You are on page 1of 16

Received: 14 January 2020 Revised: 11 May 2020 Accepted: 2 June 2020

DOI: 10.1002/rra.3671

RESEARCH ARTICLE

Investigating the turbulent flow behaviour through partially


distributed discontinuous rigid vegetation in an open channel

Naveed Anjum1 | Norio Tanaka1,2

1
Graduate School of Science and Engineering,
Saitama University, Saitama, Japan Abstract
2
International Institute for Resilient Society, Riparian vegetation does not only affect the channel flow carrying capacity, but also
Saitama University, Saitama, Japan
plays significant roles in water management, stream restoration, and river rehabilita-
Correspondence tion. This study numerically investigates the flow characteristics through longitudi-
Professor Norio Tanaka, Graduate School of
nally discontinuous rigid vegetation occupying half width of the channel, with the
Science and Engineering, Saitama University,
255 Shimo-okubo, Sakura-ku, Saitama-shi, help of three-dimensional software FLUENT in which a Reynolds stress turbulence
Saitama 338-8570, Japan.
model was adopted. Three varying conditions of vegetation were considered com-
Email: tanaka01@mail.saitama-u.ac.jp
prising of vertically double-layered vegetation (DLV), submerged vegetation (SV), and
emergent vegetation (EV) while keeping the same vegetation density, as well as a
varying discharge condition against DLV was also tested. The results indicated that
the flow distribution becomes more complex through DLV and SV followed by multi-
ple layers with an inflectional instability in the vertical velocity profile around sub-
merged canopy top, as compared to the flow through EV where the uniform
distribution of flow over the canopy column was observed. The velocity in the can-
opy zone decreased considerably because of the resistance due to vegetation, which
influenced the channel carrying capacity, in comparison to that in the non-vegetated
zone. The flow velocities through the obstructed part of the channel, that is, canopy
zone, in DLV arrangement reduced by a percentage difference of approximately
42 and 37% compared to that of SV and EV arrangements, respectively; whereas it
was reduced by approximately 55% when the discharge was twice while keeping the
same configuration of DLV. The inflectional instabilities and estimated mixing layer
over the interfacial zone suggested a stronger lateral exchange of momentum for
DLV configuration in comparison to that of SV and EV. Within the gaps between the
patch zones, the flow velocity, turbulent kinetic energy, and turbulent intensity
reduced significantly due to blockage effect and sheltering offered by the vegetation
patches, signifying a positive flow response towards aquatic life and sediment
deposition.

KEYWORDS

discontinuous vegetation, flow characteristics, flow instability, numerical modelling, partial


obstruction

River Res Applic. 2020;36:1701–1716. wileyonlinelibrary.com/journal/rra © 2020 John Wiley & Sons Ltd 1701
1702 ANJUM AND TANAKA

1 | I N T RO DU CT I O N and tall vegetation (Hamed, Sadowski, Nepf, & Chamorro, 2017; Liu
et al., 2010) with either in submerged or emerged flow state. It is,
Vegetation plays a significant part in affecting the flow turbulence of therefore, necessary to consider the effects of both short and tall veg-
a river (Nepf, 2012). Previously, the presence of vegetation in natural etation in combination, which makes the flow patterns more complex,
rivers was considered as a nuisance as it affects the flow carrying in order to properly replicate the riparian environment. Thus, the
capacity of the channel (Kouwen & Unny, 1975). However, the signifi- interaction of flow with such vegetation patterns, that is, partially dis-
cance of such vegetation has been highlighted by the recent river tributed, in an open channel may lead to the generation of complex
management as it in has great practical value in ecological engineering flow structure which is three dimensional.
perspective and restoration of rivers. It also supports biodiversity as it In addition, the study of natural waterways shows that the vege-
provides habitat and food for the growth of aquatic life (Stoesser, tation in open channels is usually longitudinally discontinuous
Neary, & Wilson, 2005; Wu & He, 2009). The influence of riparian (Tanaka & Ohmoto, 2015; Tanaka, Ohmoto, & Tanaka, 2008). In real
vegetation on flow as well as ecological processes in rivers has scenarios, however, riparian vegetation is not often continuous.
become progressively significant in flood risks, river restoration, and Rather, vegetation communities are usually patchy, with open spaces,
aquatic ecological management (Bauer, Harzer, Strobl, & due to the reason of seasonal variation (Yang, Irish, &
Kollmann, 2018; Javernick & Bertoldi, 2019; Seer, Brunke, & Socolofsky, 2015). The effect of such patch type vegetation on flow
Schrautzer, 2018). Among the hydraulic characteristics of flows with structures in an open channel is not well understood. Folkard (2005)
vegetation in open channels, the velocity dissemination has been and Okamoto and Nezu (2013) clarified the turbulence flow structure
widely studied, since it is pertinent to simulate flow patterns and dis- behind discontinuous vegetation through experimental investigation.
charge. In recent times, the flows around partly distributed vegetation Zhao and Huai (2016) adopted LES method to investigate the hydro-
in an open channel has gain larger consideration, due to the reason dynamics of flows through discontinuous vegetation. Previous studies
that such partial vegetation distribution can be found in wetlands and have been discussed separately; however, the studies elucidating the
floodplains. A strong momentum exchange takes place among the influence of different vegetation formations and their submergence
vegetation and non-vegetation regions in river flows with partial veg- while keeping the same vegetation density in the pattern of discontin-
etation distribution (Nezu & Onitsuka, 2001; Rominger & Nepf, 2011; uous and partly distributed in open channel flows are limited.
White & Nepf, 2008). However, the flow structures of such natural Although some of the flow structures through only double-layered
rivers (having partly placed vegetation) are strongly influenced and vegetation (DLV) configuration with a variety of patterns have been
become more complex. numerically clarified by Anjum and Tanaka (2020), the investigations
Numerical investigations have also been reported for partly vege- of the turbulent flow behaviours through double-layered and single-
tated open channel flows. Su and Li (2002) and Huai, Xue, and layered vegetation (with both submerged and emergent conditions)
Qian (2015) adopted a large-eddy simulation (LES) technique to while keeping the constant vegetation density is still limited. The flow
observe the coherent structures across the partially distributed can- mechanism (demonstrating the comparison) around partially distrib-
opy interface. Choi and Kang (2006) implemented a Reynolds stress uted discontinued homogenous (constant formation) and heteroge-
model (RSM) to replicate the turbulent flow structure in a large-aspect neous (varying heights) vegetation would give an extra understanding
ratio channel with partly placed submerged vegetation (SV). Several of the riparian environment and a clear idea of true representation of
previous researchers implemented flow simulation techniques through the vegetal communities along the natural river flows. Moreover, the
a uniform array of vegetation to simulate the turbulent flow (Cui & flow structure around such kind of discontinued and layered canopy
Neary, 2008; Ghani, Anjum, Pasha, & Ahmad, 2019; Lopez & becomes complicated, which is very difficult to clarify and capture
Garcia, 2001; Neary, 2003). The numerical studies on investigating through experimental investigation (Zhao & Huai, 2016). Thus, the
the flow structures in vegetated open channels utilizing CFD tool present numerical study overcomes this problem by demonstrating a
FLUENT have also been reported (Anjum & Tanaka, 2020; Lima, clear flow mechanism around such novel kind of vegetation patterns.
Janzen, & Nepf, 2015). However, in riparian environments or flood The purpose of this study is to numerically clarify the complex 3D
plains of several natural rivers, finite length vegetation can occur as flow structures through longitudinally discontinuous and partly dis-
partial distribution, with submerged as well as emergent condition of tributed vegetation in a rectangular open channel. Three vegetation
constant vegetation heights (Devi & Kumar, 2016; Devi, Sharma, & conditions, that is, submerged condition of vegetation with constant
Kumar, 2019). Sometimes the vegetation with varying heights (Huai, formation, emergent condition of vegetation with constant formation,
Wang, Hu, Zeng, & Yang, 2014; Liu, Diplas, Hodges, & and vertically DLV configuration with short vegetation submerged
Fairbanks, 2010), that is, vertically double layered with short vegeta- and tall vegetation emergent, are considered in this study. In addition,
tion such as shrubs and grasses are submerged due to flood, whereas a discharge parameter against DLV case has also been varied while
the tall vegetation such as trees remains non-submerged. Most of the keeping both the mixed array of short and tall vegetation submerged
previous researches studied the vegetated flows with the same vege- in order to replicate the high flow response against such configura-
tation height, which is not the true representation of natural rivers. tion. The primary objective of these numerical experiments is to clar-
Only a few studies are focusing on flows with a mixed culture of short ify the resulting flow structure due to the varying condition of
ANJUM AND TANAKA 1703

