You are on page 1of 24

Boundary-Layer Meteorol

DOI 10.1007/s10546-017-0244-3

RESEARCH ARTICLE

Impact of Eddy Characteristics on Turbulent Heat and


Momentum Fluxes in the Urban Roughness Sublayer

Jun Zou1,2 · Bowen Zhou1,3 · Jianning Sun1,2

Received: 8 April 2016 / Accepted: 21 February 2017


© Springer Science+Business Media Dordrecht 2017

Abstract Eddy-covariance observations above the densely built-up Centre of Nanjing were
made from December 2011 to August 2012. Separate eddy-covariance systems installed at
two levels on a 36-m tower located on a rooftop were operated simultaneously, and observa-
tions grouped into two sectors (A, B) according to the prevalent wind directions. For sector
A, where the nearby buildings are all below the lower measurement level, the sensible heat
and momentum fluxes are generally greater at the upper level. For sector B, where several
high-rise buildings are located upwind, the sensible heat and momentum fluxes at the upper
level are close to those at the lower level. The analysis shows that the turbulent eddy char-
acteristics differ between the two wind sectors, leading to a different behaviour of turbulent
exchange between the two levels. A hypothesis is proposed that addresses the vertical vari-
ation of turbulent fluxes in the urban roughness sublayer (RSL). For sector A, the buildings
block the flow, change the trajectory of scalars, and distort the footprint of scalar fluxes;
this ‘blocking effect’ is believed to lead to a smaller sensible heat flux above the canopy
layer. Such an effect should decrease with height in the RSL, explaining the increase of the
observed turbulent heat flux with height. In addition, the presence of non-uniform building
heights adversely affects turbulence organization around the canopy top, and likely elevates
the inflection point of the mean flow to a higher elevation close to the upper measurement
level, where larger shear results in a larger momentum flux. For sector B, wake effects from
the nearby high-rise buildings strongly reduce turbulence organization at higher elevations,
leading to similar sensible heat and momentum fluxes at both measurement levels.

Keywords Eddy characteristics · Roughness sublayer · Turbulent flux · Urban canopy

B Jianning Sun
jnsun@nju.edu.cn
1 School of Atmospheric Sciences, Nanjing University, Nanjing 210023, China
2 Joint International Research Laboratory of Atmospheric and Earth System Sciences, Ministry of
Education, Nanjing 210023, China
3 Key Laboratory of Mesoscale Severe Weather, Ministry of Education, Nanjing 210023, China

123
J. Zou et al.

1 Introduction

The urban surface layer is typically complex in nature, with different combinations of heights
and densities of the building roughness elements, leading to different flow regimes (Grim-
mond and Oke 1999). Generally, the urban surface layer is divided into an inertial sublayer
and an underlying roughness sublayer (RSL), with the lowest part of the RSL from the ground
up to the mean building height referred to as the urban canopy layer (Roth 2000; Rotach 2001;
Grimmond 2006). The depth of the RSL is two to five times the urban canopy height (Rau-
pach et al. 1991; Roth 2000; Zou et al. 2015). In the urban RSL, time-averaged turbulence
statistics and flux densities are horizontally and vertically inhomogeneous (Rotach 1993a;
Oikawa and Meng 1995; Kastner-Klein et al. 2001; Kastner-Klein and Rotach 2004; Christen
et al. 2009), and in particular, the Reynolds stress and sensible heat flux vary with height, in
contrast to the non-urban surface layer over flat terrain (Rotach 1993b; Oikawa and Meng
1995; Feigenwinter et al. 1999; Louka et al. 2000; Grimmond et al. 2004; Christen et al.
2009). It has long been realized that within the RSL Monin–Obukhov similarity theory is
invalid (Roth 2000). Building roughness elements exert a greater drag on boundary-layer flow
compared with open terrain, and change the behaviour of the turbulent exchange of heat and
momentum within the surface layer. This characteristic affects predictions of the evolution
of the boundary layer and surface variables if included in numerical models (Harman 2012).
Difficulties in determining representative fluxes of the broader inertial sublayer together with
the relatively few measurements in urban environments (Ao et al. 2016) implies that it is nec-
essary to extend the number of measurements in the urban RSL, so as to aid understanding
of surface–atmosphere interactions in urban areas.
Recent field campaigns to investigate turbulence structures and flow characteristics in the
urban surface layer, as well as the associated dispersion processes, include ESCOMPTE1
in Marseille (Mestayer et al. 2005), BUBBLE2 in Basel (Rotach et al. 2005), DAPPLE3
in central London (Arnold et al. 2004), URBAN 20004 in Salt Lake City (Allwine et al.
2002), the Joint Urban 20035 campaign in Oklahoma City (Allwine et al. 2004; Nelson
et al. 2007), and the COSMO6 outdoor experiments in Saitama Prefecture, Japan (Kanda
et al. 2007; Kanda and Moriizumi 2009; Inagaki and Kanda 2008, 2010). Similar field
observations in Chinese cities include those in Beijing (Quan and Hu 2009; Peng and Sun
2014; Wang et al. 2014), in the central business area of Nanjing (Liu et al. 2009; Zou
et al. 2015), and in densely built-up commercial and residential areas of Shanghai (Tan
et al. 2015; Ao et al. 2016). Besides field observations, wind-tunnel experiments are often
used to model turbulent flow over urban-like surfaces. While wind-tunnel modelling allows

1 ESCOMPTE is the Expérience sur site pour contraindre les modèles de pollution atmosphérique et de trans-
port d’émissions project. The field campaign, aiming to construct a reference data base for model validation,
was conducted in the Berre–Marseille area, France.
2 BUBBLE is the Basel UrBan Boundary Layer Experiment project, a year-long experimental effort to
investigate in detail the boundary layer structure in the City of Basel, Switzerland.
3 DAPPLE is the Dispersion of Air Pollution and its Penetration into the Local Environment project, aiming
to provide better understanding of the physical processes affecting the dispersion of pollutants at street canyon.
The first field campaign was conducted in central London.
4 URBAN 2000 is the field campaign performed in Salt Lake City in 2000, which is in terms of investigating
the urban boundary layer and focuses on pollutant dispersion processes at different scales.
5 Joint Urban 2003 is the field campaign performed in Oklahoma City, Oklahoma in 2003, aiming to provide
a more complete look at the flow within a real-world street.
6 COSMO is the Comprehensive Outdoor Scale MOdel experiments, which involves an idealized miniature
city for the study of the flow and turbulence characteristics over urban surface.

