RESEARCH ARTICLE Bumpy Topographic Effects on the Transbasin Evolution

10.1029/2018JC014837
of Large‐Amplitude Internal Solitary Wave
Special Section: in the Northern South China Sea
Recent Progresses in
Oceanography and Air‐Sea Jieshuo Xie1 , Yinghui He1 , and Shuqun Cai1,2,3
Interactions in Southeast
Asian Archipelago 1
State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences,
Guangzhou, China, 2University of Chinese Academy of Sciences, Beijing, China, 3Institution of South China Sea Ecology
Key Points: and Environmental Engineering, Chinese Academy of Sciences, Guangzhou, China
• Bumpy continental slope/shelf
topographic effects on transbasin
ISW in northern SCS are studied Abstract The bumpy continental slope/shelf topography is a quite common feature in the northern
using fully nonlinear,
South China Sea (SCS), yet its effect on the shoaling internal solitary waves (ISWs) remains poorly
nonhydrostatic model
• Transformation and energy decay understood. Therefore, numerical simulations by a fully nonlinear, nonhydrostatic model are carried out to
and partition of transbasin ISW vary explore the bumpy continental slope/shelf topographic effects on the transbasin evolution of large‐
greatly due to complexity of realistic
amplitude ISW in the northern SCS. It is found that the prominent bumps over both continental slope and
bumpy topography in northern SCS
• Both onshore‐ and shelf regions play significant roles in modulating the evolution of transbasin ISW in the northern SCS.
offshore‐propagating multimodal The bump over the continental slope is capable of triggering a solitary‐like mode‐2 internal wave packet,
waves are triggered and can be
while the bump over the continental shelf can result in three wave groups, including a leading group of
clarified by beam scattering and
local generation mechanism rank‐ordered mode‐1 ISW packet and two following groups of non‐rank‐ordered mode‐1 ISW packet and
mode‐2 internal waves. The bumps can cause a peak‐to‐peak difference of the energy decay rate of ISW up to
10–20 kW/m over continental slope region and 3–5 kW/m over continental shelf region. The wave kinetic
Correspondence to:
energy (KE) is found to exceed the available potential energy (APE) by as much as 50% over the continental
S. Cai, shelf break region. Over the shelf region, however, the bumps can first make the KE drop to as low as
[email protected] only 80% of the APE, but later the KE might bounce back to approximately 1.1–1.2 times of the APE. Both
onshore‐ and offshore‐propagating beam‐like disturbances are found to be excited by the bumps. Except
Citation: for the onshore‐propagating mode‐2 ISW packet, the reflected offshore‐propagating waves in different
Xie, J., He, Y., & Cai, S. (2019). Bumpy internal modes are also formed. These onshore‐ and offshore‐propagating multimodal internal waves can be
topographic effects on the transbasin
evolution of large‐amplitude internal
clarified by the beam scattering and local generation mechanism.
solitary wave in the northern South
China Sea. Journal of Geophysical
Research: Oceans, 124, 4677–4695. 1. Introduction
https://doi.org/10.1029/2018JC014837
The strongest internal solitary waves (ISWs) documented in the global oceans occur in the northern South
Received 4 DEC 2018 China Sea (SCS; Alford et al., 2015). Because of the associated strong currents and tremendous displacement
Accepted 10 JUN 2019
of ocean isopycnal layers, these waves pose hazards to the underwater navigation and offshore drilling (Cai
Accepted article online 22 JUN 2019
Published online 9 JUL 2019 et al., 2008; Osborne et al., 1978; Xie et al., 2011) and supply nutrients from the deep water for coral reefs and
pilot whales that forage in their wakes (Moore & Lien, 2007; Wang et al., 2007). The associated energetics of
these ISWs is also an important problem in the ocean interior that has received considerable attention
because of its important energy source for mixing (Chang et al., 2006; Jeans & Sherwin, 2001; MacKinnon
et al., 2017).
The northern SCS has been viewed as an ideal natural laboratory for studying the stages and ultimate fate of
the large‐amplitude ISWs in the ocean (Farmer et al., 2011; Li & Farmer, 2011). In the last two decades,
numerous studies use satellite and in situ observations or numerical and theoretic modelings to investigate
the behavior of ISWs in the northern SCS (Cai et al., 2012; Guo & Chen, 2014; Zhao et al., 2014; Zheng et al.,
2007) and a coherent picture including the generation, propagation, and ultimate fate of these waves can be
drawn (Alford et al., 2015). The typical ISWs here generally propagate westward from the generation source
sites at the Luzon Strait, through the deep basin, and onto the continental slope/shelf regions (Alford et al.,
2010). In situ observations (Huang et al., 2016; Klymak et al., 2006; Ramp et al., 2010) show that in the deep
basin region their downward displacement reaches more than 200 m, their phase speed can be over 3 m/s,
and their wave‐induced horizontal and vertical velocities reach up to 2.5 and 0.7 m/s, respectively. These
©2019. American Geophysical Union.
observed ISWs are among the largest measured anywhere, surpassing the measured displacement of 90 m
All Rights Reserved. and phase speed of 1.8 m/s in the Sulu and Andaman seas (Apel et al., 1985; Osborne & Burch, 1980).

XIE ET AL. 4677

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Moreover, in the deep basin region the wave energy reaches up to 1.8 GJ/m, and the wave kinetic energy
(KE) exceeds the available potential energy (APE) by about 40% (Klymak et al., 2006).

Before arriving at the middepth (i.e., ~1.5 km) of continental slope/shelf topographies, both in situ and satel-
lite observations (Alford et al., 2010; Huang et al., 2008) indicate that the detected waves are generally of the
mode‐1 type and are initially solitary. After propagating away from the middepth, however, the trailing
internal wave packets began to be formed more easily, and over the continental shelf numerous shoaling
mode‐1 ISW trains could be observed (e.g., see observations of the satellite (Jackson, 2009; Huang et al.,
2008; Zhao et al., 2004) and the Asian Seas International Acoustic Experiment (Duda et al., 2004; Lynch
et al., 2004; Orr & Mignerey, 2003; Ramp et al., 2004). Ramp et al. (2004) showed that mode‐1 ISWs could
be largely symmetric at the water depth of ~350 m and were significantly deformed with a gently sloping
front and a relatively steeper rear when reaching ~200‐m depth. Further, when near the depth of ~130 m,
overturning with the conversion of depression ISWs to elevation ISWs might happen (Orr & Mignerey,
2003). Over the shelf region, the energies of these waves were typically of about 0.3 GJ/m (Alford et al.,
2010). The in situ observations by Chang et al. (2006) suggested that these ISWs dissipated nearly all of their
energy across the Dongsha plateau before reaching the continental shelf shallower than ~100 m. During the
Asian Seas International Acoustic Experiment, one solitary‐like mode‐2 internal wave train was also
detected near the continental shelf break at ~426‐m depth, as reported by Yang et al. (2004), while it is much
less common than the solitary‐like mode‐1 internal wave. At the depth of 350 m in a nearby region, Yang
et al. (2009) further reported an in situ observation of more than 80 mode‐2 ISWs.

The polarity conversion of mode‐1 ISW in the northern SCS can be well demonstrated by adopting the
Korteweg‐de Vries (KdV)‐type equations (Cai et al., 2002; Liu et al., 1998; Orr & Mignerey, 2003). A hindcast
numerical simulation by Shen et al. (2009) reproduced the nonlinear transformation of ISWs from symmetri-
cal to severely deformed waveform as observed, and they showed that the APE might begin to exceed the KE
at the depth around 250–350 m. The recent theoretical and/or numerical simulations (e.g., Grimshaw et al.,
2014; Lamb & Warn‐Varnas, 2015; Liao et al., 2014; Xie, He, et al., 2015; Xie et al., 2016) further show that the
shoaling dynamics of mode‐1 ISWs in the northern SCS are quite sensitive to the topography, rotation, back-
ground stratification and current, initial water depth, and so forth. For instance, due to the combined effect of
rotation and topography, Grimshaw et al. (2014) show the formation of secondary trailing ISW packet asso-
ciated with the radiated inertia‐gravity waves from the leading mode‐1 ISW. Lamb and Warn‐Varnas (2015)
show that the shoaling ISWs initialized at depths of 1 and 3 km, respectively, exhibit significant differences
and thus suggest the importance of including the transbasin evolution in simulating the shoaling dynamics
of ISWs in the northern SCS. Xie et al. (2016) reported the distortion and broadening of long ISW front
observed by the synthetic aperture radar image and showed that it was due to the effects of strong back-
ground mesoscale current and variable bottom topography, respectively, in the deep basin region.

The background environmental coefficients of the KdV‐type equations were also used to illustrate the mode‐
2 ISWs observed by both in situ and satellite observations (Dong et al., 2016; Yang et al., 2009), and it is
shown that the variability of background stratification and/or current increases the relative importance of
wave nonlinearity to wave dispersion of mode‐2 waves and thus favors the evolution of mode‐2 ISWs.
Theoretical analysis and numerical simulations are also performed to investigate the generation (Lamb &
Warn‐Varnas, 2015; Liu et al., 2013; Xie, Pan, et al., 2015) or shoaling dynamics (Guo & Chen, 2012) of
mode‐2 ISWs in the northern SCS. Using a modal decomposition theory from deep water to shallow water,
Liu et al. (2013) propose that the generation of mode‐2 ISWs on the shelf region is due to the disintegration of
shoaling mode‐1 solitons from the deep ocean, a mechanism similar to that proposed by Helfrich and
Melville (1986). Lamb and Warn‐Varnas (2015) show that the interaction of a shoaling mode‐1 ISW with
a 200‐m‐high bump at ~700 m over the continental slope can result in the generation of many mode‐2 inter-
nal waves. Xie, Pan, et al. (2015) show that the mode‐2 ISWs are excited more effectively when both the tidal
Froude number and percentage contribution to a tidal internal wave beam from mode‐2 waves are high
enough. Using the MITgcm, Guo and Chen (2012) simulate the shoaling dynamics of a large‐amplitude
mode‐2 ISW in the northern SCS and show that the mode‐2 ISW generally has no essential changes until
it gets close to the continental shelf break region.
Despite the above extensive and in‐depth studies in the northern SCS, the evolution processes, the charac-
teristics, and energy variations of this transbasin large‐amplitude ISW are still not well understood

XIE ET AL. 4678

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

because of the complexity of the associated continental slope/shelf topo-

graphy. Also, there is less knowledge about the mechanism of the new-
born secondary waves. Actually, Figure 1a indicates that, at both
continental slope and shelf regions, there are numerous bumps, with
characteristic width (a few kilometers) and height (tens to hundreds of
meters) comparable to those of the typical large‐amplitude mode‐1 ISWs
in the northern SCS. However, their impacts on the shoaling transbasin
large‐amplitude mode‐1 ISW are still unclear (Lamb, 2014). Moreover,
the comparability of the scales between bumps and ISWs suggests that
the shoaling ISW should experience a very sudden change due to the
bumps, and thus, the applicability of KdV‐type equations based on the
slowly varying assumption (El et al., 2007; Grimshaw, 1979; Helfrich &
Melville, 2006) to investigate the sudden response of ISW to bumps will
be problematic. Therefore, in this study, using a fully nonlinear, nonhy-
drostatic numerical model as described in section 2, we focus on the
effects of bumpy continental slope/shelf topographies on the transbasin
evolution of large‐amplitude mode‐1 ISW in the northern SCS. The evolu-
tion processes, characteristic and energy variations of ISWs, the roles of
bumps over both continental slope and shelf regions, and the mechanism
of newborn waves will be investigated in section 3. Finally, conclusions
are shown in section 4.

