Chapter 1
Project overview
The present survey aims to investigate the 3-D velocity field of turbulent air flow above
andwithinvegetationcanopiesusinglaserDopplertechniqueinordertoanalysevelocity
statistics. This research project, which carefully follows the path previously traced by
Pietri et al. (2009) on the same subject, results from the synthesis made by Petroff
(2005) who studied atmospheric aerosol diffusion and deposit upon vegetative canopies.
Indeed, to prevent and verify the dispersion of nuclear waste, Petroff showed that some
deposit parameters depend on aerodynamical features of the flow passing through the
canopy. A specific literature about so-called canopy flows exhibited that canopy flows
are relatively well-known if canopies are dense and homogeneous but brought out the
lackofspecificinformationaboutthecharacterizationofturbulentflowspassingthrough
sparse canopy fields.
As a matter of fact, flow dynamics over vegetation canopies even influences many
other environmental aspects as the microclimate of plants and their surroundings.
Knowledge of the aerodynamic field throughout vegetation is fundamental in all cases
of air pollution, in the urban context too, since transport and collection of gaseous and
particulate matter depend on it. Moreover, there are some other involvements that may
reasonablybelinkedtotheinteractionbetweenturbulentflowsandvegetationcanopies,
as hydrology, pollution, meteorological forecasts or plant growth and maturation.
Consciousness of all these involvements has been developed more than forty years
ago and since then literature has reported several works on turbulent fluids passing
7
above and within either real canopies or wind tunnel canopy samples. We have to
consider that in the “canopy” definition even urban canopies, such as buildings, are
included.
1.1 Literature review
A first important step in order to understand canopy flows behaviour was made by
Raupach and Thom (1981) since they stated that turbulent flows in canopies are far
fromrandom. Theyunderstoodthatthesefluxesareratherdominatedbylargecoherent
structuresofwholeplantscale, resultingfrominstabilityassociatedwiththemeanshear
at the top of the canopy. Later on, Raupach et al. (1996) attributed that instability
to the Kelvin-Helmholtz instability. It has been shown how these organized motions
are responsible for a large fraction of mass, momentum and energy transport across
the canopy-atmosphere interface (e.g. Gao et al., 1989; Novak et al., 2000; Finnigan
and Shaw, 2000). Quadrant Hole analysis and other techniques brought to a widely
recognizedknowledgeofthemechanismsandthephenomenologythathidebehindthose
large coherent structures, at least in the simplest cases of steady flow conditions and
homogenous canopies. That statement has been achieved thanks to many experiments,
performed both in situ and in laboratory, by Cionco (1978), Green et al. (1995),
Raupach et al. (1996), Finnigan and Shaw (2000), Poggi et al. (2004) and several
others.
Another remarkable progress has been introduced by Raupach et al. (1996) since
they developed the concept previously anticipated by Brunet et al. (1994): the mix-
ing layer analogy. A mixing layer is by definition a turbulent shear layer that forms
when two flow streams of different velocities, initially separated by a splitting plate,
merge downstream of the trailing edge of the plate (Fig. 1.1). This analogy is based
on the idea that canopy flows generate coherent eddies and active turbulence in the
upper part of the canopies that make flow more similar to a plane mixing layer than a
boundary layer. The characteristic strong inflection point of the mean velocity profile
is the main cause of instability associated with the coherent structures in the resulting
8
fully-developed turbulent flow and itself establishes the turbulence length scale. There
are multiple properties, summarized in Table 1.1, that support the analogy between
canopy flows and mixing layers.
Figure 1.1: Mixing Layer
Table 1.1 reports in order: inflection of the streamwise velocity profilehui, normalized
standard deviations σ
u
/u
∗
and σ
w
/u
∗1
, correlation coefficient r
uw
=−
hu’v’i
σuσw
, Prandtl
number Pr
t
2
, integral length scale, skewness moments|Sk
u
|and|Sk
w
|, turbulent kinetic
energy budget; where z is the vertical axis, d is the displacement height of the logarith-
mic profile,δ is the mixing layer width, P
s
is the shear Production,ε is the dissipation,
T
t
is the turbulent transport and T
p
is the pressure transport.
