Ocean Engineering 141 (2017) 308–325

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Undular hydraulic jump formation and energy loss in a flow through MARK
emergent vegetation of varying thickness and density
⁎
Ghufran Ahmed Pashaa, Norio Tanakaa,b,
a
Graduate School of Science and Engineering, Saitama University, 255 Shimo-okubo, Sakura-ku, Saitama 338-8570, Japan
b
International Institute for Resilient Society, Saitama University, 255 Shimo-okubo, Sakura-ku, Saitama 338-8570, Japan

A R T I C L E I N F O A BS T RAC T

Keywords: Floods resulting from extreme events like tsunamis may inundate widespread inland areas, but vegetation can
Tsunami act as a natural buffer zone to reduce the inundation area and dissipate the energy of flowing water. This paper
Vegetation summarizes a series of laboratory experiments in which the energy loss through emergent vegetation in a steady
Backwater rise subcritical flow was investigated. The energy loss was determined against vegetation of variable thickness (dn,
Undular hydraulic jump
where d = diameter of cylinder, n = number of cylinders in a stream-wise direction per unit of cross-stream
Energy loss
width), density (G/d, where G = spacing of each cylinder in cross-stream direction, d = diameter of cylinder),
Drag coefficient
and initial Froude number. On the upstream side of vegetation, the backwater rise increased by increasing both
vegetation thickness and density. Contrarily, on the downstream side a breaking undular jump with a lateral
shock wave was observed for a dense vegetation arrangement (G/d = 0.25), whereas a non-breaking undular
jump with and without air bubbles was identified for intermediate (G/d = 1.09) and sparse (G/d = 2.13)
vegetation conditions, respectively. Under these conditions, the maximum energy reduction due to a jump
reached 6.4% for dense vegetation, and was reduced to 1.7% and 1.4% for intermediate and sparse vegetations,
respectively. Hence, denser vegetation offers larger resistance, thus causes significant energy loss.

1. Introduction
Various researchers have worked in recent years to derive the best
possible method for tsunami mitigation (Nateghi et al., 2016; Rahman
et al., 2017; Tanaka, 2012; Thuy et al., 2012). Both artificial (hard
solutions) and natural (soft solutions) methods can be implemented to
dissipate the energy carried by the huge currents of a tsunami. Artificial
methods include construction of sea walls and embankments, installing
tsunami gates, and putting up breakwater structures. These methods
can prove costly for developing countries because they require a huge
capital investment (Tanaka, 2009). Further, the money will be wasted if
a large tsunami arrives exceeding the designed capacity of the artificial
structure, thus causing disastrous damage to people, infrastructures
(Suppasri et al., 2012; Fraser et al., 2013; Ishigaki et al., 2013), and
even coastal forests (Tanaka et al., 2013). Currently, natural methods
such as coastal forests are widely considered to be an effective measure
to mitigate tsunami damage from both economic and environmental
points of view (Kathiresan and Rajendran, 2005; Osti et al., 2009;
Tanaka, 2009; Yanagisawa et al., 2009, 2010).
Based on previous work (Shuto, 1987; Tanaka et al., 2007), the roles of
coastal vegetation in tsunami mitigation include trapping, energy dissipa-
tion, a soft landing place, and an escape route. The effectiveness of a coastal
forest as protection depends upon various factors, including tree density
(Harada and Kawata, 2005; Irtem et al., 2009; Tanaka et al., 2011),
deficiencies in coastal forests such as open gaps (Mascarenhas and
Jayakumar, 2008; Thuy et al., 2009), the breaking moment of trees
(Tanaka et al., 2009), and magnitude of the tsunami (Tanaka, 2009).
Since the 1998 Papua New Guinea tsunami (Dengler and Preuss, 2003),
many researchers have investigated the effects of coastal vegetation on
tsunami mitigation. Hiraishi and Harada (2003) did field investigations
after the Papua New Guinea tsunami and proposed a green belt of mango
or coconut trees to reduce the tsunami force. Previous post-tsunami surveys
have also shown that coastal trees help reduce the damaging effects of
natural disasters (Danielsen et al., 2005; Kathiresan and Rajendran, 2005;
Tanaka et al., 2007; Mascarenhas and Jayakumar, 2008). To utilize the
tsunami mitigation function of a coastal forest, field plantation has also
been done along the coast of southern Asian countries (Tanaka, 2009;
Tanaka et al., 2011).
Hydraulic resistance and reflection of water by trees can reduce the
energy of flowing water, inundation depth, inundation area, and
hydraulic force behind the vegetation. The force of the water passing
through the vegetation becomes weaker, which minimizes the damage

⁎
Corresponding author at: International Institute for Resilient Society, Saitama University, 255 Shimo-okubo, Sakura-ku, Saitama 338–8570, Japan.
E-mail address: tanaka01@mail.saitama-u.ac.jp (N. Tanaka).

http://dx.doi.org/10.1016/j.oceaneng.2017.06.049
Received 29 June 2016; Received in revised form 10 May 2017; Accepted 19 June 2017
0029-8018/ © 2017 Elsevier Ltd. All rights reserved.
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

behind the vegetation. Iimura and Tanaka (2012) investigated the
effects of vegetation density both experimentally and analytically and
confirmed that both the level and velocity of the water behind the
vegetation are reduced considerably by increasing the density of
vegetation. Denser vegetation causes increased water reflection and
resistance because the water surface slope in the vegetation region is
increased. Later, Iimura and Tanaka (2013) studied the effects of
aspect ratio on tsunami mitigation using numerical simulations and
found that the effect of the collision of an inundating current behind a
forest is large if the aspect ratio is between 1 and 4. An increase in
forest width can reduce not only inundation depth but also the current
and hydraulic force behind a coastal forest (Harada and Imamura,
2005). Moreover, a few studies changed the tree density distribution in
a forest and investigated the energy reductions behind the forest (Thuy
et al., 2009; Iimura and Tanaka, 2012). However, no one has reported
the flow structure and patterns downstream of a forest and possible
energy reduction due to the formation of a hydraulic jump.
In open channels, the transition from supercritical to subcritical flow is
called a hydraulic jump. The different kinds of hydraulic jumps were
described by Chow (1959). A jump is characterized by the development of
large scale turbulence, surface waves, energy dissipation, and air entrain-
ment. For a Froude number slightly above unity, the hydraulic jump is
considered a smooth rise of the free surface, followed by a train of
stationary free-surface undulations known as an undular hydraulic jump
(Chanson, 2009). Undular jumps are often observed downstream of low
drop-structures or in a transitional region from steep to mildly sloping
channels (Ohtsu et al., 2003). An undular hydraulic jump may also occur
behind ships travelling in canals and shallow waters (Haslewood, 1985).
Various researchers have done experimental studies in an open channel to
study the flow of an undular hydraulic jump by a sluice gate (Chanson and
Montes, 1995; Ohtsu et al., 2001, 2003), but the development of an undular
hydraulic jump formed in the wake of a coastal forest, especially during a
tsunami event, has never been reported. However, a few undular flow
patterns were observed in a forest downstream during the 2011 Japan
tsunami near a bank of the Yuriage area (photo 16, Tokida and Tanimoto,
2014). The arrangement of trees in a forest can transform the flow into a
supercritical flow that results in the occurrence of a hydraulic jump. To
clarify the energy loss mechanism, not only by drag force but also the
downstream flow pattern change, flume experiments were conducted to
determine the amount of energy loss through emergent vegetation of
variable thicknesses, changing density, and flow conditions. Different flow
patterns of undular hydraulic jumps were also identified. This study will
help in selecting an optimal forest width in relation to different flow
conditions and vegetation arrangements.
2. Materials and methods
2.1. Experimental procedure and flume characteristics
2.1.1. Flume characteristics
Laboratory experiments with nine different conditions (Table 1)
were conducted in a glass-sided water flume (constant bed slope 1/
500) that is 5 m in length, 0.7 m in width, and 0.5 m in height at
Saitama University. Fig. 1a shows a schematic figure of the water
channel. In the experimental channel, the ground conditions and
tsunami characteristics implicated were not specific to any location
but were considered in general. Based on video analysis, Nandasena
et al. (2012) estimated that the maximum Froude number (Fr = V/
(gh)0.5, where V = depth-averaged velocity (m/s), g = gravitational
acceleration (m/s2), and h = water depth (m)) was around 1.5 at the
time the tsunami reached the Misawa and Hachinohe shores in Japan,
whilst the Froude number was about 1.14–1.4 at the wave front 1 km
inland from the coast of the Sendai Plain which had no tsunami
mitigation structure such as a coastal forest. At many locations
inundated by the tsunami, the tsunami's flow was subcritical and had
a Froude number between 0.7 and 1 (Spiske et al., 2010). For the Great
East Japan tsunami, Tanaka et al. (2013) estimated the Froude number
as 0.9–0.6 at 500–550 m distance from the shoreline, and a simulation
by Tanaka et al. (2014) estimated the Froude number to decrease from
1.6 near the shoreline to 0.6 around 550 m from the shoreline in the
Sendai Plain. In the current study, to set the flow conditions, Froude
similarity was applied to set the model scale of the physical experiment.
This study defined the initial Froude number (Fro) when the reference
velocity and water depth were used without a vegetation model. For
creating subcritical conditions of an inundating tsunami, the water
depths (model without vegetation) selected in the experiment were 3,
3.5, 4, 4.5, 5, 5.5, 6, 6.5 and 7 cm, setting the initial Froude number
approximately equal to 0.57, 0.62, 0.65, 0.66, 0.68, 0.69, 0.70, 0.71,
and 0.73, respectively. As a first step, a slightly lower range of initial
Froude numbers was selected due to the restricted length of the
channel. This is because flow with a higher Froude number increases
the length of the hydraulic jump. Also, the behavior of an undular
hydraulic jump downstream of vegetation is investigated better against
a low initial Froude number. As discussed in the previous studies
(Tanaka et al., 2013; Pasha and Tanaka, 2016), it is better to set a
coastal forest inland, where it is expected to trap tsunami-borne
floating debris. From that point of view, the experimental condition
is set in the lower range of Froude numbers in the post-tsunami
research (Spiske et al., 2010; Tanaka et al., 2013, 2014).

