You are on page 1of 13

Agricultural and Forest Meteorology 103 (2000) 301313

A first-order closure model for the wind flow within


and above vegetation canopies
Pingtong Zeng , Hidenori Takahashi
Laboratory of Geoecology, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0810, Japan
Received 18 March 1999; accepted 11 January 2000

Abstract
A first-order closure model that has the general utility for predicting the wind flow within and above vegetation canopies
is presented. Parameterization schemes that took into account the influence of large turbulent eddies were developed for the
Reynolds stress and the mixing length in the model. The results predicted by the model were compared with measured data for
wind speeds within and above six types of vegetation canopy, including agricultural crops, deciduous and coniferous forests,
and a rubber tree plantation during fully leafed, partially leafed and leafless periods. The predicted results agreed well with
the measured data; the root-mean-square errors in the predicted wind speeds (non-dimensionalized by the friction velocity
above the canopy) were about 0.2 or less for all of the canopies. The secondary wind maxima that occurred in the lower
canopies were also correctly predicted. The influence of foliage density on the wind profiles within and above a vegetation
canopy was successfully simulated by the model for a rubber tree plantation during fully leafed, partially leafed and leafless
periods. The bulk momentum transfer coefficients (CM ) and the values of (which are defined by z0 =(hd), where z0
is the roughness and d is the zero-displacement height of the canopy) for the vegetation canopies were also studied, and
the relationships CMh =0.0618 exp(0.792CF ) and =0.209 exp(0.414CF ) were determined; here, CMh is the bulk momentum
transfer coefficient at the canopy top; CF =Cd PAI zmax /h, where Cd is the effective drag coefficient of the canopy, PAI is the
plant area index and zmax is the height at which the plant area density is maximum. The values of ranged from 0.22 to 0.32
for the canopies studied. 2000 Elsevier Science B.V. All rights reserved.
Keywords: Numerical model; Wind velocity; Vegetation canopy; Reynolds stress; Mixing length; Turbulent eddy

1. Introduction been developed. However, accurate prediction of wind


flow is difficult due to the complexity in the array of
Wind is an important factor for scalar fluxes (heat, vegetation elements (leaves, branches and so on) and
water vapor, carbon dioxide, etc.) and movements of the complex process of air momentum transport within
spores, pollen and particles within a vegetation canopy. a vegetation canopy. Early modeling studies (e.g. In-
Information about the wind flow within the vegeta- oue, 1963; Cowan, 1968; Thom, 1971) were based on
tion canopy is important in meteorological, agricul- the K-theory or gradient-diffusion theory; that is, the
tural and ecological studies. A number of numerical momentum flux is equal to the product of an eddy
models for predicting the vegetation wind flow have viscosity and the local gradient of mean wind ve-
locity. Wilson and Shaw (1977) (hereafter WS) point
Corresponding author. Fax: +81-11-706-4867. out that a K-theory model provides little insight into
E-mail address: sht@ees.hokudai.ac.jp (P. Zeng) the nature of momentum transport processes within

0168-1923/00/$ see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 9 2 3 ( 0 0 ) 0 0 1 3 3 - 7
302 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

the vegetation canopy. Moreover, Pereira and Shaw fluences of canopy structure and foliage density on
(1980) and Watanabe and Kondo (1990) (hereafter the wind velocities within and above the canopies.
WK) show that such a model can not provide accurate
predictions of wind velocity in the lower portion of
a plant canopy, where a near-zero vertical gradient of 2. Theory and the model
mean wind velocity or a wind velocity bulge is fre-
quently observed (Shaw, 1977). In another approach, 2.1. Governing equation
WS proposed a higher-order (second-order) closure
model in which turbulent kinetic energy equations and Following Raupach et al. (1986), under neutrally
Reynolds stress equation were solved simultaneously buoyant conditions, the time- and volume-averaged
with that for mean momentum. However, WS reported equation for the mean momentum within vegetation is
that the calculated velocity profile was sensitive to the
parameterization scheme for the turbulent transport hui i
hui i hpi ij
+ uj = + +fFi + fVi (1)
term in their model. In order to avoid the deficiency t xj xi xj
in the model of WS, Meyers and Paw (1986) (here-
with
after MP) developed a third-order closure model using
third-order closure principles. However, the utility of a hui i
ij = hu0i u0j i hu00i w 00j i + (2)
higher-order closure model is still limited. The model xj
proposed by MP includes about 10 equations for only
a one-dimensional problem, and the computation cost Here, i and j are index notations (with values of 13)
is therefore high. In addition, their modeled results for and Einsteins summation is used. The overbars and
the turbulent field have large errors (MP; Meyers and single primes denote time averages and fluctuations,
Baldocchi, 1991). The parameterization schemes for while the angle brackets and double primes denote
higher-order closure models used for predicting veg- spatial volume averages and departures therefrom, re-
etation wind flow need to be further improved (MP; spectively. ui and xi are velocity and position vec-
Shaw and Seginer, 1987; Wilson, 1988). tors, respectively, t the time, p the kinematic pressure,
As an alternative approach that does not use fFi and fVi are form and viscous drag force vectors
higher-order closure principles, Li et al. (1985) pro- exerted on a unit mass of air within the averaging
posed a non-local closure scheme for the total mo- volume, respectively, ij the volume-averaged kine-
mentum flux (Reynolds stress and dispersive flux) matic momentum flux or stress tensor, and is the
and developed a first-order closure model that was kinematic viscosity. Interpretations of each term in
capable of predicting the wind velocity peaks in lower the above equations have been described by Raupach
canopies. However, several problems in their model et al. (1986).
have been pointed out by Van Pul and Van Boxel For a horizontal homogeneous vegetation, let u and
(1990). To correct these problems, Miller et al. (1991) w represent the streamwise and vertical velocity, re-
(hereafter MLL) made some modifications to their spectively, the x-coordinate in the mean streamwise
model and applied it to wind flow across an alpine direction, and the z-coordinate normal to the ground;
forest clearing. However, the utility of their model for assuming steady-state conditions and neglecting the
different types of vegetation canopy was not tested. pressure gradient and molecular transport terms, Eq.
Over the past few decades, extensive measurements (1) becomes
(e.g. Raupach et al., 1986; Shaw et al., 1988; Gao d  0 0 
et al., 1989) of turbulent flows within and above vege- hu w i + hu00 w 00 i = fx (3)
dz
tation canopies have been carried out. On the basis
of the results of the previous studies, the purpose of In this equation, hu0 w0 i is the Reynolds stress or tur-
the present study is to develop a first-order closure bulent flux of momentum, while hu00 w 00 i represents
model, which is simple in computation and has the the dispersive flux that arises from the spatial corre-
general utility for predicting wind flow within and lation of regions of mean updraft or downdraft with
above vegetation canopies, and to investigate the in- regions, where u differs from its spatial mean, and
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 303