2. Model Description
Figure 1. (a) Ocean floor topography in the northern South China Sea The impacts of bumpy topographies at the continental slope/shelf regions
with numerous bumps at both the continental slope and shelf regions. on the shoaling transbasin evolution of large‐amplitude ISW can be stu-
(b) Sketch of model domain for the transbasin evolution of large‐amplitude died numerically based on the sketch in Figure 1b. The large‐amplitude
ISW from the deep basin region to the bumpy continental slope/shelf
regions in the northern South China Sea. ISW = internal solitary wave.
mode‐1 ISW is assumed to propagate westward from the deep basin region
to the continental slope/shelf regions where the bumps may exist. Next,
the governing equations, parameter choices, and experiment configura-
tions are described.
2.1. Governing Equations and Choice of Parameters
The solution of pressure Poisson equation is the primary driver of cost in most nonhydrostatic numerical
models. Here to circumvent this, the shoaling transbasin evolutions of large‐amplitude ISWs are simulated
by the following 2.5‐dimension stream function‐vorticity equations (Vlasenko et al., 2005; Xie, Pan, & et al.,
2015),
       
ξ t þ J ðξ; ψÞ−fvz ¼ ge
ρx =ρ0 þ AH ψxx xx þ AH ψxz xz þ AV ψzz zz þ AV ψxz xz ; (1)
     
e
ρt þ J ðe
ρ; ψÞ þ ρ0 =gN 2 ðzÞψx ¼ K H e
ρx x þ K V e
ρz z þ K V ρ0z z ; (2)

where J is the Jacobian operator with J(A,B) = AxBz − AzBx, g is the acceleration due to gravity, N(z) is the
buoyancy frequency, f = 5.21 × 10−5 s−1 represents the Coriolis parameter at 21°N in the northern SCS, v
is the horizontal velocity in y direction, ρ0 is the constant average of the stationary density ρ0(z), e
ρ represents
the density disturbance because of wave motion, ξ represents the vorticity, and ψ is the two‐dimensional
stream function with the horizontal and vertical velocities in x and z directions written as u = ψz and
w = − ψx, respectively. The coefficients AH and KH represent the horizontal eddy viscosities and diffusivities,
respectively, and are defined as (Stacey & Zedel, 1986)

AH ¼ βðΔx Þ2 jux ðx; z; t Þj þ β0 ;

(3)
K H ¼ μðΔx Þ2 jux ðx; z; t Þj þ μ0 :

The coefficients AV and KV represent the vertical eddy viscosities and diffusivities, respectively, and are
defined as (Pakanowski & Philander, 1981)

XIE ET AL. 4679

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

A0
AV ¼ þ Ab ;
ð1 þ αRiðx; z; t ÞÞp
(4)
AV
KV ¼ þ K b:
ð1 þ αRiðx; z; t ÞÞp

Here the parameters β0 = Ab = 10−4 m2/s and μ0 = Kb = 10−5 m2/s describe the constant background dissi-
pation, which refer to dissipation in the absence of ISWs and are small enough to allow ISWs to propagate
over a large distance without significant damping (Vlasenko & Alpers, 2005). Moreover, the parameters
β = μ = 1, A0 = 10−2 m2/s, α = 4, and p = 2 are adjustable, and Riðx; z; t Þ ¼ N 2 ðx; z; t Þ=u2z ðx; z; t Þ is the local
Richardson number. The parameterizations of (3) and (4) make the horizontal and vertical eddy viscosities
and diffusivities increase in regions with strong shear velocities, and the Ri in (4) is set as 0 where vertical
density inversion happens because of wave breaking.
Moreover, the vorticity ξ is expressed as
ξ ¼ ψxx þ ψzz ; (5)
and the horizontal velocity v in y direction is calculated by
   
vt þ J ðv; ψÞ þ f ψz ¼ AH vx x þ AV vz z : (6)

The boundary conditions at the free surface (z = 0) and the bottom of ocean floor topography (z = H(x)) are
given by

ψ ¼ 0; ξ ¼ 0; vn ¼ 0; e
ρn ¼ 0; (7)

and the vertical open boundary conditions at two far ends (i.e., at x = L1 and x = L2) of model regions are
given by

ψ ¼ 0; ξ ¼ 0; v ¼ 0; e
ρ ¼ 0: (8)

A single large‐amplitude mode‐1 ISW shoaling toward the continental slope/shelf regions is initialized at
the deep basin region (see Figure 1b). This initial ISW is adjusted from a KdV solution, as that performed
in Xie, He, et al. (2015); similar adjustments can be referred in Vlasenko et al. (2005). The governing
equations (1)–(8) are solved numerically after performing the following transformation:

0 0
x 2 ¼ x; z2 ¼ ∫z N ðsÞds=∫H ðxÞ N ðsÞds; (9)

where N(s) is the buoyancy frequency and s is a dummy vertical integration variable. This transformation is
similar to that of the vertical terrain‐following sigma coordinate system (i.e., σ = − z/H(x)), which maps the
vertical region from sea surface to ocean bottom onto the unit interval (−1, 0). However, the difference is
that the transformation 9 stretches the vertical grid not only in areas of variable bathymetry but also in layer
near the pycnocline (more detailed descriptions of the stretching procedure can be found in Figure 4.6 in
Vlasenko et al., 2005). An advantage of using this transformation is that the internal wave disturbance
and the complex behavior of internal wave beams at water depth with strong stratification (i.e., near the pyc-
noclines) can be resolved with a much higher vertical grid resolution. The transformed equations are solved
using the alternative direction implicit method at each time semistep (see Appendix A for details of the trans-
formed equations and solving method).
As suggested in Lamb and Warn‐Varnas (2015), it is important to have high enough resolution to adequately
resolve shoaling ISWs. Here a number of 100 nonuniform vertical grids are set according to the transforma-
tion equation (9), and then even at the deep basin region the pycnocline of the stratified water can be
resolved with a vertical resolution of at least 4.8–6 m. The horizontal grid spacing Δx is set as 25 m, which
produces at least 10 points per wavelength according to observations of Alford et al. (2010) and Yang et al.
(2009) for both the mode‐1 and mode‐2 ISWs in the northern SCS. The grid leptic ratio (see definition in
Vitousek & Fringer, 2011) is also small enough to ensure that the physical dispersion dominates over the
numerical dispersion. The time step is set as Δt = 2.5 s to satisfy the Courant‐Friedrichs‐Lewy condition.

XIE ET AL. 4680

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

2.2. Experiment Configuration and Setup

Two groups of experiments with realistic and idealized topographies are
configured. The bumpy topographic effects on transbasin evolution of
ISW in the northern SCS are studied based on the first group with realistic
topographies. The specific role of each bump and the mechanism of evo-
lution are examined based on the second group with
idealized topographies.
The first experimental study is performed with three realistic cross sec-
tions (Figure 2a), and the bottom topographies of these sections are line-
arly interpolated according to the 1‐min gridded bathymetry (ETOPO1;
Amante & Eakins, 2009). The three sections are located at north of the
Dongsha Islands and are close to the cross section studied in Alford
et al. (2010; see their Figure 1b). The three realistic topographies, denoted
as R1–R3 (Figure 2b), possess the principal features of a typical ocean floor
topography, including the continental shelf, the continental slope and
rise, and the deep‐sea plain basin (Talley et al., 2011); note that here the
water depth deeper than 3,250 m is flattened to be 3,250 m. Moreover,
many undulating bumps with different sizes are located at both continen-
tal shelf and slope regions of the ocean floor topography in the northern
SCS. At the continental shelf region (e.g., sections R1 and R2), the most
prominent bumps have a characteristic height h of ~150 m and a charac-
teristic width σ of ~5 km. At the continental slope region (e.g., sections R2
and R3), the most prominent bumps have a characteristic height h of
approximately 80–250 m, and they appear at depths of approximately 1
Figure 2. (a) The three lines from south to north denote three realistic cross‐ and 1.6 km, respectively. On the basis of the realistic topographies along
sections R1–R3, with topographies extracted from the ETOPO1 1‐arc minute sections R1–R3, we will study the bumpy topographic effects on the evolu-
gridded bathymetry in the northern South China Sea. (b) Bathymetries of tion processes, and characteristics and energy variations of ISW in the
the three realistic sections R1–R3 and the idealized smooth section I0. northern SCS (see section 3.1).
Principal features of ocean floor topography, including plain deep basin,
continental rise, and continental slope and shelf, are labeled in (b). The second experimental study is performed with four idealized topogra-
Prominent bumps over sections R1–R3 are denoted in (b) by arrows. Note phies. These idealized topographies are configured with the following
that the depth of realistic bathymetry deeper than 3,250 m is flattened to be expression of smooth topography,
3,250 m.
H 0 −h0
H smooth ðx Þ ¼ −H 0 ; (10)
1 þ exp½τ ðx−x 0 Þ
where H0 and h0 represent the depth of deep‐sea basin and continental shelf, respectively; τ represents the
steepness coefficient of continental slope; and x0 denotes the position of middepth of topography
(Figure 1b). In this study, the idealized smooth topography along section I0 (thick dashed line in Figure 2
b) is configured by setting H0 = 3,250 m, h0 = 300 m, τ = 4 × 10−5, and x0 = −270 km. The idealized smooth
topography along section I0 not only possesses the principal features of a typical ocean floor topography but
also is an overall good approximation to all of the three realistic topographies along sections R1–R3 (except
at the continental rise region, where the topographic impacts on the evolution of transbasin ISW are actually
negligible). Furthermore, the specific impacts of bumps over continental slope and/or shelf regions on evo-
lution of transbasin ISW are studied with the expression:
 
