow zones of
turbility near the river mouths. (b) First high-quality image 3 days after landfall of Mitch showing a
large-scale plume that covered most of the Bay Islands and extended to 200 km from its origin. (c)
The coastal water plume extended further northward to reach Glovers atoll on the Belize shelf.
The plume dissipated by dilution.
(d)
26
Figure 3: Selected bathymetry features for the triply nested-grid modeling system consisting of (a) an
outer model covering western Caribbean Sea (WCS); (b) a middle model including the southern
meso-American Barrier Reef System (MBRS); and an inner model zooming in the north coast of
Honduras and Bay Islands. Abbreviations are used for the Meso-American Barrier Reef System
(MBRS), Yucatan Strait (YS), and
meters.
Gulf of Honduras (GOH). Contours are labeled in units of
27
Figure 4: Simulated currents at (a) 1 m and (b) 75 m at day 295.5 (12:00, October 23) of 1998 when
Hurricane Mitch was upgraded quickly from a tropic depression to a hurricane with the sustained wind
speeds of about 95 km h-1 in the southern Caribbean Sea. The red line represents the storm track and
solid green circle represents the location of the storm center at this time. Velocity vectors are plotted at
every second model grid point.
28
Figure 5: Simulated currents at (a) 1 m and (b) 75 m at day 298.5 (12:00 October 26) of 1998 when
Mitch was strengthened significantly with the maximum sustained wind speeds of about 290 km h-1 in
the Nicaragua Rise of the western Caribbean Sea. The red line represents the storm track and solid
green circle represents the location of the storm center at this time. Velocity vectors are plotted at
every second model grid point.
29
Figure 6: Simulated currents at (a) 1 m and (b) 75 m at day 301.0 (0:00 October 29) right before Mitch
made landfall at the north Honduras coast with the sustained wind speeds of 205 km h-1. The red line
represents the storm track and solid green circle represents the location of the storm center at this time.
Velocity vectors are plotted at every second model grid point.
30
Figure 7: Simulated currents at (a) 1 m and (b) 75 m atday 304.5 (12:00 November 1) of 1998 when
Mitch moved through southwest Nicaragua and decreased to a tropical depression. The red line
represents the storm track. Velocity vectors are plotted at every second model grid point.
31
Figure 8: Simulated sea surface temperature (SST) associated with Hurricane Mitch at different times
produced by the outer model. Contour intervals are 1oC. The red dashed line represents the storm track
and the storm symbol shows the position of the storm center.
32
Figure 9: Simulated changes in sea surface temperature (SST) and currents associated with Hurricane
Mitch at different times produced by the outer model. Contour intervals are 2oC. The red lin represents
the storm track and the storm symbol represents the location of the storm center.
33
Figure 10: Comparison of spatial patterns of river plume characterized by sea surface salinity between
the remote sensing data processed through SeaWiFS data (a,c,e) and the outer model results (b,d,f)
during Hurricane Mitch. Clouds are masked as purple color in a,c,e.
34
Figure 11: Model-simulated spatial patterns of river plumes during Hurricane Mitch with the different
parameterization of lower salinity. The minimum salinity value is set to 20 psu in Exp-Slow (a,c,e)
and 28 psu in Exp-High (b,d,f). Other model parameterization is same as in the control run.