fx (= fF1 + fV1 ) is the total streamwise drag per unit larger-scale eddies, which is determined by the mean
mass of air within the averaging volume. These terms wind speed differences between heights with large dis-
must be parameterized in order to solve the equation tance. Hence, the Reynolds stress is parameterized by
and estimate the wind velocity hui.
d hui z
hu0 w0 i = K + Cg hur i (hur i hui) (5)
2.2. Closure schemes dz h
where K is the eddy viscosity, h the vegetation height,
2.2.1. Reynolds stress ur the wind velocity at a reference height above the
In K-theory models, the Reynolds stress is param- vegetation, and Cg is a coefficient. On the right-hand
eterized as side of the equation, the first term (defined as Rs ),
d hui which has been formed according to the conventional
hu0 w 0 i = KM (4) K-theory, is responsible for small-eddy diffusion,
dz
and the second term (defined as Rl ) is responsible
where KM is the eddy viscosity. This model shows that for non-local transport. Since non-local transport is
the turbulent flux of momentum results from the local caused by the shear between the wind flows above and
gradient in the mean wind velocity. Hence, it is also within the canopy, we used hur i hui to represent the
called a small-eddy closure technique (Stull, 1988). intensity of the shear. hur i outside of the parentheses
K-theory models have been widely used for studies was used to account for the intensity of the wind flow
in the surface layer and have been proven to be reli- above the canopy. As it is easier for turbulent eddies
able for the inertial sublayer above a surface (Raupach to penetrate into a sparse canopy than into a dense
et al., 1980). However, Corrsin (1974) has pointed out canopy (Shaw et al., 1988), the coefficient Cg in the
that the application of K-theory is limited to the place equation is defined to be a function of the integrated
where the length scales of flux-carrying motions have plant area density and is expressed by
to be much smaller than the scales associated with
 Z h 
average gradients. It is unfortunate that such a condi-
tion is often violated within vegetation canopies (Rau- Cg = C1 exp C2 Cd A dz (6)
z
pach and Thom, 1981; Baldocchi and Meyers, 1988b).
Many measurements have shown that wind flow within where A is the plant area density (m1 ), Cd the ef-
and just above a plant canopy is dominated by turbu- fective drag coefficient of the plant elements, and C1
lence with vertical length scales at least as large as the and C2 are constants that are determined by numerical
vegetation height (Kaimal and Finnigan, 1994). These experiments. The second term in Eq. (5) is an addi-
large-scale turbulent eddies are intermittent and ener- tional term that we have introduced. This term is sim-
getic, and they can penetrate the canopy crowns and ilar to the term that represents dispersive flux in the
enter the subcanopy trunk space to generate non-local model proposed by MLL, but instead of Cg defined in
turbulent transport. Most of the transport of momen- the present study, MLL used a constant coefficient. In
tum and scalar properties within the canopy are gen- the present study, we have not use the constant coef-
erated by these large-scale turbulent eddies (Raupach ficient proposed by MLL because we believe that the
et al., 1986; Baldocchi and Meyers, 1988a). Baldoc- momentum transported by large eddies is modified by
chi and Meyers (1988b) report that the Reynolds stress the structure of a canopy. Results of the numerical ex-
within a plant canopy is influenced not only by the periments revealed that wind velocities predicted by
product of KM and the local vertical gradient in the MLLs model are very sensitive to the selected values
mean wind velocity but also by the non-local turbulent of the coefficient, and we failed to find a universal
transport through the activities of large-scale eddies. constant that can produce correct predictions for the
Based on the results obtained from previous studies, wind velocities in canopies with different structures.
the turbulent momentum flux is divided into two parts Although there is little information about the disper-
in the present study: one diffused by the smaller-scale sive flux in a real vegetation canopy, a wind-tunnel
eddies, which depends on the local gradient of the study using a model canopy (Raupach et al., 1986) has
mean wind velocity; and the other transported by the shown that the amount of dispersive flux is very small
304 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

and negligible compared with the amount of turbu- by MLL, which is a function of z and the total leaf
lent flux. In addition, wind velocities within vegetation area in the portion below d and does not reflect the
canopies have been successfully predicted by WS and effect of local canopy structure.
MP, who used higher-order closure models in which
a term of dispersive flux is not included. Therefore, 2.2.3. Drag force
dispersive flux is considered to be negligible and was The drag of plant elements fx is parameterized ac-
omitted in the present study. cording to WS