H bumpy ðx Þ ¼ H smooth þ ∑i hi exp −ðx−x i Þ2 =2σ i ; (11)

which represents the bumpy topographies configured by superimposing some Gaussian bumps with the
characteristic height hi and characteristic width σi over the smooth topography Hsmooth at the position xi
(Figure 1b). Here i (= 1, 2, …) represents the sequence number of bumps. On the basis of idealized topogra-
phies configured by expressions (10) and (11), we will examine the specific roles of bumps over continental
slope or shelf region and study the mechanism of newborn waves (see section 3.2).
The overall topographic characteristics of the two groups of experiments are summarized in Table 1. All
simulations in the two groups of experiments are carried out with a summer climatological stratification,

XIE ET AL. 4681

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Table 1 obtained by the WOA13 climatological temperature and salinity in sum-

Description of Topographic Characteristics of the Two Groups of Experiments mer in the northern SCS (Locarnini et al., 2013; see Figures 3a and 3b
With Realistic Sections R1–R3 and Idealized Sections I0–I3, Respectively
for the profiles of density and buoyancy frequency). Figures 3c–3e present
Section Description the initial density, and the horizontal and vertical velocity fields of a typi-
R1 Realistic topography with two prominent bumps QR1 and NR1 cal large‐amplitude ISW adjusted from a KdV solution as mentioned
only over the continental shelf above. This initial ISW is designed with a total energy of approximately
R2 Realistic topography with two prominent bumps QR2 and PR2 2 GJ/m, consistent with the prototypical soliton observed in the deep
over the continental shelf and slope, respectively
basin region in the northern SCS (Klymak et al., 2006).
R3 Realistic topography with one prominent bump QR3 over the
continental shelf and two prominent bump PR3 and QR3 In this study, for simplicity of comparison and without loss of generality,
over the continental slope the model initial fields in both realistic and idealized experimental studies
I0 Idealized continental topography without any bumps
are set by adopting this typical large‐amplitude ISW shown in Figures 3c–
I1 Idealized topography with two bumps Q and P over the
continental shelf and slope, respectively 3e to focus on the study of the bumpy topographic effects on the associated
I2 Idealized topography with one bump P only over the transbasin evolution. To show the model's ability in solving the large‐
continental slope amplitude ISWs, a model test is given in Appendix B. It shows that the
I3 Idealized topography with one bump Q only over the simulated large‐amplitude ISW can basically retain its wavelength, ampli-
continental shelf
tude, and total energy when it propagates over a uniform depth with the
Note. The bumps are marked over the related sections (see above). constant background dissipation.

3. Results
Model results show that the shoaling transbasin ISW in the northern SCS generally undergoes no obvious
transformation only if it arrives at a position near the middepth of ocean floor topography, consistent with
both in situ and satellite observations (Alford et al., 2010; Huang et al., 2008; Zhao et al., 2004). Therefore,
unless otherwise stated, we mainly present model results in domain (−500 km, −250 km) × (−0.4 km, 0
km) for clarity. In this domain, the shoaling transbasin ISW has
approached and transited the middepth of the topography and exhibited
obvious transformation.

3.1. Realistic Bumpy Topographies in the Northern SCS

Model results of both density and horizontal velocity fields (Figure 4) for
the realistic section R2 are illustrated first as the standard case. For refer-
ence, the two associated most prominent bumps over the continental
slope and shelf regions are marked over the topography (see PR2 and
QR2 shown at bottom panels in Figure 4).
The result at t = 23 hr shows that, before approaching the prominent
bumps, the waveform of the shoaling transbasin ISW has no obvious
adjustment. However, after transiting the first prominent bump PR2,
which is located at a water depth of ~1,000 m over the continental slope
of section R2, the original single ISW transforms into two groups of new
waves (see result at t = 30 hr), which are denoted as S and T, respectively.
Specifically, the group S is the typical mode‐1 packet with at least two
solitary‐like internal waves, while the group T can be regarded as a
mode‐2 internal wave.
At time from 40 to 43 hr, the wave group S of mode‐l ISW packet reaches
and then passes the crest of the second prominent bump QR2 over the con-
tinental shelf of section R2. This bump has a characteristic height and
width of ~150 m and ~5 km, respectively. It is seen that a much more dra-
matic adjustment happens to both the front and rear shoulders of the lead-
ing ISW during this period. The result at t = 46 hr shows that, after this
dramatic adjustment, another three groups of new waves, denoted as S1,
Figure 3. Model background stratification including profiles of (a) density
S2, and S3, respectively, are formed. Here the newborn waves in groups
and (b) buoyancy frequency and model initial internal solitary wave fields S1 and S2 are resulted from the fission of the depressed front shoulder
of (c) density, and (d) horizontal and (e) vertical velocities. and elevated rear shoulder, respectively, of the leading ISW in group S,

XIE ET AL. 4682

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 4. Model results of density and horizontal velocity (m/s) for the shoaling transbasin internal solitary wave
along the realistic section R2 at eight snapshots: t = 23, 30, 40, 43, 46, 50, 54, and 57 hr, respectively. The waves with
overhead letters S (including S1, S2, and S3) and T represent the newborn waves after shoaling transbasin evolution. For
reference, the bathymetry is shown at the bottom of each panel and the two prominent bumps over continental slope
and shelf regions are labeled as PR2 and QR2, respectively (note the difference in vertical scale of bathymetry at bottom
panels with that of model results).

and it is seen that these two groups of waves are both rank‐ordered mode‐1 ISW packet at t = 46 hr.
Moreover, similar to the previously formed mode‐2 wave T triggered by bump PR2 over the slope, it is
seen that another mode‐2 wave group (denoted as S3) is also triggered by the bump QR2 over the shelf.
For the leading group S1 of mode‐1 packet, it is seen that it presents no essential adjustment and remains the
well rank‐ordered waveform during the following period from 50 to 57 hr. However, the ISWs in the
following group S2, which are originally excited as a rank‐ordered packet at t = 46 hr, gradually lose their
regular order and transform into a non‐rank‐ordered packet. This non‐rank‐ordered packet is actually due
to the modulation associated with the leading wave group S1. It is seen that during this following period
the ISWs in the packet of group S2 gradually catch up with and ride on the elevated tail of wave packet in
the group S1, resulting in an interaction with the disturbances of background stratification/current asso-
ciated with this elevated tail. This interaction promotes the energy transfer and inhomogeneous dissipation
among ISWs in the following packet and thus leads to the irregular order of ISWs in the packet of group S2.
The mechanism of this non‐rank‐ordered packet can also be found similarly in Vlasenko and Stashchuk
(2006), although there the energy exchange among ISWs in one packet is due to the interaction between tidal
flow and excited internal waves by variable topography. In addition, it is noticed that the two groups of
mode‐2 internal waves T and S3, triggered after transiting bumps PR2 and QR2, respectively, both split up into
solitary‐like waves (see result at t = 57 hr) because of the dominant effect of wave nonlinearity over wave
dispersion in the shoaling process from deep water to shallow water (Xie et al., 2016; Yang et al., 2009).
Overall, the model results in Figure 4 reveal that four groups of new waves are eventually formed after pas-
sing the two prominent bumps PR2 and QR2 over the topography R2, including the leading packet of rank‐
ordered mode‐1 ISWs (S1), the following packet of non‐rank‐ordered mode‐1 ISWs (S2), and two groups of

XIE ET AL. 4683

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 5. Results of simulated density and horizontal velocity (m/s) of shoaling transbasin internal solitary wave for realistic sections R1 (left panel) and R3
(middle panel) and idealized smooth section I0 (right panel) at four snapshots: t = 30, 43, 50, and 57 hr, respectively. The waves with overhead letters S (including
S1, S2, S3, S11, S12, and S13) and T (including T1 and T2) represent the new formed waves after shoaling transbasin evolution. The arrows denote the radiated
inertia‐gravity wave. The bathymetry and related prominent bumps are shown at the bottom for reference (note the difference in vertical scale of bathymetry at
bottom panels with that of model results).

solitary‐like mode‐2 waves (S3 and T). For comparison, Figure 5 further shows the shoaling transbasin
evolutions along the other two realistic sections R1 (left panel) and R3 (middle panel) and along the
idealized smooth section I0 (right panel).
The section R1 is located at south of section R2 (Figure 2). Different from the standard case of section R2,
results for section R1 (left panel in Figure 5) show that no obvious mode‐2 disturbance is excited at the slope
region because of the relative smooth topography here. However, at the shelf region of section R1, the topo-
graphy undulation becomes much more complex. The two most prominent bumps here, denoted as NR1 and
QR1, are also marked (see bottom of left panel in Figure 5). Due to their impacts, a much more complex
adjustment than those along section R2 happens. At least two groups of mode‐1 packet (S1 and S2) and three
groups of mode‐2 packet (S3, T1, and T2) are formed at t = 43 hr after passing the bump NR1. Next, the fol-
lowing wave groups S2 and S3 are greatly damped and dissipated, especially when they interact with the rela-
tively larger bump QR1. For instance, the wave group S2 is almost disappeared because of the possibly strong
local dissipation near bump QR1. However, as shown at t = 57 hr, the leading group S1 successfully passes the
bump QR1, and it transforms into another three groups of new waves eventually (denoted as S11, S12, and S13,
respectively), similar to the transformation of the former leading ISWs.
The transformation along section R3 (see middle panel in Figure 5) is comparatively similar to that along
section R2 in the standard case, whereas there are two obvious mode‐2 disturbances T1 and T2 triggered
at the slope region of section R3 because of the successive impacts by two prominent bumps (OR3 and
PR3) at depths of approximately 1.6 and 1 km, respectively. Moreover, after passing bump QR3 at the shelf