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

the evaluation of the mean flow, turbulence statistics, coherent turbulence structures, and
their dependence on the morphological characteristics of the urban canopy in a controlled
environment (MacDonald 2000; Kastner-Klein et al. 2001; Cheng and Castro 2002; Kastner-
Klein and Rotach 2004; Castro et al. 2006; Böhm et al. 2013), the effects of atmospheric
stability cannot be easily simulated. Numerical simulations are also used to investigate the
flow dynamics and turbulence characteristics in the urban RSL (Coceal et al. 2006, 2007;
Kanda 2006; Jiang et al. 2008; Nakayama et al. 2011; Kanda et al. 2013; Giometto et al.
2016), though almost all such simulations are made under neutral conditions, and the thermal
effects of the urban canopy are not included.
For the above field campaigns, the observed turbulent fluxes often increase with height, not
only within the urban canopy layer, but beyond the spatially-averaged building height z H , to
roughly z = 2−2.5z H into the upper portion of the RSL (e.g. Grimmond et al. 2004; Christen
et al. 2009). While turbulent fluxes do naturally decrease by a small fraction (<10%) in the
constant-flux layer, the presence of an urban canopy affects surface-layer structure through
additional heating by walls and roofs of buildings. Rooftop heating is usually more significant
than surface heating, and the urban canopy disrupts the vertical transfer of momentum and
heat. As a consequence, a decrease in the magnitude of turbulent fluxes below the canopy top
is often observed for vegetation canopies (Kaimal and Finnigan 1994) and urban canopies
alike (Christen et al. 2009). Turbulent fluxes are typically a maximum at the vegetation canopy
height, becoming roughly constant above (Dupont and Patton 2012). For an urban canopy
consisting of rigid and bluff bodies, the magnitude of turbulent fluxes increases above the
canopy height to about z = 2z H before the recovery of the constant-flux layer.
For canopy flow, both turbulent and dispersive fluxes are present (Cheng and Castro
2002; Belcher 2005; Coceal et al. 2006; Martilli and Santiago 2007; Böhm et al. 2013;
Giometto et al. 2016), with numerical simulations showing dispersive fluxes to be important
for the overall momentum transfer in the RSL above urban-like surfaces (Coceal et al. 2006;
Martilli and Santiago 2007). While wind-tunnel experiments for a uniform canopy show the
importance of the dispersive stress to the Reynolds stress for computation of the momentum
flux within the canopy, the dispersive stress can be ignored above the canopy (Böhm et al.
2013). Large-eddy simulations (LES) of heat transfer from a single cube (with different
temperatures at the roof, leeward wall and side walls) and a patch on the ground, show
that the dispersive heat flux dominates heat transfer below the cube top, becoming smaller
immediately above the cube top, and decreasing rapidly with height (Boppana et al. 2013).
By comparing the heat flux measured in the inertial sublayer using an eddy-covariance
system, and the weighted value of roof and canyon heat fluxes measured in the RSL by
two scintillometers, Salmond et al. (2012) found that the derived heat flux in the RSL is
smaller than the local-scale heat flux measured in the inertial sublayer. They argued that
the non-measurable dispersive heat flux partly explains the difference. The recent numerical
study of Giometto et al. (2016) shows that spatial variations of time-averaged variables and
non-measurable dispersive terms are important in the RSL above a real urban surface. Hence,
flux measurements in the urban RSL may be directly influenced by the dispersive terms and
should, therefore, be interpreted cautiously.
While previous tower measurements have provided evidence for the increase of the heat
flux with height in the RSL above the urban canopy, the reason is still unclear. We analyze
here the impact of eddy characteristics on the turbulent fluxes at two heights on a rooftop
tower in the downtown area of Nanjing, China. The organization of the paper is as follows:
in Sect. 2, the observational site and the data processing procedures are introduced, and in
Sect. 3, turbulent momentum and heat transfer at the two measurement levels above the
mean urban canopy height are compared. The relationship between the turbulent fluxes and

123
J. Zou et al.

Fig. 1 a Aerial view of the study site, b oblique view of the tower taken from a building south of the tower.
Image (a) is from Google Earth. The tower is located at the centre of (a). The radii of the two concentric
circles around the tower are 500 and 800 m. Colours denote the building height, blue below the rooftop of
the measurement building, cyan above the rooftop of the measurement building, but below the lower flux-
measurement level, green between the two flux-measurement levels, red above the upper flux-measurement
level. The three tall buildings in the lower right background of (b) are about 500 m north of the tower and the
wider tall building is about 800 m north-north-west of the tower. The tower is installed on the roof of a 22-m
building. L1 and L2 are the two measurement heights above ground level

the eddy characteristics is analyzed, and a possible mechanism is presented that explains the
increase of the sensible heat flux with height in the urban RSL. Conclusions are given in
Sect. 4.

2 Methods

2.1 Field Site and Instruments

We made field observations of turbulence and the surface energy balance in an urban area in
Nanjing, which is located to the north-west of the Yangtze River Delta in the most densely
urbanized region in China. Our observational site (32◦ 02 N, 118◦ 79 E) is in downtown Nan-
jing (Liu et al. 2009), and is surrounded by residential and commercial buildings (Fig. 1a).
The mean heights of the buildings (z H , building weighted) are 19.7 and 21.3 m within radii
of 500 and 800 m, respectively. Vegetation covers only about 7–8% of the area within an
800-m radius, with trees 10–15 m high and irregularly dispersed.
To investigate the vertical variation of turbulent fluxes above the urban canopy, a 36-m tri-
angular tower (with a side width of 0.35 m), was installed on the roof of a 22-m high building,
with two measurement heights at 33.5 m (z ≈ 1.7z H ) and 48.5 m (z ≈ 2.5z H ) above ground
level, where z H is 19.7 m within a 500-m radius. Instruments were mounted on 1.5-m long
booms extended from the tower to the south-east (the azimuth angle is about 150◦ , Fig. 1b),

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

and refer to the lower and higher levels hereafter as level 1 (L1) and level 2 (L2), respectively
(Fig. 1b). The relative height range is slightly greater than the ‘down-mode’ measurements
made between z = 1.83z H and 2.22z H in Grimmond et al. (2004), who reported increases in
the turbulent flux with height. At each level, an ultrasonic anemometer (CSAT3, Campbell
Scientific Inc.) was used to measure the three-dimensional velocity components (u, v, w)
and virtual air temperature (Tv ); a net radiometer (CNR4, Kipp and Zonen) measured the
incoming and outgoing shortwave and longwave radiation fluxes. Instrumental data were
collected and stored on a data logger (CR3000, Campbell Scientific Inc.), with signals for
the eddy-covariance calculations sampled at 10 Hz, and subjected to spike detection and
rejection algorithms. Following Kaimal and Finnigan (1994), the coordinates were rotated
twice to align with the streamwise direction for flux computation. At 30-min intervals, block
averages were used to compute the instantaneous fluctuations, momentum and sensible heat
fluxes. Corrections were applied to the sonic virtual temperature according to Schotanus et al.
(1983).
Continuous measurements started on December 1 2011 and ended on August 8 2012, at
which point the eddy-covariance system at level 2 became unreliable, possibly associated
with damage caused by the passage of Typhoon Haikui (2012) on that particular day. The
measurement period covers winter, spring, and most of the summer. To make meaningful
comparison of the ratio of momentum and sensible heat fluxes between the two measurement
levels, we consider only fluxes that have the same sign at both levels. In addition, data from
rainy days are also removed, leaving a total of 8157 out of 9653 30-min averaged data for
analysis.

2.2 Urban Morphology

The flux footprints are estimated for the two measurement heights with the analytical footprint
(K–M) model of Kormann and Meixner (2001), which is widely used by the urban microm-
eteorological community (e.g. Christen et al. 2011; Kotthaus and Grimmond 2012; Nordbo
et al. 2013). The idealized K–M model is strictly applicable to flat and homogeneous ter-
rain under stationary conditions with height-independent cross-wind dispersion. LES results
show that it may underestimate (overestimate) the length of the along-wind (cross-wind) flux
footprint for measurements made over an urban-like canopy (Hellsten et al. 2015). Never-
theless, for lack of a better alternative, the K–M model is used here to estimate the footprint,
with calculations indicating that the far upwind end of the 75% source area is slightly less
than 500 m for all wind directions at level 1, and around 800 m for most wind directions at
level 2.
To update the building information in Liu et al. (2009) for the same site, in situ investi-
gations were made in September 2012 to collect the height and orientation of each building
in the study area within the 800-m radius around the observational site. The length, width
and coordinate information of the buildings were obtained through the Geographic Informa-
tion System following the procedures described in Grimmond and Souch (1994). The mean
building height (zH ), standard deviation of the building height ( σH ), displacement height
(
z d ), aerodynamic roughness length ( z 0 ), plan aerial fraction (λp ) and frontal area index (λf )
were calculated for buildings in the source area, where tildes indicate that the source area
is dependent on wind direction. Following Nordbo et al. (2013), we took into account 20◦
sectors with a 1◦ step for radii of 500 and 800 m. The aerodynamic parameters z d and z 0
were derived following Raupach (1994, 1995), which performs well at this site (Liu et al.
2009). The morphological parameters z H , σH , λp and λf (see Grimmond and Oke 1999) are
displayed in Fig. 2.