2.2.2. Mixing length fx = Cd A hui2 (10)


The eddy viscosity K is parameterized according to
the Prandtlvon Krmn mixing-length theory This parameterization scheme has been widely used
in studies on wind flow within a vegetation canopy

2 d hui (e.g. MP; WK; Wang and Takle, 1996). The effective
K=l (7)
dz drag coefficient (Cd ) of a single leaf measured in wind
where l is the mixing length. Above the vegetation tunnel changes with the leaf orientation and turbu-
canopy surface, the mixing length is expressed as lent scales and intensity around the leaf (Raupach and
Thom, 1981). However, MP reported that the value of
l = (z d), zh (8) Cd of a vegetation canopy is a constant, i.e. it does
not depend on wind speed and the position within the
where is the von Krmns constant equal to 0.4,
canopy.
and d is the zero-plane displacement (m). The mix-
ing length within a canopy is complicated due to the
effects of canopy elements. In conventional K-theory
models, Inoue (1963) suggested that l is constant 3. Numerical aspects
throughout the canopy layer, while Seginer (1974),
Kondo and Akashi (1976) and WK considered l to The selected reference height for the reference wind
be constrained by the ground surface and the local speed hur i is twice the vegetation height. This selec-
internal structure of the canopy and defined it as a tion is based on measurements showing that there is
function of the local plant area density, drag coeffi- a roughness sublayer within which the eddy viscos-
cient and height. As the conventional K-theory model ity is found to be enhanced and the semi-logarithmic
is used in the present model to represent only the law is not obeyed over a rough surface (e.g. Garrat,
small-eddy diffusion within the canopy, the definition 1978; Raupach et al., 1980; Simpson et al., 1998).
of mixing length within the canopy in the present Three new parameters are introduced in the present
model is different from that in conventional K-theory model: C1 and C2 in Eq. (6) and Cl in Eq. (9). The
models. However, the effects of ground surface and values of these three parameters were firstly fitted to
canopy structure pointed out by Seginer (1974) and be 0.01, 1.0 and 5.0, respectively, by using the corn
other researchers will be qualitatively the same. In canopy data, but C2 was adjusted to 2.0 in the simula-
addition, it is considered that the value of l within tions for the rubber canopy data. Variations in C2 only
the canopy can not be larger than that at the canopy affect the modeled wind profile at the lower canopy.
top lh (= (h d) according to Eq. (8)). Thus, the The final values of 0.01, 2.0 and 5.0 for C1 , C2 and
mixing length within the canopy is parameterized as Cl , respectively, were used for all the canopies in the
z present study. The computation grid included 60 equal
l= (9) intervals from the ground surface to three times the
1 + Cl Cd Az
vegetation height. All variables in the model were
with non-dimensionalized by scales of h and the wind speed
at 3h height. Non-dimensional wind speed was given
l (h d), z<h
by 1 at the upper boundary and 0 at zg /h=0.001, where
where Cl is a constant determined by numerical ex- zg is the roughness length of the ground surface. Ex-
periments. This model is different from that proposed perimental results showed that the solution was not
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 305

sensitive to the chosen of the value zg . The initial wind


profile was assumed, and the solution was found iter-
atively until the differences in the computed hui were
less than the control level (104 in this study).

4. Results and discussion

4.1. Wind profiles in various types of vegetation


canopy

Wind speeds were predicted by the present model


within and above six types of vegetation canopy, which
included agricultural crops, and deciduous and conif-
erous forests with leaf area indices (LAI) from about
two to near seven and vegetation heights from about
1 m to more than 20 m (Table 1). The predicted and
measured wind-speed profiles ranging from ground
surface to twice the canopy height for three canopies
and to three times the canopy height for the other three
canopies, are shown together with the distributions of
plant area density (PAD) in Fig. 1. The measured wind
profiles in the pine and oakhickory forests were av-
erages. The leaf area density of the pine forest was
the average of pine A and pine B in Amiro (1990a),
and the tree height of 18.5 m was used in the present
study, which was based on the reported tree height
of 1520 m for the pine forest. Though the measured
wind profiles within and above the canopies show large
differences according to the species and structure of
the canopies, the predicted wind profiles matched them
excellently. The measured wind profiles in most of Fig. 1. Comparison of predicted (solid lines) and measured (circles)
the canopies show small gradients or slight reversals wind speeds within and above six types of vegetation canopy.
Vertical distributions of plant area density (PAD, dashed lines) are
also shown. Wind profiles ranging to 2h (h is vegetation height) are
Table 1 shown for corn, pine and oakhickory canopies, and those ranging
List of vegetation canopies and the references. Oakhickory and to 3h are shown for wheat, bean and aspenmaple canopies. u is
aspenmaple are mixed deciduous forests with principal species the friction velocity above the canopy.
of oak and hickory, and aspen and maple, respectivelya
Vegetation h (m) LAI References in gradient in the lower portion of the canopies, and
canopies
an obvious secondary maximum can be seen in the
Corn 2.8 3 Wilson and Shaw (1977) oakhickory canopy. K-theory models are incapable
Pine 18.5 2.3 Amiro (1990a) of predicting these features. The present model suc-
Oakhickory 23 4.9 Meyers and Baldocchi (1991)
Bean 1.18 6.3 Thom (1971)
cessfully demonstrated these features, revealing that
Wheat 1.25 6.6 Legg (1975a, b) the non-local transport was accurately considered in
Aspenmaple 18 5 Neumann et al., (1989); the model. It was revealed by the present model that
Gao et al. (1989) though the small-eddy diffusion Rs was near zero or
a h is the vegetation height, and LAI is the leaf area index. even upwardly transported the momentum when the
306 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