XIE ET AL. 4684

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

region of section R3, there are three similar wave groups (i.e., S1, S2, and
S3) are eventually formed (see result at t = 57 hr) as those along section R2.
An implication according to the results above is that, even for a single
large‐amplitude mode‐1 ISW from the deep basin region in the northern
SCS, the shoaling transbasin transformation is quite complex and differ-
ent along different realistic topographies. In addition to a rank‐ordered
mode‐1 packet in leading wave group, the following secondary wave
groups may exhibit as non‐rank‐ordered mode‐1 and mode‐2 waveforms.
These transformations provide explanations of the complexity and rich-
ness of satellite‐detected ISWs near the continental shelf region relative
to the deep basin region in the northern SCS (e.g., Dong et al., 2016;
Huang et al., 2008; Jackson, 2009). Besides, the comparisons of model
results along three different realistic topographies indicate that the ISW
transformations depend greatly on the topographic details over both the
continental slope and shelf regions. This can be further confirmed by
the model simulations along the idealized smooth section I0 (see right
panel in Figure 5). In stark contrast to those with realistic bumpy topogra-
phies above, it is seen that, except for the leading packet S with two
soliton‐like internal waves, no any other secondary waves are excited,
even though the idealized smooth section I0 is indeed an overall good
approximation to all three realistic sections R1–R3 at both slope and shelf
regions (as mentioned in section 2.2). Nevertheless, it is worth mentioning
that an inertia‐gravity wave following the tail of leading packet S is
radiated as that in Grimshaw et al. (2014) and a weak steepening also hap-
pens (see, e.g., t = 57 hr), although this radiated wave is much weaker
than the other secondary waves resulting from the bumpy topographic
effects and has no dispersed packet formed because of the strong eddy dis-
sipations in this study (see a sensitivity study with weaker eddy dissipation
in Appendix B, which presents a much obvious steepening and breaking
of ISW packet due to the radiation of inertia‐gravity wave).
The transformation of shoaling transbasin ISW along different sections
can be further studied from the view point of energy variation. The total
wave energy (E), obtained by summing the wave KE and APE as that in
Lamb and Nguyen (2009) and Xie, He, et al. (2015) over the whole model
domain (−500 km, 0 km), is shown in Figure 6a for the sections R1–R3
and I0. The model results show that the total energies at the end of each
simulation are approximately 0.2–0.3 GJ/m, that is, approximately 90%
of the wave energies are dissipated in its shoaling transbasin processes.
This is consistent with estimations by Chang et al. (2006), Klymak et al.
(2006), and Alford et al. (2010) using in situ observations. Actually, the
Figure 6. Energy variations of shoaling transbasin internal solitary wave
along the realistic sections R1–R3 and idealized section I0. (a) Total wave model results show that, at t = 35 hr before ISW reaches the shelf region,
energy. (b) Energy decay rate. (c) kinetic energy (KE)/available potential the reserved wave energy decreases to only about 0.4–0.5 GJ/m, that is,
energy (APE). most of the energy (up to ~80%) of transbasin large‐amplitude ISW in
the northern SCS has already lost at the slope region. The energy decay
rate (dE/dt), as shown in Figure 6b, indeed exhibits a very rapid loss of energy at the slope region, reducing
gradually from ~30 to ~5 kW/m; however, at the shelf region (i.e., after t = 35 hr), the energy decay rate
reduces to an average as low as around 2–3 kW/m. In addition, Figure 6b indicates that at the three realistic
sections the decay rate of shoaling ISW oscillates greatly due to the bumpy topographic effects. It may have a
peak‐to‐peak difference up to 10–20 kW/m over the slope region and 3–5 kW/m over the shelf region.
The partition of wave energies, measured by the ratio KE/APE, also varies greatly during the shoaling
transbasin process, as seen in Figure 6c. It is shown that the overall tendencies of energy partition along
all of the three realistic sections are close to that along the ideal section I0. Generally, the maximum ratio
KE/APE (= ~1.5) reaches at t = ~27 hr, when the ISW is just propagating at the continental shelf break

XIE ET AL. 4685

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 7. Bathymetries of three idealized bumpy sections. (a) Idealized bumpy section I1 with two bumps P and Q over
continental slope and shelf, respectively. (b) Idealized bumpy section I2 with the bump P over continental slope. (c)
Idealized bumpy section I3 with the bump Q over continental shelf.

region in the northern SCS (i.e., at a water depth around 400–600 m). This ratio is slightly larger than the
ratio of 1.4 estimated by Klymak et al. (2006) in the deep ocean basin region, where the wavefields are
constructed by fitting the KdV solution to in situ observations. Next, this ratio gradually decreases at the
continental shelf region, and eventually the ISW energies tend to be equally partitioned between KE and
APE (i.e., the ratio KE/APE is close to 1). Moreover, it is seen that the energy partition also varies greatly
due to the influence of bumpy bottom topography. In particular, at the shelf region, the bumps can make
the KE reduce to as low as 80% of the APE (for example, see results along the realistic section R2 at t =
~39.5 hr when the ISW just arrives at the bump crest of QR2 in Figure 4). However, the KE bounces back
to approximately 1.1 times of the APE at t = ~45 hr when the wave groups almost leave the bump QR2.
Along the section R1, the KE bounced back can even reach up to 1.2 times of APE at t = ~49 hr.
Therefore, the results above show that the transformation, the energy decay, and partition of ISW could vary
greatly during the shoaling transbasin process because of the complexity of the realistic topography in the
northern SCS. The comparison with the idealized smooth section I0 further confirms that the bumps over
the three realistic topographies in the northern SCS have significant influences on the wave transformation
and energy variation. However, the specific influence of each bump is still unclear due to their random sizes
and irregular connections. To have a better understanding of the specific role of these bumps, we next carry
out simulations along another three idealized bumpy sections.

3.2. Idealized Bumpy Topographies: Role of Bumps Over Continental Slope/Shelf

In this subsection we present model results along three idealized bumpy sections I1–I3 (Figure 7). The three
sections are constructed using the expression (11) by including specific idealized bumps P and/or Q (see
Table 2 for associated parameters) over continental slope and/or shelf regions, respectively, based on the
idealized smooth section I0. Specifically, the idealized bump Q is located over the shelf region at x = ~
−400 km, and it has a characteristic height of approximately 150 m; the bump P is located over the slope
region at a water depth of 1,000 m, and it has a characteristic height of approximately 350 m. It is seen that
the size of the idealized bump Q is very close to those of the two prominent realistic bumps QR1 and QR2, while
the idealized bump P is ~100 m higher than the most prominent realistic bump PR3. We distinguish the influ-
ence of each bump and investigate the mechanism of newborn waves on
the basis of simulations along the three idealized bumpy sections.
Table 2 In Figure 8, model simulation corresponding to the idealized bumpy sec-
The Position xi, Height hi, and Width σi of the Gaussian Bumps Used in the tion I1 is shown. The overall result for idealized bumpy section I1 with
Expression (11) for the Idealized Bumps Q and P Located Over the Three both bumps P and Q over continental slope and shelf regions, respectively,
Idealized Bumpy Sections I1–I3 in Figure 7
can indeed show a very similar transformation to that for the realistic sec-
Parameters Position xi (km) Height hi (m) Width σi (km) tion R2. As that in Figure 4, four groups of newborn waves are formed,
Idealized bump Q −400 150 5 that is, the solitary‐like mode‐2 wave group T is formed after transiting
Idealized bump P −305 −250 5 the bump P, and the rank‐ordered mode‐1 wave group S1, non‐rank‐
−290 400 5 ordered mode‐1 wave group S2, and solitary‐like mode‐2 wave group S3
Note. The idealized bump P consists of one negative and one positive are formed after transiting the bump Q. The similarity reflects the signifi-
Gaussian bump. cance of the two prominent bumps PR2 and QR2 over the realistic section

XIE ET AL. 4686

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 8. Model results of density and horizontal velocity (m/s) for the shoaling transbasin internal solitary wave along
the idealized bumpy section I1 at eight snapshots: t = 23, 30, 40, 43, 46, 50, 54, and 57 hr, respectively. The waves
labeled with overhead letters S (including S1, S2, and S3), T and C represent the newborn waves after shoaling transbasin
evolution. For reference, the bathymetry is shown at bottom of each panel and the two prominent bumps P and Q over
slope and shelf regions are labeled (note the difference in vertical scale of bathymetry at bottom panels with that of model
results).

R2 in affecting the shoaling transbasin evolution. Also, it is seen that, because of the relatively bigger size of
the idealized bump P over the idealized section I1, the eventually formed mode‐2 wave group T, including a
number of at least 3–4 solitary‐like internal waves (see result at t = 57 hr in Figure 8 for reference), is much
stronger than that excited over the slope region of the realistic section R2.
The roles of bumps P and Q can further be seen according to model results for the idealized bumpy sections
I2 and I3, which have only one isolated bump over the continental slope or shelf region. It is seen that, due to
the isolated effect of bump P over the slope region of idealized section I2, a solitary‐like mode‐2 wave group
(T) that is almost the same as that along the idealized section I1 is finally excited, together with only one
wave group S over the shelf region (see left panel in Figure 9). Moreover, the three wave groups (S1, S2,
and S3), also almost the same as those along the idealized section I1, are excited over the shelf region of idea-
lized section I3 after ISW passes the isolated bump Q (see right panel in Figure 9); however, over the slope
region of section I3, no solitary‐like mode‐2 wave is excited due to the absence of any bump there. The result
along the idealized section I3 actually reflects a relatively very similar scenario to that along the realistic
section R1.
To understand the formation of solitary‐like mode‐2 waves excited by the bump over the slope region, model
results for the idealized bumpy section I2 in the model domain (−375 km, −250 km) × (−1.2 km, 0 km) are
depicted in Figure 10. It shows that, when the shoaling transbasin ISW interacts with the continental slope
with a bump, upward‐ and downward‐propagating beam‐like disturbances are excited, emanating from the
bump crest toward both onshore and offshore. Overall, these disturbances are consistent with the character-
istic lines depicted by the dashed curves in Figure 10. The behavior of these beam‐like disturbances is actu-
ally similar to that of tidally induced beams, which have been studied intensively during the last decades in