Table 1
Experimental conditions.

Case Initial Froude number ‘Fro’ Vegetation density D (cm) W (cm) Vegetation thickness ‘dn’ Vegetation
No. ‘G/d’ (No. cm) type

1 0.57, 0.62, 0.65, 0.66, 0.68, 0.25 1 3.88 179.21 Dense

0.69, 0.70, 0.71, 0.73
2 0.57, 0.62, 0.65, 0.66, 0.68, 1.09 1.67 10.55 174.72 Intermediate
0.69, 0.70, 0.71, 0.73
3 0.57, 0.62, 0.65, 0.66, 0.68, 2.13 2.5 24.27 179.36 Sparse
0.69, 0.70, 0.71, 0.73
4 0.57, 0.62, 0.65, 0.66, 0.68, 0.25 1 8.23 380.13 Dense
0.69, 0.70, 0.71, 0.73
5 0.57, 0.62, 0.65, 0.66, 0.68, 1.09 1.67 23.60 390.85 Intermediate
0.69, 0.70, 0.71, 0.73
6 0.57, 0.62, 0.65, 0.66, 0.68, 2.13 2.5 52.48 387.83 Sparse
0.69, 0.70, 0.71, 0.73
7 0.57, 0.62, 0.65, 0.66, 0.68, 0.25 1 12.58 581.05 Dense
0.69, 0.70, 0.71, 0.73
8 0.57, 0.62, 0.65, 0.66, 0.68, 1.09 1.67 35.2 582.96 Intermediate
0.69, 0.70, 0.71, 0.73
9 0.57, 0.62, 0.65, 0.66, 0.68, 2.13 2.5 78.52 580.27 Sparse
0.69, 0.70, 0.71, 0.73

309
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Fig. 1. Experimental set-up: (a) schematic figure of channel with vegetation model, (b, c) set-up of load cell at vegetation front, (d) details of vegetation arrangement and its definition
(e) dense vegetation model with dn-380.13 No. cm, (f) intermediate vegetation model with dn-390.85 No. cm, (g) sparse vegetation model with dn-387.83 No. cm, and (h) flow structure
scheme and definition of various parameters.