Table 2
gradient of mean wind was reversed, the non-local
Mean errors in predicted wind speeds (non-dimensionalized by
transport Rl was large and maintained the Reynolds the friction velocity above the canopy)
stress in the lower canopies (see also Section 4.4).
Vegetation Root-mean- Mean absolute Mean
Shaw (1977) has pointed out that the mean wind gra-
canopies square errors errors errors
dient will be reversed if the non-local turbulent trans-
port momentum is large enough in the lower portion Corn 0.203 0.168 0.074
Pine 0.126 0.114 0.020
of a vegetation canopy, i.e.
Oakhickory 0.205 0.171 0.106
  Bean 0.163 0.138 0.010
w 0 u0 w 0  0 
u w0 Wheat 0.135 0.098 0.039
> p0 + Aspenmaple 0.233 0.168 0.026
z z x
He also suggested that in the region where there is a re-
versal in mean wind gradient, larger scales of turbulent displacement (d) is needed. This value is estimated by
motion transport momentum downward, while smaller the model based on an additional input value of mea-
scales transport momentum upward according to the sured wind speed above the canopy. Experimental re-
local gradient. Very strong wind shear appears near the sults showed that the predicted wind profile within the
canopy top in two deciduous canopies (oakhickory canopy was not sensitive to the choice value of d if d
and aspenmaple canopies) because there are very is lower than 0.85 h.
dense foliage layers at the upper portion of these two Predicted results according to higher-closure mod-
canopies. The foliage is also very dense at the upper els (e.g. MP; Shaw and Seginer, 1987; Meyers and
portion of the bean canopy, but the wind shear is not Baldocchi, 1991) often show larger errors near the
as large as that in the deciduous canopies, because canopy top, where strong shear in the flow field
the drag coefficient of the bean canopy is very small occurs. The present model does not have this prob-
(Table 3). The wind speed within the wheat canopy, lem. Table 2 shows the mean, mean absolute and
which has the greatest foliage density, is very low. The root-mean-square errors in predicted results calcu-
predicted wind profile within the pine canopy is also lated by comparing predicted and measured values.
very similar to those measured in other pine forests The largest root-mean-square error is 0.23 (for the
that have similar distributions of foliage densities (e.g. oakhickory canopy), and other root-mean-square
Allen, 1968; Halldin and Lindroth, 1986). errors are 0.2 or less. The mean absolute errors are
Shaw and Pereira (1982) reported that the wind pro- smaller than 0.2, and the mean errors are almost zero
file predicted by their second-order closure model was for all canopies. MP compared their predicted results
not logarithmic immediately above the canopy sur- by using a higher-order closure model with measured
face. However, many researchers have observed loga- data for six canopies (in which corn and bean canopies
rithmic or near-logarithmic wind profiles above many were the same as those used in the present study).
vegetation canopies (e.g. Thom, 1971; Oliver, 1971; Their root-mean-square errors were larger than 0.2 for
Dolman, 1986). Fig. 1 shows that the predicted and most canopies, and the largest error was 0.54; these
measured wind profiles above the bean canopy are al- values were larger than those in the present study.
most the same. The measured profiles were the profiles In the present study, triangular distributions for the
of DF in Fig. 7 of the report by Thom (1971), which plant area density were used to approximate the ob-
were reported to be logarithmic above the canopy. served distributions for all of the canopies except the
The wind profile above a canopy computed by the aspenmaple canopy. This technique has been used by
present model is approximately logarithmic because Pereira and Shaw (1980), who reported that the wind
the non-local transfer (Rl ) is small above a canopy (see profile for a corn canopy (the same corn canopy as that
also Section 4.4). The roughness sublayer over a plant used in the present study) calculated by the triangular
canopy is believed to be shallow (e.g. Simpson et al., distribution was virtually indistinguishable from that
1998) except in cases where the canopy is very sparse calculated using the observed distribution. The data
(e.g. Garrat, 1978). In order to predict the wind profile of observed plant area density of a vegetation canopy
above the canopy, a correct value for the zero-plane generally shows great scattering. The good agreement
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 307

Table 3
0.15, respectively; published values of Cd for these
Comparison of drag coefficients (Cd ) yielded by the present model
and those estimated in other studies two canopies are not available for comparison. The re-
sults of the present study also support the claim made
Vegetation Present Other References
canopies study studies by MP that the use of a constant effective drag coef-
ficient is appropriate for a plant canopy.
Corn 0.20 0.20 Wilson and Shaw (1977)
Pine 0.20 0.10.25 Amiro (1990b)
Oakhickory 0.18 0.15 Lee et al. (1994) 4.2. Influence of foliage density on the wind profile
Bean 0.04 0.030.04 Thom (1971)
Influence of foliage density on the wind profile was
investigated using the data measured in a rubber plan-
between the modeled results and the measured data in tation located in the Hainan Island, China. The rub-
the present study suggests that this technique is use- ber trees were about 12 m in height and the foliage
ful for numerical studies of wind flow within vege- density were estimated from the leaf-fall collections
tation canopies. A beta distribution, which was used and the measured solar radiations. Detailed informa-
by Meyers and Baldocchi (1991), was used for the tion of the instrumentation and site has been described
aspenmaple canopy in the present study. by Takahashi et al. (1986) and Yoshino et al. (1988).
As was the case in the study by MP, a constant Fig. 2 shows the simulated (using the present model)
effective drag coefficient (Cd ) for each canopy was and measured wind profiles in the rubber tree plan-
used in the present study. The effective drag coeffi- tation during fully leafed, partially leafed and leaf-
cient for a vegetation canopy is difficult to measure less periods. The measured profiles were averages in
due to problems of the shelter effect (Thom, 1971) and near-neutral conditions and when the wind direction
leaf orientation. In numerical studies, it is usually de- was ESE, in which the effect of the surrounding wind-
termined by trial-and-error to produce the best agree- breaks was smallest and the fetch was the longest. The
ment with observations. The effective drag coefficient plant area indices (PAIs) during the three periods were
for each canopy in the present study was determined about 5, 3 and 1, respectively.
by trial-and-error. Table 3 shows a comparison of the Again, the simulated results agreed with the mea-
values of Cd yielded by the present model and those sured data excellently; the root-mean-square errors
estimated in other studies for the same canopies. It were less than 0.2, and the mean errors were almost
is shown that those yielded by the present model are zero for the canopy during the three periods. As can
almost the same as or within the range of those es- be seen in the figure, the wind profiles change little
timated in other studies. The values of Cd used for in form during the three periods, because there was
the wheat and aspenmaple canopies were 0.12 and little change in the distribution patterns of the PAD