vegetation configuration and submergence by keeping the same vege- turbulence, is examined. The flow characteristics measurement at var-
tation density. The simulations are carried out by a CFD tool FLUENT ious positions and sections are taken in order to elucidate the turbu-
by adopting a Reynolds stress turbulence model. The influence of lence structure as the flow moves through the considered vegetation
novel vegetation configuration, particularly on the flow velocity and configuration.

F I G U R E 1 Schematic diagrams of
computational domain (a) top view of
domain showing vegetation arrangement
and specified positions and sections, and
(b–e) vertical section view for Case A,
B, C, and D, respectively. The dashed lines
(in red) shows the sections, whereas the
dots (in red) shows the specified locations
for the flow structure investigation
[Colour figure can be viewed at
wileyonlinelibrary.com]
1704 ANJUM AND TANAKA

2 | MATERIALS AND METHODS

T A B L E 1 Hydraulic parameters, where hs and ht are the shorter and taller vegetation heights, Ss/d and St/d are the short (staggered) and tall (staggered) vegetation spacings, Q is the discharge, H

1,900

2,670
Re*

-
-
2.1 | Modelling setup and boundary conditions

37,800

75,240
A 1.5 m long and 0.3 m wide computational domain was modelled.

Re

-
-
The domain comprised of partly distributed (covering half the width
of the domain) and longitudinally discontinued rigid vegetation

0.27

0.32
(Figure 1a). The vegetation arrangements of vertically double-layered

Fr

-
-
configuration, and single-layered configuration with submerged as
well as the emergent condition were considered in this study as pres-

U (m/s)

0.42
ented in Figure 1b–e. The x-axis, y-axis, and z-axis indicate longitudi-

0.3
-
-
nal, lateral, and vertical directions, respectively. The vegetation was
modelled as rigid circular cylinders. The layered vegetation configura-

H (cm)
12.66
tion of Liu et al. (2010) was adopted. The short and tall cylinders

is the flow depth, U is the initial average velocity, Fr is the Froude number, Re is the flow Reynolds number, and Re* is the cylinder Reynolds number

18
(diameter ‘d’ of 0.635 cm) were 7.6 and 15.2 cm long, respectively.

-
-
The hydraulic conditions for the four cases (Case A, B, C, and D) are

Aspect ratio (Lv/Lg)


presented in Table 1.
The schematic diagram (top and vertical views) for the modelled
computational domains showing vegetation arrangement is pres-
ented in Figure 1. Two longitudinal sections (LS1 and LS2), that is,
one passing in between the gap of short and tall cylinders (to avoid

1
-
-
-
the significant effect of the cylinder wake), and the other passing
directly through short and tall vegetation arrays (to elucidate the

Q (×10−3 m3/s)
direct influence of vegetation) were adopted. Two cross-sections (C1
and C2) passing through the middle of the last vegetation gap and

11.4

22.8
patch were also considered for investigating the flow structures.

-
-
Moreover, seven important locations, that is, x1–x7, (located in the
centre of patches and gaps) were also investigated for the vertical

N/A
St/d
10

10
flow distribution.

5
A tri-pave mesh with 150 × 150 × 60 nodes in longitudinal, lat-
eral, and vertical directions, respectively, was adopted for the mod-

N/A
Ss/d
5
5

5
elled domain, which gave approximately 1.35 million grid points. This
small computational domain was adopted to reduce the mesh struc-
ture and computational time as well as cost. The unstructured mesh Short submerged, tall submerged
Short submerged, tall emergent

with tetrahedral cells (having triangular faces) used in this study is


Submergence condition

presented in Figure 2a. The local mesh around the vegetation cylin-
ders was refined in order to ensure computational accuracy. This
mesh refinement (mainly within the vicinity of the cylinder) was per-
formed so that the cylinder boundaries could be identified. A trial for
Submerged
Emergent

grid independency was also run to get the quality simulation results.
The mesh grids of 0.2 million (relatively coarser), 1.35 million (fine),
and 2.4 million (finer) were tested initially. The difference in the depth
averaged velocities across the cross-section (C2) amongst the coarse
Staggered

Staggered
Staggered
Vegetation arrangement

and fine grids was around 5%; however, the variation in results by
N/A

more refinement was around 1% which was not noteworthy


Tall

(Figure 2b). Thus, the fine mesh with 1.35 million grids was adopted.
The velocity values in case of coarse grid are observed to be slightly
Staggered
Staggered