123
J. Zou et al.

Fig. 2 a Mean building height z H (thick line) and the sum of the mean building height and standard deviation
of building height z H+σ H (thin line), b displacement height zd (thick line) and roughness length z0 (thin
line), c plan aerial fraction λp , d frontal area index λf as functions of wind direction. The calculation methods
are described in the text. The green and blue lines represent statistics computed within the 500-m and 800-m
circles in Fig. 1a, respectively. The areas in the red frame represent sector A and sector B in Fig. 1a. The black
dashed lines denote the two measurement levels L1 (lower one) and L2 (upper one) shown in Fig. 1b

The mean building height within 800 m of our site rarely exceeds the lower measurement
height level 1 (Fig. 2a). Since the mean building height does not correspond to the top of
the urban canopy, because building heights may vary significantly within a real city, we use
zH + σH as an upper limit estimate of the height of the urban canopy for different wind
directions. As shown in Fig. 2a, when the wind direction is in the range of 000◦ −090◦ , i.e.
in sector A, the canopy top in the source area within the 500-m (800-m) radius is lower than
level 1 (L2), suggesting that the two measurement heights are both above the canopy top. As
the canopy top corresponding to level 2 is significantly higher than for level 1 because of the

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

taller buildings between 500-m and 800-m radii in this sector (see Fig. 1a), the morphological
characteristics in the source areas of the two heights are different. However, when the wind
direction is 180◦ −250◦ , i.e. in sector B, the canopy top in the source area within the 500-
m radius is generally higher than level 1 because of the high-rise buildings (see Fig. 1a).
Although the estimated canopy top in the source area within 800 m is lower than level 2
in this sector, the wakes of three upstream high-rise buildings (Fig. 1a) may significantly
influence the flux measurements.
The wind-direction dependence of the displacement height and roughness length within
500 and 800 m is almost identical to that of the mean building height (Fig. 2b). The plan
aerial fraction λp is defined as the ratio of the plan area of buildings to the total surface area,
and is about 0.3 in most directions within both distances of 500 and 800 m (Fig. 2c). The
wind-direction dependence of λp suggests that the surrounding building density is mostly
homogeneous, except for the north-eastern sector where λp increases to about 0.5, and in the
south-eastern sector where λp < 0.2 because of the presence of open spaces. The frontal
area index λf is defined as the ratio of the total area of the building facets facing a particular
wind direction to the plane area, and is elongated in the north–south directions and limited
in the west–east directions (Fig. 2d), because most of the rectangular buildings have their
major axes orientated along the west–east direction as shown in Fig. 1a.

2.3 Transfer Efficiency and Dispersive Flux

The vertical transfer efficiency rwϕ of the turbulent flux w  ϕ  is defined as the correlation
coefficient,
w ϕ 
rwϕ = , (1)
σw σϕ
where w is the vertical velocity, ϕ is a velocity component or temperature, and σw and σϕ
are their standard deviations. Primes denote deviations from the ensemble average (which
can only be calculated from the time average). The ensemble average is here approximated
by the temporal block average, which is denoted by the overbar. For a given product of σw
and σϕ , a higher transfer efficiency leads to a larger turbulent flux.
However, the presence of the urban canopy leads to a heterogeneous spatial distribution of
mean flow and turbulence quantities. In response, the double-averaging approach decomposes
a general variable ϕ (x, y, z, t) into a time–space average ϕ̄(z) (the overbar and triangular
brackets denote temporal and spatial averages, respectively), giving a fluctuation of the time-
averaged variable with respect to its time–space value ϕ̄  (x, y, z) and a turbulent fluctuation
ϕ  (Belcher 2005; Böhm et al. 2013; Giometto et al. 2016),
ϕ (x, y, z, t) = ϕ̄(z) + ϕ̄  (x, y, z) + ϕ  (x, y, z, t) . (2)
Consequently, w  ϕ  is the turbulent flux, and w̄  ϕ̄   is the dispersive flux. Dispersive fluxes
transport momentum and heat both within the urban canopy and the urban RSL above the
canopy (Belcher 2005). Hence, RSL tower measurements can be severely biased because of
spatial heterogeneity as expressed by the dispersive contribution to the flow (Giometto et al.
2016).

123
J. Zou et al.

Fig. 3 Definitions of quadrant numbering for a u  w  , b w  Tv following Christen et al. (2007) and Dupont
and Patton (2012)

2.4 Quadrant Analysis

A quadrant analysis assists in the identification of turbulence characteristics (Christen et al.


2007), where the quadrant flux for momentum in the i-th quadrant is
 
 
Q u  w  = u a wb , (3)
i
with the subscripts a and b denoting either ‘+’ or ‘−’ values, depending on the quadrant.
Figure 3a shows the quadrant numbers for u  w  based on the definitions of Christen et al.
(2007) and Dupont and Patton (2012). Quadrant 2 denotes ‘ejections’ (u − w+  , low momen-
 
tum fluid transported upwards), and Quadrant 4 denotes ‘sweeps’ (u + w− , high momentum
fluid transported downwards); both of these two quadrants represent organized events. Rep-
resenting counter-flux events are Quadrant 1, referred to as an ‘outward interaction’ (u + w+
 ),
 
 interaction’ (u − w− ).
and Quadrant 3 as an‘inward
A flux fraction Si u  w  can then be calculated as
        
Si u  w  = Q i u  w  / Q i u  w  = Q i u  w  /u  w  (4)
i
to quantify the contribution from a particular quadrant
 to the overall flux. Following Christen
et al. (2007), we use two parameters based on Si u  w  : (i) the difference between the flux
fractions of Quadrants 4 and 2,
     
S u  w  = S4 u  w  − S2 u  w  , (5)
 
where momentum transport is dominated by ejection events when S u  w  < 0, and
 
sweeps when S u  w  > 0; and (ii) the ratio of unorganized to organized contributions to
the average flux, called the ‘exuberance’,
           
E x u  w  = S1 u  w  + S3 u  w  / S2 u  w  + S4 u  w  (6)

as an indicator for the transfer efficiency (Shaw et al. 1983). An Ex value closer to zero sug-
gests less unorganized counter-flux events, and hence a more efficient transfer of momentum.

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

The quadrant numbers for w  Tv  are defined following Dupont and Patton (2012), and are
shown in Fig. 3b. Quadrant 1 represents a warm upward plume (w+  T  ), while Quadrant 3
v+
 T  ); Quadrants 2 and 4 represent warm downward
represents a cool downward plume (w− v−
 T  ) and cool upward (w  T  ) motions, respectively.
(w− v+ + v−  
Analogous to the momentum flux, Si w  Tv  can be calculated as
         
Si w  Tv  = Q i w  Tv  / Q i w  Tv  = Q i w  Tv  /w  Tv , (7)
i
and in contrast to momentum transfer, organized heat transfer depends on the thermal stability.
Since events with a positive contribution to heat transfer are the organized events, Quadrants
1 and 3 are the organized events under unstable conditions, whereas  Quadrants
 2 and
 4 are
the organized events under stable conditions. Thus, we define S w  Tv  and E x w  Tv 
as follows,
     
SU w  Tv  = S3 w  Tv  − S1 w  Tv  , (8a)
     
SS w  Tv  = S4 w  Tv  − S2 w  Tv  , (8b)
           
E xU w  Tv  = S2 w  Tv  + S4 w  Tv  / S1 w  Tv  + S3 w  Tv  , (9a)
           
E x S w  Tv  = S1 w  Tv  + S3 w  Tv  / S2 w  Tv  + S4 w  Tv  , (9b)

where the subscripts U and S represent unstable and stable conditions, respectively.
Under unstable
 conditions, heat transport is dominated by  warm  upward plumes when
SU w  Tv  < 0, and cool downward plumes when SU w  Tv  > 0. Under stable con-
 
ditions, heat transport is dominated by warm downward motion when SS w  Tv  < 0, and
   
cool upward motion when SS w  Tv  > 0. As with the momentum flux, an E xU w  Tv 
 
or E x S w  Tv  value closer to zero suggests less unorganized counter-gradient events, and
hence a more efficient transfer of heat.

3 Results and Discussion

While the focus here is mainly on daytime conditions, results are presented for both daytime
and nighttime conditions for completeness, though it is difficult to discuss the nighttime
situation due to the complexity of the nighttime stable boundary layer (SBL). As the SBL
can be quite shallow (i.e. 100 m or lower) on clear days with low wind shear (Zhou and Chow
2011), measurements at heights from z = 1.7z H to 2.5z H are potentially affected by the
outer-layer flow in the SBL. In addition, the nighttime boundary layer is not always stable
in urban areas (Oke et al. 1999; Grimmond et al. 2004; Christen and Vogt 2004; Ao et al.
2016), and our observations show that the nighttime urban surface layer is usually stable in
the winter, while unstable in the summer.