Fig. 2. Simulated (solid lines) and measured (circles) wind profiles in the rubber tree plantation during (a) fully leafed, (b) partially leafed
and (c) leafless periods. Profiles of PAD are shown by dashed lines.
308 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

of the rubber trees during the three periods. How- 4.3. Bulk momentum transfer coefficients
ever, the normalized wind speed (u/u ) within the
canopy increases as the foliage density decreases, It can be seen in Figs. 1 and 2 that values of the
because it is easier for momentum to penetrate the normalized velocities u/u (u is the friction velocity
canopy from above the canopy if the canopy is thin. above the canopy) at the top of all the canopy tops are
Although the PAI decreased about five-fold from the very similar, but changes of them according to the type
fully leafed to leafless periods, the normalized wind of canopy are discernible. The same phenomenon was
speed within the leafless canopy, except that at the also shown by MP. If the normalized velocity above
very upper portion of the canopy, increased less than the canopy is inverted and squared, it becomes the bulk
two-fold. This discrepancy can be explained by the momentum transfer coefficient CM , which is defined
effect of the effective drag coefficient. Seginer et al. as
(1976) has shown, from the results of wind-tunnel
experiments, that the effective drag coefficient of a = u2 = CM u2 (11)

dense canopy is smaller than that of a thin canopy.
The effective drag coefficients used by the model for where is the shear stress, the air density, u the fric-
the fully leafed and leafless canopies were 0.15 and tion velocity, and u is the mean wind speed at a height
0.36, respectively, showing that the effective drag co- above the vegetation. CM is also called drag coeffi-
efficient of tree branches in a fully leafed canopy is cient by some researchers. Dolman (1986) showed that
smaller than that in a leafless canopy. The same fact CM above an oak forest was about two-times larger
was also found by Sato (personal communication) in on foliated conditions than that on defoliated condi-
his studies on wind flow through windbreaks. Each tions. On the other hand, WK showed that CM over a
of the three wind profiles showed a weak secondary rice canopy was scattered in a wide range over peri-
wind maximum in the lower portion of the canopy ods with differing leaf area indices. It is probable that
because the trunk space of the rubber plantation was the variations in the CM above a vegetation canopy do
very open. The wind profiles are near logarithmic not depend only on the foliage density of the canopy.
above the three canopies, and the slope of the wind Fig. 3a shows the variations in the CM at the canopy
profile during fully leafed period is larger than that top versus Cd PAI, where PAI is the plant area index,
during leafless period. These findings are consistent for the nine canopies, which used in the above two sec-
with those observed above a Japanese larch forest tions, according to the modeled results. Although the
(Allen, 1968) and an oak forest (Dolman, 1986). values of CM are scattered at large values of Cd PAI,

Fig. 3. Variation in the bulk momentum transfer coefficients at the canopy top (CMh ) vs. (a) CD (=Cd PAI) and (b) CF (=Cd PAI zmax /h)
for nine vegetation canopies.
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 309

