You are on page 1of 18

PUBLICATIONS

Journal of Geophysical Research: Atmospheres


RESEARCH ARTICLE Interaction between turbulent flow and sea breeze front
10.1002/2016JD026247
over urban-like coast in large-eddy simulation
Key Points: Ping Jiang1, Zhiping Wen1,2, Weiming Sha3, and Guixing Chen1,2
• Fine-scale turbulent flow and sea
breeze front are explicitly resolved in 1
Center for Monsoon and Environment Research, School of Atmospheric Sciences, and Guangdong Province Key
large-eddy simulation
• Streaky turbulent structures form over Laboratory for Climate Change and Natural Disaster Studies, Sun Yat-sen University, Guangzhou, China, 2Zhuhai Joint
urban surface with strong wind shear Innovative Center for Climate-Environment-Ecosystem and Key Laboratory of Urban Climate and Ecodynamics, Beijing
and moderate buoyancy Normal University, Zhuhai, China, 3Department of Geophysics, Tohoku University, Sendai, Japan
• Streaky turbulent structures render
sea breeze front into 3-D structures
with strengthened updrafts
Abstract Turbulent flow and its interaction with a sea breeze front (SBF) over an urban-like coast with a
regular block array were investigated using a building-resolving computational fluid dynamics model. It
was found that during daytime with an offshore ambient flow, streaky turbulent structures tended to grow
Correspondence to:
within the convective boundary layer (CBL) over a warm urban surface ahead of the SBF. The structures
G. Chen,
chenguixing@mail.sysu.edu.cn were organized as streamwise streaks at an interval of a few hundred meters, which initiated at the rooftop
level with strong wind shear and strengthens in the CBL with moderate buoyancy. The streaks then
interacted with the onshore-propagating SBF as it made landfall. The SBF, which was initially characterized
Citation:
Jiang, P., Z. Wen, W. Sha, and G. Chen as a shallow and quasi-linear feature over the sea, developed three-dimensional structures with intensified
(2017), Interaction between turbulent updrafts at an elevated frontal head after landfall. Frontal updrafts were locally enhanced at intersections
flow and sea breeze front over
where the streaks merged with the SBF, which greatly increased turbulent fluxes at the front. The frontal
urban-like coast in large-eddy
simulation, J. Geophys. Res. Atmos., 122, line was irregular because of merging, tilting, and transformation effects of vorticity associated with streaky
5298–5315, doi:10.1002/2016JD026247. structures. Inland penetration of the SBF was slowed by the frictional effect of urban-like surfaces and
turbulent flow on land. The overall SBF intensity weakened after the interaction with turbulent flow. These
Received 17 NOV 2016
findings aid understanding of local weather over coastal cities during typical sea breeze conditions.
Accepted 8 MAY 2017
Accepted article online 10 MAY 2017
Published online 27 MAY 2017
1. Introduction
The sea breeze is one of the atmospheric phenomena that dominate local weather in coastal areas [e.g.,
Masselink and Pattiaratchi, 1998; Miller et al., 2003]. It is mainly induced by differential heating from land-
sea contrast [Simpson, 1994]. The leading edge of the sea breeze often behaves like a shallow cold front with
a sudden change in winds, temperature, and humidity [Simpson et al., 1977; Rao et al., 1999]. The sea breeze
front (SBF) brings in cool marine air to relieve hot weather, sometimes induces cloud formation, and even
triggers convective weather [Dias and Machado, 1997; Papanastasiou et al., 2010; Ryu et al., 2016]. The sea
breeze is also believed to affect the dispersion of air pollutants [Britter and Hanna, 2003; Thompson et al.,
2007], airborne allergens [Raynor, 1974], and crop-threatening insects [Berry and Taylor, 1968]. Therefore,
further understanding of SBF intrusion in coastal areas is important to both local weather forecasts and
environmental monitoring.
The sea breeze is a typical form of gravity current with complex structures, showing many similarities to those
in laboratory experiments [Charba, 1974; Lapworth, 2000]. Its fine-scale structures may greatly modify or inter-
act with the general behavior of the front. For instance, Kelvin-Helmholtz billows (KHBs), which are associated
with turbulent mixing and wave perturbations over the sea breeze head, can act as top friction that decele-
rates the SBF’s inland penetration [Sha et al., 1991]. An overhanging nose forms at the advancing SBF because
of a combined effect of friction and inertia [Simpson, 1982]. Continental warm air overrunning the nose favors
gravitational instability, resulting in lobes/clefts at the front [Simpson, 1972]. These lobes/clefts are often asso-
ciated with strong vertical motion and thus have a strong effect on the flux transport at the SBF [Allen, 1971].
Because the aforementioned structures of the mesoscale SBF have a scale of a few hundred meters, they chal-
lenge both observation and modeling studies. Although general SBF-related features have been extensively
studied, SBF fine-scale dynamics and structures are less understood [e.g., Crosman and Horel, 2010].
The sea breeze is a shallow circulation with a depth of several hundred meters, which is comparable to the
©2017. American Geophysical Union. height of the boundary layer at the coast [Simpson, 1969; Reible et al., 1993]. When penetrating inland, the
All Rights Reserved. SBF is subject to the thermal and dynamical effects of underlying surfaces [Miller et al., 2003]. Surface