310
G.A. Pasha, N. Tanaka
2.1.2. Scaling
The scale of the current experimental study was 1/100. Because the
flow conditions considered in the experiment are subcritical, they
represent tsunami flow in an inland area at some distance from the
shore line. According to Tanaka and Suzuki (2013), the water depth
behind the coastal forest in the Miyagino district was 3.0–3.6 m, while
the point at which the water depth was measured was about 800 m
from the shoreline; and the water depth at the front area of the house
group was 2.5–4.6 m at 1200 m from the shoreline. If the length ratio
between prototype (Lp) and model (Lm) is Ls = Lp /Lm, then scaling an
experimental water depth of 5 cm (average of selected water depths) to
a tsunami water height of 4.5–5 m requires a scaling factor, Ls, that is
approximately equal to 100.
2.1.3. Experimental conditions
Vegetation models covering the full width of the channel were
mounted in the water flume bed about 2.5 m from the upstream inlet.
The water level was measured throughout the center of the channel as
indicated in Fig. 1a by using a point gauge at 2–5 cm intervals
(depending on the variation in the water surface). Before mounting
the vegetation model in a channel, the discharge was measured using a
flow meter (Signet 8150 Flow Totalizer) against each selected initial
water depth (ho), and the discharge values against each initial water
depth (ho) were noted. Subsequently, after placing the vegetation
model in a channel, the discharge values were changed according to
the selected nine flow conditions. The depth-averaged velocity was
computed using the measured water depth and discharge. A two-axis
load cell (force gauge) (SSK Co., model LB60) having a resolution of 1/
1000 and capacity to measure a maximum load of 1 N was used to
measure the drag force on an individual cylinder (Fig. 1b). The drag
coefficient Cd for each solid circular cylinder was defined as:
Cd =
2F
ρV 2dH (1)
where F (N) is the drag force on individual centrally positioned
cylinders at the forest front and forest back, ρ (kg/m3) is the fluid
density, d (m) is the diameter of the cylinder, and H (m) is the water
depth in front of the cylinder. V (m/s) is the calculated mean velocity
against water depth H (m) and discharge Q (m3/s). In addition, to
measure the drag force, a gap is required between individual cylinders
and the bottom of the channel as indicated in Fig. 1c. The experiments
were conducted with the smallest possible gap, i.e., 0.001 m, similar to
the measurements taken by Takemura and Tanaka (2007). The drag
coefficient of individual cylinders was determined both at vegetation
front and back, and the position of load cell is indicated in Fig. 1a.
2.2. Vegetation conditions
The tree species selected for the vegetation model was the Japanese
pine tree (average tree height = 15 m and trunk diameter = 0.4 m)
found on the Sendai Plain; its tree crown was high relative to the
tsunami height, and the trees can be considered as circular cylinders
(Tanaka et al., 2014). For a 1/100 scale model, wooden cylinders with a
diameter of 0.004 m were used as tree models in a staggered arrange-
ment. Fig. 1d and Table 1 show details of the vegetation arrangement,
where D is the distance between cylinders and W is the width of the
vegetation model.
According to Takemura and Tanaka (2007), flow structures differ
depending on the G/d arrangement of the vegetation model (vegetation
density), where G represents the spacing between each cylinder in a
cross-stream direction, and d is the diameter of a cylinder (Fig. 1d).
The G/d ratio indicates whether vegetation density is sparse or dense.
In order to investigate the energy loss through vegetation, vegetation
models with three different G/d values (0.25, 1.09, and 2.13) were
made. A model with G/d of 0.25 represents dense vegetation, while G/d
of 1.09 and 2.13 characterize intermediate and sparse vegetations,
311
Ocean Engineering 141 (2017) 308–325
respectively (Fig. 1e, f, g). D and W were determined under the same
vegetation thickness: dn (No. cm), which is expressed as a function of
summed tree diameter and is defined as a product of the diameter of a
tree (d) at breast height and the number of trees (n) in a rectangle with
a frontage of unit length along the shoreline and depth equal to the
width of the forest (W) (Fig. 1d), (Shuto, 1987). In this study, dn was
calculated as:
2
dn = Wd × 10 2
3 D2 (2)
2
where 10 in the above equation adjusts a unit in Shuto's definition of
dn (No. cm) on a 1:100 scale, because d, D, and W are in centimeters.
In the experiment dn was set to three values, i.e., ~ 180, ~ 380, and ~
580 (No. cm). Shuto (1987) classified the effectiveness of a forest and
the degree of damage to the forest and land in terms of tsunami
inundation depth, vegetation thickness, and presence of undergrowth.
Referring to his study, here dn ~ 180 No. cm represents areas where
some trees on weak soil or at the forest fringe may be damaged and soil
around the trees may be scoured; dn ~ 380 No. cm corresponds to an
area where soil in the forest may be scoured and damaged to some
extent; dn ~ 580 No. cm represents an area where neither damage to
trees nor damage to soil occurs.
2.3. Non-dimensional pi groups
Scaling experimental results to prototype conditions is challenging.
Fig. 1h shows a flow structure scheme and important parameters. Flow
conditions and vegetation arrangement are the main parameters that
decide the amount of energy lost. Using Buckingham's pi theorem, the
following dimensionless groups were developed:
⎡ Δh Vo V1 ρw hoVo G
f⎢ , , , , , dn*,
⎢⎣ ho gho gh1 μ d
ho y2 b LJ V2 ΔE1 ΔE2 ⎤
, , , , , , ⎥=0
L y1 y1 y1 V1 E1 y1 ⎥⎦ (3)
where Δh = backwater rise, ho = initial water depth without vegetation,
Vo = velocity at ho, Fro = Froude number at ho (initial Froude number),
FUS = Froude number on the upstream of vegetation against water
depth h1 and velocity V1, g = gravitational acceleration, ρw = density of
water, μ = viscosity of water, L = wavelength, dn* = dn/D is non-
dimensional vegetation width, y1 = minimum water depth during the
jump (water depth at the toe of a hydraulic jump), y2 = mean water
depth after the hydraulic jump, b = channel width, LJ = length of
hydraulic jump, V2 = mean velocity after hydraulic jump, HGL =
hydraulic grade line, EGL = energy grade line, E1 = specific energy at
vegetation front, E2 = mean specific energy after formation of a
hydraulic jump where the flow is subcritical, ΔE1 = total energy loss
(E1-E2), and ΔE2 = energy loss due to hydraulic jump [(y2-y1)3/4y1y2]
(Fig. 1d, h). (For the definitions of G, d, dn, and D, see Section 2.2).
A smooth glass-sided channel with an extremely small bed slope
was used for the experiment, and therefore the effects of channel slope
and wall roughness were excluded. Because the viscosity and density of
water were same for every case and Froude scaling is commonly used
for free surface gravity flows, the π4 group (the Reynolds number) was
ignored. Therefore, the backwater rise (Δh/ho), total energy loss (ΔE1/
E1), and energy loss due to the hydraulic jump (ΔE2/y1) are a function
of the initial Froude number (Fro = Vo / gho ), vegetation density (G/d),
and non-dimensional vegetation width (dn*):
Δh ΔE1 ΔE2 ⎡ V G ⎤
, , = f ⎢ o , , dn*⎥
ho E1 y1 ⎣⎢ gho d ⎦⎥ (4)
However, traditionally, vegetation thickness (dn) is commonly used
to express the resistance of vegetation (Shuto, 1987; Iimura and
Tanaka, 2012); thus, instead of non-dimensional vegetation width
G.A. Pasha, N. Tanaka
Fig. 2. Flow phenomenon describing backwater rise and hydraulic jump formation.
(dn*), vegetation thickness (dn) was used to plot the relationships
between backwater rise and energy loss in the current study.
3. Results and discussion
3.1. Flow structure around vegetation model
As seen in Fig. 2, the initial water depth (ho) without vegetation,
which was constant throughout the channel, was raised upstream of the
vegetation model, and the water surface slope inside the vegetation
became larger due to the resistance offered by the trees. However, on
the downstream side, an undular hydraulic jump was observed.
3.1.1. Upstream of vegetation model – backwater rise
Increasing the vegetation density raises the maximum water level in
front of the vegetation, and the water surface slope inside the
vegetation increases (Iimura and Tanaka, 2012). The initial flow
conditions without vegetation models placed in the channel varied
between Fro = 0.57 and Fro = 0.73. Fig. 3a, b, c shows that the
backwater rise increased almost linearly with increasing Froude
number. The increased velocity for the higher Froude number, i.e.,
higher energy head for higher velocities, consequently increased the
backwater rise. However, the relative backwater rise was increased only
slightly for different Froude conditions (Fig. 3d). This shows that the
initial flow depth had very little effect on the relative backwater rise for
a given Froude number.
Other than the Froude number, the backwater rise greatly depends
on the vegetation arrangement, i.e., vegetation density and thickness.
Fig. 3a, b, c also shows the back water rise for the three types of
vegetation, i.e., dense (G/d = 0.25), intermediate (G/d = 1.09), and
sparse (G/d = 2.13), respectively. While keeping the spacing between
cylinders (G/d) constant and increasing vegetation thickness (dn), the
relative backwater rise was increased by 30–35% when dn was raised
from 180 No. cm to 380 No. cm and was further raised by 20–25%
while increasing dn from 380 No. cm to 580 No. cm. Similarly, the
relative backwater rise was also increased by 25–30% by reducing the
distance between cylinders from dense to sparse and keeping the
vegetation thickness constant. It can be seen in Fig. 3 that the
backwater rise was lower for the sparse vegetation condition (Fig. 3c)
compared to the dense condition (Fig. 3a) for a given Froude number
and thickness. A dense arrangement of trees caused a higher flow
resistance compared to a sparse arrangement and consequently
resulted in a higher backwater rise.
312
Ocean Engineering 141 (2017) 308–325
3.1.2. Downstream of vegetation model – undular hydraulic jumps
and classification
As the water flows through the vegetation, it faces great resistance,
resulting in decreased water depth as it progresses. Depending upon
the resistance offered by the trees, if the water depth (at the vegetation
end) is less than the critical depth, then the flow becomes supercritical.
This supercritical flow will be transformed into a subcritical flow by the
formation of a hydraulic jump. In the current study, undular hydraulic
jumps were observed downstream of the vegetation model depending
on the initial Froude number, vegetation density, and vegetation
thickness (Fig. 2). Undular hydraulic jumps were classified by the ratio
of initial water depth to wavelength (ho/L) (Fig. 4a), using detailed
analysis of identical pictures of every case and with the help of water
profiles drawn for the center cross-section of the flow by using a point
gauge. Fig. 4b shows the flow chart of classification of undular
hydraulic jumps. Here, the undular jump is classified into three
categories; non-breaking undular (NBU), breaking undular (BU)
jumps, and no jump. The non-breaking undular jump is subdivided
into three groups depending on the initial water depth to wavelength
ratio; namely, non-breaking undular jumps with a small wavelength
(0.25 ≤ ho/L ≤ 0.50) without air bubbles (NBS) and with air bubbles
(NBSA), medium wavelength (0.20 ≤ ho/L < 0.25) without air bubbles
(NBM) and with air bubbles (NBMA), and large wavelength without air
bubbles (0.15 ≤ ho/L < 0.20) (NBL). In some cases of low Froude
number flow, a few undulations were observed, which was classified as