Staggered

higher as compared to those of medium and fine grid. This may be


Short

due to the reason that a coarse mesh could not easily identify the
N/A

boundaries of small barriers like vegetation cylinders, and the resis-


tance or drag offered by these structures may not have been clearly
Case

estimated; and thus, have consequently resulted in comparatively


D
A

C
B

larger flow velocity magnitudes. Whereas, a fine mesh (refinement


ANJUM AND TANAKA 1705

F I G U R E 2 (a) Computational
grid on x–y plane of the domain,
where the zoomed part shows the
local mesh around the vegetation
cylinder vicinity, (b) Depth averaged
velocity along the cross-section (C2)
for mesh independence trial (Case
A). The dashed line shows the
interface between the canopy zone
and free zone [Colour figure can be
viewed at wileyonlinelibrary.com]

approach, especially around the vicinity of the cylinders) overcomes pressure = 0.3, momentum = 0.7, turbulent kinetic energy = 0.8, tur-
this problem by predicting a reasonable flow structure, and ensures bulent dissipation rate = 0.8, and Reynolds stresses = 0.5, were uti-
computational accuracy (Stoesser, Liang, Rodi, & Jirka, 2006). lized in the simulation process. In FLUENT (ANSYS), the standard
The boundary conditions that were adopted in the model are: under-relaxation parameters for all the variables are set to values that
(a) periodic boundary condition at inlet/outlet of the domain to are near optimum for maximum type of flows. The solution was con-
achieve the uniform flow conditions (translational periodicity), which sidered to be converged when all the residuals reached below
is adopted when the geometry of concern and the probable flow pat- 1 × 10−5. With further iterations, the residuals were not changing,
tern has a reiterating nature; (b) symmetry condition at free surface and the solution with primary flow velocity was also checked, which
(which allows the flow surface to become flat), as has been was no longer noticeably changed with the subsequent iterations.
implemented by several researchers (Lima et al., 2015; Zhao & Thus, using these criteria, it was assumed that the solution had
Huai, 2016) in their vegetated flow modeling studies; and (c) wall reached a steady state. The governing equations and the turbulence
boundary with the non-slip condition at adjacent walls, domain bed, model details can be seen from the FLUENT user's guide (2019;
and the cylinder walls. The simulation and post-processing were car- Versteeg & Malalasekera, 2008).
ried in a computational fluid dynamics (CFD) tool FLUENT. For the
closure of turbulence, RSM was adopted. The SIMPLE algorithm was
adopted for achieving the pressure–velocity coupling. The standard 2.2 | Validation of numerical model
wall function with a near wall treatment was applied, which is needed
to better predict the flow in the boundary layer. The standard wall The validation of the numerical model was performed with the data of
functions are appropriate for a wide variety of turbulent flows. The Liu et al. (2010). They conducted flume experiments for investigating
dimensionless wall distance (y+) values in the considered mesh ranged the flow hydrodynamics through DLV in a 4.3 m long and 0.3 m wide
between 30 < y+ <100, which was observed to be reasonable for the open channel. The rigid layered vegetation, that is, tall (15.2 cm) and
applicability of the wall function (Versteeg & Malalasekera, 2008). The short (5.1 cm), with a d of 0.635 cm, was mounted on the channel
default (standard) values of under-relaxation factors, that is, bed, at 1.3 m distance from the inlet. The cylinders were organized in
1706 ANJUM AND TANAKA

F I G U R E 3 Experimental setup (Liu


et al., 2010) and validation of the
numerical model (a) top view of modelled
domain showing vegetation arrangement
and specified positions (in red colour),
(b) isometric view of domain, and
(c) comparison of experimental and
numerical results [Colour figure can be
viewed at wileyonlinelibrary.com]

a staggered pattern with Ss/d = 5 and St/d = 10 (where Ss is the spac- The domain modelled for computation comprised of periodic
ing between the short layer cylinders and St is the spacing between length of the vegetation arrangement (Figure 3a,b). The simulated
the tall cylinders), respectively. The mean flow was measured with a results of turbulent intensity at the given locations were compared
1-D Laser Doppler Velocimetry (LDV) system. The flow rate and the with those of experimental data (Figure 3c). The x-axis displays the
flow depth were 0.0114 m3/s and 0.1212 m, respectively. longitudinal turbulent intensity (urms), while the y-axis displays the
ANJUM AND TANAKA 1707

flow depth. The results indicate that the turbulent intensity in the zone numerical results illustrate a close agreement, which confirmed the
directly downstream of tall vegetation (Location 1) is very high due to validity of the numerical model. However, an insignificant dissimilarity
large flow resistance. Moreover, the intensity in the free zone (Location (an average of approximately 3%) in the experimental and numerical
2) is comparatively lower than in the zones right downstream and results occurred, which may be due to the simplifications in modelling
upstream of the vegetation (Locations 1 and 3). The experimental and or unavoidable measurement error of LDV.

F I G U R E 4 Vertical distribution of mean stream-wise velocity at specified locations for (a) Case A, (b) Case B, (c) Case C, and (d) Case D. For
the locations, see Figure 1a,c,d. The horizontal dotted line shows the top of shorter submerged vegetation, whereas the horizontal dashed line
shows the top of tall emergent vegetation. (e) Schematic diagram of the velocity structure for the considered vegetation configurations [Colour
figure can be viewed at wileyonlinelibrary.com]
1708 ANJUM AND TANAKA