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5298


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

properties such as steep topography and complex geometry may strongly regulate local winds, temperature,
and dispersion in the boundary layer [Collier, 2006]. In particular, over urban areas with tall buildings, various-
scale turbulent flow may develop in the convective boundary layer (CBL). Small eddies with strong instanta-
neous flow form in the urban canopy layer, owing to the discrete nature of buildings [Eliasson et al., 2006].
Given favorable conditions, such eddies may grow coherently to form turbulent organized structures of large
size in the CBL [Inagaki et al., 2012]. This organized structures manifest as low-speed streaks coupled with
updrafts that are aligned with a streamwise orientation [Kanda, 2006; Park and Baik, 2013]. Such turbulent
flow is related to the characteristic length of roughness elements [Finnigan et al., 2009] and reaches a size
several times larger than an individual building [Castillo et al., 2011; Inagaki et al., 2012]. A common type is
horizontal convective rolls that can be observed over urban and/or coastal areas [Miao and Chen, 2008;
Iwai et al., 2008; Chen et al., 2015a]. These are believed to influence the behavior of the inland-moving SBF
[Stephan et al., 1999]. Therefore, the formation of turbulent flow over urban surfaces and its impacts on
SBF fine-scale structures need to be clarified.
Some studies have attempted to draw a connection between turbulent flow and the SBF. Wakimoto and
Atkins [1994] and Atkins et al. [1995] revealed that vertical motion and clouds vary substantially along the
SBF when it merges with preexisting roll convection. Dailey and Fovell [1999, 2001] used a three-dimensional,
cloud-resolving model to illustrate the function of roll convection in modulating SBF. They noted that roll
convection perpendicular to the SBF locally enhances (suppresses) frontal convection where roll updrafts
(downdrafts) collide with the front, whereas parallel to the front alternately enhances/suppresses overall
convective activity along the front. Ogawa et al. [2003] also showed that the frontal structures (shape, vertical
velocity, and temperature) vary periodically as they are affected by vertical motions associated with prefron-
tal convective cells. These studies have focused on turbulence-SBF interaction over flat or homogeneous
surfaces. However, it is of particular interest to examine the interaction over urban areas with buildings which
are large roughness elements.
In the present study, we performed an idealized simulation of the SBF over an urban-like coast with a regular
block array using a building-resolving computational fluid dynamics (CFD) model. The goal is to assess how
urban-induced turbulent flow develop over coastal cities and then modifies the fine-scale structures of an
onshore-propagating SBF. The remainder of the paper is organized as follows. Section 2 describes the model
and experiment configuration. Section 3 overviews the general features of turbulent activities and sea breeze
intrusion. Detailed structures and statistics of turbulent flow are analyzed in section 4. In section 5, interaction
between streaky turbulent structures and the SBF is discussed. Finally, conclusions are presented.

2. Configuration of Numerical Model and Experiment


2.1. Local Meteorological Model Based on Large-Eddy Simulation
To conduct numerical experiments, we used a parallelized CFD model for local meteorology [Sha, 2002]. The
model explicitly resolves steep topography and buildings in three-dimensional Cartesian coordinates, with a
blocking-off technique for surface geometries [Patankar, 1980; Sha, 2002]. The model is based on the fully
compressible Navier-Stokes equations, which are discretized by finite volumes and solved by the
Semi-Implicit Method for Pressure-Linked Equations revised algorithm [Patankar, 1980; Sha et al., 1991].
The variable fields are corrected iteratively until reaching a converged solution at each time step
[Patankar, 1980; Sha, 2002], and thus the physics consistency among the variables is warranted. The third-
order upwind Quadratic Upstream Interpolation for Convective Kinematics (QUICK) scheme is used for advec-
tion [Leonard, 1979]. Subgrid-scale turbulence is handled by a classic Lilly-Smagorinsky subgrid model [Lilly,
1962; Smagorinsky, 1963]. A two-layer approach is applied for near-wall treatment, which provides the CFD
region with the wall shear stress extracted from a separate modeling process applied to the near-wall layer.
For more model details, please refer to Sha [2002, 2008].
The CFD model has been used to conduct a series of idealized experiments on topography and dense
buildings, and the calculation of turbulent flow can be quite stable with dense roughness [Sha, 2008]. It
has also been used to reproduce local winds, temperature, and roll convection in the sea breeze over com-
plex surfaces, favorably consistent with lidar observations [Chen et al., 2015a, 2015b]. It is therefore applicable
to modeling turbulent flow and the SBF on an urban-scale domain as in the current study. Here we focus on
the phenomena on land (urban surface) where sensible heat flux is dominant and humidity is relatively