XIE ET AL. 4687

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 9. Results of simulated density and horizontal velocity (m/s) for shoaling transbasin internal solitary wave along
the idealized bumpy sections I2 (left panel) and I3 (right panel) at four snapshots: t = 30, 43, 50, and 57 hr, respectively.
The waves labeled with overhead letters S (including S1, S2, and S3), T, and C represent newborn waves. The bathymetry
and related bumps are shown at bottom of each panel for reference (note the difference in vertical scale of bathymetry at
bottom panels with that of model results).

many theoretical, numerical, and laboratory studies (e.g., Akylas et al., 2007; Diamessis et al., 2014;
Grisouard et al., 2011; Mercier et al., 2012; Sarkar & Scotti, 2017). A local generation mechanism (e.g., da
Silva et al., 2007; Gerkema, 2001; New & Pingree, 1992) can be adopted here to illustrate the formation of
the solitary‐like mode‐2 waves. It is seen that the onshore‐propagating beams impinge on the pycnocline
from below. The beams get refracted and scattered first, resulting in disturbances of different modes
trapped within the pycnocline. Then these trapped disturbances can further steepen and split up into
ISWs when propagating onshore because of the nonlinear and nonhydrostatic effects (Grimshaw et al.,
2004; Farmer et al., 2009; Xie, Pan, et al., 2015). For instance, the mode‐2 wave T is formed at around

Figure 10. Model fields of density and horizontal velocity (m/s) in domain (−375 km, −250 km) × (−1.2 km, 0 km) for the
idealized bumpy section I2 at t = 30, 40, and 50 hr, respectively, which illustrate the local generation of onshore‐propa-
gating second (T) and third (C) internal modes by the beam‐like disturbances emanating from the bump P. Dashed curves
represent the characteristic lines of the beam‐like disturbances.

XIE ET AL. 4688

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Figure 11. Depth‐integrated total wave energy, E(x, t), in the x‐t plane for idealized sections I0–I3. Numbers 1, 2, and 3 in
(b) and (c) represent slanting energy lines of onshore‐ and offshore‐propagating internal waves of the first, second, and
third modes, respectively, that are scattered form internal beams originating from bumps P and Q over the slope region of
sections I1 and I2.

t = 30 hr (Figure 10a) and steepens at around t = 40 hr (Figure 10b), while it finally can evolve into a packet
of rank‐ordered ISWs at around t = 50 hr (Figure 10c). In addition, Figure 10 reveals that a mode‐3 internal
wave C with three zero crossings in velocity profile is also similarly scattered from the internal beams,
whereas no obvious steepening and breaking happen due to the associated weak wave nonlinearity
(see also results of idealized sections I1 and I2 in Figures 8 and 9 for the formation of the third‐mode
internal wave).
Actually, except for the onshore‐propagating internal waves above, the internal waves propagating offshore
are also similarly scattered from offshore‐propagating internal wave beams. Figure 11 shows the depth‐
integrated total wave energy, E(x, t), in the x‐t plane for the four idealized
sections I0–I3. The different results among four idealized sections can be
seen according to the patches of the minima and maxima in the wave
energy contours along the different slanting lines. It is clear that, in addi-
tion to onshore‐propagating energies, the offshore propagation of
reflected internal wave energies can also be seen, originating from the
position where a bump exists. For example, the reflected internal waves
of the first, second, and third modes, originating from the bump P over
the slope region of sections I1 and I2, are labeled over the slanting energy
lines in Figures 11b and 11c. The associated offshore speeds, reflected by
the slope of the slanting lines, are around 2.9, 1.57, and 0.93 m/s, respec-
tively, which are very close to the linear characteristic internal wave
speeds in the deep basin region (i.e., 3.25 km) obtained by solving the
Figure 12. Results of simulated density and horizontal velocity (m/s) in related eigenvalue problems (e.g. Apel et al., 2006; Xie et al., 2016) of inter-
model domain (−300 km, −150 km) × (−3.25 km, 0 km) for the idealized nal gravity waves.
bumpy section I2 at t = 36 and 47 hr, respectively, which show the local
generation of reflected offshore‐propagating first (S), second (T), and third The link between these reflected offshore‐propagating internal waves and
(C) internal modes by offshore‐propagating beam‐like disturbances ema- offshore‐propagating internal beams emanating from the bump P can
nating from bump P. further be seen from model result in Figure 12, which shows the

XIE ET AL. 4689

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

horizontal velocity field for the idealized section I2 in model domain

(−300 km, −150 km) × (−3.25 km, 0 km). The offshore‐propagating
first‐mode (S), second‐mode (T), and third‐mode (C) internal waves,
which can be discerned according to the number of their zero crossings,
are scattered successively from the offshore internal beams, after the
beams impinge on the pycnocline from below. However, none of them
steepen and break up into solitary‐like waves because of their associated
weak wave nonlinearity.
Moreover, Figure 13 shows variations of the total energy, energy decay,
and partition along the four idealized sections I0–I3. Figure 13a shows
that the difference in total wave energy among different sections is less
than ~0.04 GJ/m, which suggests that the total energy of shoaling ISW
over different idealized sections is generally insensitive to the existence
of bumps at the continental slope/shelf region. Nevertheless, the bumps
can greatly alter the energy decay rate and energy partition. Figure 13b
shows that the energy decay rate of shoaling ISW increases greatly on
the offshore side of the bump, while it decreases rapidly after passing
the bump crest and later bounces back to the normal decay rate as that
along the idealized smooth section I0. Figure 13c shows that the ratio
KE/APE first reduces when ISW passes over the bumps and later it
bounces back, while over the shelf region it oscillates around the equipar-
tition point (KE/APE = 1). This variation of energy partition suggests a
mutual conversion of energies between KE and APE, resulting from the
interaction of shoaling transbasin ISW with bumps over the
continental shelf.

4. Conclusion and Discussion

Bumps over continental slope/shelf regions, with characteristic height
and width comparable to those of the shoaling large‐amplitude mode‐1
ISWs, are very common in the northern SCS. The comparable size of
these local bumps with ISWs suggests that a shoaling ISW should experi-
ence a very sudden change associated with a bump. Therefore, we use a
fully nonlinear, nonhydrostatic numerical model to examine the influ-
ences of these bumps on the shoaling transbasin evolution of ISW in
the northern SCS. Two groups of numerical experiments are performed
to study (1) the evolution processes, characteristics, and energy variations
of the shoaling transbasin ISW; (2) the specific roles of bumps over con-
tinental slope and shelf regions; and (3) the mechanism of newborn
waves excited by bumps.
Figure 13. Energy variations of shoaling internal solitary wave along the
idealized sections I0–I3. (a) Total wave energy. (b) Energy decay rate. (c)The first experimental study is performed with realistic topographies
kinetic energy (KE)/available potential energy (APE). along three different cross sections near the Dongsha Islands. It is
shown that the transbasin evolution of shoaling ISW over different
realistic sections in the northern SCS is rather complex and different.
In addition to the leading group of rank‐ordered mode‐1 wave packet, the following secondary wave
groups may exhibit as non‐rank‐ordered mode‐1 packet or mode‐2 waveforms. Model results further
show that up to ~80% of the ISW energy is lost over the continental slope region with a gradual decrease
in energy decay rate from ~30 to ~5 kW/m, while over the shelf region another 10% of ISW energy is lost
with a decrease in energy decay rate to around 2–3 kW/m. The energy partition also varies greatly dur-
ing the shoaling processes. Generally, the ratio KE/APE of the shoaling ISW reaches maximum when
ISWs are at the continental shelf break region (i.e., at the bathymetry around 400–600 m) in the north-
ern SCS. After the shoaling ISW propagates away from the continental shelf break and into the conti-
nental shelf region, this ratio gradually decreases and eventually tends to be unity. Moreover, both the

XIE ET AL. 4690

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

energy decay rate and the ratio KE/APE of shoaling ISW oscillate greatly due to the bumpy topographic
effects in the northern SCS. Specifically, the bumps can cause a peak‐to‐peak difference of the energy
decay rate of ISW up to 10–20 kW/m over continental slope region and 3–5 kW/m over continental shelf
region. Over continental shelf region, the bumps can make the KE of ISW drop to as low as only 80% of
the APE, while after the drop the KE immediately bounces back to approximately 1.1–1.2 times of
the APE.
The second experimental study includes simulations with four idealized topographies that are designed on
the basis of the realistic topographies in the northern SCS. It is shown that an idealized smooth topography
can only result in a leading packet with two soliton‐like internal waves, while no secondary waves as those
over realistic topographies can be excited, even though this smooth topography is a good approximation to
all realistic topographies over the continental slope and shelf regions except the bumps. Indeed, an idealized
topography with prominent bumps over both continental slope and shelf regions can produce a very similar
transformation to that of the realistic topographies. The specific roles of each bump are further studied based
on two idealized simulations with only one isolated bump over continental slope or shelf region. The simu-
lations show that the bumps over continental slope are capable of triggering a solitary‐like mode‐2 internal
wave packet, while the bumps over continental shelf are responsible for the three wave groups, that is, the
leading group of rank‐ordered mode‐1 packet and the following two groups of non‐rank‐ordered mode‐1
packet and mode‐2 internal waves. Simulations over the idealized bumpy topographies also show that the
energy decay rate of shoaling ISW increases greatly on the offshore side of the bump, while after transiting
bump crest it decreases rapidly and later it bounces back to the normal decay rate. The ratio KE/APE of
shoaling ISW is reduced when passing over the bumps, but later it also bounces back, and over the continen-
tal shelf it oscillates around the equipartition point (KE/APE = 1). This variation of energy partition suggests
a conversion of energy between KE and APE, resulting from the interaction of ISWs with bumps over
continental slope/shelf.
Over the shelf region, the leading group of rank‐ordered mode‐1 packet and the following group of non‐
rank‐ordered mode‐1 packet are resulted from the fission of the depressed front shoulder and elevated
rear shoulder, respectively, of the shoaling ISW. The non‐rank‐ordered characteristic of the following
mode‐1 packet is due to the transfer and inhomogeneous dissipation of energy among the solitons in
the wave packet, resulting from the interaction of waves with the background stratification/current asso-
ciated with the elevated tail of the leading group. Moreover, both upward‐ and downward‐propagating
beam‐like disturbances, emanating from the bump crest to both onshore and offshore, can be excited
when ISW interacts with a bump at continental slope or shelf region. A local generation mechanism,
similar to that for the tidally induced internal beams, is adopted for illustrating the onshore‐propagating
solitary‐like mode‐2 internal wave packet. These internal beams, induced by the shoaling ISW and
impinging on the pycnocline from below, refract and scatter into both onshore and offshore internal
waves of different modes trapped near the pycnocline and propagating away from the bumps.
However, it is shown that only the trapped onshore disturbance of the second internal mode successfully
steepens and splits up into nonlinear ISWs because of strong nonlinear effects, while all the reflected
offshore‐propagating first‐, second‐, and third‐mode internal waves are not able to split up into solitary‐
like waves.
Nevertheless, our emphasis here is on the role of the prominent bumps in modulating the response of
shoaling ISWs. Because of the strong bottom currents induced by shoaling ISWs, many sand dunes com-
posed of fine to medium sand are reported to be formed on the upper continental slope region in the
northern SCS (Ma et al., 2016; Reeder et al., 2011). The further response of shoaling ISWs to this kind
of topographic bumps, which are at least 1 order of magnitude smaller than the prominent bumps stu-
died here, is unclear as yet due to the limit of model resolution and the complexity of associated sedi-
ment resuspension and transport processes (Boegman & Stastna, 2019; Stastna & Lamb, 2008).
Besides, due to the influence of background current, distinct processes of convective breaking and
trapped core phenomena of shoaling ISWs (Lamb, 2002, 2003) are observed near the Dongsha region
between the realistic sections R1 and R2 (Lien et al., 2012; Lien et al., 2014). However, here we do
not take into account background current, and thus, in the future work, the combined effect of back-
ground currents and local topographic bumps on the evolution of shoaling ISWs in the northern SCS
should be considered.