there is a clear tendency for CM to increase as Cd PAI ments using a higher-order closure model. From Eq.
increases. Furthermore, when the variations in CM are (15), we can get
plotted against Cd PAI zmax /h, where zmax is the height
1
at which the plant area density is maximum, a good z0 = (h d) (17)
correlation between them is revealed (Fig. 3b). Shaw exp((uh /u ))
and Pereira (1982) have shown that zmax is an impor- where uh is the mean wind speed at the canopy top.
tant factor in determining the aerodynamic roughness From Eqs. (11), (16) and (17) we find that
of a plant canopy. The value of Cd PAI zmax /h is small-
est (0.146) in the case of the leafless rubber planta- 1 1
= = (18)
tion canopy. It seems that Cd PAI zmax /h can accurately exp((uh /u )) exp( CMh )
represent the momentum absorption ability of a vege-
Thus, is a function of CMh , also changes with CF
tation canopy. Let
(according to Eq. (14)), and is not a constant when
CD Cd PAI, (12) CF >0.2. It was difficult to obtain a simple expression
for the relationship between and CF from Eqs. (14)
zmax
CF CD (13) and (18). However, the results of least-squares analysis
h for the values of and CF for the nine canopies showed
and CMh represent the bulk momentum transfer coef- that
ficient at the canopy top, then the fitting-curve for the
= 0.209 exp(0.414CF ) (19)
nine canopies in Fig. 3b is expressed as
with the correlation coefficient equaling 0.892. The
CMh = 0.0618 exp(0.792CF ) (14) values of for the nine canopies ranged from 0.22 (the
and the correlation coefficient was equal to 0.889. Us- bean canopy) to 0.32 (the oakhickory canopy), and
ing a K-theory model, WK predicted that CM increased the average for all of the canopies was 0.26, which is
with an increase in CD when CD <0.3, but it decreased the same value as that obtained by Moore (1974) and
with an increase in CD when CD increased further for Shaw and Pereira (1982). For comparison, Bruin and
a rice canopy. However, their prediction was not sup- Moore (1985) reported =0.22 for a pine forest (the
ported well by the measured data. Thetford Forest).
If the logarithmic law is obeyed above a canopy
surface, the wind profile above the canopy surface is 4.4. Reynolds stress
expressed as
  The computed Reynolds stress for the corn canopy
u zd showed that more than 80% of the momentum was ab-
u= ln (15)
z0 sorbed in the upper half of the canopy (Fig. 4), which
was the same as that shown by the measured data. The
where z0 is the roughness of the canopy surface. Thom
computed Reynolds stress decreased to almost zero
(1971) suggested that
near the ground. Though there was no measured data
z0 = (h d) (16) for comparison in the lower canopy, the modeled re-
sults were almost the same as those showed by WS,
where h is the canopy height, and the value of was who computed using a higher-order closure model.
estimated to be 0.36 for an artificial crop. Seginer Fig. 5 shows the computed profiles of two com-
(1974) estimated the value of to be 0.37 on the basis ponents of the Reynolds stress, Rl and Rs , for the
of the canopy wind model of Inoue (1963) and an ob- corn canopy. Though Rs is larger than Rl above the
servation by Kondo (1971). However, the value of canopy and in about the upper 70% of the canopy,
was estimated by Moore (1974) to be 0.26 according it decreases rapidly with height within the canopy,
to 105 published d, z0 and h data, and it was also es- and it is almost zero in the lower 40% canopy; the
timated by Shaw and Pereira (1982) to be 0.26 (when Reynolds stress in the lower portion of the canopy
CD >0.2) on the basis of results of numerical experi- is maintained by the non-local transport momentum
310 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

(non-local transport of the Reynolds stress) in many


canopies (e.g. Shaw and Seginer, 1987; Baldocchi and
Meyers, 1988a, b; Amiro, 1990a). Above the canopy,
Rs is more than three times larger than Rl and is the
main source of the Reynolds stress, implying that the
predicted mean wind profile above the canopy is ap-
proximately logarithmic. For the layer above 2h, the
non-local transfer momentum is parameterized to be
zero in the model. Though only the profiles for the
corn canopy are shown, those for the other canopies
are qualitatively the same.

Fig. 4. Profile of the computed Reynolds stress (line) and the 5. Conclusions
measured data (closed circles) for the corn canopy.

Taking into account the non-local turbulent trans-


port, we have developed a first-order closure model
Rl . Meyers and Baldocchi (1991) have pointed out
for predicting the wind flow within and above vege-
that the shear production, which is generated by the
tation canopies. The high accuracy and the universal
interaction between the turbulent field and the mean
utility of the model were verified by comparisons of
velocity gradient, is small and turbulence imported
modeled results and measured data in six types of
from above the canopy is a strong source for the
vegetation canopy and a rubber tree plantation dur-
turbulent kinetic energy in the lower canopy. Thus,
ing fully leafed, partially leafed and leafless periods.
the present model can account for the phenomena of
The root-mean-square errors in the predicted wind
counter-gradient momentum transport and secondary
speeds were about 0.2 or less for all of the canopies;
wind maxima that occurs in the lower portions of veg-
these errors are smaller than those results from a
etation canopies. Rl peaks at about 0.8h and decreases
higher-order closure model. The wind speeds in the
above and below this height, and this distribution
lower canopies, which appeared to be almost constant
pattern is similar to those of the measured hw 0 u0 w 0 i
or reversed in gradient, were also correctly predicted.
In addition to its high accuracy and universal utility,
the present model costs little computation time due to
its simplicity. The model would be very useful for the
applications for predicting vegetation wind flows or
scalar fluxes (e.g. heat and water vapor) between the
atmosphere and vegetated surfaces when it is coupled
with other models.
A simulation study on the influence of foliage den-
sity on the wind profiles within and above a vegetation
canopy was performed for a rubber tree plantation
during fully leafed, partially leafed and leafless peri-
ods. The simulated wind profiles within the canopy
changed little in form during the three periods but
that the normalized wind speed (normalized by the
friction velocity above the canopy) within the canopy
Fig. 5. Profiles of two components of the Reynolds stress in the increased as the foliage density decreased. The slope
corn canopy: small-eddy diffusion Rs (solid line) and non-local of the wind profile above the fully leafed canopy was
transfer Rl (dash line). larger than that above the leafless canopy.
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 311