no jump (NJ). Also, the cases in which the length of a jump exceeded
the available channel length was classified as a jump exceeding the
channel (JEC) in the current classification. Fig. 5a–f shows different
patterns of undular hydraulic jumps. Fig. 5g shows the flow in a
subcritical state with no hydraulic jump (NJ) due to low Froude
number (Fr1 < 1), while Fig. 5h represents a supercritical flow with
the length of the jump exceeding the channel length with Fr1 > 1
(JEC). The water profiles of five out of nine flow conditions of (i) dense,
(ii) intermediate, and (iii) sparse vegetation against dn-180, -380, and
-580 (No. cm) are shown in Fig. 6a, b, and c, respectively. The
classification of the hydraulic jump along with the water depth without
vegetation and critical depth are also shown for every water profile in
Fig. 6. Table 2 shows the values of Froude number (Fr1) at minimum
water depth during the jump (y1) for Fig. 6.
3.1.2.1. Vegetation conditions for formation of undular jumps. Fig. 7
shows the patterns of undular hydraulic jumps against different
vegetation densities and initial Froude number (Fro). For vegetation
having a constant thickness and sparse conditions (G/d = 2.13), mostly
G.A. Pasha, N. Tanaka
Fig. 3. Backwater rise Δh for various initial Froude numbers Fro: (a) dense vegetation (G/d = 0.25), (b) intermediate vegetation (G/d = 1.09), (c) sparse vegetation (G/d = 2.13), and (d)
relative backwater rise Δh/ho, where ho is initial water depth without vegetation and dn has units of No. cm.
non-breaking undular hydraulic jumps were observed. At a very low
initial Froude number, no hydraulic jump (NJ) was observed; however,
by increasing Fro, the initial water depth to wavelength ratio of undular
waves started to decrease along with increases in the length of the
jump. Lateral shock waves of extremely small intensity and located very
far from the vegetation model were seen, but they did not affect the
flow pattern (represented by dashed lines in Fig. 5).
In intermediate vegetation (G/d = 1.09), non-breaking undular
jumps with a small wavelength (NBS) were observed for low Fro. As the
Fro increased, the intensity of jumps also increased and included
prominent air bubbles. As the Fro increased further, air bubbles started
to diminish with increased length of jumps. At the maximum Fro value
(0.73), lateral shock waves were observed, which changed the pattern
of undular hydraulic jumps from non-breaking to breaking undular
jumps.
In the dense vegetation arrangement, non-breaking undular jumps
with a small wavelength (NBS) were observed for very low Fro. In this
case, lateral shock waves formed, which turned the non-breaking
undular jumps into breaking undular jumps with the increase in Fro.
The shock waves were initiated upstream of the first wave crest and
propagated downstream across the flume. According to Montes (1986),
the presence of lateral shock waves is associated with the sidewall
boundary layers. He suggested that the lateral boundary layers retard
the fluid near the wall and develop the critical conditions faster than on
the channel centerline. Chanson (2009) argued that the shock waves
area form of flow separation at the sidewalls, since flow recirculation is
313
Ocean Engineering 141 (2017) 308–325
observed immediately downstream of the shock wave near the free
surface. For the last three flow conditions (Fro = 0.70, 0.71, and 0.73),
the length of the hydraulic jump was so large that it crossed the
available channel length, and supercritical flow conditions prevailed on
the downstream side of the vegetation model. A jump exceeding the
channel (JEC) was observed in this case due to the limitation of the
channel length. Moreover, no air bubbles were identified in the dense
vegetation arrangement.
In the current study, the shock waves were generated due to
channel limitations; however, they didn't affect the characteristics of
the undular hydraulic jump. In Fig. 5, dashed lines represent the
direction of shock waves, and it can be seen that wavelength
characteristics do not change due to shock waves (Fig. 5a, c e).
Moreover, the water depth and energy loss were calculated only at
the center of the channel where the effect of shock waves was much
less. However, the shock wave slightly affected the continuity of
undular patterns in the case of a breaking undular hydraulic jump,
but due to calculation of mean value of energy loss, the effect of
shock waves became negligible.
3.1.2.2. Hydraulic conditions for formation of undular jumps. Fig. 8
summarizes the hydraulic conditions other than the vegetation
conditions required to form various flow conditions of undular
jumps. The vertical axis is the ratio of channel width (b) to water
depth at the toe of a hydraulic jump (y1), and the horizontal axis is the
Froude number (Fr1) against y1. The toe of a hydraulic jump is the
minimum water depth during the length of a jump and was determined
G.A. Pasha, N. Tanaka
Fig. 4. (a) Definition of initial water depth to wavelength ratio (ho/L). NBS: non-
breaking undular jump with small wavelength, NBM: non-breaking undular jump with
medium wavelength, NBL: nonbreaking undular jump with large wavelength, (b)
classification of undular hydraulic jumps.
at the section where the pressure becomes hydrostatic; thus, the
streamline curvature can be neglected (Fig. 1h). Fig. 8 was plotted by
comparing all the values of b/y1 irrespective of the vegetation
conditions (vegetation density and thickness) with Fr1. The overall
trend shows that, against a constant b/y1, increasing Fr1 changes the
undular hydraulic jumps patterns from:
No jump → few undulations → non-breaking hydraulic jump of
small wavelength → non-breaking hydraulic jump of medium wave-
length → non-breaking hydraulic jump of large wavelength → breaking
undular hydraulic jump.
Air bubble formation in the non-breaking undular jump was
observed for flows having Fr1 between 1.0 and 1.45 and b/y1 values
less than 25 (approximately). For vegetation conditions G/d = 2.13 and
dn-180 No. cm, the Fr1 calculated was less than 1; however, the
undular hydraulic jump classification using ho/L was categorized as an
undular hydraulic jump of small or medium wavelength (Fig. 7a) and is
marked by section A in Fig. 8. Section B is also a portion where Fr1 was
less than 1, and it includes vegetation conditions of G/d = 2.13 and dn-
580 No. cm and a few cases of G/d = 1.09 and dn-580 No. cm. The Fr1
was calculated by using the measured water level and steady state
discharge. However, the local velocities downstream of the vegetation
are assumed to be slightly higher, which resulted in development of few
undulations even with Fr1 < 1 for some cases.
In the current hydraulic conditions for formation of undular
hydraulic jumps, against 10 < b/y1 < 40, Fr1 = 1.7 is the upper limit
of non-breaking undular jumps (Line C in Fig. 8), whereas Fr1 = 1.9 is
314
Ocean Engineering 141 (2017) 308–325
the upper limit of breaking undular jumps (Line D1 in Fig. 8). The
maximum Fr1 obtained in this study is 2.0 (Line D2 in Fig. 8), which
was against highest initial Froude number with vegetation G/d = 1.09
and dn-580 No. cm, but due to channel limitations, the length of
hydraulic jump exceeded the available channel length, and thus it was
not classified.
Ohtsu et al. (2003) identified the flow conditions of undular
hydraulic jumps in a horizontal channel controlled by a sluice gate
and classified the jumps based on lateral shock waves and inflow
Froude number (F1). He classified the undular jumps into non-break-
ing undular hydraulic jump, breaking undular hydraulic jump, and
weak classical jump. According to his findings, the upper limit of the
inflow Froude number for a non-breaking undular hydraulic jump is
1.78 (b/y1 ≥ 10) and for a breaking undular hydraulic jump is 2.1 (b/y1
≥ 12), which is very close to the upper limits of Fr1 obtained in the
current experimental study.
3.2. Variation of drag coefficient with vegetation density and Froude
number
Fig. 9 shows the relationship between drag coefficient (Cd) and
initial Froude number (Fro) with different vegetation densities (dense,
intermediate, and sparse) and thicknesses (dn-180, 380, and 580 (No.
cm)). Against all the subcritical flow conditions (Froude No. 0.57–
0.73), the drag coefficient increased slightly with increasing Froude
number. However, for the last three Fro values i.e., Fro = 0.70, 0.71,
and 0.73, the drag coefficient abruptly increased due to cylinder
oscillations. During oscillations, additional mass acted on the cylinder
and resulted in higher drag force as a result of a higher drag coefficient.
Nepf (1999) pointed out that drag coefficient changes with the density
of vegetation, and this coefficient is decreased by reducing the spacing
between vegetation (i.e., from sparse to dense vegetation arrange-
ments). However, Nepf (1999) set the vegetation model array to a
constant width of five meters, whereas in the current study, the width
of the vegetation model (W) was also changed by changing the spacing
between cylinders against a constant vegetation thickness. Thus, the
drag coefficient trend obtained was opposite that of the study of Nepf
(1999), i.e., the drag coefficient was increased by reducing spacing
between cylinders, as shown in Fig. 9.
For low initial Froude numbers, the drag coefficient at the vegetation
front observed during the experiment was almost the same at all three
vegetation densities (Fig. 9a, c, e), which started to increase as cylinders
started to oscillate. The drag coefficient acting on cylinders was greater for
vegetation with dn-180 No. cm as compared to vegetation with dn-380 No.
cm and dn-580 No. cm. The width of the vegetation in the current study
was much less than that of Nepf (1999); therefore, the flow structure was
different. Because the backwater rise was highest for dense vegetation
compared to intermediate and sparse vegetation against the same discharge
(Fig. 3), the approach velocity V in Eq. (1) was lowest for dense vegetation
and highest for sparse vegetation. This resulted in a maximum drag
coefficient in dense vegetation, which decreased as the spacing between
the cylinders was increased (sparse vegetation) (Fig. 9). The drag coefficient
at the vegetation back was slightly higher than that at the vegetation front.
This difference was higher for the dense vegetation arrangement and
started to decline as spacing between cylinders increased. For sparse
conditions, the difference between drag coefficients was extremely low.
This is because the water surface slope inside the vegetation was large in
cases of dense vegetation compared to sparse vegetation. Also, due to
presence of slight oscillations at vegetation back in dense vegetation, the
drag coefficient values at vegetation back were a little higher than at front.
Cylinder oscillations are caused due to the measurement method of the
drag force by load cell which was set with a small gap between cylinders and
the bed. So, the values of drag coefficient which were greatly influenced by
cylinder oscillations are marked inside dashed circle line A in Fig. 9 due to
less reliability.
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Fig. 5. Classification of undular jumps: (a) non-breaking undular jump with small wavelength (NBS), (b) non-breaking undular jump with small wavelength and air bubbles (NBSA), (c)
non-breaking undular jump with medium wavelength (NBM), (d) non-breaking undular jump with medium wavelength and air bubbles (NBMA), (e) non-breaking undular jump with
large wavelength (NBL), (f) breaking undular jump with lateral shock wave (BU), (g) no jump or very few undulations (NJ), and (h) hydraulic jump exceeded the channel length (JEC).
Dashed lines show lateral shock waves.