XIE ET AL. 4691

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Appendix A: Transformed Equations and

Solving Method
After performing the transformation equation (9), the new governing
equations of the shoaling transbasin ISW are written as

 
ξ t þ C ðx 2 ; z2 Þ J ðξ; ψÞ−fvz2 ¼ gℜ1 ðe
ρÞ=ρ0 þ AH ℜ2 ðξ Þ þ AV ℜ3 ðξ Þ; (A1)

ρt þ C ðx 2 ; z2 ÞJ ðe
e ρ; ψÞ þ ρ0 =gN 2 ðx 2 ; z2 Þℜ1 ðψÞ
¼ K H ℜ2 ðe
ρÞ þ K V ℜ3 ðe
ρ þ ρ0 Þ; (A2)

ξ ¼ ℜ2 ðψÞ þ ℜ3 ðψÞ; (A3)

Figure B1. Model simulated variation of wave total energy E, amplitude A,  

phase speed c, and wavelength λ with distance for the initialized large‐ vt þ Cðx 2 ; z2 Þ J ðv; ψÞ þ f ψz2 ¼ AH ℜ2 ðvÞ þ AV ℜ3 ðvÞ; (A4)
amplitude Dubriel‐Jacotin‐Long‐internal solitary wave over the uniform
depth. where

−H ðx Þ
Cðx 2 ; z2 Þ ¼ N ðx 2 ; z2 Þr ðx 2 Þ; r ðx 2 Þ ¼ 1=hðx 2 Þ; hðx 2 Þ ¼ ∫ N ðsÞds; pðx 2 Þ ¼ r ðx 2 Þdhðx 2 Þ=dx 2 ; qðx 2 Þ
0

¼ r ðx 2 Þd2 hðx 2 Þ=dx 22 ;

and ℜ1, ℜ2, and ℜ3 represent the differential operators and are expressed as follows:

∂ ∂
ℜ1 ¼ −z2 pðx 2 Þ ;
∂x 2 ∂z2
∂2 ∂2 ∂2   ∂
ℜ2 ¼ 2 −2z2 pðx 2 Þ þ z2 p2 ðx 2 Þ 2 þ z2 2p2 ðx 2 Þ−qðx 2 Þ ;
∂x 2 ∂x 2 ∂z2 ∂z2 ∂z2
∂2 dN ∂
ℜ3 ¼ r 2 ðx 2 ÞN 2 ðx 2 ; z2 Þ 2 þ r ðx 2 Þ :
∂z2 dz ∂z2

The alternative direction implicit method, which utilizes a rectangular

grid with second‐order approximation to the spatial derivatives and
first‐order approximation to the temporal derivatives in every temporal
semilayer, is then used to solve the transformed equations (A1)–(A4).
Using this method, the time integration of the equations is split into
two temporal semisteps at n + 1/2 and n + 1, and the finite difference
is performed for the spatial discretization of the transformed equations;
more details of this method can be referred in Kantha and Clayson
(2000). In this study, all the four equations including elliptic
equation (A3) are solved at each semilevel. At the first (second) semistep
the differential scheme in the x direction is implicit (explicit), while in the
y direction it is explicit (implicit). The implicit system at each temporal
semistep is a tridiagonal matrix and can then be solved using the
standard techniques.

Figure B2. Density and horizontal velocity (m/s) of shoaling ISW with
weaker eddy dissipation (β = μ = 0.5) for idealized smooth section I0 at Appendix B: Model Test and Sensitivity Study
two snapshots: t = 50 and 57 hr, respectively. The waves with overhead let- We test the model's ability in solving the propagation of the initialized
ters S represent the mode‐1 packet formed after shoaling transbasin evolu-
large‐amplitude ISW over a uniform depth (i.e., −3,250 m). This large‐
tion. The radiated inertia‐gravity wave with the consequent formation of
solitary‐like internal waves is denoted by the arrows. The bathymetry is amplitude ISW is initialized into the model using the solution of the fully
shown at the bottom for reference (note the difference in vertical scale of nonlinear, Dubriel‐Jacotin‐Long equation (Dunphy et al., 2011;
bathymetry at bottom panels with that of model results). Turkington et al., 1991):

XIE ET AL. 4692

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

N 2 ðz−ηÞ
∇2 η þ η ¼ 0; (B1)
c20

η ¼ 0; atz ¼ 0; H and as x→∞; (B2)

where η(x,z) and c0 represent the isopycnal displacement and phase speed, respectively, of the fully non-
linear ISW. This Dubriel‐Jacotin‐Long equation can generate an exact solution to the incompressible
Euler equations subject to the Boussinesq approximation. For a specified APE, the equation is solved itera-
tively by minimizing KE. Here we set APE as 0.9 GJ/m. This generates an initial large‐amplitude ISW with
wave amplitude A0 of ~163.3 m and phase speed of ~3.4 m/s, and the associated wavelength λ0 as indicated
in Aghsaee et al. (2010) is about 5,238.5 m. The horizontal and vertical eddy viscosities and diffusivities and
the Coriolis parameter are set to 0, and the periodic boundary condition is used. Figure B1 shows the model
results of total energy, wave amplitude, phase speed, and wavelength of this large‐amplitude ISW, normal-
ized by the associated initial values. It shows that, after a long propagation distance of about 150 km, all these
wave characteristics change by about 1%, which suggests that the simulated ISW could basically retain its
initial waveform, amplitude, speed, and wavelength.
Moreover, we study the sensitivity of evolution of radiated inertia‐gravity wave to the magnitude of the eddy
dissipation based on the simulation over idealized section I0. Figure B2 shows the model result with weaker
eddy dissipation (i.e., β = μ = 0.5, which are reduced by half), when ISWs arrive at the continental shelf
region of the idealized section I0. It is seen that, except for a much stronger leading mode‐1 packet S, the
radiated inertia‐gravity wave successfully breaks with the consequent formation of at least 2–3 solitary‐like
internal waves (at t = 57 hr). This result suggests that the evolution of radiated inertia‐gravity wave due to
the combined effect of topography and rotation (Grimshaw et al., 2014) is sensitive to the eddy dissipation.