Property Surface layer Mixing Layer Canopy
Inflection inhui No Yes Yes
σ
u
/u
∗
2.5÷3.0 1.8 1.8÷2.0
σ
w
/u
∗
1.2÷1.3 1.4 1.0÷1.2
r
uw
∼0.3 ∼0.4 ∼0.5
Pr
t
∼1.0 ∼0.5 ∼0.5
Integral length scale _(z-d) _ δ _(h-d)
|Sk
u
|,|Sk
w
| small o(1) o(1)
TKE budget P
s
∼
=ε large T
t
, T
p
large T
t
, T
p
Table 1.1: Comparison of statistical flow properties between surface layer, mixing layer
and canopy flow [from Finnigan, 2000]
Following investigations (e.g. Poggi et al., 2004; Dupont and Brunet, 2008; Pietri
1
σ, r and Sk will be presented in chapter 2; u
∗
is the friction velocity, see par. 5.2.3
2
Pr = ν/α, with ν kinematic viscosity and α thermal diffusivity
9
et al., 2009; Huang et al., 2009) showed how the validity of this analogy is reliant on
canopy density. Of particular importance are recent studies of Pietri et al. (2009)
and Huang et al. (2009), the only ones that even considered sparse and very sparse
cases. In 2000, Finnigan stated that when plant elements become sparser the mixing
layer analogy ceases to be valid. In that condition, canopy flow is more likely seen as a
rough-wall boundary layer, reflecting that vegetation turbulence undergoes transition
from mixing-layer-like to boundary-layer-like. On this basis Pietri et al. developed a
study to understand this particular transition. Traces of this phenomenon have been
found in the evolution of the inflectional point in the mean longitudinal velocity profile,
in the sign of the skewness statistical moment and in the mixing length at the top of
the canopy.
In the same year, Huang et al., with a large-eddy simulation work on five wide
range vegetation densities, confirmed that survey. They outlined several effects on flow
statistics, a denser canopy tends to result in:
1. agreaterinflectionstrengthfortheverticalprofileofthemeanstreamwisevelocity
u/u
h
at the plant top;
3
2. the scaled shear stressτ/u
∗2
and the scaled velocity variances, σ
u
/u
∗
andσ
w
/u
∗
,
are damped more rapidly with increasing depth inside canopy;
4
3. anhigherefficiencyoftheverticalmomentumtransportacrossthecanopy–atmosphere
interface as the correlation coefficient|r
uw
| increases;
4. a greater amplitude of skewness (|Sk
u
|and|Sk
w
|) and kurtosis (|Kr
u
|and|Kr
w
|)
in the upper part of the canopy;
5. a smaller amplitude of the two-point integral length scale;
6. an higher scaled longitudinal convection velocity within the canopy.
Thanks to this knowledge Wang (2012) has recently developed a fine analytical
model for streamwise wind profiles into sparse canopies.
3
u
h
velocity at tree height h
4
τ and Kr will be presented in chapter 2
10
1.2 Historical references
Many past works contributed to the comprehension of canopy flows (e.g. Cionco, 1978;
Raupach and Thom 1981; Gao et al., 1989; Brunet et al., 1994; Green et al., 1995;
Finnigan, 2000; Finnigan and Shaw, 2000; Novak et al., 2000; Poggi et al., 2004, 2008,
2010; Zhu at al., 2006; Dupont and Brunet, 2008; Pietri et al., 2009), carried out either
in field or wind tunnel, but among all of them, only some characteristic survey has
been chosen as reference for our experimental results. Here we release a short synthesis
about them.
Green et al. in 1995 made measurements of the three-component velocity field
in a 8 meters high Scottish spruce forest, using six anemometers. Trees were widely
spaced at intervals of 8, 6 and 4 m and measures were taken between 0.25h and 1.25h,
where h was the height of trees. They understood how influential was the foliage
density on their results in terms of turbulence statistics. Turbulence intensity, skewness
and kurtosis factors increased with tree density and depth into canopy. Tree density
really influenced their horizontal spatial variation in statistics. Tree spacing had a
major role in momentum penetration and turbulent transport. They found traces of
the coherent large-scale turbulent motions, that they said to be intermittent. Through
a conditional sampling of the shear stress u
0
w
0
they saw that turbulence motions in a
sparser vegetation were less dominated by extreme and occasional events than had been
found before in denser one. After a careful analysis of their data, we must say that
all our comparison with their work should take into account the fact that they have
overestimated the quantity u
02
.
Novak et al. in 2000 made a similar survey in Canada, but they enriched it with a
wind tunnel experience; both in field and in wind tunnel they measured the 3-D velocity
and analysed their results with turbulence statistics. The open air experience was made
in a Sitka spruce forest, similar to Green’s one, with an hot wire anemometer and three
different tree densities. In the wind tunnel experiment they considered four densities
(four Leaf Area Index, cf. paragraph 3.2.1) of a model forest 15 cm high. In this way
they were able to make an important match analysis from the two studies, showing
11
that statistics of both of them was basically in agreement. As already known, they
recognized the dominating role of the turbulent large-scale structures within the canopy
and reported variations of many parameters with density variation. In this widespread
work they even studied the drag coefficient C
d
and the power spectra of the three
velocity components. They concluded that, as density increased: skewness, kurtosis
and C
d
increased; and the longitudinal mean velocity, Reynolds stresses u
0
w
0
/u
∗2
, the
three standard deviations, Sk
v
, the integral length scales decreased.
Figure 1.2: Statistical data from Novak et al., 2000. WTA, WTB, WTC, WTD (Wind
Tunnel)andGGH4, GGH6, GGH8(openfield)representarrangementsfromthedensest
to the sparsest
Poggi et al. in 2004, made measurements with a Laser Doppler technique in order
12