3.3. Energy reduction behind vegetation
As the tree density increases (i.e., space between vegetation
becomes smaller), the maximum water level and maximum velocity
behind the vegetation decrease (Iimura and Tanaka, 2012). In the
current study, the water level was measured at the center of the
channel and throughout the length of the channel. Using the
measured water depth, detailed water profiles were drawn against
each case. The specific energy is defined as the energy per unit
volume of water at any section of a channel measured with respect to

315
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Fig. 6. Distribution of water level for vegetation thickness (i) dn-180 No. cm, (ii) dn-380 No. cm, and (iii) dn-580 No. cm, (a) dense vegetation (G/d = 0.25), (b) intermediate vegetation
(G/d = 1.09), and (c) sparse vegetation (G/d = 2.13). ho/L: initial water depth to wavelength ratio. For the meanings of NJ, NBS, NBSA, NBM, NBMA, NBL, BU, and JEC, see the caption
of Fig. 5.

the channel bottom (Chow, 1959). Thus, the value of α is taken as 1. Because the flow is steady, the depth-
averaged velocity against the known discharge (Q) was calculated
V2 using the relationship V = Q/A. Fig. 6 shows a significant difference
E=y+α
2g (5) between the effects of water depths on upstream and downstream
vegetation, which resulted in a certain loss of energy. The energy loss
where E is specific energy, y is water depth, V is velocity, and α is a (ΔE) through vegetation is the difference between specific energy
coefficient to account for variations in velocity. In the current study, upstream and downstream of vegetation.

316
G.A. Pasha, N. Tanaka
Fig. 6. (continued)
With increasing vegetation density, the water level was increased in
front of the vegetation by reflection and damming up, and the water
surface slope inside the vegetation was increased (Iimura and Tanaka,
2012; Pasha and Tanaka, 2016). Under the current conditions, the
water flow passing through the vegetation became critical just down-
stream of the vegetation, which resulted in formation of an undular
hydraulic jump (Fig. 6). Although the intensity of the hydraulic jump
317
Ocean Engineering 141 (2017) 308–325
was small, it still contributed to energy loss.
In the present study, the energy loss is subdivided into two
portions:
1. total energy loss (ΔE1 = E1 – E2), and
2. energy loss due to a hydraulic jump (ΔE2 = (y2-y1)3/4y1y2)
G.A. Pasha, N. Tanaka
Fig. 6. (continued)
where E1 is the specific energy at the forest front, E2 is the mean
specific energy after formation of a hydraulic jump where the flow is
subcritical, y1 = the minimum water depth during the jump, and y2 =
mean water depth after the hydraulic jump. The mean values of E2 and
y2 are considered because of the undulations in the water surface after
the hydraulic jump.
318
Ocean Engineering 141 (2017) 308–325
3.3.1. Total energy loss
3.3.1.1. Effect of Froude number on total energy loss. Fig. 10a shows
the relationship between the relative total energy loss (ΔE1/E1) and the
initial Froude number, which was varied between Fro = 0.57 and Fro =
0.73 for (i) dense, (ii) intermediate, and (iii) sparse vegetation
arrangements. In each subplot of Fig. 10a, the data is plotted against
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Table 2
Froude number (Fr1) at y1 for Fig. 6.

dn (No. cm) G/d Fro

0.58 0.65 0.68 0.70 0.73

180 0.25 1.11 1.74 1.82 1.72 1.84

1.09 0.82 1.01 1.09 1.29 1.46
2.13 0.59 0.73 0.81 0.89 0.94
380 0.25 1.14 1.86 1.81 1.81 1.88
1.09 0.85 1.16 1.25 1.46 1.50
2.13 1.17 1.29 1.36 1.38 1.35
580 0.25 1.15 1.45 1.87 2.01 2.01
1.09 0.77 0.88 1.03 1.25 1.43
2.13 0.79 0.89 1.00 1.14 1.18

Fig. 8. Hydraulic conditions for formation of undular jumps. For the meaning of NJ,
NBS, NBSA, NBM, NBMA, NBL, BU, and JEC, see the captions of Fig. 5. A and B
represent vegetation having small undulations even with Fr1 < 1. C represents the upper
limit of a non-breaking undular jump, and D is the upper limit of a breaking undular
jump (D1 - jump observed, D2 - jump not observed).

Against all the values of vegetation density and thickness, Fig. 10a-i,
a-ii, and 10a-iii shows that the relative total energy loss remained
almost constant when the initial Froude number was increased from
0.57 to 0.73. The initial flow depth (ho) has very little effect on total
relative energy loss. In a dense arrangement (Fig. 10a-i), the last three
points plotted against all the three thickness values showed compara-
tively low energy reduction. This is because the length of the hydraulic
jump increased with an increase in the Froude number, and for the
highest tested initial Froude number values (Fro = 0.70, 0.71, 0.73), the
length of the jump exceeded the available channel length downstream
of the vegetation model. In this case, E2 was calculated against
supercritical flow conditions, which resulted in low energy reduction.
However, it is important to calculate the damage due to the occurrence
of a hydraulic jump downstream of the vegetation in future studies.
Because a breaking undular hydraulic jump was formed against
vegetation having G/d = 0.25, the length of the hydraulic jump was
increased by increasing the Froude number (Fig. 11a). In dense vegetation
(Fig. 10a-i), the relative length of the hydraulic jump for the highest tested
initial Froude number values (Fro = 0.70, 0.71, 0.73) was so large that it
crossed the available channel length downstream of the vegetation model.
In such a case, the flow remained supercritical, and no hydraulic jump was
observed within the channel. In the case of intermediate vegetation
(Fig. 11b), the relative length of jumps remained almost the same, except
for the highest tested initial Froude number, where a slight increase in the
relative length was observed. The non-breaking undular jump in cases of
sparse vegetation (Fig. 11c) showed a slightly higher value of relative jump
length (maximum for dn-380 No. cm) compared to intermediate vegetation
(Fig. 11b) due to the longer wave length of undulations in sparse vegetation.

3.3.1.2. Effect of vegetation density on total energy loss. The density

Fig. 7. Vegetation conditions for formation of undular jumps: (a) dn-180 No. cm, (b) dn-
380 No. cm, and (c) dn-580 No. cm. For the meanings of NJ, NBS, NBSA, NBM, NBMA,
NBL, BU, and JEC, see the captions of Fig. 5.
three values of vegetation thickness (dn = 180, 380, and 580 (No. cm)).
Vertical bars represent standard deviation. It is important to mention
here that total energy loss includes losses due to vegetation itself and
the undular hydraulic jump forming downstream of vegetation.
of vegetation greatly affects the downstream flow because increasing the
density of vegetation reduces velocity of the water behind the vegetation
(Iimura and Tanaka, 2012), contributing to energy loss. Fig. 10b shows the
relationship between the relative total energy loss (ΔE1/E1) and vegetation
densities (G/d = 0.25, 1.09, 2.13) for three values of vegetation thickness (i)
dn-180 No. cm, (ii) dn-380 No. cm, and (iii) dn-580 No. cm. Since the
relative total relative energy loss remained almost constant with different
initial Froude numbers (Fig. 10a), therefore, Fig. 10b was plotted against
the five out of nine Fro, i.e., 0.58, 0.65, 0.68, 0.70, 0.73, to improve the
readability of the figure. Also, the trend line and R2 value are included for
the single flow condition with Fro = 0.68.
It is clear from all the subplots of Fig. 10b that the total relative energy
loss is greatest with the denser vegetation (G/d = 0.25) and least with the
sparse vegetation (G/d = 2.13). Against a constant vegetation thickness,

319
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Fig. 9. Change of drag coefficient (Cd) on individual cylinder located at center of forest model and flume: (a) G/d = 0.25 – front, (b) G/d = 0.25 – back, (c) G/d = 1.09 – front, (d) G/d =
1.09 – back, (e) G/d = 2.13 – front, and (f) G/d = 2.13 – back. Where, dn has units of No. cm, and the data plotted within dashed circle line (A) represents data which is not reliable due
to cylinder oscillations.