3 | RESULTS AND DISCUSSION velocities in the lower canopy layer also significantly reduced in DLV
configuration of fully submerged state, whereas the magnitudes
3.1 | Mean flow characteristics increased in the upper canopy layer due to sparser arrangement of vege-
tation. While further moving towards the flow surface, that is, above the
3.1.1 | Vertical distribution of velocity upper canopy layer, the velocity is relatively higher and almost constant
for different locations due to no vegetation drag in this logarithmic layer.
The vertical distribution of stream-wise velocity at specified locations is In general, with a constant vegetation density, the velocities in
presented in Figure 4a–d. The mean stream-wise velocity was made the canopy zone (patch as well as gap zones) of DLV arrangement
dimensionless with reference to the initial average velocity ‘U’. The (while keeping short vegetation submerged and tall emerged; Case A)
velocities are reduced to minimum close to the bed zone because of the are reduced by a percentage difference of approximately 42, 37, and
channel bed resistance. The depth integrated flow velocities in the gap 55% as compared to that of SV arrangement (Case B), EV arrange-
zones, that is, locations x2, x4, and x6, are reduced by a percentage differ- ment (Case C), and DLV arrangement (while keeping both short and
ence of 21, 6, 25, and 4% as compared to their following patch zones tall vegetation submerged; Case D), respectively. Whereas, the flow
that is, locations x3, x5, and x7, for Case A, B, C, and D, respectively. For velocities are reduced by a percentage difference of 7% in the case of
the first patch location, that is, x1, the profiles show relatively larger EV (Case C) as compared to that of case SV (Case B).
velocities within the height of short vegetation zone (which is affected
more in DLV configuration). This profile may exist because the drag influ- Velocity structure scheme
ence of individual vegetation structures on the flow has not significantly The flow is observed to be developing in the leading patch for almost
outspread (in the leading patch) to interact with the effects of other veg- all the cases. However, the flow developed after passing through the
etation structures. Such zone is named as intrusion zone, as has also first patch of vegetation and achieved an equilibrium state in almost all
been discussed by Zeng and Li (2014) and Zhao and Huai (2016), after the considered configurations. The schematic diagram for the velocity
which the velocity is reduced within the short vegetation height, that is, profile structure of vertical distribution through the DLV, SV, and EV
for locations x2–x7. A noteworthy difference in velocity magnitudes after achieving the developed state is represented in Figure 4e. The
between the lower canopy layer and upper canopy layer can be sketch describes the general idea of the vegetated flow over vertically
observed in cases of DLV, that is, Case A and D (Figure 4a,d). The veloc- double-layered and single-layered (submerged as well as emergent)
ity magnitudes are low in lower canopy layer, for the locations x3–x7, rigid vegetation along with the flow depth for the particular vegetation
which began to rise near the top of short vegetation and showed slightly arrangement. The velocity profile in SV conditions (Case A, B, and D)
increasing trends towards the flow surface. The higher velocities above seems to be complex due to inflection around the boundary layer.
the short vegetation indicate faster flow due to the sparse arrangement Thus, multiple layers, that is, low velocity zone within the height of
of tall vegetation layer. Moreover, for both the patch as well as the gap short vegetation, inflection zone around the top boundary of SV, and
zones, the non-dimensional velocity magnitudes relatively decreased high velocity zone above the short vegetation height, exist in the
within the short vegetation height, as the flow travelled further down- velocity profile through DLV and SV. This indicates a similarity with
stream, whereas the trend is opposite in the overlying flow, indicating the velocity profiles found by previous researcher (Huai et al., 2014;
that the flow develops and becomes stable farther away from the leading Singh, Rahimi, & Tang, 2019). The velocity structure scheme through
patch. Whereas, the velocity magnitudes are reduced within the vegeta- DLV (Case A) shows that the mixing layer region due to inflection
tion canopy, followed by a steep gradient of velocity over the canopy nearby the top of short vegetation layer is limited because of the
top in case of SV only (Case B). The velocity within the canopy layer in external drag by the tall vegetation layer, which results in reduced
SV case (Figure 4b) varied significantly with respect to the locations, that sharpness of velocity gradient over this region of flow. However; on
is, decreased as the flow travelled further downstream from the leading the contrary, the velocity profile experiences a steep sharpness over
patch. While a stronger rise in velocity is observed above the short vege- the mixing layer region in SV configuration (Case B) due to the absence
tation height, followed by a logarithmic profile distribution in this region of external drag by tall vegetation layer, which increases the length of
due to absence of the vegetation drag, where the profiles obtained the mixing layer, that is, approximately twice to that of DLV case.
similar trend, that is, no difference in magnitudes at all the locations. This Whereas, an almost constant spatial vertical distribution of velocity
is similar to the findings reported by Lopez and Garcia (2001) and results in case of EV (Case C) due to constant drag offered by the only
Righetti and Armanini (2002). On the contrary, the velocity profiles are tall vegetation with emerged condition, signifying a uniform flow distri-
almost constant above the bed zone towards the flow surface at almost bution within the canopy column. Similar results were found by the
all the locations for emergent vegetation (EV) configuration (Case C: past researchers (Nepf & Vivoni, 2000; Tsujimoto, Shimizu, Kitamura, &
Figure 4c). Furthermore, the velocity near the flow surface slightly Okada, 1992). Moreover, when the discharge was high that is, both
decreased (especially for the locations x2, x3, x4, and x5), which may be short and tall vegetation were submerged (Case D), the mixing layer
due to the reason of external drag offered by the emerged vegetation around the upper canopy top became slightly higher as compared to
structures. The velocity profiles showed twice increments, that is, around the lower canopy top. Hence, the velocity structure is even more com-
the top of both short and tall vegetation, in case of higher discharge plex with the complete submergence of heterogeneous vegetation
value where both configurations were submerged (Case D). The configuration.
ANJUM AND TANAKA 1709

3.1.2 | Contour plot distribution of velocity the flow encountered significant variations as it reached the vegeta-
tion patches. The velocities in the gap zones are lower in comparison
Longitudinal distribution to those in the patch zones due to the sheltering influence of it
The spatial distribution of the dimensionless stream-wise velocity at (Figure 5a,c,e,g). Thus, a noteworthy difference in the flow velocity is
longitudinal sections, that is, LS1 (passing through the gap of short found between the vegetation patch and the gap zones. The velocity
and tall vegetation array) and LS2 (passing directly through the short profiles failed to obtain their usual shapes due to the reason that the
and tall vegetation array) are presented in Figure 5a–h. It shows that effect of the patches prevails for a specific distance behind the

F I G U R E 5 Contour plots of the


spatial distribution of non-
dimensional stream-wise velocity ‘u/
U’ along the longitudinal sections, that
is, LS1 and LS2, for (a and b) Case A,
(c and d) Case B, (e and f) Case C, and
(g and h) Case D. The dashed boxes
show the locations and the area
occupied by the vegetation patches.
For the longitudinal sections, see
Figure 1 [Colour figure can be viewed
at wileyonlinelibrary.com]
1710 ANJUM AND TANAKA

downstream edge. Overall, the flow velocities in DLV arrangement arrangement (Case A) are comparatively reduced to a minimum (in the
are observed to be very low (Figure 5a,b) due to the mutual influence wake zones) in comparison to that observed in SV and EV arrange-
of short and tall vegetation, as compared to those resulted in SV and ments. Past researches have proposed that the wake zones behind
EV arrangements (Figure 5c–f). Contrarily, the flow resistance through the vegetation support deposition of fine particle that further stimu-
DLV at higher discharge value (Case D; Figure 5g,h), is observed to be lates the growth and expansion of the vegetated zone (Gurnell,
relatively low due to submergence of both vegetation configurations. Thompson, Goodson, & Moggridge, 2008; Shi, Jiang, & Nepf, 2016;
Furthermore, the velocities directly behind the cylinders in DLV Tsijimoto, 1999). This type of finite patches of vegetation can produce

F I G U R E 6 Contour plots of the spatial distribution of non-dimensional stream-wise velocity ‘u/U’ along the cross-sections, that is, C1 and C2,
for (a and b) Case A, (c and d) Case B, (e and f) Case C, and (g and h) Case D. The dashed boxes show the locations and the area occupied by the
vegetation patches. For the cross-sections, see Figure 1 [Colour figure can be viewed at wileyonlinelibrary.com]
ANJUM AND TANAKA 1711