Acknowledgments References
The authors acknowledge the valuable
and constructive comments and Aghsaee, P., Boegman, L., & Lamb, K. G. (2010). Breaking of shoaling internal solitary waves. Journal of Fluid Mechanics, 659, 289–317.
suggestions of the reviewers. This work https://doi.org/10.1017/S002211201000248X
was jointly supported by the Frontier Akylas, T. R., Grimshaw, R. H. J., Clarke, S. R., & Tabaei, A. (2007). Reflecting tidal wave beams and local generation of solitary waves in
Science Key Research Program of the the ocean thermocline. Journal of Fluid Mechanics, 593, 297–313. https://doi.org/10.1017/S0022112007008786
CAS (QYZDJ‐SSW‐DQC034); NSFC Alford, M. H., Lien, R. C., Simmons, H., Klymak, J., Ramp, S., Yang, Y. J., et al. (2010). Speed and evolution of nonlinear internal waves
Grants 41430964, 41521005, and transiting the South China Sea. Journal of Physical Oceanography, 40(6), 1338–1355. https://doi.org/10.1175/2010JPO4388.1
41776008; Pearl River S&T Nova Alford, M. H., Peacock, T., MacKinnon, J. A., Nash, J. D., Buijsman, M. C., Centurioni, L. R., et al. (2015). The formation and fate of internal
Program of Guangzhou waves in the South China Sea. Nature, 521(7550), 65–69. https://doi.org/10.1038/nature14399
(201806010091); Guangzhou S&T Amante, C., and B. W. Eakins (2009), ETOPO1 1 arc‐minute global relief model: Procedures, data sources and analysis, NOAA Tech.
program (201804010373); Guangdong Memo. NESDIS NGDC‐24, 25 pp.
province Sicence Foundation Apel, J., L. Ostrovsky, Y. Stepanyants, and J. Lynch (2006), Internal solitons in the ocean, Tech. Rep., Woods Hole Oceanogr. Inst. Tech.
(2018A030313432); Youth Innovation Rept., WHOI‐2006‐04.
Promotion Association CAS (2019336); Apel, J. R., Holbrook, J. R., Liu, A. K., & Tsai, J. J. (1985). The Sulu Sea internal soliton experiment. Journal of Physical Oceanography,
ISEE2018PY05, CAS; the Innovation 15(12), 1625–1651. https://doi.org/10.1175/1520‐0485(1985)015<1625:TSSISE>2.0.CO;2
Group Program LTOZZ1703 of LTO Boegman, L., & Stastna, M. (2019). Sediment resuspension and transport by internal solitary waves. Annual Review of Fluid Mechanics,
(SCSIO, CAS). The simulations were 51(1), 129–154. https://doi.org/10.1146/annurev‐fluid‐122316‐045049
performed on the HPCC at the SCSIO, Cai, S., Gan, Z., & Long, X. (2002). Some characteristics and evolution of the internal soliton in the northern South China Sea. Chinese
CAS. The ETOPO1 Global Relief Model Science Bulletin, 47(1), 21–27. https://doi.org/10.1360/02tb9004
(2009) can be downloaded from https:// Cai, S., Long, X., & Wang, S. (2008). Forces and torques exerted by internal solitons in shear flows on cylindrical piles. Applied Ocean
www.ngdc.noaa.gov/mgg/global/ research, 30(1), 72–77. https://doi.org/10.1016/j.apor.2008.03.001
global.html. The climatological Cai, S., Xie, J., & He, J. (2012). An overview of internal solitary waves in the South China Sea. Surveys in Geophysics, 33(5), 927–943. https://
seasonal‐mean data are from the World doi.org/10.1007/s10712‐012‐9176‐0
Ocean Atlas (2013, https://www.nodc. Chang, M.‐H., Lien, R.‐C., Tang, T. Y., D'Asaro, E. A., & Yang, Y. J. (2006). Energy flux of nonlinear internal waves in northern South China
noaa.gov/OC5/woa13/woa13data. Sea. Geophysical Research Letters, 33, L03607. https://doi.org/10.1029/2005gl025196
html). da Silva, J. C. B., New, A. L., & Azevedo, A. (2007). On the role of SAR for observing “local generation” of internal solitary waves off the
Iberian Peninsula. Canadian Journal of Remote Sensing, 33(5), 388–403. https://doi.org/10.5589/m07‐041
Diamessis, P., Wunsch, S., Delwiche, I., & Richter, M. (2014). Nonlinear generation of harmonics through the interaction of an internal
wave beam with a model oceanic pycnocline. Dynamics of Atmospheres and Oceans, 66, 110–137. https://doi.org/10.1016/j.
dynatmoce.2014.02.003
Dong, D., Yang, X. F., Li, X. F., & Li, Z. W. (2016). SAR observation of eddy‐induced mode‐2 internal solitary waves in the South China Sea.
IEEE Transactions on Geoscience and Remote Sensing, 54(11), 6674–6686. https://doi.org/10.1109/TGRS.2016.2587752
Duda, T. F., Lynch, J. F., Irish, J. D., Beardsley, R. C., Ramp, S. R., Chiu, C.‐S., et al. (2004). Internal tide and nonlinear internal wave
behavior at the continental slope in the northern South China Sea. IEEE Journal of Oceanic Engineering, 29(4), 1105–1130. https://doi.
org/10.1109/JOE.2004.836998
Dunphy, M., Subich, C., & Stastna, M. (2011). Spectral methods for internal waves: Indistinguishable density profiles and double‐humped
solitary waves. Nonlinear Processes in Geophysics, 18(3), 351–358. https://doi.org/10.5194/npg‐18‐351‐2011
El, G. A., Grimshaw, R. H. J., & Kamchatnov, A. M. (2007). Evolution of solitary waves and undular bores in shallow‐water flows over a
gradual slope with bottom friction. Journal of Fluid Mechanics, 585, 213–244. https://doi.org/10.1017/S0022112007006817

XIE ET AL. 4693

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

Farmer, D. M., Alford, M. H., Lien, R.‐C., Yang, Y. J., Chang, M.‐H., & Li, Q. (2011). From Luzon Strait to Dongsha Plateau: Stages in the life
of an internal wave. Oceanography, 24(4), 64–77. https://doi.org/10.5670/oceanog.2011.95
Farmer, D. M., Li, Q., & Park, J.‐H. (2009). Internal wave observations in the South China Sea: The role of rotation and nonlinearity.
Atmosphere‐Ocean, 47(4), 267–280. https://doi.org/10.3137/OC313.2009
Gerkema, T. (2001). Internal and interfacial tides: Beam scattering and local generation of solitary waves. Journal of Marine Research, 59(2),
227–255. https://doi.org/10.1357/002224001762882646
Grimshaw, R. (1979). Slowly varying solitary waves. I Korteweg‐de Vries equation. Proceeding of the Royal Society of London Series A,
368(1734), 359–375. https://doi.org/10.1098/rspa.1979.0135
Grimshaw, R., Guo, C., Helfrich, K., & Vlasenko, V. (2014). Combined effect of rotation and topography on shoaling oceanic internal
solitary waves. Journal of Physical Oceanography, 44(4), 1116–1132. https://doi.org/10.1175/JPO‐D‐13‐0194.1
Grimshaw, R., Pelinovsky, E., Talipova, T., & Kurkin, A. (2004). Simulation of the transformation of internal solitary waves on oceanic
shelves. Journal of Physical Oceanography, 34(12), 2774–2791. https://doi.org/10.1175/JPO2652.1
Grisouard, N., Staquet, C., & Gerkema, T. (2011). Generation of internal solitary waves in a pycnocline by an internal wave beam: A
numerical study. Journal of Fluid Mechanics, 676, 491–513. https://doi.org/10.1017/jfm.2011.61
Guo, C., & Chen, X. (2012). Numerical investigation of large amplitude second mode internal solitary waves over a slope‐shelf topography.
Ocean Modelling, 42, 80–91. https://doi.org/10.1016/j.ocemod.2011.11.003
Guo, C., & Chen, X. (2014). A review of internal solitary wave dynamics in the northern South China Sea. Progress in Oceanography,
121(2014), 7–23. https://doi.org/10.1016/j.pocean.2013.04.002
Helfrich, K., & Melville, W. (1986). On long nonlinear internal waves over slope‐shelf topography. Journal of Fluid Mechanics, 167(1),
285–308. https://doi.org/10.1017/S0022112086002823
Helfrich, K., & Melville, W. (2006). Long nonlinear internal waves. Annual Review of Fluid Mechanics, 38(1), 395–425. https://doi.org/
10.1146/annurev.fluid.38.050304.092129
Huang, W., J. Johannessen, W. Alpers, J. Yang, and X. Gan (2008), Spatial and temporal variations of internal wave sea surface signatures
in the South China Sea studied by spaceborne SAR imagery, In: Proc. SeaSAR, Frascati, Italy, January 21‐25, pp. 1–6.
Huang, X., Chen, Z., Zhao, W., Zhang, Z., Zhou, C., Yang, Q., & Tian, J. (2016). An extreme internal solitary wave event observed in the
northern South China Sea. Scientific Reports, 6(1), 30041. https://doi.org/10.1038/srep30041
Jackson, C. R. (2009). An empirical model for estimating the geographic location of nonlinear internal solitary waves. Journal of
Atmospheric and Oceanic Technology, 26(10), 2243–2255. https://doi.org/10.1175/2009JTECHO638.1
Jeans, D. R. G., & Sherwin, T. J. (2001). The evolution and energetics of large amplitude nonlinear internal waves on the Portuguese shelf.
Journal of Marine Research, 59(3), 327–353. https://doi.org/10.1357/002224001762842235
Kantha, L. H., & Clayson, C. A. (2000). Numerical models of oceans and oceanic processes (p. 940). San Diego, CA: Academic Press.
Klymak, J. M., Pinkel, R., Liu, C.‐T., Liu, A. K., & David, L. (2006). Prototypical solitons in the South China Sea. Geophysical Research
Letters, 33, L11607. https://doi.org/10.1029/2006GL025932
Lamb, K. G. (2002). A numerical investigation of solitary internal waves with trapped cores formed via shoaling. Journal of Fluid
Mechanics, 451, 109–144. https://doi.org/10.1017/S002211200100636X
Lamb, K. G. (2003). Shoaling solitary internal waves: On a criterion for the formation of waves with trapped cores. Journal of Fluid
Mechanics, 478, 81–100. https://doi.org/10.1017/S0022112002003269
Lamb, K. G. (2014). Internal wave breaking and dissipation mechanisms on the continental slope/shelf. Annual Review of Fluid Mechanics,
46(1), 231–254. https://doi.org/10.1146/annurev‐fluid‐011212‐140701
Lamb, K. G., & Nguyen, V. T. (2009). Calculating energy flux in internal solitary waves with an application to reflectance. Journal of
Physical Oceanography, 39(3), 559–580. https://doi.org/10.1175/2008JPO3882.1
Lamb, K. G., & Warn‐Varnas, A. (2015). Two‐dimensional numerical simulations of shoaling internal solitary waves at the ASIAEX site in
the South China Sea. Nonlinear Processes in Geophysics, 22(3), 289–312. https://doi.org/10.5194/npg‐22‐289‐2015
Li, Q., & Farmer, D. (2011). The generation and evolution of nonlinear internal waves in the deep basin of the South China Sea. Journal of
Physical Oceanography, 41(7), 1345–1363. https://doi.org/10.1175/2011JPO4587.1
Liao, G., Xu, X. H., Liang, C., Dong, C., Zhou, B., Ding, T., et al. (2014). Analysis of kinematic parameters of internal solitary waves in the
northern South China Sea. Deep Sea Research, Part I, 94, 159–172. https://doi.org/10.1016/j.dsr.2014.10.002
Lien, R.‐C., D'Asaro, E., Henyey, F., Chang, M. H., Tang, T. Y., & Yang, Y. J. (2012). Trapped core formation within a shoaling nonlinear
internal wave. Journal of Physical Oceanography, 42(4), 511–525. https://doi.org/10.1175/2011JPO4578.1
Lien, R.‐C., Henyey, F., Ma, B., & Yang, Y. J. (2014). Large‐amplitude internal solitary waves observed in the northern South China Sea:
Properties and energetics. Journal of Physical Oceanography, 44(4), 1095–1115. https://doi.org/10.1175/JPO‐D‐13‐088.1
Liu, A. K., Chang, Y. S., Hsu, M.‐K., & Liang, N. K. (1998). Evolution of nonlinear internal waves in the East and South China Seas. Journal
of Geophysical Research, 103(C4), 7995–8008. https://doi.org/10.1029/97JC01918
Liu, A. K., Su, F. C., Hsu, M. K., Kuo, N. J., & Ho, C. R. (2013). Generation and evolution of mode‐two internal waves in the South China
Sea. Continental Shelf Research, 59, 18–27. https://doi.org/10.1016/j.csr.2013.02.009
Locarnini, R. A., Mishonov, A. V., Antonov, J. I., Boyer, T. P., Garcia, H. E., Baranova, O. K., et al. (2013). World Ocean Atlas 2013, Volume
1: Temperature. S. Levitus, Ed., A. Mishonov Technical Ed.; NOAA Atlas NESDIS (Vol. 73, 40 pp.) Silver Spring MD: National
Oceanographic Data Center, User Services Team, NOAA/NESDIS E/OC1, SSMC III.
Lynch, J. F., Ramp, S. R., Chiu, C.‐S., Tang, T. Y., Yang, Y. J., & Simmen, J. A. (2004). Research highlights from the Asian Seas International
Acoustics Experiment in the South China Sea. IEEE Journal of Oceanic Engineering, 29(4), 1067–1074. https://doi.org/10.1109/
JOE.2005.843162
Ma, X., Yan, J., Hou, Y., Lin, F., & Zheng, X. (2016). Footprints of obliquely incident internal solitary waves and internal tides near the
shelf break in the northern South China Sea. Journal of Geophysical Research:Oceans, 121, 8706–8719. https://doi.org/10.1002/
2016jc012009
MacKinnon, J. A., Alford, M. H., Ansong, J. K., Arbic, B. K., Barna, A., Briegleb, B. P., et al. (2017). Climate process team on internal‐
wave driven ocean mixing. Bulletin of the American Meteorological Society, 98(11), 2429–2454. https://doi.org/10.1175/BAMS‐D‐16‐
0030.1
Mercier, M. J., Mathur, M., Gostiaux, L., Gerkema, T., Magalhães, J. M., da Silva, J. C. B., & Dauxois, T. (2012). Soliton generation by
internal tidal beams impinging on a pycnocline: Laboratory experiments. Journal of Fluid Mechanics, 704, 37–60. https://doi.org/
10.1017/jfm.2012.191
Moore, S., & Lien, R.‐C. (2007). Pilot whales follow internal solitary waves in the South China Sea. Marine Mammal Science, 23(1), 193–196.
https://doi.org/10.1111/j.1748‐7692.2006.00086.x