decreasing the vegetation density (increasing the spacing between cylin-
ders) decreased the resistance to the flow, which resulted in a decrease in
the relative energy loss. Because the range of initial Froude numbers was
limited, the difference in relative energy loss between the different initial
Froude numbers also was not high. For a vegetation thickness of dn-180
No. cm, dense vegetation dissipated the relative total energy of 33–37%,
which was decreased to 29–32% and 26–28% in intermediate and sparse
conditions, respectively (Fig. 10b-i). For dn-380 No. cm, the maximum
relative total energy losses were 48%, 42%, and 39% for dense, inter-
mediate, and sparse vegetation, respectively (Fig. 10b-ii). Similarly, for dn-
580 No. cm, the dense, intermediate, and sparse vegetation dissipated
maximum relative total energy losses of 52%, 48%, and 44%, respectively
(Fig. 10b-i). According to Fig. 10b, the overall difference of relative total
energy loss between dense and sparse vegetation against all the vegetation
thickness is 5–9%.
3.3.1.3. Effect of vegetation thickness on total energy loss. Fig. 10c
shows the relationship between the relative total energy loss (ΔE1/E1)
and vegetation thickness (dn = 180, 380, 580 (No. cm)) for three values
of vegetation densities (i) dense vegetation, G/d = 0.25, (ii)

320
G.A. Pasha, N. Tanaka
Fig. 10. Relationship between relative total energy loss (ΔE1/E1) for (a) Initial Froude number (Fro), (i) G/d = 0.25, (ii) G/d = 1.09, and (iii) G/d = 2.13, (b) vegetation density (G/d) (i)
dn = 180 No. cm, (ii) dn = 380 No. cm, and (iii) dn = 580 No. cm, and (c) vegetation thickness (dn), (i) G/d = 0.25, (ii) G/d = 1.09, and (iii) G/d = 2.13.
intermediate vegetation, G/d = 1.09, and (iii) sparse vegetation, G/d =
2.13. Similar to Fig. 10b, five out of nine flow conditions were selected
to determine the effect of vegetation thickness on total energy loss with
Fro = 0.68 showing a trend line.
Against a constant vegetation density, the relative energy reduction
was increased by increasing the thickness of vegetation from dn-180 to
580 No. cm. Wider vegetation had a larger resistance and reflection of
water and thus resulted in higher energy loss. Against all the vegetation
densities, the relative total energy loss increased 9–12% when the
vegetation thickness was increased from dn-180 No. cm to dn-380 No.
cm. Further increasing the vegetation thickness to dn-580 No. cm also
increased the relative total energy loss by 5–6%. Thus, the overall
difference of relative total energy loss between dn-180 No. cm and dn-
580 No. cm against all the vegetation densities was 15–17% (Fig. 10c).
3.3.2. Energy reduction behind vegetation due to hydraulic jump
Although it is difficult to determine the exact energy loss due to an
undular hydraulic jump formed downstream of vegetation, it still
contributes to the energy loss that needs to be calculated. The
important factors that affect the energy loss include Froude number,
vegetation density, and thickness.
3.3.2.1. Effect of Froude number on energy loss due to hydraulic
jump. Fig. 12a shows the relationship between the relative energy
321
Ocean Engineering 141 (2017) 308–325
losses due to a hydraulic jump (ΔE2/y1) and the initial Froude number
(Fro = 0.57–0.73) for (i) dense, (ii) intermediate, and (iii) sparse
vegetation arrangements. The data in each subplot of Fig. 12a is plotted
against three values of vegetation thickness (dn = 180, 380, and 580
(No. cm)). Vertical bars represent standard deviation.
As shown in to Fig. 7, the characteristics of an undular hydraulic
jump for vegetation with fixed density and thickness changes with
increasing initial Froude number. This also increases the energy
loss due to a hydraulic jump. Unlike the total relative energy loss,
the relative energy loss due to a jump increased with the increase in
the initial Froude number. The increase was more prominent in
dense (Fig. 12a-i) and intermediate vegetation (Fig. 12a-ii) due to
the higher intensity of the hydraulic jump, like a breaking undular
jump and non-breaking undular jump with air bubbles (Fig. 7).
Whereas in sparse vegetation, the relative energy loss due to a jump
remained almost constant with the increase in the initial Froude
number (Fig. 12a-iii) because the water depth downstream of the
sparse vegetation crossed the critical water depth in very few cases
(Fig. 6-iii).
3.3.2.2. Effect of vegetation density on energy loss due to hydraulic
jump. Fig. 12b shows the relationship between the energy loss due to a
hydraulic jump (ΔE2/y1) and vegetation densities (G/d = 0.25, 1.09,
2.13) for three values of vegetation thickness (i) dn-180 No. cm, (ii) dn-
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

380 No. cm, and (iii) dn-580 No. cm against five out of nine Fro, i.e.,
0.58, 0.65, 0.68, 0.70, and 0.73. The trend line and R2 value are cited
for the single flow condition having Fro = 0.68.

Comparing Fig. 12b-i, b-ii, and b-iii shows that the maximum
energy loss was observed in the dense vegetation arrangement (G/d =
0.25) compared to intermediate (G/d = 1.09) and sparse arrangements
(G/d = 2.13). For most of the experimental conditions, a breaking
undular hydraulic jump was formed in cases of dense vegetation, while
a non-breaking undular jump occurred against the intermediate and
sparse vegetation. The inflow Froude number (Fr1) for formation of a
breaking undular jump was higher than Fr1 for a non-breaking undular
jump (Fig. 8). Therefore, the maximum energy loss due to a hydraulic
jump in the case of dense vegetation was larger (6.3%) than that in
cases of intermediate (1.7%) and sparse vegetation (1.4%).

3.3.2.3. Effect of vegetation thickness on energy loss due to hydraulic

jump. Fig. 12c shows the relationship between the relative energy loss
due to a jump (ΔE2/y1) and vegetation thickness (dn = 180, 380, 580
(No. cm)) for three vegetation densities: (i) dense vegetation, G/d =
0.25, (ii) intermediate vegetation, G/d = 1.09, and (iii) sparse
vegetation, G/d = 2.13. Because JEC was observed for the last three
flow conditions of dense vegetation (Fig. 7), the remaining six flow
conditions were selected to make Fig. 12c-i with Fro = 0.66 having a
trend line and R2 value. However, for intermediate and sparse
vegetation conditions, five out of nine flow conditions were selected
to determine the effect of vegetation thickness on total energy loss with
Fro = 0.68 having a trend line and R2 value.

According to Fig. 7a, an undular hydraulic jump was formed for the
narrow vegetation (dn-180 No. cm). For vegetation thickness, dn-180
No. cm, the undular jump contributed to energy loss to a maximum of
5% in dense vegetation (Fig. 12c-i), 1.5% in intermediate vegetation
(Fig. 12c-ii), and 0% in sparse vegetation (Fig. 12c-iii). As the vegetation
thickness was increased to dn-380 No. cm, the energy loss due to the
hydraulic jump reached the maximum values of 6.3%, 1.7%, and 1.4% in
dense, intermediate, and sparse vegetation, respectively. However,
contrary to the previous phenomena, a further increase in the vegetation
thickness (dn-580 No. cm) resulted in less energy loss due to the
hydraulic jump i.e., 5.0%, 1.1%, and 0.2% maximum energy loss in
dense, intermediate, and sparse vegetation, respectively. For lower
values of the initial Froude number, the difference in energy loss due
to a change in vegetation thickness was small, but as the initial Froude
Fig. 11. Relative length of jump (LJ/y1) for three vegetation densities and initial Froude
number increased, the difference became prominent, especially in dense numbers: (a) dense vegetation (G/d = 0.25), (b) intermediate vegetation (G/d = 1.09),
and intermediate cases. Many textbooks (e.g., Chow, 1959) indicate a and (c) sparse vegetation (G/d = 2.13) where, dn has units of No. cm.
low energy loss due to an undular hydraulic jump. Also, Streeter and
Wylie (1979) specified that the fraction of energy dissipated by an
undular hydraulic jump is less than 5%. In the current study, except for a
few dense cases, the amount of energy dissipation was well below 5%.
3.3.3. Comparison of total energy loss and energy loss due to
For dn-580 No. cm, the width of the vegetation was so large that it
hydraulic jump
dissipated the maximum energy within the vegetation. When the flow
The energy reduction due to vegetation and a hydraulic jump was
reached the end of the vegetation, the maximum energy had been
explained separately in the previous sections, but it is important to
dissipated, and the inflow Froude number decreased too. Although the
quantify the contribution of energy loss due to a hydraulic jump to the
backwater rise was increased by increasing the vegetation thickness,
total energy loss. Thus, a comparison of total energy loss and energy
the water surface slope inside the vegetation decreased slightly with
loss due to a hydraulic jump is made in Table 3. Out of nine flow
increasing vegetation thickness as the width of vegetation also in-
conditions, the maximum values of both the total energy loss and
creases by increasing vegetation thickness. The drag coefficient at the
energy loss due to a hydraulic jump were used for the comparison in
back of dense vegetation (Fig. 9b) showed a low value for dn-580 No.
Table 3. In Table 3, the relative total energy loss (ΔE1/E1) is compared
cm compared to vegetation with dn-180 and dn-380 (No. cm). This
with the energy loss due to a hydraulic jump relative to y1 (ΔE2/y1) and
difference in drag coefficient decreased as the vegetation type changed
E1 (ΔE2/E1).
from dense to sparse (Fig. 9f). The length of the hydraulic jump also
In the case of dense vegetation (G/d = 0.25), the maximum
was greatest for dn-380 No. cm and least for dn-580 No. cm, as shown
contribution to energy loss by the hydraulic jump calculated in
in Fig. 11a, b, c. All the above-mentioned points suggest that the
terms of y 1, i.e., (ΔE2/y1 ) of the total energy loss (ΔE1 /E1) was
optimum value for vegetation thickness was dn-380 No. cm in the
around 13.2% against dn-380 No. cm, and was reduced to 12.3%
current experimental conditions.
and 9.4% for dn-180 No. cm and dn-580 No. cm, respectively