the bio-geomorphic responses between the flow and vegetation, vegetation. In general, the ratio of the discharge (calculated using the
deposition of sediments, and further growth (Gurnell, 2014; Gurnell, lateral velocity distribution) passing through the canopy zone to that
Bertoldi, & Corenblit, 2012; Marion et al., 2014; Nepf, 2012). Thus, in the non-vegetated part is found to be 0.33, 0.85, 0.48, and 0.70 for
the vegetation patches with double-layered configuration can offer DLV (while keeping short cylinder submerged and tall emergent), SV,
more sheltering to the gap zones compared to that of single-layered EV, and DLV (while keeping both short and tall cylinders submerged),
configurations, signifying positive feedback towards aquatic life as respectively. Thus, it also identifies that the presence of vertically
well as sediment deposition. DLV, while keeping the only short vegetation in the submerged state,
A rise in the flow velocity around the top of submerged short can offer more resistance to the flow in the canopy zone and
vegetation can be clearly observed, with high velocity magnitudes in increases the main channel flow carrying capacity.
the upper zone of short vegetation height, and low magnitudes of The velocity vectors of the secondary currents (lateral ‘v’ and ver-
velocity within the lower zone (Figure 5b,d,h). Barrios-Piña, Ramírez- tical ‘w’) along the cross-section (C2) are presented in Figure 7. A clear
León, Rodríguez-Cuevas, and Couder-Castañeda (2014) also found in difference between the canopy and no-canopy zone can be observed.
their numerical modelling of flow through layered vegetation array The strength of the secondary currents (a relatively minor flow sup-
that the velocity rises above the SV zone. This type of flow structures erimposed on the primary flow, with relatively different speed and
in the locality of cylinders is not easy to attain with an experimental direction) is visibly larger in the canopy area as compared to that in
investigation, which shows the importance of the present numerical the no-canopy zone. The drag offered by the vegetation cylinders
study. A stronger rise in the flow velocity is found above the canopy induced secondary currents. The predicted secondary currents
layer in SV configuration (Figure 5c,d), signifying a strong momentum showed a relatively large magnitude near the interface close to the
transfer between the canopy top and the overlying flow. Moreover, vegetation zone and moved slightly away from the interface to show
the vertical velocity distribution appeared uniform through EV due to a circulation, that is, the vectors of v and w show a pattern of second-
the constant vegetation formation with the emerged condition, ary circulations around the vegetation. Thus, this rotation of second-
followed by no inflectional instability in the flow (Figure 5e,f). The ary currents resulted in swirling flows around the cylindrical bodies.
flow velocities through DLV configuration also showed rising trends The retarding effect of the vegetation zone generates the point of
around the edge of submerged canopies at higher discharge value inflection, which induces the flow instability to generate the vortex at
(Figure 5g,h), indicating a stronger velocity rise around upper canopy the interface zone (Jahra, Kawahara, Hasegawa, & Yamamoto, 2011;
top as compared to that of the lower canopy top. Nezu & Onitsuka, 2001). Thus, by flow visualization, vortices have
been observed mainly at the vertical and lateral canopy interfaces;
Lateral distribution and hence, dominating the vertical and lateral transport of momentum
The spatial distribution of the dimensionless stream-wise velocity and are advected between the canopy and no-canopy zone by the
along the cross-sections, that is, C1 (passing through the gap between secondary currents, generated by the anisotropy of the turbulence.
the patches) and C2 (passing through the centre of vegetation patch), The detailed description of mixing and coherent vortices depending
is presented in Figure 6a–h. It shows that the flow velocity distribu- upon the formation and submergence level of vegetation configura-
tion is significantly influenced due to vegetation formation. The veloc- tions is discussed in the next section.
ities in the canopy zone (obstructed part) greatly reduced because of
the drag offered by the vegetation in all the cases. The presence of
vegetation along one side of the channel significantly offered resis- 3.1.3 | Description of mixing along the cross-
tance to the flow and affected the channel conveyance. The velocities section
increased significantly while moving from the canopy zone to the
non-obstructed part, that is, non-vegetated zone, trailed by a notice- In the previous sections, it is noticeably shown that the velocity is
able alteration along the interface between the two zones. Thus, the diminished in the canopy zone. Although it showed an increment rela-
fast-moving flow in the non-vegetated zone and slow-moving flow in tive to the initial average velocity in the overlying flow of canopy
the canopy zone (due to the velocity difference between the two which is pronounced more in the non-vegetated zone. This is because
zones) results in a larger momentum exchange along the interface of the drag associated with the vegetation structures, such that the
between the two zones, accompanied by a free shear layer. This vegetation and water interface is a zone of strong shear that shows a
shows consistency with the outcomes presented by previous mixing layer. A mixing layer is formed when the momentum absorp-
researchers (Huai, Zhang, Wang, & Katul, 2019; Tang et al., 2019). tion by the canopy can generate a sudden variation in the velocity
The flow velocities in the non-vegetated zone increased by a per- profile, which is essential to produce the Kelvin–Helmholtz instability
centage difference of 75, 54, 68, and 59% compared to that in the (Finnigan, 2000; Ghisalberti & Nepf, 2005). Thus, vortices govern the
canopy zone for DLV with short cylinder submerged and tall emergent mass and momentum exchange between the canopy and the overly-
(Case A), SV (Case B), EV (Case C), and DLV with both short and tall ing flow or the flow in the non-vegetated part of the channel. The
cylinders submerged (Case D), respectively. This indicates that high mixing zone between the two almost-uniform mean velocity sections
flow resistance in the canopy zone is compensated with the faster define an effective vortex size and is supposed to initiate from
flow in the no-canopy zone, showing the importance of side Kelvin–Helmholtz instability (Huai et al., 2019). This coherent vortex
1712 ANJUM AND TANAKA

F I G U R E 7 Secondary flow, that is, v,w velocity vector fields (m/s), along the cross-section (C2) for (a) Case A, (b) Case B, (c) Case C, and
(d) Case D. Where the black and red dotted lines shows the location of cylinders just before and after the selected cross-section, respectively
[Colour figure can be viewed at wileyonlinelibrary.com]

size is regarded as the shear length scale in the canopy turbulence instabilities between the vegetation formations of both cases, that is,
(Raupach, Finnigan, & Brunet, 1996), and it represents the characteris- DLV and SV. The vortex size along the lateral interface for DLV is sig-
tics of the turbulence length scale for the vegetated flows at the can- nificantly reduced and affected with the submergence of both short
opy/open water interface (Yan, Wai, & Li, 2016). and tall vegetation during high flow discharge (Figure 8d). The results
Figure 8a–d shows the schematic description of vertical and lat- show that the size of these coherent vortices and the mixing length
eral mixing in the partly obstructed channel by vegetation along the growth around the interface depends on the canopy configuration as
cross-section. Thus, the mixing layers at the vertical and lateral inter- well as the submergence condition of vegetation formations.
faces adopted from the lateral velocity distributions (Figure 6) were
used to estimate the vortex size. The integrated size of vertical and
transverse coherent vortices estimated from the velocity distributions 3.2 | Turbulence characteristics
demonstrate that the efficiency of lateral transport of momentum at
the interfacial zone, that is, interface of vegetated and non-vegetated 3.2.1 | Turbulent kinetic energy
part, for DLV case (Figure 8a) was observed to be notably high as
compared to that of vertical momentum transfer between the SV top The distribution of dimensionless depth averaged turbulent kinetic
and its overlying flow. In contrast to this, the mixing length is energy (TKE) along the longitudinal section (LS1) is presented in
observed to be restricted at the lateral interfacial zone in case of SV Figure 9. The TKE is the mean kinetic energy per unit mass affiliated
(Figure 8b), whereas a larger growth of vortex size or mixing length is with the eddies in a turbulent fluid flow. The turbulent kinetic energy
observed over the submerged canopy top and its overlying flow. This is characterized by the fluctuations of the root-mean square velocity
signifies a larger inflectional instability over this zone of flow which (Baldocchi, 2005; Pope, 2000). It can be observed that the flow is
governs and sufficiently increases the efficiency of vertical exchange disrupted and showed non-uniformity as it passed through the patch
of momentum as compared to that of lateral transport of momentum zones. A significantly rising trend in TKE is found through the patch
over the interfacial zone. Furthermore, the mixing length along the zones, in comparison to that in the gap zones for all the cases. This is
interfacial zone is witnessed to be slightly lower for EV case due to the reason of higher drag by the vegetation structures in the
(Figure 8c) in comparison to that of DLV. Whereas in the contrast, it patch zones, which produces higher turbulent energy. While passing
was found to penetrate more in comparison to that of SV, which can through the patch zones, the flow entered in the low and high velocity
be a compensation of difference in the drag and inflectional zones where it followed rising and falling trends; thus, replicated a
ANJUM AND TANAKA 1713