Based on the modeled wind speeds using the present p kinematic pressure (Pa)
model, we were able to clarify the effects of canopy PAD plant area density (m2 m3 )
density, structure and effective drag coefficient on the PAI plant area index (m2 m2 )
bulk momentum transfer coefficient and the coefficient Rl non-local momentum transport (m2 s2 )
, and we were also able to determine the correlations Rs small-eddy diffusion (m2 s2 )
between CM and CF and between and CF . t time (s)
The Reynolds stress in the present model was T time average interval (s)
parameterized by two terms: one representing the u streamwise velocity (m s1 )
small-eddy diffusion, and one representing non-local ui wind velocity vector (m s1 )
transport through large-scale turbulent eddies. The ur wind velocity at a reference height
modeled results showed that the non-local transfer (m s1 )
component was small above the canopy but large, and u friction velocity above vegetation
the main source of the Reynolds stress, in the lower canopy (m s1 )
portion of the canopy. The model can also account for hu0 w0 i Reynolds stress (m2 s2 )
counter-gradient momentum transport occurs in the V volume (m3 )
lower portion of a vegetation canopy. A parameteri- w vertical velocity (m s1 )
zation scheme was developed for the mixing length xi position vector (m)
within a canopy. z height (m)
z0 roughness of vegetation surface (m)
zmax the height at which the plant area
6. Nomenclature
density is maximum (m)
A plant area density (m2 m3 )
Greek letters
C1 , C2 , Cl constants equal to 0.01, 2 and 5,
kinematic viscosity of the air (m2 s1 )
respectively
von Karmans constant (equal to 0.4)
CD a coefficient equal to Cd PAI
the coefficient between z0 and
Cd canopy effective drag coefficient
air density (kg m3 )
CF a coefficient defined as Cd PAI zmax /h
shear stress (m2 s2 )
Cg coefficient of non-local momentum
ij kinematic momentum flux (m2 s2 )
transport
a scalar or vector variable
CM bulk momentum transfer coefficient
CMh bulk momentum transfer coefficient at
Other symbols
the canopy top
overbar ( ) time average
d zero plane displacement height (m)
prime ( 0 ) deviation from time average
fx total (form and viscous) streamwise drag
angle brackets h i volume average
force (N)
double prime ( 00 ) deviation from volume average
fF form drag force (N)
fV viscous drag force (N)
h mean vegetation height (m)
i, j index notations, with values of 1, 2 or 3 Acknowledgements
K momentum eddy viscosity (defined in
present study) (m2 s1 ) Micrometeorological measurements in a rubber
KM conventional momentum eddy viscosity plantation were conducted as a part of the project
(m2 s1 ) A Study on the Geoecological Factors of Land Use
l mixing length (m) and Vegetation in the Tropics and Subtropics in the
LAI leaf area index (m2 m2 ) Peoples Republic of China, which was supported by
lh mixing-length at vegetation canopy grants-in-aid from the Japanese Ministry of Educa-
top (m) tion, Science and Culture (leader: Masatoshi Yoshino,
312 P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313