3.1 Ratio of Turbulent Fluxes at the Two Measurement Heights

Figure 4 shows ratios (L2/L1, where levels 1 (L1) and 2 (L2) are the upper and lower
measurement heights as shown in Fig. 1b) of turbulent heat and momentum fluxes as a

123
J. Zou et al.

Fig. 4 Variation of the L2/L1 ratio of a, b kinematic heat flux and c, d streamwise momentum flux with
wind direction under different stability conditions. The blue dots represent 30-min data (the data with negative
ratios have been removed). The red lines are the median values calculated for 10◦ intervals. The red shadings
represent the inter-quartile range. There are 6940 (85.1%) data points under unstable while 1217 (14.9%)
under stable conditions

function of the wind direction under unstable and stable conditions, respectively, as quantified
by the Monin–Obukhov stability parameter ζ = z  /L, where z  is the height above the
displacement height, and L is the Obukhov length. Here, ζ > 0 indicates stable and ζ < 0
unstable conditions. Medians of flux ratios calculated at 10◦ intervals (as well as the inter-
quartile ranges) fluctuate significantly under stable conditions due to the fewer data points.
Under both unstable and stable conditions, the median ratios are larger than unity for most
wind directions, and the general trends for sensible heat and momentum fluxes are similar.
Large flux ratios appear for easterly (000◦ −180◦ ) flows, while small ratios are found for
south-westerly flows. For wind directions 000◦ −090◦ , the analysis in Sect. 2.2 suggests that

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

the two measurement heights are both above the canopy top. Since the high-rise buildings
are located more than 500 m upwind of the tower, their wake effects have a minimal impact
on the measurements. Figure 4 shows that both sensible heat and momentum fluxes at level
2 are larger than those at level 1 for wind directions between 000◦ −090◦ . While the ratios of
sensible heat and momentum fluxes are also larger than unity for the 090◦ −180◦ quadrant,
level 1 is higher than the canopy top for some wind directions, while level 2 is very close to the
canopy top for most wind directions (Fig. 2a). Overall, observations in these two quadrants
(000◦ −180◦ ) are consistent with Grimmond et al. (2004) and Christen et al. (2009), who
showed that both sensible heat and momentum fluxes increase with height from the surface
up to approximately z = 2z H .
In the 180◦ −270◦ quadrant, the sensible heat-flux ratio stays closely below unity under
unstable conditions, but changes from above 1 to below 1 under stable conditions, while
the momentum-flux ratio changes from above 1 to below 1 under both unstable and stable
conditions. As shown in Fig. 1a, there are a number of high-rise buildings in this quadrant, with
two (marked in red) especially close to the site. The relatively small values of the turbulent
flux ratios are likely caused by the wake effects of these high-rise buildings. Finally, in the
270◦ −360◦ quadrant, the flux ratios change from below 1 to above 1. Although this quadrant
has the fewest tall buildings in the source area, there is another measurement boom positioned
on the tower to the west of the sensor at both levels. Since the tower is to the north-west
of the sensors (see Fig. 1b), the wakes from the booms may well have contaminated the
observations in this quadrant.
The morphological characteristics shown in Figs. 1 and 2 suggest that levels 1 and 2
are above the urban canopy top in the 000◦ −090◦ quadrant (hereafter sector A). Therefore,
sector A is selected to compare the vertical heat and momentum transport, and to analyze
the underlying eddy characteristics in the upper RSL above the urban canopy height. The
180◦ −250◦ sector (hereafter sector B) is chosen to investigate the wake effects of the high-rise
buildings on the eddy characteristics at levels 1 and 2. The 250◦ −270◦ sector is intentionally
neglected to avoid wake effects of upwind instruments.

3.2 Transfer Efficiencies of Heat and Momentum

Figure 5 shows the median and inter-quartile range of σw , σTv , σw σTv , w  Tv and rwTv at the
two measurement levels in sectors A and B. First, we note that the median value of σw at level
2 is systematically larger than that at level 1 for both sectors (Fig. 5a, b), because the intensity
of vertical turbulent motions generally decreases towards the surface due to the restriction of
the ground. Given the overlap of inter-quartiles at the two measurement levels, the ensemble
data in each 30-min bin are subjected to the two-sample Student’s t test to determine the
significance of the difference in [σw ] (where square brackets represent ensemble averaging
for the data in sector A or B for a given sample, because the ensemble mean rather than
median is used when applying the t test). Most of the ensemble means in the daytime pass
the t test (marked by grey shading) for a statistical significance threshold of 0.05. As the
ensemble mean is almost the same as the ensemble median, the results of the significance
test are representative for the ensemble median values, suggesting that [σw ] at level 2 is
statistically larger than that at level 1 to a 95% confidence level during most of the daytime.
For temperature, σTv at level 2 is smaller than that at level 1 in both sectors, although the
differences are small for measurements in sector A (Fig. 5c, d), consistent with temperature
fluctuations being larger near the surface due to strong buoyancy gradient production, and
decreasing with height in the surface layer (Stull 1988).

123
J. Zou et al.

Fig. 5 Variations of median (solid curves) and inter-quartiles (shadings) of, from top to bottom panels,
σw , σTv , σw σTv , w  Tv and rwTv with local standard time (LST) under all stability conditions. The left panels
represent observations from sector A, and the right panels from sector B. Red lines represent the higher level
L2 and blue the lower level L1. The time interval is 30 min. The grey shadings represent periods when the
mean of the two levels are statistically different from each other at the 95% confidence level (see the text for
details)

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

The products of σw and σTv at levels 1 and 2 are


almost identical in both sectors (Fig. 5e, f).
For the heat flux, measurements show that w  Tv at level 2 is systematically larger in sector
A during most daytime hours (Fig. 5g), with the difference being statistically significant
between 1000 and 1500 LST when daytime convection is most vigorous. In sector B, the
median value of w  Tv at level 2 is smaller in the morning (although the difference between the
two levels is not significant), and becomes close to that at level 1 after midday (Fig. 5h). The
calculated transfer efficiencies rwTv at the two measurement
 heights also display different
characteristics in the two sectors. For sector A, rwTv at level 2 is larger than that at level 1
to a 95% confidence level throughout much of the daytime (Fig. 5i). For sector B, the median
value of rwTv at level 2 is smaller in the morning and approaches that at level 1 after midday,
when the results cease to be statistically significant (Fig. 5j).
The ensemble median and inter-quartiles of σu , σu σw , u  w  and ruw at both levels for
sectors A and B are presented in Fig. 6. Given that σu is similar for both sectors and levels,
the difference of σu σw at the two levels in Fig. 6c, d mostly reflects the difference in σw ,
and hence σu σw at level 2 is consistently larger than that at level 1 in both sectors during the
daytime (c.f. Fig. 5a, b). While the difference between the two levels in sector B (Fig. 6d)
is comparable to that in sector A (Fig. 6c), it is not statistically significant due to the lack
of data and relatively large standard deviation. In Fig. 6e, the momentum flux at level 2 is
significantly larger than that at level 1 in sector A, similar to the observations of the sensible
heat flux displayed in Fig. 5g. In comparison, momentum fluxes in sector B (Fig. 6f) at
both heights are nearly the same. While the transfer efficiency ruw at level 2 is significantly
larger for sector A throughout the day (Fig. 6g), ruw is the same at both heights for sector B
(Fig. 6h).
The effect of stability on the transfer efficiencies of the sensible heat
and momentum
fluxes is shown in Fig. 7. For the heat flux, the transfer efficiencies rwTv at level 2 are
larger