322
G.A. Pasha, N. Tanaka Ocean Engineering 141 (2017) 308–325

Fig. 12. Relationship between relative energy loss due to hydraulic jump (ΔE2/y1) for (a) initial Froude number (Fro), (i) G/d = 0.25, (ii) G/d = 1.09, and (iii) G/d = 2.13, (b) vegetation
density (G/d) (i) dn = 180 No. cm, (ii) dn = 380 No. cm, and (iii) dn = 580 No. cm, and (c) vegetation thickness (dn), (i) G/d = 0.25, (ii) G/d = 1.09, and (iii) G/d = 2.13.

Table 3 The contribution of energy loss by a jump calculated in terms of E1

Comparison of relative total energy loss (ΔE1/E1) with energy loss due to hydraulic jump was much lower than calculated in terms of y1. The maximum effect of
relative to y1 (ΔE2/y1) and E1 (ΔE2/E1), plotted against maximum values among initial
ΔE2/E1 in ΔE1/E1 was 3.5% for dense vegetation with dn-180 (No. cm)
Froude numbers.
and was reduced to 3.0% and 2.1% for dn-380 No. cm and 580 No. cm,
G/d dn (No. cm) ΔE1/E1 ΔE2/y1 ΔE2/E1 respectively. However, the contribution of energy loss by a jump
relative to E1 in the total energy loss was very much smaller in the
0.25 179.21 37.0% 4.6% 1.3%
cases of intermediate and sparse vegetation, as the maximum con-
380.12 48.0% 6.3% 1.4%
581.05 52.1% 4.9% 1.1%
tributions were 1.5% and 1.2% in intermediate and dense vegetation
1.09 174.72 31.7% 1.4% 0.5% conditions against dn-180 and 380 (No. cm), respectively (Table 3).
390.85 42.3% 1.7% 0.5% The findings of current experimental study make it evident that
582.96 47.8% 1.1% 0.3% denser vegetation resulted in greater energy loss. However, it is difficult
2.13 179.36 28.1% 0.0% 0.0%
to construct dense vegetation in the field, and thus it is recommended
387.83 39.5% 1.4% 0.5%
580.27 44.8% 0.2% 0.1% to favor intermediate vegetation over sparse vegetation for dissipating
tsunami energy. Similarly, the wider vegetation, i.e., greater vegetation
thickness, leads to more total energy losses, but the contribution to
(Table 3). However, increasing the spacing between vegetation energy loss of a hydraulic jump relative to y1 to total energy loss was
cylinders reduced the intensity of the hydraulic jump, as was its not negligible, especially in intermediate (max. 4.4%) and dense
contribution in the total energy loss. The hydraulic jump that was vegetation (max. 13.2%), as can be seen in Table 3. Considering losses
formed downstream of the intermediate vegetation (G/d = 1.09) due to an undular hydraulic jump, the concept of designing vegetation
contributed to maximum energy losses of 4.4%, 4.1%, and 2.2% for mitigation that produces a hydraulic jump downstream can be
against dn-180, 380, and 580 (No. cm), respectively, whereas the utilized when available space is limited.
contribution to energy loss by a hydraulic jump in the case of sparse
vegetation (G/d = 2.13) was the least. In sparse vegetation, the 4. Conclusions
maximum contribution to energy loss by a jump relative to y 1 in
total energy loss was 3.5% against dn-380 No. cm, while the The flow structure around vegetation and the energy loss in a steady
contribution was reduced to 0.1% and 0.4%, for dn-180 No. cm subcritical flow (Froude number 0.55–0.75) through emergent vegeta-
and 580 No. cm, respectively (Table 3). tion were investigated experimentally in a water flume with variation in