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5299


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 1. Schematic of simulation setup. Domain size is 18 km (1800 grids) in streamwise direction (x), 6 km (600 grids) in
spanwise direction (y), and z = 1675 m (80 grids) in the vertical (z). Land area is at left side, while sea area is at right.
Coastline is located at x = 12 km from the left boundary. A regular array of blocks (120 m (x) × 120 m (y) × 80 m (z)) is
established on land. The dark blue area to the right side indicates a subarea of sea breeze perturbation.

deficient. The turbulence usually develops in the urban boundary layer below cloud base where cloud
macrophysics/microphysics is not involved [Simpson, 1969; Reible et al., 1993]. Thus, moist processes are
excluded in the CFD model. Previous studies noted that humidity may produce some buoyancy inducing
cloud streets especially over ocean where latent heat is strong [e.g., Liu et al., 2004]. Nevertheless, we may
expect slightly stronger turbulence if the influence of humidity is considered.
2.2. Establishment of Idealized Numerical Experiment of SBF
The notational convention adopted throughout the study is that x, y, and z denote the streamwise, spanwise,
and vertical coordinates, and u, v, and w denote the streamwise, spanwise, and vertical components of wind
velocity, respectively. Figure 1 shows the setup of the model domain and block array. The total grid number
was 1800 (x) × 600 (y) × 80 (z). The horizontal resolution was uniformly 10 m, while the vertical interval was
5 m in the lowest 30 layers and increases from 6 to 55 m aloft, giving a domain size of 18 km (x) × 6 km
(y) × 1675 m (z). Such a domain is comparable to a typical city scale and represents the mesoscale features
of local weather. The grid spacing is fine enough to resolve small-scale turbulent flow and fine-scale struc-
tures of the SBF [Atkinson and Zhang, 1996]. To depict land-sea contrast, the domain was divided into two
subdomains, land (12 km (x) × 6 km (y)) and sea (6 km (x) × 6 km (y)). As shown later, such a setting ensured
that the turbulent flow over land reached a quasi steady state within one spin-up time of 15 min, which was
approximately two eddy turnover times. The results of sensitive experiment with longer sea domain also
showed little differences, indicating that the length of sea domain was long enough for turbulent flow on
urban area to be fully developed before interacting with the SBF.
An array of 50 × 25 rectangular blocks was aligned regularly on land. Each block was 120 m in both length and
width with a height of 80 m. This mimics a group of buildings rather than individual buildings. The street
spacing between two blocks was 120 m, giving an area density of 25%. Thus, 144 grid points were used to
resolve an individual block or the canyon between two blocks. We noted that the turbulent flow around
the block such as the canyon vortex can be explicitly resolved (figure not shown, but similar to Figure 5a
in Inagaki et al. [2012]). The roughness length z0 for the CFD was estimated as ~6.4 m using the morphometric
method of Macdonald et al. [1998]. And z0 for the mean wind profile was further estimated as ~10.8 m based

on the simulated u  lnz plot in surface layer (figure not shown). Here a uniform array of blocks was applied to
facilitate the analysis, which is similar to other idealized simulations [e.g., Inagaki et al., 2012]. This provides an
understanding of the structures of turbulent flow and its interaction with SBF over urban-like surface from a
fundamental perspective. We note that uneven buildings or land use may impose a spatial contrast of
dynamic and thermal forcing, which can influence the features of turbulent flow [e.g., Chen et al., 2015b;
Giometto et al., 2016]. Further introducing complex surface with variable block’s shape and/or height is a
potential work in the future [Jurelionis and Bouris, 2016].

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5300


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

In the typical SBF condition, the atmo-


spheric field (F) should at least consist
of a background prevailing wind (FPW),
local-scale perturbation of sea breeze
circulation (FSB), and small-scale turbu-
lent flow (FTurb) [Miller et al., 2003]:

F ¼ F PW þ F SB þ F Turb : (1)