for sector A under both stable and unstable conditions (Fig. 7a, c), and for sector B, rwTv at
level 1 is larger under unstable conditions (Fig. 7b), but similar at both heights under stable
conditions (Fig. 7d). For the momentum flux, the transfer efficiencies |ruw | at level 2 are
larger for sector A (Fig. 7e, g), which is similar to rwTv in this sector, and for sector B
however, |ruw | at level 2 is almost the same as that at level 1 under both unstable and stable
conditions (Fig. 7f, h). Variations in the transfer efficiencies of sensible heat and momentum
fluxes with stability are in agreement with previous studies (e.g. Roth and Oke 1995; Quan
and Hu 2009; Wood et al. 2010; Wang et al. 2014).
For sector A, observations indicate that both the sensible heat and momentum fluxes
increase with height in the daytime, in agreement with Rotach (1993a), Grimmond et al.
(2004), and Christen et al. (2009). Statistical analysis suggests higher heat and momentum
transfer efficiencies at level 2, implying that turbulence is more organized at the greater height,
without necessarily implying larger turbulent fluxes. While numerical simulations and wind-
tunnel experiments indicate that the maximum momentum flux occurs at height z = z H
for roughness elements of uniform height (Coceal et al. 2006), the maximum momentum
flux occurs above z = z H for buildings of different heights (Kastner-Klein and Rotach 2004;
Kanda 2006). The results of LES modelling over an urban surface show that both the domain-
averaged and time-averaged locally-sampled turbulent momentum fluxes can increase with
height to nearly z = 2z H for a standard deviation of building height = 0.4z H (Giometto et al.
2016), supporting our measurements showing turbulent fluxes increasing with height in the
RSL up to z = 2.5z H for sector A.
For sector B, where wake effects of nearby upwind tall buildings are expected, the dif-
ferences of both sensible heat and momentum fluxes between the two levels are small, and

123
J. Zou et al.

Fig. 6 As for Fig. 5, but for σu , σu σw , −u  w  , and −ruw from top to bottom panels, respectively. Plots for
σw are given in Fig. 5

the transfer efficiencies rwTv and ruw at both elevations are nearly the same. Compared with
sector A, the nearly height-invariant transfer efficiencies at levels 2 and 1 for sector B can be
attributed to the additional wake effects of the nearby high-rise buildings, which introduce
additional turbulent mixing and reduce turbulence organization, consistent with our results
at level 2.

3.3 Eddy Characteristics for Heat and Momentum Exchange

To further investigate eddy characteristics, Fig. 8 presents flux fractions (i.e. S) and the
exuberance Ex at both two heights for the sensible heat flux under unstable conditions. For

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

Fig. 7 Variations of median (solid curves) and inter-quartiles (shadings) of transfer efficiencies of a–d sensible
heat and e–h momentum fluxes with stability (ζ ). The left panels represent observations from sector A, and
the right panels from sector B in Fig. 1a. Red lines represent level 2 and green lines represent level 1. The grey
shadings represent periods when the mean of the two levels are statistically different from each other at the
95% confidence level

 
sector A, SU w  Tv at both heights are statistically smaller than zero in the daytime. The
majority of the daytime data points (0800–1600 LST)  passed the one-sample Student’s t
test, indicating that the difference between SU w  Tv and zero is significant to a 95%
confidence level, while nighttime data points fluctuate around zero (Fig. 8a). The data show
that the warm upward plumes (Quadrant 1 events)dominate  the daytime heat transfer for this
sector. While the absolute median value of E xU w  Tv at level 2 is systematically smaller

123
J. Zou et al.

 
Fig. 8 Time series of the median (solid curves) and inter-quartiles (colour shadings) of S w  Tv and
 
E x w  Tv for sector A (left panels) and sector B (right panels) under unstable conditions. The circles
indicate instances when the mean is statistically different from zero. The grey shadings represent periods
when the mean of the two levels are statistically different from each other at the 95% confidence level

than that at level 1 to a 95% confidence level for most of the daytime hours, it is nearly
the same as that at level 1 in the nighttime (Fig. 8c), implying that turbulence at the higher
level is better organized and more efficient under unstable conditions, consistent with rwTv
being larger at level 2 in this sector (Fig.
 7a). However, this result contrasts with that reported
in Christen et al. (2007), where E x w  Tv decreases slightly with height from a value of
−0.2 at z = 1.2z H to −0.3 at z = 2.2z H under convective conditions. The larger absolute
values in the nighttime suggest that turbulence organization is reduced under weakly unstable
conditions in comparison with the stronger
 unstable conditions in the daytime. Under stable
 
conditions, the absolute value of E x S w Tv at level 2 is smaller for sector A than that at level
1 (Fig. 9c),
 which is consistent with the results of rwTv (Fig. 7c). More negative data points
of SS w  Tv at level 1 suggest that the warm downward motion dominates heat transfer
at the lower level under stable conditions (Fig. 9a). However, the warm downward motion
and the
 coolupward motion are comparable at level 2, as interpreted from the fluctuation of
SS w  Tv around zero.
 
For sector B, SU w  Tv is around zero in the daytime but increases from small negative
values in the early morning to small positive values in the late afternoon (Fig. 8b), suggesting
an increase in Quadrant 3 events (cool downward plumes) and/or a decrease in Quadrant 1
events (warm upward plumes), as most likely induced by the wakes  of the upstream high-
 
rise buildings. During the nighttime, the medians of SU w Tv at both heights fluctuate
 
significantly due to the lack of data points. In Fig. 8d, the absolute value of E xU w  Tv
at level 2 is larger initially in the morning, with the two curves
 for levels 1 and 2 gradually
converging towards noon. The larger negative value of E xU w  Tv implies less organized

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

Fig. 9 As Fig. 8, but for stable conditions

turbulence and consequently less efficient turbulent heat exchange at level 2, consistent with
rwTv being smaller
 atlevel 2 in the morning (Fig. 5j). During the night, the absolute median
value of E xU w  Tv at level 2 is systematically larger than that at level 1, although the
difference between the two heights does not achieve the 95% confidence level, suggesting
that the turbulent transfer of sensible heat at level 2 is less efficient than at level 1 under
weakly unstable conditions (c.f. Fig. 7b). Under stable conditions, the median of SS w  Tv
fluctuates significantly with time, with more negative data points implying
 warm
 downward
motion is more active at both the heights (Fig. 9b). The values of E x S w  Tv at both heights
are nearly the same (Fig. 9d).    
For the momentum flux, the median and inter-quartiles of S u  w  and E x u  w  at
 
both heights are shown in Fig. 10. For sector A, the median value of S u  w  at level
1 (z = 2.0zH ) fluctuates slightly around zero in the daytime, suggesting that this level
is the cross-over height of ejections (above this level) and sweeps (below this level) for
the momentum exchange. This cross-over level is higher than z = 1.25z H from the direct
numerical simulations of boundary-layer flow over a regular array of cubical roughness
elements under neutral conditions (Coceal et al. 2007). The difference here is largely attributed
to atmospheric stability. In the field observations of Christen et al. (2007), the cross-over
height was found at z = 1.6z H for convective conditions  and at z = 1.9z H for unstable
 
(strongly convective) conditions. The negative S u w at level 2 (z = 2.5zH ) in the
daytime indicates that ejections contribute more to the momentum exchange, which supports
Coceal et al. (2007) who found that above the cross-over height, shear stress is mainly
generated byfewer,  but more energetic, ejection events. Figure 10c shows that the absolute
value of E x u  w  at level 2 is significantly smaller than that at level 1 throughout the day,
suggesting that turbulent motion at the higher level is better organized, consistent
 with  ruw at
 
level 2 being larger in this sector (see Fig. 6g). For sector B, the positive S u w at both
heights indicates sweep events contribute more to the momentum exchange, consistent with

123
J. Zou et al.

Fig. 10 As Fig. 8, but for u  w  under all stability conditions

the influence
 of wakes from the high-rise buildings acting at both heights. Figure 10d shows
 
that E x u w at both heights are nearly the same, which is consistent with the behaviour
of ruw (c.f. Fig. 6h).
The above analyses show that for sector A, during the daytime, turbulent eddies at level
2 exhibit stronger characteristics of warm upward plumes for heat exchange, and ejection
contributions for momentum exchange. While heat exchange is largely influenced by warm
upward plumes at level 1, momentum exchange is governed nearly equally by ejection and
sweep events. Such dissimilarity in heat and momentum exchange was also observed in
the measurements in Basel (Christen et al. 2007). Meanwhile, the absolute value of E x
for heat and momentum exchange at level 2 is systematically smaller than that at level 1,
suggesting that turbulent eddies are better organized and have a higher transfer efficiency for
heat and momentum at higher elevations in the RSL, supporting the understanding that heat
and momentum fluxes increase with height in the urban RSL. For sector B, the measurements
suggest that the upstream high-rise buildings likely distort the airflow and change the turbulent
eddy structures, leading to stronger sweep characteristics at levels 1 and 2. In contrast, the
momentum flux observations for sector A show ejection events contributing more at level 2.