F I G U R E 8 Schematic diagram of vortex behavior within the partly obstructed channel by vegetation describing the vertical and lateral mixing
along the cross-section for (a) Case A, (b) Case B, (c) Case C, and (d) Case D [Colour figure can be viewed at wileyonlinelibrary.com]

and EV (Figure 5c,e) as compared to those in DLV case (Figure 5a), the
production of TKE is also observed to be comparatively high in these
cases. Moreover, the relatively large production of TKE in the patch
zones for Case C might be due to the reason of greater velocity fluc-
tuations due to high resistance and drag by the tall EV. On the other
hand, the logarithmic layer of flow in SV case experienced minimum
production of TKE due to absence of the drag and turbulence offered
by the vegetation in this region of flow.

3.2.2 | Turbulent intensity

The distribution of turbulent intensity (ratio of the root-mean square


F I G U R E 9 Longitudinal distribution of depth averaged turbulent
kinetic energy ‘TKE’ along the domain length. For the longitudinal velocity fluctuations to the mean flow velocity) along the longitudi-
section (LS1), see Figure 1a [Colour figure can be viewed at nal section (LS1) is presented in Figure 10 to characterize the turbu-
wileyonlinelibrary.com] lence in the channel flow. The turbulent intensity (also often
referred to as turbulence level) significantly rose in the vegetation
saw-tooth distribution. This result is consistent with that observed in patch zones, in comparison to that in the gap between the patches,
previous research work (Zhao & Huai, 2016). Furthermore, the magni- indicating high flow turbulence in the areas covered by canopy.
tudes of TKE through DLV are relatively higher in the lower canopy Thus, these gap zones (between the vegetation patches) of low flow
layer in comparison to that in the upper canopy layer (Case A and velocities and low flow turbulence due to flow resistance and shel-
Case D). This effect resulted because of relatively low porosity of veg- tering by the vegetation patches may be suitable for aquatic life in
etation in the lower canopy layer, which caused larger turbulence to terms of their nourishment and physical environment. Moreover, the
the flow, by producing more turbulent eddies. As the flow velocities, turbulent intensity appears to be comparatively very high for DLV
as well as its fluctuations, are large within the vegetation zones for SV configuration (Case A) due to low flow velocities in the canopy zone
1714 ANJUM AND TANAKA

• The presence of discontinuous rigid vegetation with partial distri-


bution in an open channel results in multiple flow mechanisms. The
flow velocity reduces within the height of short SV, followed by an
inflectional instability around the canopy top resulting in a mixing
layer over this region of flow. The length of the mixing layer is
influenced and inhibited to half in case of vertically double-layered
(shorter submerged and taller emergent) vegetation compared to
that in submerged single-layered vegetation. Furthermore, the ver-
tical velocity profile becomes even more complex due to double
inflectional instabilities through layered canopy when both short
and tall vegetation is submerged.
• The sheltering effect of the vegetation patches reduces the flow
F I G U R E 1 0 Longitudinal distribution of depth averaged turbulent velocities within the gap zones, the influence of which is observed
intensity along the domain length [Colour figure can be viewed at
to be significantly high for the flow through DLV configuration
wileyonlinelibrary.com]
(with short vegetation submerged and tall emerged) compared to
those of single-layered (either in submerged or emerged condition)
and very high flow resistance offered by the combined effect of vegetated flows. This categorizes that the discontinuous patches
shorter submerged and taller EV. The interaction of short and tall with layered vegetation formation can give better positive feed-
vegetation results in the production of larger wakes and vortices that back in terms of the physical environment of the aquatic organisms
promote the generation of more flow turbulence (Raupach for their nourishment and further growth of vegetal communities
et al., 1996). On the contrary, although the flow is observed to be as well as deposition of sediments.
comparatively non-uniform and velocity fluctuations are found to be • Due to the partial vegetation obstruction in the channel, the veloc-
very high in SV (Case B) and EV (Case C); however, the flow resis- ity in the vegetation part reduces considerably because of the veg-
tance due to vertically single-layered vegetation (either in sub- etation drag; whereas the fluid passes with higher flow velocities
merged or emerged condition) is found to be comparatively less, through the non-vegetation part of the channel, thus affects the
which resulted in low turbulent intensity compared to that of verti- channel conveyance. The difference in flow velocity between the
cally DLV. Furthermore, the turbulent intensity significantly reduced vegetated and non-vegetated zones results in a free shear layer at
when both the short and tall vegetation were submerged in Case D the interface of the vegetation and non-vegetation zones, which
as compared to that of Case A, which is due to the reason of reduced results in the domination of the lateral exchange of momentum.
drag and resistance offered by this heterogeneous vegetation forma- While keeping the vegetation density and porosity constant, the
tion under fully submergence state. The turbulent intensity is found ratio of the discharge through the vegetation part to the non-
to be minimum in the logarithmic layer in Case B due to no flow vegetation part becomes almost 0.33, 0.85, 0.48, and 0.70 for DLV
obstruction in this region. (by keeping short cylinder vegetation while tall emergent), SV, EV,
The results identify that the patch type vegetation with layered and DLV (while keeping both short and tall vegetation submerged),
configuration (while keeping short vegetation submerged and tall respectively.
emergent) or only emerged condition of vegetation can provide large • The observed mixing layers for flows through DLV suggest a com-
resistance to the flow in the canopy part that can increase the dis- plex and stronger lateral exchange for partly vegetated channels in
charge carrying capacity of the main channel by influencing its effi- comparison to flows through simpler configurations of canopy like
ciency. The reduced percentage of turbulence in the gap zones SV or EV only. In contrast to this, the vertical mixing and transport
between these patches can provide positive feedback for the aquatic of momentum in SV configuration dominate the lateral mixing over
life and sediment deposition; that is expected to be pronounced more the interfacial zone. Thus, it indicates that the plant morphology
in case of DLV. along with its elevation or submergence in a partly vegetated chan-
nel strongly affects the interface turbulence with implications on
the lateral momentum transport.
4 | C O N CL U S I O N S • The drag offered by vegetation in the patch zones caused the flow
turbulence to noticeably increase, whereas on the contrary, the
A CFD tool based on RANS technique utilizing FLUENT was turbulence in the flow reduced as it passed through the gap zones.
applied in this study to investigate the turbulent flow behaviour This clarifies that the gaps between the patch zones acquire less
through longitudinally discontinuous vegetation with a partial dis- percentage of turbulence, where the environment to the aquatic
tribution in an open channel under varying vegetation formations life may become suitable.
and submergence levels. The results have been analysed qualita-
tively as well as quantitatively. Following conclusions were drawn The study showed that the vegetation with vertically layered for-
from this study: mation (while keeping short cylinder submerged and tall emergent)
ANJUM AND TANAKA 1715