We designed an initial field including


these three major components. The
ambient wind is one of the critical fac-
tors that determines sea breeze devel-
opment [Adams, 1997]. Here we
focused on a type of sea breeze with
FPW blowing offshore. This would result
in a SBF with strong frontal lifting, while
the inland extent of the front is hin-
dered [Finkele et al., 1995; Miller et al.,
2003]. A moderate westerly wind
(u < 6 m/s below z = 800 m) was applied
throughout the domain (Figure 2a).
Besides, a similar potential temperature
(θ) profile was applied over both land
and sea, except for a slight difference
of ~2 K near the surface (Figure 2b).
The potential temperature remained
almost unchanged below z = 700 m to
depict the mixed layer and increases
with a rate of ~0.015 K/m from z = 700
Figure 2. Initial profiles of (a) streamwise velocity (u, m/s), (b) potential
temperature (θ, K) used in the simulation; (c) horizontal plane to 1000 m to create the capping inver-
(z = 100 m) and (d) vertical x-z cross section (y = 3 km) of initially sion [Inagaki et al., 2012]. The initial
perturbed θ fields. Small solid boxes in Figures 2c and 2d indicate the height of CBL was set at z = ~800 m.
block array. Dashed box in Figure 2c outlines computational area in The setting gives a moderately unstable
Figure 7.
internal boundary layer on land but a
relatively stable one over sea.
FSB is generally characterized by a relatively cool air mass and onshore flow within a shallow internal boundary
layer [Miller et al., 2003]. Such features can be superimposed on ambient conditions to represent an advancing
sea breeze. We assigned the sea breeze a temperature perturbation of 5 K and onshore wind of 2 m/s (red
lines in Figures 2a and 2b). These values were prescribed in a subarea of 1.2 km (x) × 6 km (y) around the east
lateral boundary. A narrow transition area of 240 m (x) × 6 km (y) ahead of the sea breeze perturbation was set
to represent the frontal zone. Thus, the offshore ambient flow and sea breeze onshore flow were converged at
x = ~16.7 km. Experiment shows that the simulation results were insensitive to the width of the transition area.
The perturbation was largely confined to a 400 m layer to represent the onshore flow of shallow sea breeze
circulation based on observations [Miller et al., 2003; Iwai et al., 2008; Sweeney et al., 2014].
For bottom boundary condition, prescribed values of temperature were given on land (310.15 K) and sea
(295.15 K) surfaces. A land-air temperature difference of ~8 K was employed to represent the surface heating
in the day. Following Kawai et al. [2007], it is interpreted as a magnitude of ~0.028 K·m/s in surface sensible
heat flux (SHF) on land. Based on surface SHF, the characteristic vertical velocity (w*) can be estimated as
~0.9 m/s. The radiation boundary condition was applied for all four lateral boundaries; the area at least
1 km away from lateral boundaries was used for analysis.
Various-scale turbulent flow is one of the important features in the CBL [Inagaki and Kanda, 2010]. To gener-
ate the turbulent flow, CFD model with smooth inflow often takes hours of integration [Mirocha et al., 2010].

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5301


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 3. Horizontal planes of vertical velocity (w, m/s) at z = 100 m (left column) and z = 306 m (middle column), and θ (K)
at z = 150 m (right column) for six simulation times: (a–c) for fcst = 5 min, (d–f) for fcst = 10 min, (g–i) for fcst = 15 min,
(j–l) for fcst = 20 min, (m–o) for fcst = 30 min, and (p–r) for fcst = 40 min. Small boxes show the block array.

In our simulation, the atmosphere conditions are not uniform and the time scale of the sea breeze intrusion at
coastal city is also limited. Thus, it is inappropriate to generate turbulent flow using a long-time integration,
and it demands the methods to accelerate the turbulent growth. Some studies employed the turbulence
recycling method to generate turbulent inflow data for the realistic boundary conditions [Lund et al., 1998;
Kataoka and Mizuno, 2002; Park et al., 2015]. This needs a recycling subdomain added in the upstream of
the main domain to gain the turbulent signals, which might lead to more computational burden. Muñoz-
Esparza et al. [2014] proposed a simple and efficient method, in which cell perturbation is added at the
inflow boundaries to accelerate the development of turbulent flow. Here we used a similar method and
applied cell perturbation over the full domain. An amplitude of ±0.5 K was randomly perturbed in the θ
field, which was added to a cell of 8 × 8 grids for each level below 350 m (Figures 2c and 2d). The thermal
perturbations with spatial variations were continuously added to four boundaries during the simulation. As
shown later in section 3.1, this modified perturbation method was more effective than previous schemes
in generating turbulent flow in the CBL.
We focused on SBF inland intrusion in a limited urban-scale domain, ignoring the built-up processes of sea
breeze circulation at larger scale and its evolution early in the day, which have been addressed in many
studies [e.g., Antonelli and Rotunno, 2007]. Thus, the model was run for 1 h during the SBF’s landfall, with a
time step of 1 s. With perturbed initiations of the SBF and random disturbances, the onshore-propagating
SBF at the sea surface and turbulent flow within a typical daytime CBL on land can be incorporated into
numerical experiments.