323
G.A. Pasha, N. Tanaka
inflow, vegetation density, and thickness. The following conclusions
have been drawn from the current study:
1. Upstream of the vegetation model, keeping the spacing between
cylinders (density) constant while increasing the thickness (width) of
vegetation from dn-180 to 580 No. cm increases the backwater rise
relative to water depth without vegetation up to 60%. Similarly, it is
also increased up to 30% by reducing the spacing between cylinders
from G/d = 2.13 to 0.25 and keeping the vegetation thickness
constant. A dense arrangement of tree models exhibits a higher flow
resistance compared to a sparse arrangement, which consequently
results in a higher back water rise. In addition, the relative back-
water rise increases slightly with increase in Froude number.
2. An undular hydraulic jump is formed just downstream of the
vegetation model. In the case of a dense vegetation arrangement,
mostly a breaking undular jump is observed with prominent lateral
shock wave, whereas a non-breaking undular jump with air bubbles
is seen in intermediate vegetation. However, with sparse vegetation,
a non-breaking undular jump without air bubbles is documented.
Against a constant b/y1 ratio (where b is the channel width and y1 is
the water depth at the toe of a hydraulic jump), patterns of undular
hydraulic jump are also changed by increasing the Fr1 (Froude
number against water depth y1) from 1.0 to 2.0 irrespective of
vegetation conditions in the following order: a non-breaking hy-
draulic jump of small wavelength (0.25 ≤ ho/L ≤ 0.50), a non-
breaking hydraulic jump of medium wavelength (0.20 ≤ ho/L <
0.25), a non-breaking hydraulic jump of large wavelength (0.15 ≤
ho/L < 0.20), and a breaking undular hydraulic jump.
3. Both the backwater rise and the water level downstream of the
vegetation depend on vegetation density. The reduction in water
level and velocity behind the vegetation (after the hydraulic jump)
increases with increase in vegetation density, resulting in energy
loss. Also, the water surface slope results from the resistance due to
vegetation is increased by increasing vegetation density. The denser
and wider the vegetation, the greater is the total energy loss
(vegetation and hydraulic jump). However, for a very large thickness
such as dn-580 No. cm, the maximum energy is dissipated within the
large vegetation width, and much less energy is dissipated by the
hydraulic jump.
4. Here, it is recommended that conditions leading to a hydraulic jump
be utilized when available space is limited, though maximum
vegetation thickness would lead to even more total energy losses.
Although dense vegetation dissipates the maximum energy, it is hard
to construct it in a field with closely spaced trees. Under the current
conditions, vegetation with density G/d = 1.09 and thickness ~ 380
No. cm produces the most favorable conditions for energy loss due to
an undular hydraulic jump.
The findings are important for optimum vegetation design. In the
future, more experimental study of high Froude number flow is
required in order to further investigate the phenomena.
Acknowledgements
This study was funded by a Japan Society for the Promotion of
Science (JSPS) Grant-in-Aid for Scientific Research (No. 15H02987).
References
Chanson, H., 2009. Current knowledge in hydraulic jumps and related phenomena. Eur.
J. Mech. B/Fluids 28 (2009), 191–210.
Chanson, H., Montes, J.S., 1995. Characteristics of undular hydraulic jumps:
experimental apparatus and flow patterns. J. Hydraul. Eng. 121 (2), 129–144.
Chow, V.T., 1959. Open Channel Hydraulics. McGraw-Hill Publishing Co., New York.
Danielsen, F., Sorensen, M.K., Olwig, M.F., Selvam, V., Parish, F., Burgess, N.D.,
Hiraishi, T., Karunagaran, V.M., Rasmussen, M.S., Hansen, L.B., Quarto, A.,
Suryadiputra, N., 2005. The Asian tsunami: a protective role for coastal vegetation.
Science 310, 643.
324
Ocean Engineering 141 (2017) 308–325
Dengler, L., Preuss, J., 2003. Mitigation lessons from the July 17, 1998, Papua New
Guinea tsunami. Pure Appl. Geophys. 160, 2001–2031.
Fraser, S., Raby, A., Pomonis, A., Goda, K., Chian, S.C., Macabuag, J., Offord, M., Saito,
K., Sammonds, P., 2013. Tsunami damage to coastal defences and buildings in the
March 11th, 2011 Mw9.0 Great East Japan earthquake and tsunami. Bull. Earthq.
Eng. 11, 205–239.
Harada, K., Imamura, F., 2005. Effects of coastal forest on tsunami hazard mitigation–a
preliminary investigation. In: Satake, K. (Ed.), Tsunamis: case studies and recent
development. Advances in Natural and Technological Hazards Research. Springer,
279–292.
Harada, K., Kawata, Y., 2005. Study on tsunami reduction effect of coastal forest due to
forest growth. Ann. Disaster Prev. Res. Inst. 48C, 161–165.
Haslewood, D., 1985. Ships in restricted water (critical conditions). Proc. Inst. Civ. Eng. 2
(79), 275–293.
Hiraishi, T., Harada, K., 2003. Greenbelt tsunami prevention in South-Pacific region.
Rep. Port Airpt. Res. Inst. 42 (2), 23.
Iimura, K., Tanaka, N., 2012. Numerical simulation estimating the effects of tree
density distribution in coastal forest on tsunami mitigation. Ocean Eng. 54,
223–232.
Iimura, K., Tanaka, N., 2013. Dangerous zone formation behind finite–length coastal
forest for tsunami mitigation. J. Earthq. Tsunami 7 (4), 1–18.
Irtem, E., Gedik, N., Kabdasli, M.S., Yasa, N.E., 2009. Coastal forest effects on tsunami
run-up heights. Ocean Eng. 36, 313–320.
Ishigaki, A., Higashi, H., Sakamoto, S., Shibahara, S., 2013. The Great East–Japan
earthquake and devastating tsunami: an update and lessons from the past great
earthquakes in Japan since 1923. Tohoku J. Exp. Med. 229 (4), 287–299.
Kathiresan, K., Rajendran, N., 2005. Coastal mangrove forests mitigated tsunami. Estuar.
Coast. Shelf Sci. 65 (3), 601–606.
Mascarenhas, A., Jayakumar, S., 2008. An environmental perspective of the post-
tsunami scenario along the coast of Tamil Nadu, India: role of sand dunes and
forests. J. Environ. Manag. 89, 24–34.
Montes, J.S., 1986. A study of the undular jump profile. In: Proceedings of the 9th
Australasian Fluid Mechanics Conference AFMC. Auckland, New Zealand, pp. 148–
151.
Nandasena, N.A.K., Sasaki, Y., Tanaka, N., 2012. Modeling field observations of the 2011
Great East Japan tsunami: efficiency of artificial and natural structures on tsunami
mitigation. Coast. Eng. 67, 1–13.
Nateghi, R., Bricker, J.D., Guikema, S.D., Bessho, A., 2016. Statistical analysis of the
effectiveness of seawalls and coastal forests in mitigating tsunami impacts in Iwate
and Miyagi prefectures. PLoS One 11 (8), e0158375.
Nepf, H.M., 1999. Drag, turbulence, and diffusion in flow through emergent vegetation.
Water Resour. Res. 35 (2), 479–489.
Ohtsu, I., Yasuda, Y., Gotoh, H., 2001. Hydraulic condition for undular-jump formations.
J. Hydraul. Res. 39 (2), 203–209.
Ohtsu, I., Yasuda, Y., Gotoh, H., 2003. Flow conditions of undular hydraulic jumps in
horizontal rectangular channels. J. Hydraul. Res. 129 (12), 948–955.
Osti, R., Tanaka, S., Tokioka, T., 2009. The importance of mangrove forest in tsunami
disaster mitigation. Disasters 33 (2), 203–213.
Pasha, G.A., Tanaka, N., 2016. Effectiveness of finite length inland forest in trapping
tsunami-borne wood debris. J. Earthq. Tsunami 10 (2), 1–26.
Rahman, M.M., Schaab, C., Nakaza, E., 2017. Experimental and numerical modeling of
tsunami mitigation by canals. J. Waterw. Port Coast. Ocean Eng. 143 (1),
(04016012-1-11).
Shuto, N., 1987. The effectiveness and limit of tsunami control forests. Coast. Eng. Jpn.
30 (1), 143–153.
Spiske, M., Weiss, R., Bahlburg, H., Roskosch, J., Amijaya, H., 2010. The TsuSedMod
inversion model applied to the deposits of the 2004 Sumatra and 2006 Java tsunami
and implications for estimating flow parameters of palaeo-tsunami. Sediment. Geol.
224, 29–37.
Streeter, V.L., Wylie, E.B., 1979. Fluid Mechanics. McGraw-Hill Publishing Co., New
York.
Suppasri, A., Koshimura, S., Imai, K., Mas, E., Gokon, H., Muhari, A., Imamura, F., 2012.
Developing tsunami fragility curves from the surveyed data of the 2011 Great East
Japan tsunami in Sendai and Ishinomaki Plains. Coast. Eng. J. 54 (1),
1250008–1 –1250008–16.
Takemura, T., Tanaka, N., 2007. Flow structures and drag characteristics of a colony-type
emergent roughness model mounted on a flat plate in uniform flow. Fluid Dyn. Res.
39, 694–710.
Tanaka, N., 2009. Vegetation bioshields for tsunami mitigation: review of the
effectiveness, limitations, construction, and sustainable management. Lands Ecol.
Eng. 5, 71–79.
Tanaka, N., 2012. Effectiveness and limitations of coastal forest in large tsunami:
conditions of Japanese pine trees on coastal sand dunes in tsunami caused by Great
East Japan Earthquake. J. Jpn. Soc. Civil. Eng. Ser. B1 (Hydraul. Eng.) 68 (4),
(II_7–II_15).
Tanaka, N., Suzuki, I., 2013. Comparison between drag force on a house behind coastal
forest and impulsive force by tsunami-driven woody debris from the forest at the
Great East Japan Earthquake. In: Proceedings of the 35th IAHR World Congress.
Chengdu, China, 8–13 September, pp. 891–898.
Tanaka, N., Sasaki, Y., Mowjood, M.I.M., Jinadasa, K.B.S.N., Homchuen, S., 2007.
Coastal vegetation structures and their functions in tsunami protection: experience
of the recent Indian Ocean tsunami. Lands Ecol. Eng. 3, 33–45.
Tanaka, N., Jinadasa, K.B.S.N., Mowjood, M.I.M., Fasly, M.S.M., 2011. Coastal
vegetation planting projects for tsunami disaster mitigation—effectiveness evaluation
of new establishments. Lands Ecol. Eng. 7, 127–135.
Tanaka, N., Yagisawa, J., Yasuda, S., 2013. Breaking pattern and critical breaking
condition of Japanese pine trees on coastal sand dunes in huge tsunami caused by
Great East Japan Earthquake. Nat. Hazards 65, 423–442.
Tanaka, N., Yasuda, S., Iimura, K., Yagisawa, J., 2014. Combined effects of coastal forest
and sea embankment on reducing the washout region of houses in the Great East
Japan tsunami. J. Hydro-Environ. Res. 8, 270–280.
G.A. Pasha, N. Tanaka
Thuy, N.B., Tanimoto, K., Tanaka, N., Harada, K., Iimura, K., 2009. Effect of open gap in
coastal forest on tsunami run-up – investigations by experiment and numerical
simulation. Ocean Eng. 36, 1258–1269.
Thuy, N.B., Tanaka, N., Tanimoto, K., 2012. Tsunami mitigation by coastal vegetation
considering the effect of tree breaking. J. Coast Conserv. 16 (1), 111–121.
Tokida, K., Tanimoto, R., 2014. Lessons for countermeasures using earth structures
against tsunami obtained in the 2011 off the Pacific coast of Tohoku earthquake.
Soils Found. 54 (4), 523–543.
325
Ocean Engineering 141 (2017) 308–325
Yanagisawa, H., Koshimura, S., Goto, K., Miyagi, T., Imamura, F., Ruangrassamee, A.,
Tanavud, C., 2009. The reduction effects of mangrove forest on a tsunami based on
field surveys at Pakarang Cape, Thailand and numerical analysis. Estuar. Coast. Shelf
Sci. 81, 27–37.
Yanagisawa, H., Koshimura, S., Miyagi, T., Imamura, F., 2010. Tsunami damage
reduction performance of a mangrove forest in Banda Aceh, Indonesia inferred from
field data and a numerical model. J. Geophys. Res. 115, C06032.