reduced the flow in the vegetation part of the channel to a great patches in open channel flow. River Research and Applications, 36(1),
extent. Whereas it sufficiently increased the discharge carrying capac- 115–127. https://doi.org/10.1002/rra.3546
Baldocchi, D. (2005). Lecture 16, wind and turbulence, part 1, surface bound-
ity of the main channel (without vegetation). This type of vegetation
ary layer: Theory and principles, ecosystem science division. Berkeley,
pattern could be utilized over the floodplains that can become very CA: Department of Environmental Science, Policy and Management,
effective during higher flows like floods. In general, the outcomes of University of California.
this study may increase the comprehensive knowledge of the vegeta- Barrios-Piña, H., Ramírez-León, H., Rodríguez-Cuevas, C., & Couder-
Castañeda, C. (2014). Multilayer numerical modeling of flows through
tion effects on flow structures in partially vegetated channels
vegetation using a mixing-length turbulence model. Water, 6,
(e.g., rivers or streams) with discontinuity, and may also give some 2084–2103.
guidance for the management and practice of river ecological restora- Bauer, M., Harzer, R., Strobl, K., & Kollmann, J. (2018). Resilience of ripar-
tion on river banks or floodplains. The low flow velocities as well as ian vegetation after restoration measures on river inn. River Research
and Applications, 34(5), 451–460.
low turbulence in the gaps between the vegetation patches indicates
Choi, S. U., & Kang, H. (2006). Numerical investigations of mean flow and
a positive aspect for aquatic life, stabilizing the bed, promoting habitat turbulence structures of partly vegetated open-channel flows using
diversity as well as sediment deposition that may drive the pattern of the Reynolds stress model. Journal of Hydraulic Research, 44, 203–217.
future vegetation growth in open channel flows. In addition, this study Cui, J., & Neary, V. S. (2008). LES study of turbulent flows with submerged
vegetation. Journal of Hydraulic Research, 46(3), 307–316.
will prove helpful in a logical distribution of vegetation in the flow to
Devi, T. B., & Kumar, B. (2016). Flow characteristics in an alluvial channel
minimize the flood disaster, ecological restoration of a river in order
covered partially with submerged vegetation. Ecological Engineering,
to restore the function of the river system and designing ecological 94, 478–492.
habitats for aquatic organisms. This mixed culture of vegetation could Devi, T. B., Sharma, A., & Kumar, B. (2019). Flow characteristics in a partly
be a feasible solution as a natural approach towards reducing flood vegetated channel with emergent vegetation and seepage.
Ecohydrology & Hydrobiology, 19(1), 93–108.
risk or to manage flooding over the floodplains. However, the mixing
Finnigan, J. (2000). Turbulence in plant canopies. Annual Review of Fluid
of tall trees with short vegetation for the river management and fores- Mechanics, 32(1), 519–571.
tation can influence the intrusion process which is an important factor Fluent Inc. (2019). User's guide, Release 19.2.
for the variation in river morphology (such landscape can also influ- Folkard, A. M. (2005). Hydrodynamics of model Posidonia oceanic patches
in shallow water. Limnology and Oceanography, 50, 1592–1600.
ence flooding). Therefore, it is further recommended to experimen-
Ghani, U., Anjum, N., Pasha, G. A., & Ahmad, M. (2019). Investigating the
tally or numerically investigate such influencing factors that may turbulent flow characteristics in an open channel with staggered vege-
affect flood risk management as well as river morphology. In addition, tation patches. River Research and Applications, 35(7), 966–978.
a more detailed study is needed to illustrate the vertical and lateral https://doi.org/10.1002/rra.3460
Ghisalberti, M., & Nepf, H. (2005). Mass transport in vegetated shear
momentum exchange (with the help of related physical quantities
flows. Environmental Fluid Mechanics, 5(6), 527–551.
such as Reynolds stress, etc.) along-with the quantification of momen- Gurnell, A. M., Thompson, K., Goodson, J., & Moggridge, H. (2008). Propa-
tum thickness and shear layer growth through the considered vegeta- gule deposition along river margins: Liking hydrology and ecology.
tion configurations to further clarify the phenomena. Journal of Ecology, 96, 553–565.
Gurnell, A. M., Bertoldi, W., & Corenblit, D. (2012). Changing river chan-
nels: The roles of hydrological processes, plants and pioneer fluvial
ACKNOWLEDGEMEN TS landforms in humid temperate, mixed load, gravel bed rivers. Earth-
The authors acknowledge the anonymous reviewers for their useful Science Reviews, 111(1–2), 129–141. https://doi.org/10.1016/j.
comments to improve this manuscript. The authors are also thankful earscirev.2011.11.005
Gurnell, A. M. (2014). Plants as river system engineers. Earth Surface Pro-
to Mr. Rowan De Costa for his help in improving the English language
cesses and Landforms, 39, 4–25. https://doi.org/10.1002/esp.3397
of the manuscript.
Hamed, A. M., Sadowski, M. J., Nepf, H. M., & Chamorro, L. P. (2017).
Impact of height heterogeneity on canopy turbulence. Journal of Fluid
DATA AVAI LAB ILITY S TATEMENT Mechanics, 813, 1176–1196.
The authors confirm that the data supporting the findings of this Huai, W., Wang, W., Hu, Y., Zeng, Y., & Yang, Z. (2014). Analytical model
of the mean velocity distribution in an open channel with double-
study are available within the article.
layered rigid vegetation. Advances in Water Resources, 69, 106–113.
Huai, W., Xue, W., & Qian, Z. (2015). Large-eddy simulation of turbulent
CONF LICT S OF INTE R ES T rectangular open-channel flow with an emergent rigid vegetation
The authors declare no conflicts of interest for this study. patch. Advances in Water Resources, 80, 30–42.
Huai, W., Zhang, J., Wang, W. J., & Katul, G. G. (2019). Turbulence struc-
ture in open channel flow with partially covered artificial emergent
ORCID vegetation. Journal of Hydrology, 573, 180–193.
Naveed Anjum https://orcid.org/0000-0002-5469-3434 Jahra, F., Kawahara, Y., Hasegawa, F., & Yamamoto, H. (2011). Flow–
Norio Tanaka https://orcid.org/0000-0002-0592-2173 vegetation interaction in a compound open channel with emergent
vegetation. International Journal of River Basin Management, 9(3–4),
247–256.
RE FE R ENC E S Javernick, L., & Bertoldi, W. (2019). Management of vegetation encroach-
Anjum, N., & Tanaka, N. (2020). Hydrodynamics of longitudinally discon- ment by natural and induced channel avulsions: A physical model. River
tinuous, vertically double layered and partially covered rigid vegetation Research and Applications, 35(8), 1257–1268.
1716 ANJUM AND TANAKA