3. General Characteristics of Turbulent Flow and Sea Breeze Intrusion


3.1. Development and Evolution of Turbulent Flow
We first examined the general features of turbulent flow and sea breeze intrusion simulated by the CFD
model. Figure 3 shows the temporal evolution of vertical velocity (w) and temperature on horizontal planes.
Weak disturbances of vertical motion emerged in urban area in the first 5 min of forecast (fcst = 5 min)

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5302


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

(Figures 3a and 3b). They behaved like


small cells and were scattered randomly
across the entire urban area. There was
a similar pattern in the temperature
field. Such random disturbances indi-
cate a rapid initiation of the small-scale
turbulent flow [Mitsumoto et al., 1983],
a result of the perturbed temperature
field. At fcst = 10 min, turbulent struc-
tures of vertical motion became more
intense both near the surface and in
Figure 4. Energy spectrum (m /s ) at different times at z = 306 m. The the mixed layer (Figures 3d and 3e).
3 2

analysis is carried out by obtaining the spectrum at each spanwise The updrafts in mixed layer tended to
section and then calculating the streamwise average from x = 2 to 12 km. form regular streaky turbulent struc-
5/3
Dashed line indicates K inertial subrange slope. tures [Inagaki and Kanda, 2010; Park
and Baik, 2013]. The streamwise extent
of these streaks was several hundred meters (comparable to the mixed layer depth zi = ~700 m), which
was much smaller than the typical roll convection [e.g., Weckwerth et al., 1997]. Such streaks can be referred
to as transitional structures between rolls and cells [Park and Baik, 2014; Salesky et al., 2017]. Such structures
mainly formed on land, while some may extend offshore ~500 m under the steering of ambient westerly flow.
At fcst = 15 min, turbulent structures of vertical motion in the surface layer has not changed much, while
those in the mixed layer grew larger and stronger (Figures 3g and 3h). It appears that the organization of tur-
bulence coherent structure continued in the mixed layer. At this time step, offshore streaks began to interact
with the advancing SBF at x = 13 km (Figure 3h). At fcst = 20 min, spatial aspects and turbulence magnitudes
(Figures 3j and 3k) were nearly the same as those in Figures 3g and 3h. This indicates that prefrontal turbulent
flow became fully developed and reached a quasi-static condition, as discussed later. Meanwhile, three-
dimensional structures of updrafts became distinct at the frontal zone. In contrast, turbulent flow was quite
weak behind the SBF. After the SBF passes, the stratification became stable, which may help the accumula-
tion of pollutants [Nakane and Sasano, 1986; Thompson et al., 2007]. The sea breeze layer may become mod-
erately unstable due to the strong surface heating at noon and gradually favor the turbulent growth [Iwai
et al., 2008; Chen et al., 2015a].
To further show turbulent activity, we examined the energy spectrum on land in the mixed layer. Figure 4
shows that at the beginning (fcst = 0–5 min), turbulent energy was mostly in a high-frequency range (wave
number ~ 100), corresponding to active small-scale turbulent flow (wavelength ~ 50 m). This turbulence
quickly established a cascading rate resembling the K5/3 inertial subrange slope. During fcst = 3–11 min,
the energy gradually reached a magnitude of ~1 m3/s2. At fcst = 11–17 min, the wavelength of the energy
peak shifted from ~130 m to ~250 m. This indicates the formation of turbulence coherent structure that
was confined within the CBL top (zi = ~700 m) [e.g., Park and Baik, 2013]. The energy spectrum showed little
change after fcst = 15 min, suggesting the fully developed turbulent flow. As in Muñoz-Esparza et al. [2014],
the duration for the simulation to reach a quasi steady state is used as the criterion to assess the efficiency of
perturbation method. Thus, it takes ~15 min in our simulation to generate the fully developed turbulent flow
with cell perturbation employed in full domain. This is much shorter than the duration of a few hours using
the lateral perturbation method [Muñoz-Esparza et al., 2014]. Therefore, the modified perturbation method in
the present study is very efficient.