3.4 Discussion

In the urban setting, vertically-distributed heat sources from roofs, walls and the ground in
the daytime potentially enhance turbulent heat flux divergence, especially inside the street
canyons, consistent with classic heat flux profiles within plant canopies (Fig. 3.5 in Kaimal
and Finnigan 1994). However, in the real urban canopy, building heights are non-uniform,
implying that both the roofs and walls of taller-than-average buildings can act to increase the
sensible heat flux with height above the mean building height.
However, even for an urban canopy with a uniform building height, the sensible heat flux
can still increase with height in part of the RSL above z = z H because of the ‘blocking

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

effect’ of the ground sources. Based on data from coupled Lagrangian stochastic and LES
models, Hellsten et al. (2015) evaluated the flux and concentration footprints of passive
scalars in and above an urban-like canopy for a source height 1 m above the ground in the
street canyons. They found the horizontally-averaged scalar flux increases with height with
a maximum value at z = 1.8z H , and a negative flux footprint area upwind of the sensor
position for a measurement height (at the location above a building) at z = 1.25z H ; for a
higher measurement height at z = 1.8z H , the flux footprint in this area becomes positive.
A negative flux footprint implies that an emission source in the upstream area may have a
negative contribution to the flux measurement (Cai et al. 2010). A larger flux at a greater height
suggests the arrival of particles from upstream sources at elevated levels. As particles such as
these may be present in larger concentrations at higher elevations (Hellsten et al. 2015), the
local particle flux may be negative at lower heights due to dispersion. This phenomenon is
likely caused by the ‘blocking effect’ of buildings, which block the flow, change the trajectory
of scalars, and distort the footprint of scalar fluxes. The blocking effects may also act on the
wall and even roof sources in real cities. For sensors located above z = z H , but within the
RSL, such blocking effects should be the largest at z = z H , while decreasing with distance
from the buildings. In other words, as the height increases, the sensor’s upwind footprint
becomes less blocked, till further above the RSL, the blocking effects become negligible.
Accessing a greater proportion of the idealized flux footprint (assuming flat terrain) at higher
elevations may contribute to the increase of heat flux with height.
Another reason for the vertical variation of the measured turbulent fluxes in the urban RSL
may be the presence of dispersive fluxes. For a rigid canopy with the same element height,
the dispersive fluxes play an important role in the canopy layer, but can be neglected above
the canopy top (Coceal et al. 2006; Böhm et al. 2013). For a real urban canopy with different
building heights, Giometto et al. (2016) found dispersive momentum fluxes to be the same
order of magnitude as their Reynolds counterparts in the canopy layer, and non-negligible in
the urban RSL, with the horizontal mean dispersive momentum flux maximum at z = z H ,
decreasing to a small value at z = 2z H , while remaining almost constant with height in the
upper part of the RSL. Our measurement heights levels 1 and 2 for sector A are 2z H and
2.5z H , respectively, which according to the results in Giometto et al. (2016), should mean
the dispersive fluxes at both heights are nearly the same. Hence, it is likely that for sector A,
the difference in the turbulent fluxes between the two heights is not caused by the presence
of dispersive fluxes.
In addition to blocking, buildings also affect turbulent transport through wake production.
In numerical simulations (Cheng and Porté-Agel 2015) and laboratory measurements (Castro
et al. 2006), wake effects are strongest around the mean building height, and decrease towards
the top of the RSL. The wake effects reduce the organization of turbulence,
 and lower turbulent
transfer efficiencies in the canopy layer. Based on E x u  w  calculated from the data of
wind-tunnel experiments, Böhm et al. (2013) found that, in the canopy layer, the turbulence
in the non-wake regions is more organized than in wake regions. They argued that the bluff-
body wakes add energetic, but less coherent turbulence, to the organized eddies (as in the
wakeless regions), thereby reducing the efficiency of momentum transfer. This effect acts in
competition with surface (ground, roofs and walls) heating, where thermal convection tends
to increase the organization of turbulent flows and improve turbulent transfer efficiency. In
a realistic urban setting, building heights are not uniform, building densities decrease with
height, and so do the wake effects, leading to more organized turbulence at higher elevations.
Taking these factors together, a hypothesis for our observed heat-flux profiles emerges. For
sector A, the wake effects of tall buildings on the flux measurements are expected to be small,

123
J. Zou et al.

because their locations are far from the tower, and the measurements of turbulent fluxes at
both levels are determined mainly by the morphological characteristics of the buildings within
a 500-m radius. As wake effects are strong at the building top, and the mixing efficiencies
are lowered, there is less turbulence organization at level 1. As wake effects decrease with
height, turbulence at level 2 becomes better organized, and the mixing efficiencies improve.
For the turbulent transfer of sensible heat at both levels, the ‘blocking effects’ can (at least
partially) explain why the sensible heat flux at level 2 is larger than at level 1. As for the
fact that the turbulent momentum flux is larger at level 2 than at level 1, we can point to
the fact that, over the real urban surface with different building heights, the inflection point
of the mean flow is located above the mean building height z H (Giometto et al. 2016),
consistent with LES modelling results for flow over an urban-like canopy with different
building heights (Nakayama et al. 2011). It is likely that the inflection point is close to level 2
in our measurement site, which leads to greater shear and consequently a larger momentum
flux at this height.
For sector B, the high-rise buildings are fairly close to our measurement tower. Heating
from the building walls and wakes of the buildings may directly affect the flux measurements
at both measurement levels. While the heat input from the walls may improve the local
turbulent mixing efficiencies by better organizing thermal convection at elevated levels, wakes
may generate energetic, but less coherent, turbulence in the case of the presence of high-rise
buildings. When the positive effects due to heating prevail over the adverse wake effects,
the overall mixing efficiency is expected to improve. At level 2, the positive wall heating
effects are still present, but the wake effects from those same high-rise buildings dominate the
overall turbulence characteristics (because of the stronger wakes induced by the tall buildings’
rooftop), such that the mixing efficiency decreases. With the positive heating effects at level
1 and negative wake effects at level 2, the mixing efficiencies and organization at both levels
converge, leading to similar sensible heat fluxes at both levels. For the momentum transfer
at both levels, the situation is similar and hence of similar magnitude. When compared with
sector A, the turbulent momentum transfer at both heights for sector B becomes sweep-event
dominating (see Fig. 10a, b), demonstrating the wake effects of the high-rise buildings.
As the above hypothesis has its limitations, it should be tested against field observations,
numerical simulations, wind-tunnel and/or outdoor experiments. The most serious limitation
here is the single measurement location. Since the urban canopy is spatially inhomogeneous,
large spatial variations of vertical flux profiles may likely exist. It is possible that when fluxes
are found to increase with height over the rooftop, they decrease with height over the nearby
street canyon. However, Christen et al. (2009) and Grimmond et al. (2004) found similar
results for sidewalk and rooftop towers, lending some confidence to the regular appearance
of such vertical flux profiles.