Kouwen, N., & Unny, T. E. (1975). Flexible roughness in open channels. Stoesser, T., Liang, C., Rodi, W., & Jirka, G. H. (2006). Large eddy simula-
Journal of the Hydraulics Division, 101(NHY1), 194–196. tion of fully-developed turbulent flow through submerged vegetation.
Lima, P. H. S., Janzen, J. G., & Nepf, H. M. (2015). Flow patterns around In Riverflow (pp. 227–234). London: Taylor & Francis Group.
two neighboring patches of emergent vegetation and possible implica- Su, X., & Li, C. (2002). Large eddy simulation of free surface turbulent flow
tions for deposition and vegetation growth. Environmental Fluid in partly vegetated open channels. International Journal for Numerical
Mechanics, 15(4), 881–898. Methods in Fluids, 39, 919–937.
Liu, D., Diplas, P., Hodges, C. C., & Fairbanks, J. D. (2010). Hydrodynamics Tanaka, T., Ohmoto, T., & Tanaka, T. (2008). Flow resistance and momen-
of flow through double layer rigid vegetation. Geomorphology, 116, tum transport in open channel with longitudinaly discontinuous vege-
286–296. tation. Proceedings of Hydraulic Engineering, 52, 763–768. https://doi.
Lopez, F., & Garcia, M. H. (2001). Mean flow and turbulence structure of org/10.2208/prohe.52.763
open-channel flow through non-emergent vegetation. Journal of Tanaka, T., & Ohmoto, T. (2015). Turbulent structure in open channel with
Hydraulic Engineering, 127(5), 392–402. permeable and impermeable side cavities. Journal of Applied Water
Marion, A., Nikora, V., Puijalon, S., Bouma, T., Koll, K., Ballio, F., … Engineering and Research., 4, 44–53. https://doi.org/10.1080/
Statzner, B. (2014). Aquatic interfaces: A hydrodynamic and ecological 23249676.2015.1090354
perspective. Journal of Hydraulic Research, 52, 744–758. https://doi. Tang, X., Rahimi, H., Singh, P., Wei, Z., Wang, Y., Zhao, Y., & Lu, Q. (2019).
org/10.1080/00221686.2014.968887 Experimental study of open-channel flow with partial double-layered
Neary, V. S. (2003). Numerical solution of fully developed flow with vege- vegetation. The 1st International Symposium on Water Resource and
tative resistance. Journal of Engineering Mechanics, 129(5), 558–563. Environmental Management (WREM 2018), Kunming, China, Edited
Nepf, H. M. (2012). Flow and transport in regions with aquatic vegetation. by Wang, Y.; E3S Web of Conferences, Volume 81, id.01010.
Annual Review of Fluid Mechanics, 44, 123–142. Tsujimoto, T., Shimizu, Y., Kitamura, T., & Okada, T. (1992). Turbulent
Nepf, H. M., & Vivoni, E. R. (2000). Flow structure in depth-limited, vege- open-channel flow over bed covered by rigid vegetation. Journal of
tated flow. Journal of Geophysical Research, 105(C12), 547–557. Hydroscience and Hydraulic Engineering, 10(2), 13–25.
Nezu, I., & Onitsuka, K. (2001). Turbulent structures in partly vegetated Versteeg, H. K., & Malalasekera, W. (2008). An introduction to computa-
open-channel flows with LDA and PIV measurements. Journal of tional fluid dynamics. Retrieved from https://ekaoktariyantonugroho.
Hydraulic Research, 39, 629–642. files.wordpress.com/2008/04/an-introduction-to computational-fluid
Okamoto, T., & Nezu, I. (2013). Spatial evolution of coherent motions in dynamics-versteeg.pdf
finite-length vegetation patch flow. Environmental Fluid Mechanics, 13, White, B. L., & Nepf, H. M. (2008). A vortex-based model of velocity and
417–434. shear stress in a partially vegetated shallow channel. Water Resources
Pope, S. B. (2000). Turbulent flows (pp. 122–134). Cambridge: Cambridge Research, 44(1), W01412. https://doi.org/10.1029/2006WR005651
University Press. Wu, W., & He, Z. (2009). Effects of vegetation on flow conveyance and
Raupach, M. R., Finnigan, J. J., & Brunet, Y. (1996). Coherent eddies and sediment transport capacity. International Journal of Sediment Research,
turbulence in vegetation canopies: The mixing-layer analogy. Bound- 24, 247–259.
ary-Layer Meteorology, 78, 351–382. Yan, X., Wai, W. O., & Li, C. (2016). Characteristics of flow structure of
Righetti, M., & Armanini, A. (2002). Flow resistance in open channel flows free-surface flow in a partly obstructed open channel with vegetation
with sparsely distributed bushes. Journal of Hydrology, 269, 55–64. patch. Environmental Fluid Mechanics, 16, 807–832. https://doi.org/10.
Rominger, J. T., & Nepf, H. M. (2011). Flow adjustment and interior flow 1007/s10652-016-9453-4
associated with a rectangular porous obstruction. Journal of Fluid Yang, Y., Irish, J. L., & Socolofsky, S. A. (2015). Numerical investigation of
Mechanics, 680, 636–659. wave-induced flow in mound-channel wetland systems. Coastal Engi-
Seer, F. K., Brunke, M., & Schrautzer, J. (2018). Mesoscale river restoration neering, 102, 1–12.
enhances the diversity of floodplain vegetation. River Research and Zeng, C., & Li, C. W. (2014). Measurements and modeling of open-channel
Applications, 34(8), 1013–1023. flows with finite semi-rigid vegetation patches. Environmental Fluid
Shi, Y., Jiang, B., & Nepf, H. M. (2016). Influence of particle size and den- Mechanics, 14(1), 113–134.
sity, and channel velocity on the deposition patterns around a circular Zhao, F., & Huai, W. (2016). Hydrodynamics of discontinuous rigid sub-
patch of model emergent vegetation. Water Resources Research, 52, merged vegetation patches in open-channel flow. Journal of Hydro-
1044–1055. https://doi.org/10.1002/2015WR018278 Environment Research, 12, 148–160.
Singh, P., Rahimi, H. R., & Tang, X. (2019). Parameterization of the model-
ing variables in velocity analytical solutions of open-channel flows with
double-layered vegetation. Environmental Fluid Mechanics, 19(3),
765–785. https://doi.org/10.1007/s10652-018-09656-8 How to cite this article: Anjum N, Tanaka N. Investigating the
Stoesser, T., Neary, V., & Wilson, C. (2005). Modeling vegetated channel turbulent flow behaviour through partially distributed
flows: Challenges and opportunities. In WSEAS (the world scientific and
discontinuous rigid vegetation in an open channel. River Res
engineering academy and society) conference on fluid mechanics, Corfu,
Greece, Corfu, Greece: Online Research @ Cardiff (ORCA). http://
Applic. 2020;36:1701–1716. https://doi.org/10.1002/
orca.cf.ac.uk/id/eprint/21362. rra.3671

You might also like