3.2. Evolution of Sea Breeze Intrusion


In this subsection, we analyzed the detailed inland intrusion of the SBF, with emphasis on its propagation
speed. First we identified the maximum zonal gradient of temperature (|Δθ/Δx|max) at z = 150 m to define
the SBF location. Based on SBF location variation with time, we estimated the inland penetration speed of
the front (USBF). The results with other definitions of the SBF’s location such as zero zonal wind speed show
little difference. Figure 5a shows that this penetration can be classified into three stages. The SBF was mainly
over the sea in the first stage (fcst = 0–11 min) and had a relatively high penetration speed (~4.5 m/s) and
weak temperature gradient (~0.8 K/100 m). In the second stage (fcst = 12–18 min) when the SBF made

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5303


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

landfall, its speed and temperature


gradient dramatically changed. In the
third stage (after fcst = 19 min), the
SBF advanced on land slowly
(~2.5 m/s) with a very strong tempera-
ture gradient (~3.5 K/100 m).
Following Simpson [1969] and Simpson
and Britter [1980], the penetration speed
of the sea breeze can be estimated by:
rffiffiffiffiffiffi
Δθ
USBF ≈k gd  0:59ug ; (2)
θ

where k is a constant (0.62), g is gravita-


tional acceleration, d is the height of the
sea breeze head, θ and Δθ are the tem-
perature in warm air and the tempera-
ture difference between warm air and
sea breeze cool marine air, and ug is
the cross shore geostrophic wind com-
ponent. The first term on the right side
Figure 5. Temporal evolution of spanwise average location (blue line) is the rate at which the sea breeze
and inland penetration speed of SBF (USBF, green line, m/s) in (a) the
advances inland as a gravity current.
original run and (b) the test run with half the block height. SBF location
was defined as position of the maximum absolute value of temperature The second term is the retardation of
gradient (|Δθ/Δx|max, red line, K/100 m) along x axis at z = 150 m. the opposing offshore wind. Based on
the initial conditions set in section 2,
we took d as ~400 m (the height of sea
breeze perturbation), θ and Δθ as 301.15 K and ~5 K (the values at z = ~200 m in Figure 2b), and ug as
~2 m/s (the velocity at z = ~200 m in Figure 2a), giving the theoretical speed USBF as ~3.8 m/s. This speed
is within the reasonable range 1–7 m/s reported in other studies [Simpson et al., 1977; Stull, 1988; Finkele,
1998]. It is also close to the simulated one (~4.5 m/s) at the sea surface in Figure 5a, indicating that the
combined effect of the ideal gravity current and ambient wind accounts for the inland progression of the
SBF on the sea.
The inland penetration speed and temperature gradient changed substantially in the SBF landfall stage
(fcst = 11–18 min). At fcst = 11–13 min, the speed rapidly increased from 5 to 13 m/s. This sudden change
may be attributed to the first contact of the SBF with offshore-advected turbulent flow (Figures 3e and 3f).
At fcst = 14–18 min, the speed decreased rapidly from 13 m/s to near 0, with a great increase of temperature
gradient from 0.5 to 4.9 K/100 m. In this stage, the SBF made landfall and presented a direct contrast to the
warm air on land. The blockage of coastal buildings may temporarily hinder the SBF from penetrating inland,
which is similar to the barrier effect of topography on sea breeze intrusion [Ookuchi et al., 1978].
After traversing the coastline (fcst = 19–60 min), the SBF penetration speed became nearly constant on land
(2.5 m/s), which is ~56% of that over the sea (~4.5 m/s). This deceleration can be attributed partly to the
enhanced drag effect of urban surfaces [Bornstein and Thompson, 1981]. Urban-induced turbulence and
frontal fine-scale structures (as shown later), acting as a friction-like force, are also likely to slow SBF pene-
tration [Sha et al., 1991; Ogawa et al., 2003]. Along with the deceleration, both the temperature gradient
(Figure 5a) and the vertical velocity (figure not shown) slightly declined, showing a gradual SBF weakening
on land. To quantify the impact of surface roughness on the deceleration of SBF, we conducted one simula-
tion with half block height. The temporal evolution of inland penetration speed of the SBF is given in
Figure 5b. It is shown that the SBF can penetrate inland faster than that in the original run. For instance,
the SBF arrived at x = ~6.5 km at fcst = 50 min, in contrast to that at x = ~8 km in the original run.
Besides, the USBF after landfall (fcst = 17 min) was ~3 m/s, which was larger than that of ~2.5 m/s in the
original run. This change of SBF intrusion speed may be informative to those mesoscale modelers to
parameterize the urban-drag effect.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5304


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 6. (a, b) The x-z cross sections (y = 3 km) of Richardson number (Ri, shades) and θ (red lines, 0.5 K interval), and (c, d)
SBF location (defined by isotherms of 299–300 K) at four levels (z = 50, 100, 125, and 150 m). In Figures 6a and 6c,
fcst = 10 min, and in Figures 6a and 6c, fcst = 20 min. Dashed box in Figure 6b is region of fluctuating structures mentioned
in the text.