4 Conclusions

Field observations are used to analyze the turbulence characteristics and their influence on
the turbulent transport of heat and momentum at z = 1.7zH and z = 2.5zH in the urban
RSL. Two wind sectors are selected for the purpose of comparison. The daytime data show
that, for sector A, the sensible heat and momentum fluxes increase with height in agreement
with previous studies (e.g. Rotach 1993a; Grimmond et al. 2004; Christen et al. 2009). Heat
and momentum transfer efficiencies at the upper level are larger than at the lower level, with
quadrant analyses indicating that heat and momentum transfer are better organized at the

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

upper level. For sector B, the sensible heat and momentum flux are nearly the same at both
levels, and heat and momentum transfer efficiencies and turbulence organization are also
similar. With increasing instability (stability) under unstable (stable) conditions, the transfer
efficiency of the sensible heat and momentum fluxes increases and decreases, respectively,
consistent with previous studies (e.g. Roth and Oke 1995; Quan and Hu 2009; Wood et al.
2010; Wang et al. 2014).
We explain the observed vertical-flux-profile behaviour for sectors A and B from the
combined effects of non-uniform building heights, blocking, heating from tall building walls
and the wakes of tall buildings. For sector A, where both measurement levels are above the
mean upwind building height, and any high-rise buildings are far away, the momentum and
sensible heat fluxes are mostly influenced by the wakes and blocking effects of buildings of
non-uniform building height. While wake effects elevate the inflection point of the mean flow
to a greater height that is likely close to level 2, where the stronger shear leads to the larger
momentum flux at this level, blocking effects likely distort the footprints of the sensible heat
flux at different heights above the urban canopy, resulting in sensible heat fluxes increasing
with height in the RSL over the urban canopy.
For sector B, the influence of three nearby high-rise buildings is significant. The wall heat-
ing from these tall buildings may directly enhance the vertical flux divergence, and indirectly
contribute to more organized turbulent convection. The increase of mixing efficiencies at the
lower level is consistent with these indirect effects. At the upper measurement level, trailing
wakes from the tall buildings dominate and the mixing efficiencies largely decrease, leading
to similar fluxes observed at both levels in this sector. The validity of our hypothesis needs to
be tested with the help of more comprehensive field observations, wind-tunnel and outdoor
experiments involving multiple measurement locations, and/or numerical simulations.

Acknowledgements This study was supported by the State Key Basic Program under Grant No.
2010CB428501 and the Jiangsu Provincial Collaborative Innovation Center of Climate Change. Support
for Bowen Zhou was obtained from the Deng Feng Program of Nanjing University and the National Natural
Science Foundation of China (Grant 41505047). Jun Zou was also supported by the program for Outstanding
PhD Candidate of Nanjing University. The authors thank the anonymous reviewers, whose comments helped
improve the manuscript.

References
Allwine KJ, Shinn JH, Streit GE, Clawson KL, Brown M (2002) Overview of URBAN 2000: a multiscale
field study of dispersion through an urban environment. Bull Am Meteorol Soc 83(4):521–536
Allwine KJ, Leach MJ, Stockham LW, Shinn JS, Hosker RP, Bowers JF, Pace JC (2004) Overview of joint urban
2003—an atmospheric dispersion study in Oklahoma City. In: Symposium on planning, nowcasting, and
forecasting in the urban zone. American Meteorological Society, Seattle, p J7.1
Ao X, Grimmond CSB, Chang Y, Liu D, Tang Y, Hu P, Wang Y, Zou J, Tan J (2016) Heat, water and carbon
exchanges in the tall megacity of Shanghai: challenges and results. Int J Climatol 36:4608–4624
Arnold SJ, Apsimon H, Barlow J, Belcher S, Bell M, Boddy JW, Britter R, Cheng H, Clark R, Colvile RN,
Dimitroulopoulouh S, Dobre A, Greally B, Kaur S, Knights A, Lawton T, Makepeace A, Martin D,
Neophytou M, Neville S, Nieuwenhuijsen M, Nickless G, Price C, Robin A, Shallcross D, Simmonds
P, Smalley RJ, Tate J, Tomlin AS, Wang H, Walsh P (2004) Introduction to the DAPPLE air pollution
project. Sci Tot Environ 332:139–153
Belcher SE (2005) Mixing and transport in urban areas. Phil Trans R Soc A 363:2947–2968
Böhm M, Finnigan JJ, Raupach MR, Hughes D (2013) Turbulence structure within and above a canopy of
bluff elements. Boundary-Layer Meteorol 146(3):393–419
Boppana VBL, Xie Z-T, Castro IP (2013) Large-eddy simulation of heat transfer from a single cube mounted
on a very rough wall. Boundary-Layer Meteorol 147(3):347–368

123
J. Zou et al.

Cai X, Chen J, Desjardins R (2010) Flux footprints in the convective boundary layer: large-eddy simulation
and Lagrangian stochastic modelling. Boundary-Layer Meteorol 137(1):31–47
Castro IP, Cheng H, Reynolds R (2006) Turbulence over urban-type roughness: deductions from wind-tunnel
measurements. Boundary-Layer Meteorol 118(1):109–131
Cheng H, Castro IP (2002) Near wall flow over urban-like roughness. Boundary-Layer Meteorol 104(2):229–
259
Cheng WC, Porté-Agel F (2015) Adjustment of turbulent boundary-layer flow to idealized urban surfaces: a
large-eddy simulation study. Boundary-Layer Meteorol 155(2):249–270
Christen A, Vogt R (2004) Energy and radiation balance of a central European city. Int J Climatol 24:1395–1421
Christen A, van Gorsel E, Vogt R (2007) Coherent structures in urban roughness sublayer turbulence. Int J
Climatol 27:1955–1968
Christen A, Rotach MW, Vogt R (2009) The budget of turbulent kinetic energy in the urban roughness sublayer.
Boundary-Layer Meteorol 131(2):193–222
Christen A, Coops NC, Crawford BR, Kellett R, Liss KN, Olchovski I, Tooke TR, van der Laan M, Voogt
JA (2011) Validation of modeled carbon-dioxide emissions from an urban neighborhood with direct
eddy-covariance measurements. Atmos Environ 45:6057–6069
Coceal O, Thomas TG, Castro IP, Belcher SE (2006) Mean flow and turbulence statistics over groups of
urban-like cubical obstacles. Boundary-Layer Meteorol 121(3):491–519
Coceal O, Dobre A, Thomas TG, Belcher SE (2007) Structure of turbulent flow over regular arrays of cubical
roughness. J Fluid Mech 589:375–409
Dupont S, Patton EG (2012) Momentum and scalar transport within a vegetation canopy following atmospheric
stability and seasonal canopy changes: the CHATS experiment. Atmos Chem Phys 12(2):6363–6418
Feigenwinter C, Vogt R, Parlow E (1999) Vertical structure of selected turbulence characteristics above an
urban canopy. Theor Appl Climatol 62:51–63
Giometto MG, Christen A, Meneveau C, Fang J, Krafczyk M, Parlange MB (2016) Spatial characteristics of
roughness sublayer mean flow and turbulence over a realistic urban surface. Boundary-Layer Meteorol
160(3):425–452
Grimmond CSB (2006) Progress in measuring and observing the urban atmosphere. Theor Appl Climatol
84:3–22
Grimmond CSB, Oke TR (1999) Aerodynamic properties of urban areas derived from analysis of surface
form. J Appl Meteorol 38(9):1262–1292
Grimmond CSB, Souch C (1994) Surface description for urban climate studies: a GIS based methodology.
Geocarto Int 9(1):47–59
Grimmond CSB, Salmond JA, Oke TR, Offerle B, Lemonsu A (2004) Flux and turbulence measurements at
a densely built-up site in Marseille: heat, mass (water and carbon dioxide), and momentum. J Geophys
Res Atmos 109(D24):2561–2580
Harman IN (2012) The role of roughness sublayer dynamics within surface exchange schemes. Boundary-
Layer Meteorol 142:1–20
Hellsten A, Luukkonen SM, Steinfeld G, Kanani-Sühring F, Markkanen T, Järvi L, Lento J, Vesala T, Raasch
S (2015) Footprint evaluation for flux and concentration measurements for an urban-like canopy with
coupled Lagrangian stochastic and large-eddy simulation models. Boundary-Layer Meteorol 157(2):1–
27
Inagaki A, Kanda M (2008) Turbulent flow similarity over an array of cubes in near-neutrally stratified
atmospheric flow. J Fluid Mech 615:101–120
Inagaki A, Kanda M (2010) Organized structure of active turbulence over an array of cubes within the loga-
rithmic layer of atmospheric flow. Boundary-Layer Meteorol 135:209–228
Jiang D, Jiang W, Liu H, Sun J (2008) Systematic influence of different building spacing, height and layout
onmean wind and turbulent characteristics within and over urban building arrays. Wind Struct 11:275–289
Kaimal JC, Finnigan JJ (1994) Atmospheric boundary layer flows: their structure and measurement. Oxford
University Press, Oxford, 289 pp
Kanda M (2006) Large-eddy simulations on the effects of surface geometry of building arrays on turbulent
organized structures. Boundary-Layer Meteorol 118:151–168
Kanda M, Moriizumi T (2009) Momentum and heat transfer over urban-like surfaces. Boundary-Layer Mete-
orol 131:385–401
Kanda M, Kanega M, Kawai T, Moriwaki R, Sugawara H (2007) Roughness lengths for momentum and heat
derived from outdoor urban scale models. J Appl Meteorol Climatol 46(7):1067–1079
Kanda M, Inagaki A, Miyamoto T, Gryschka M, Raasch S (2013) A new aerodynamic parametrization for
real urban surfaces. Boundary-Layer Meteorol 148:357–377
Kastner-Klein P, Rotach MW (2004) Mean flowand turbulence characteristics in an urban roughness sublayer.
Boundary-Layer Meteorol 111:55–84