Nos. 61041012 and 62043011). We are grateful to Legg, B.J., 1975a. Turbulent diffusion within a wheat canopy: I.
Prof. T. Maitani (Okayama University), Mr. T. Sato Measurement using nitrous oxide. Quart. J. R. Meteorol. Soc.
(Hokkaido Head Office, Japan Weather Association) 101, 591610.
Legg, B.J., 1975b. Turbulent diffusion within a wheat canopy:
and Mr. Natsume (Hokkaido Institute of Environmen- II. Results and interpretation. Quart. J. R. Meteorol. Soc. 101,
tal Science) for their helpful comments. Mr. Natsume 611628.
also checked our computing programs. Li, Z.J., Miller, R., Lin, J.D., 1985. A first-order closure scheme
to describe counter-gradient momentum transport in plant
canopies. Boundary-Layer Meteorol. 33, 7783.
References Meyers, T.P., Baldocchi, D.D., 1991. The budgets of turbulent
kinetic energy and Reynolds stress within and above a deciduous
forest. Agric. For. Meteorol. 53, 207222.
Allen, L.H. Jr., 1968. Turbulence and wind speed spectra within
a Japanese larch plantation. J. Appl. Meteorol. 7, 7378. Meyers, T.P., Paw, U.K.T., 1986. Testing of a higher-order
Amiro, B.D., 1990a. Comparison of turbulence statistics within closure model for airflow within and above plant canopies.
three boreal forest canopies. Boundary-Layer Meteorol. 51, Boundary-Layer Meteorol. 37, 297311.
99120. Miller, D.R., Lin, J.D., Lu, Z.N., 1991. Air flow across an alpine
Amiro, B.D., 1990b. Drag coefficients and turbulence spectra forest clearing: a model and field measurements. Agric. For.
within three boreal forest canopies. Boundary-Layer Meteorol. Meteorol. 56, 209225.
52, 227246. Moore, C.J., 1974. A comparative study of forest and grassland
Baldocchi, D.D., Meyers, T.P., 1988a. Turbulence structure in a micrometeorology. Ph.D. Thesis, The Flinders, University of
deciduous forest. Boundary-Layer Meteorol. 43, 345364. South Australia, p. 237.
Baldocchi, D.D., Meyers, T.P., 1988b. A spectral and Neumann, H.H., Den Hartog, G., Shaw, R.H., 1989. Leaf area
lag-correlation analysis of turbulence in a deciduous forest measurements based on hemispheric photographs and leaf-letter
canopy. Boundary-Layer Meteorol. 45, 3158. collection in a deciduous forest during autumn leaf-fall. Agric.
Bruin, H.A.R., Moore, C.J., 1985. Zero-plane displacement and For. Meteorol. 45, 325345.
roughness length for tall vegetation, derived from a simple mass Oliver, H.R., 1971. Wind profiles in and above a forest canopy.
conservation hypothesis. Boundary-Layer Meteorol. 31, 3949. Quart. J. R. Meteorol. Soc. 97, 548553.
Corrsin, S., 1974. Limitations of gradient transport model in Pereira, A.R., Shaw, R.H., 1980. A numerical experiment on
random walks and in turbulence. Adv. Geophys. 18A, 2560. the mean wind structure inside canopies of vegetation. Agric.
Cowan, I.R., 1968. Mass, heat and momentum exchange between Meteorol. 22, 303318.
stands of plants and their atmospheric environment. Quart. J. Raupach, M.R., Thom, A.S., 1981. Turbulence in and above plant
R. Meteorol. Soc. 94, 523544. canopies. Ann. Rev. Fluid Mech. 13, 97129.
Dolman, A.J., 1986. Estimates of roughness length and zero plane Raupach, M.R., Coppin, P.A., Legg, B.J., 1986. Experiments
displacement for a foliated and non-foliated oak canopy. Agric. on scalar dispersion within a model plant canopy: Part I.
For. Meteorol. 36, 241248. The turbulence structure. Boundary-Layer Meteorol. 35, 21
Gao, W., Shaw, R.H., Paw, U.K.T., 1989. Observation of organized 52.
structure in turbulent flow within and above a forest canopy. Raupach, M.R., Thom, A.S., Edwards, I., 1980. A wind-tunnel
Boundary-Layer Meteorol. 47, 349377. study of turbulent flow close to regularly arrayed rough surfaces.
Garrat, J.R., 1978. Flux-profile relations above tall vegetation. Boundary-Layer Meteorol. 18, 373397.
Quart. J. R. Meteorol. Soc. 104, 199211. Seginer, J., 1974. Aerodynamic roughness of vegetated surfaces.
Halldin, S., Lindroth, A., 1986. Pine forest microclimate simulation Boundary-Layer Meteorol. 5, 383393.
using different diffusivities. Boundary-Layer Meteorol. 35, Seginer, J., Mulhearn, P.J., Bradley, E.R., Finnigan, J.J., 1976.
103123. Turbulent flow in a model plant canopy. Boundary-Layer
Inoue, E., 1963. On the turbulent structure of air flow within crop Meteorol. 10, 423453.
canopies. J. Meteorol. Soc. Jpn. 41, 317325. Shaw, R.H., 1977. Secondary wind speed maxima inside plant
Kaimal, J.C., Finnigan, J.J., 1994. Atmospheric Boundary Layer canopies. J. Appl. Meteorol. 16, 514521.
Flows: Their Structure and Measurement. Oxford University Shaw, R.H., Pereira, A.R., 1982. Aerodynamic roughness of a plant
Press, New York, p. 289. canopy: a numerical experiment. Agric. Meteorol. 26, 5165.
Kondo, J., 1971. Relationship between the roughness coefficient Shaw, R.H., Seginer, I., 1987. Calculation of velocity skewness
and other aerodynamic parameters. J. Meteorol. Soc. Jpn. 49, in real and artificial plant canopies. Boundary-Layer Meteorol.
121124. 39, 315332.
Kondo, J., Akashi, S., 1976. Numerical studies on the Shaw, R.H., Den Hartog, G., Neumann, H.H., 1988. Influence of
two-dimensional flow in horizontally homogeneous canopy foliar density and thermal stability on profiles of Reynolds stress
layers. Boundary-Layer Meteorol. 10, 255272. and turbulence intensity in a deciduous forest. Boundary-Layer
Lee, X., Shaw, R.H., Black, T.A., 1994. Modelling the effect of Meteorol. 45, 391409.
mean pressure gradient on the mean flow within forests. Agric. Simpson, I.J., Thurtell, G.W., Neumann, H.H., Den Hartog, G.,
For. Meteorol. 68, 201212. Edwards, G.C., 1998. The validity of similarity theory in the
P. Zeng, H. Takahashi / Agricultural and Forest Meteorology 103 (2000) 301313 313

roughness sublayer above forests. Boundary-Layer Meteorol. Wang, H., Takle, E.S., 1996. On three-dimensionality of shelterbelt
87, 6999. structure and its influences on shelter effects. Boundary-Layer
Stull, B.S., 1988. An Introduction to Boundary Layer Meteorology. Meteorol. 79, 83108.
Kluwer Academic Publishers, Boston, p. 666. Watanabe, T., Kondo, J., 1990. The influence of canopy structure
Takahashi, H., Makita, H., Nakagawa, K., Hayashi, Y., Hao, Y., and density upon the mixing length within and above vegetation.
Chin, L., Wang, P., Wei, F., Yao, M., 1986. Micrometeorology J. Meteorol. Soc. Jpn. 68, 227235.
in a rubber plantation during cold wave weather conditions. Wilson, J.D., 1988. A secondary-order closure model for flow
In: Yoshino, M. (Ed.), Climate, Geoecology and Agriculture through vegetation. Boundary-Layer Meteorol. 42, 371392.
in South China, Vol. 1. Climatological notes, Institute Wilson, N.R., Shaw, R.H., 1977. A higher order closure model
of Geoscience, University of Tsukuba, Tsukuba, Japan, for canopy flow. J. Appl. Meteorol. 11971205.
pp. 3758. Yoshino, M., Jiang, A., Hao, Y., Takahashi, H., Nakagawa, K.,
Thom, A.S., 1971. Momentum absorption by vegetation. Quart. J. Yamakawa, S., Makita, H., Chujo, H., Xie, L., Wang, P., 1988.
R. Meteorol. Soc. 97, 414428. System of micrometeorological observation in and out of a
Van Pul, A., Van Boxel, J.H., 1990. Comment on a first-order rubber plantation in Hainan Island. In: Yoshino, M. (Ed.),
closure scheme to describe counter-gradient momentum Climate, Geoecology and Agriculture in South China, Vol. 2.
transport in plant canopies by Z.J. Li, D.R. Miller and J.D. Climatological notes, Institute of Geoscience, University of
Lin. Boundary-Layer Meteorol. 51, 313315. Tsukuba, Tsukuba, Japan, pp. 516.

You might also like