We further examined the difference of SBF before and after landfall. Figure 6a shows that the front was rela-
tively flat with slope π/12 at the sea surface, somewhat smaller than theoretical slopes [e.g., Von Kármán,
1940]. Over the smooth water surface (low Reynolds number), the sea breeze head was shallow (~250 m high)
and no overhanging nose appeared [Simpson, 1982]. Figure 6c shows that the frontal line and vertical struc-
tures of the SBF were barely disturbed, and so they had a quasi two-dimensional aspect. For comparison, after
SBF landfall, the sea breeze head rose to 350 m (Figure 6b), owing to the frictional drag of underlying urban
surfaces [Thompson et al., 2007]. An overhanging structure (nose) of temperature was seen at z = 160 m.
Heterogeneous structures along the frontline at x = 10.5–11.5 km were distinct at all four given levels
(z = 50, 100, 125, and 150 m) in Figure 6d. The marine air mainly penetrates inland along streamwise streets,
while it was blocked at the building array. Disturbances of the frontal line were also seen in the layer above
the buildings. These inhomogeneous structures reveal an instantaneous impact of complex urban buildings
on local flow and temperature [Eliasson et al., 2006].
Another interesting feature is wave-like structures in the upper portion of the sea breeze head (dashed box in
Figure 6b), analogous to KHBs [Sha et al., 1991; Rao et al., 1999; Thompson et al., 2007]. To understand the
formation of these structures, we estimated the Richardson number (Ri), i.e., the ratio of static stability to
the kinetic energy of shear:
 
g ∂θ ∂u 2
Ri ¼ = ; (3)
θ ∂z ∂z

where u is the horizontal wind component. KHBs mainly form as vortices in the presence of Kelvin-Helmholtz
instability (0 < Ri < ~2.5) [Sha et al., 1991]. Such a condition did occur at the upper boundary of the sea
breeze head (Figure 6b), where low-level onshore flow and return flow aloft generated strong wind shear
[Nielsen, 1992]. KHBs can be also clearly seen in the streamlines (figure not shown). In contrast, the wave-like
structures were much weaker over the sea with a relatively shallow sea breeze head (Figure 6a). This
difference suggests that the block array and turbulent flow over land may play a role in activating the
KHBs, potentially by modulating the height of the sea breeze head [Thompson et al., 2007].

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5305


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

4. Structures and Statistics of


Turbulent Flow
4.1. Evolution of Turbulent
Kinetic Energy
We examined the prefrontal turbulent
flow in detail as it has a great impact
on the sea breeze intrusion. Figure 7
shows time-height variations of turbu-
lent kinetic energy (TKE) estimated
from the dashed box in Figure 2c. The
small-scale turbulent flow initially
2 2 formed at the rooftop (z = 80 m) in
Figure 7. Temporal evolution of turbulent kinetic energy (TKE, m /s )
averaged over dashed box in Figure 2c. Dotted line indicates the the first 5 min and then increased
transient decrease band of TKE mentioned in text. quickly in the mixed layer during
fcst = 5–15 min. After fcst = 15 min,
TKE became nearly constant at a magni-
tude of 0.3 m2/s2, suggesting a quasi steady state of turbulent activity (Figures 3c–3f). Fully developed
turbulent flow was mostly confined to the CBL. At fcst = 25 min, some overshooting updrafts [Liu and
Sang, 2011] associated with strong convection were obvious at z = ~600 m and quickly weakened in the
next 5 min.
Figure 7 also shows that just before the arrival of SBF, a temporarily reduced band of TKE (0.3 m2/s2) was tilted
upward with time ahead of the inclined SBF, as marked by the dotted line. The TKE then increased ~10 times
to a magnitude of 3.0 m2/s2 at the SBF. This suggests that the turbulent flow was strongly activated upon
interacting with the SBF. In turn, the SBF was transformed into three-dimensional structures, with an elevated
sea breeze head and strong localized updrafts (Figures 6b and 6d). Frontal updrafts are believed to remove
near-surface pollutants by a so-called “translocation” effect [Lyons et al., 1995]. The enhanced updrafts initiate
moist convection at the frontal area and sometimes form clouds [Wakimoto and Atkins, 1994]. After SBF
passage (after fcst = 40 min), TKE decreased greatly because of a stable stratification in the cool marine air.
As a result, the pollutants might be concentrated in a shallow density current behind the SBF, degrading
air quality [Thompson et al., 2007].

4.2. Structures of Turbulent Flow


Figure 8a shows a snapshot of the prefrontal turbulent flow over land. The upward (downward) motions
coincided well with low-speed (high-speed) streaks. Such relatively slow (u < 1 m/s) updrafts (e.g., (x,
z) = (4.3 km, 300 m)) were characteristic structures mostly appearing above urban surface as in Park and
Baik [2014]. Some updrafts were vertically extensive in the CBL, while others at rooftop were relatively small.

Figure 8. (a) Snapshot of w (contours, solid line for 0.5 m/s and dashed line for 0.5 m/s) and u (shading, m/s) along
2 2
x = 9850 m at fcst = 20 min. (b) Spectral density (shading, m /s ) of u along y axis averaged from x = 9850 to 11,100 m
(black line indicates the 5% significance level).