123
Impact of Eddy Characteristics on Turbulent Heat and Momentum…

Kastner-Klein P, Fedorovich E, Rotach MW (2001) A wind tunnel study of organised and turbulent air motions
in urban street canyons. J Wind Eng Ind Aerodyn 89(9):849–861
Kormann R, Meixner FX (2001) An analytical footprint model for non-neutral stratification. Boundary-Layer
Meteorol 99(2):207–224
Kotthaus S, Grimmond CSB (2012) Identification of micro-scale anthropogenic CO2 , heat and moisture
sources–processing eddy covariance fluxes for a dense urban environment. Atmos Environ 57:301–316
Liu G, Sun J, Jiang W (2009) Observational verification of urban surface roughness parameters derived from
morphological models. Meteorol Appl 16(2):205–213
Louka P, Belcher SE, Harrison RG (2000) Coupling between air flow in streets and the well-developed boundary
layer aloft. Atmos Environ 34:2613–2621
MacDonald RW (2000) Modelling the mean velocity profile in the urban canopy layer. Boundary-Layer
Meteorol 97:25–45
Martilli A, Santiago JL (2007) CFD simulation of airflow over a regular array of cubes. Part II: analysis of
spatial average properties. Boundary-Layer Meteorol 122(3):635–654
Mestayer PG, Durand P, Augustin P, Bastin S, Bonnefond J-M, Bénech B, Campistron B, Coppalle A, Delbarre
H, Dousset B, Drobinski P, Druilhet A, Fréjafon E, Grimmond CSB, Groleau D, Irvine M, Kergomard C,
Kermadi S, Lagouarde J-P, Lemonsu A, Lohou F, Long N, Masson V, Moppert C, Noilhan J, Offerle B,
Oke TR, Pigeon G, Puygrenier V, Roberts S, Rosant J-M, Sad F, Salmond J, Talbaut M, Voogt J (2005)
The urban boundary-layer field campaign in Marseille (UBL/CLU-ESCOMPTE): set-up and first results.
Boundary-Layer Meteorol 114(2):315–365
Nakayama H, Takemi T, Nagai H (2011) LES analysis of the aerodynamic surface properties for turbulent
flows over building arrays with various geometries. J Appl Meteorol 50:1692–1712
Nelson MA, Pardyjak ER, Klewickl JC, Pol SU, Brown MJ (2007) Properties of the wind field within the
Oklahoma City Park Avenue street canyon. Part I: mean flow and turbulence statistics. J Appl Meteorol
Climatol 46:2038–2054
Nordbo A, Järvi L, Haapanala S, Moilanen J, Vesala T (2013) Intra-city variation in urban morphology and
turbulence structure in Helsinki, Finland. Boundary-Layer Meteorol 146(3):469–496
Oikawa S, Meng Y (1995) Turbulence characteristics and organized motion in a suburban roughness sublayer.
Boundary-Layer Meteorol 74(3):289–312
Oke TR, Spronken-Smith RA, Jauregui E, Grimmond CSB (1999) The energy balance of central Mexico City
during the dry season. Atmos Environ 33:3919–3930
Peng Z, Sun J (2014) Characteristics of the drag coefficient in the roughness sublayer over a complex urban
surface. Boundary-Layer Meteorol 153(3):568–580
Quan L, Hu F (2009) Relationship between turbulent flux and variance in the urban canopy. Meteorol Atmos
Phys 104:29–36
Raupach MR (1994) Simplified expressions for vegetation roughness length and zero-plane displacement as
functions of canopy height and area index. Boundary-Layer Meteorol 71:211–216
Raupach MR (1995) Corrigenda. Boundary-Layer Meteorol 76:303–304
Raupach MR, Antonia RA, Rajagopalan S (1991) Rough-wall turbulent boundary layers. Appl Mech Rev
44(1):1–25
Rotach MW (1993a) Turbulence close to a rough urban surface. Part I: Reynolds stress. Boundary-Layer
Meteorol 65:1–28
Rotach MW (1993b) Turbulence close to a rough urban surface. Part II: variances and gradients. Boundary-
Layer Meteorol 66:75–92
Rotach MW (2001) Simulation of urban-scale dispersion using a Lagrangian stochastic dispersion model.
Boundary-Layer Meteorol 99(3):379–410
Rotach MW, Vogt R, Bernhofer C, Batchvarova E, Christen A, Clappier A, Feddersen B, Gryning S-E, Martucci
G, Mayer H, Mitey V, Oke TR, Parlow E, Richner H, Roth M, Roulet Y-A, Ruffieux D, Salmond JA,
Schatzmann M, Voogt JA (2005) BUBBLE—an urban boundary layer meteorology project. Theor Appl
Climatol 81:231–261
Roth M (2000) Review of atmospheric turbulence over cities. Q J R Meteorol Soc 126:941–990
Roth M, Oke TR (1995) Relative efficiencies of turbulent transfer of heat, mass, and momentum over a patchy
urban surface. J Atmos Sci 52:1864–1874
Salmond JA, Roth M, Oke TR, Christen A, Voogt JA (2012) Can surface-cover tiles be summed to give
neighborhood fluxes in cities. J Appl Meteorol Climatol 51:133–149
Schotanus P, Nieuwstadt FTM, DeBruin HAR (1983) Temperature measurement with a sonic anemometer
and its application to heat and moisture fluxes. Boundary-Layer Meteorol 26(1):81–93
Shaw RH, Tavangar J, Ward DP (1983) Structure of the Reynolds stress in a canopy layer. J Appl Meteorol
22:1922–1931
Stull RB (1988) An introduction to boundary layer meteorology. Kluwer, Dordrecht, 666 pp

123
J. Zou et al.

Tan JG, Yang LM, Grimmond CSB, Shi JP, Gu W, Chang YY, Hu P, Sun J, Ao XY, Han ZH (2015) Urban
integrated meteorological observations: practice and experience in Shanghai, China. Bull Am Meteorol
Soc 96:85–102
Wang L, Li D, Gao Z, Sun T, Guo X, Bou-Zeid E (2014) Turbulent transport of momentum and scalars above
an urban canopy. Boundary-Layer Meteorol 150:485–511
Wood CR, Lacser A, Barlow JF, Padhra A, Belcher SE, Nemitz E, Helfter C, Famulari D, Grimmond CSB
(2010) Turbulent Flow at 190 m height above London during 2006–2008: a climatology and the appli-
cability of similarity theory. Boundary-Layer Meteorol 137(1):77–96
Zhou B, Chow FK (2011) Large-eddy simulation of the stable boundary layer with explicit filtering and
reconstruction turbulence modeling. J Atmos Sci 68(9):2142–2155
Zou J, Liu G, Sun J, Zhang H, Yuan R (2015) The momentum flux-gradient relations derived from field
measurements in the urban roughness sublayer in three cities in China. J Geophys Res Atmos 120:10797–
10809

123

You might also like