XIE ET AL. 4694

 Journal of Geophysical Research: Oceans 10.1029/2018JC014837

New, A. L., & Pingree, R. D. (1992). Local generation of internal soliton packets in the central Bay of Biscay. Deep Sea Research, 39(9),
1521–1534. https://doi.org/10.1016/0198‐0149(92)90045‐U
Orr, M. H., & Mignerey, P. C. (2003). Nonlinear internal waves in the South China Sea: Observation of the conversion of depression to
elevation internal waves. Journal of Geophysical Research, 108(C3), 3064. https://doi.org/10.1029/2001JC001163
Osborne, A. R., & Burch, T. L. (1980). Internal solitons in the Andaman Sea. Science, 208(4443), 451–460. https://doi.org/10.1126/
science.208.4443.451
Osborne, A. R., Burch, T. L., & Scarlet, R. I. (1978). The influence of internal waves on deepwater drilling. Journal of Petroleum Technology,
30(10), 1497–1504. https://doi.org/10.2118/6913‐PA
Pakanowski, R. C., & Philander, S. G. H. (1981). Parameterisation of vertical mixing in numerical models of tropical oceans. Journal of
Physical Oceanography, 11(11), 1443–1451. https://doi.org/10.1175/1520‐0485(1981)011<1443:POVMIN>2.0.CO;2
Ramp, S. R., Tang, T. Y., Duda, T. F., Lynch, J. F., Liu, A. K., Chiu, C.‐S., et al. (2004). Internal solitons in the northeastern South China Sea.
Part I: Sources and deep water propagation. IEEE Journal of Oceanic Engineering, 29(4), 1157–1181. https://doi.org/10.1109/
JOE.2004.840839
Ramp, S. R., Yang, Y. J., & Bahr, F. L. (2010). Characterizing the nonlinear internal wave climate in the northeastern South China Sea.
Nonlinear Processes in Geophysics, 17(5), 481–498. https://doi.org/10.5194/npg‐17‐481‐2010
Reeder, D. B., Ma, B. B., & Yang, Y. J. (2011). Very large subaqueous sand dunes on the upper continental slope in the South China Sea
generated by episodic shoaling deep‐water internal solitary waves. Marine Geology, 279(1‐4), 12–18. https://doi.org/10.1016/j.
margeo.2010.10.009
Sarkar, S., & Scotti, A. (2017). From topographic internal gravity waves to turbulence. Annual Review of Fluid Mechanics, 49(1), 195–220.
https://doi.org/10.1146/annurev‐fluid‐010816‐060013
Shen, C. Y., Evans, T. E., Oba, R. M., & Finette, S. (2009). Three‐dimensional hindcast simulation of internal soliton propagation in the
Asian Seas International Acoustics Experiment area. Journal of Geophysical Research, 114(C1), C01014. https://doi.org/10.1029/
2008jc004937
Stacey, M. W., & Zedel, L. (1986). The time‐dependent hydraulic flow and dissipation over the sill of Observatory Inlet. Journal of Physical
Oceanography, 16(6), 1062–1076. https://doi.org/10.1175/1520‐0485(1986)016<1062:TTDHFA>2.0.CO;2
Stastna, M., & Lamb, K. G. (2008). Sediment resuspension mechanisms associated with internal waves in coastal waters. Journal of
Geophysical Research, 113(C10), C10016. https://doi.org/10.1029/2007JC004711
Talley, L. D., Pickard, G. L., Emery, W. J., & Swift, J. H. (2011). Data analysis concepts and observational methods. In Descriptive physical
oceanography: An introduction (Chap. 2.2, p. 13). London, UK: Elsevier. https://doi.org/10.1016/B978‐0‐7506‐4552‐2.10001‐0
Turkington, B., Eydeland, A., & Wang, S. (1991). A computational method for solitary internal waves in a continuously stratified fluid.
Studies in Applied Mathematics, 85(2), 93–127. https://doi.org/10.1002/sapm199185293
Vitousek, S., & Fringer, O. B. (2011). Physical vs. numerical dispersion in nonhydrostatic ocean modeling. Ocean Modelling, 40(1), 72–86.
https://doi.org/10.1016/j.ocemod.2011.07.002
Vlasenko, V., & Alpers, W. (2005). Generation of secondary internal waves by the interaction of an internal solitary wave with an under-
water bank. Journal of Geophysical Research, 110(C2), C02019.
Vlasenko, V., & Stashchuk, N. (2006). Amplification and suppression of internal waves by tides over variable bottom topography. Journal of
Physical Oceanography, 36(10), 1959–1973. https://doi.org/10.1175/JPO2958.1
Vlasenko, V., Stashchuk, N., & Hutter, K. (2005). Baroclinic tides: Theoretical modeling and observational evidence (p. 351). Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511535932
Wang, Y. H., Dai, C. F., & Chen, Y. Y. (2007). Physical and ecological processes of internal waves on an isolated reef ecosystem in the South
China Sea. Geophysical Research Letters, 34, L18609. https://doi.org/10.1029/2007GL030658
Xie, J., He, Y., Chen, Z., Xu, J., & Cai, S. (2015). Simulations of internal solitary wave interactions with mesoscale eddies in the northeastern
South China Sea. Journal of Physical Oceanography, 45(12), 2959–2978. https://doi.org/10.1175/JPO‐D‐15‐0029.1
Xie, J., He, Y., Lü, H., Chen, Z., Xu, J., & Cai, S. (2016). Distortion and broadening of internal solitary wavefront in the northeastern South
China Sea deep basin. Geophysical Research Letters, 43, 7617–7624. https://doi.org/10.1002/2016GL070093
Xie, J., Pan, J., & David, J. (2015). Multimodal internal waves generated over a subcritical ridge: Impact of the upper‐ocean stratification.
Journal of Physical Oceanography, 45(3), 904–926. https://doi.org/10.1175/JPO‐D‐14‐0132.1
Xie, J., Xu, J., & Cai, S. (2011). A numerical study of the load on cylindrical piles exerted by internal solitary waves. Journal of Fluids and
Structures, 27(8), 1252–1261. https://doi.org/10.1016/j.jfluidstructs.2011.04.007
Yang, Y. J., Fang, Y. C., Chang, M.‐H., Ramp, S. R., Kao, C.‐C., & Tang, T. Y. (2009). Observations of second baroclinic mode internal
solitary waves on the continental slope of the northern South China Sea. Journal of Geophysical Research, 114(C10), C10003. https://doi.
org/10.1029/2009JC005318
Yang, Y. J., Tang, T. Y., Chang, M.‐H., Liu, A. K., Hsu, M.‐K., & Ramp, S. R. (2004). Solitons northeast of Tung‐Sha Island during the
ASIAEX pilot studies. IEEE Journal of Oceanic Engineering, 29(4), 1182–1199. https://doi.org/10.1109/JOE.2004.841424
Zhao, Z., Klemas, V., Zheng, Q., & Yan, X.‐H. (2004). Remote sensing evidence for baroclinic tide origin of internal solitary waves in the
northeastern South China Sea. Geophysical Research Letters, 31, L06302. https://doi.org/10.1029/2003gl019077
Zhao, Z., Liu, B., & Li, X. (2014). Internal solitary waves in the China seas observed using satellite remote‐sensing techniques: A review and
perspectives. International Journal of Remote Sensing, 35(11–12), 3926–3946. https://doi.org/10.1080/01431161.2014.916442
Zheng, Q., Susanto, R. D., Ho, C.‐R., Song, Y., & Xu, Q. (2007). Statistical and dynamical analyses of generation mechanisms of solitary
internal waves in the northern South China Sea. Journal of Geophysical Research, 112(C3), C03021. https://doi.org/10.1029/
2006jc003551

XIE ET AL. 4695