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5306


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 9. Composite structure of the streaks in the mixed layer. (a, d) Horizontal planes (z = 306 m); (b, e) y’-z cross sections
(x’ = 0 m); and (c, f) x’-z cross sections (y’ = 0 m). Shading in Figures 9a–9c and Figures 9d–9f indicate w (m/s) and
u’ (m/s), respectively. Contours in Figures 9a–9c indicate θ’ (solid line for 0.2 K and dashed for 0.2 K).

The spectral density field of vertical velocity (Figure 8b) shows that the turbulent flow in the CBL had a span-
wise wavelength of ~500 m, which were organized by the structure and stratification of CBL itself. At the roof-
top, the spanwise wavelength was ~250 m, which was twice the horizontal size of the blocks [Park and
Baik, 2014].
Over the urban area, there were a number of streaky structures in the CBL as shown in Figure 3. A composite
of these streaks was made to examine their dynamical and thermal structures. We identified the convective
updrafts with a streaky feature larger than 500 m (x) × 150 m (y) at z = 306 m. The composite of the selected
streaks was made with respect to their centroids, which are defined as the grid points of the maximum
updrafts of individual streaks. The original coordinates (x, y, z) is transformed to a new one (x’, y’, z) based
on the center locations of the composited turbulent flow (the same for composited figures below).
Figure 9a shows that the turbulent flow was characterized by a streamwise streak of updrafts with down-
drafts at both sides. Such streaks are the typical form of boundary layer convection in the condition of
obvious ambient wind [Weckwerth et al., 1997; Inagaki and Kanda, 2010; Castillo et al., 2011; Park and Baik,
2013, 2014]. The strongest updrafts (~1.1 m/s) and downdrafts (0.5 m/s) were in the middle of the CBL
(Figure 9b). The updrafts were confined in a narrow width in y plane, while the downdrafts were more exten-
sive, producing an asymmetric feature [Raasch and Harbusch, 2001]. The updrafts corresponded to the warm
and low-speed perturbations (cf. Figures 9a–9c and 9d–9f), which therefore played a role in the vertical trans-
port of momentum and heat in the CBL [Park and Baik, 2014]. The updrafts were clearly accompanied by a
pair of counterrotating circulations on both sides (Figures 9d and 9e). Such a secondary circulation may gain
sensible heat (and/or latent heat over ocean) from the low levels to refuel the updrafts. These dynamics and
thermodynamics are similar to those of the roll-type convection revealed in previous studies [Iwai et al., 2008;
Chen et al., 2015a].
Figure 9a also shows that the positive temperature perturbation was centered at (x’, y’) = (0, 150), which
was located upstream of the maximum updrafts ((x’, y’) = (0, 0)). Such a leading feature of temperature
perturbation probably indicates an active role of thermodynamics (buoyancy) in forcing the streaks. The
warm perturbation had an inclined vertical structure (Figure 9c), probably owing to the vertical shear in
streamwise flow. This inclined structure, characterized by relatively warm air below and cool air aloft,
may produce an unstable stratification in the western portion of the eastward moving streaks in the
ambient westerly flow.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5307


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

0 3 2 3
Figure 10. Profiles of horizontal average TKE budget terms over land at fcst = 18 min: (a) buoyancy (g=θ θ w 0 , 10 m /s ),
3 2 3 0 3 2 3
(b) wind shear (u0 w 0 ∂u= ∂z, 10 m /s ), (c) turbulent transport ( ∂w 0 E = ∂z, 10 m /s ), and (d) pressure
0 0 3 2 3 0
transport (1=ρ ∂w p = ∂z, 10 m /s ); and four corresponding flux terms, i.e., (e) turbulent heat flux (θ w 0 , K·m/s),
0 0 2 2 0 0 3 3 0 0
(f) turbulent momentum flux (u w , m /s ), (g) energy flux (w E , m /s ), and (h) pressure flux (w p , hPa m/s).

4.3. Turbulent Kinetic Energy Budget


To estimate the mechanisms forcing the turbulent flow, we analyzed the generation rates of TKE [Weckwerth
et al., 1997]. Following LeMone [1973] and Moeng and Sullivan [1994], the TKE budget is given as
  0 0
∂ g 0 0 0 0 ∂u 0 0 ∂v ∂w 0 E 1 ∂w p
E¼ θw  uw þvw    D; (4)
∂t θ ∂z ∂z ∂z ρ ∂z
0
where overbars denote a regional mean, E represents TKE, θ w 0 is turbulent heat flux, and u0 w 0 and v 0 w 0 are
turbulent momentum fluxes. The first term on the right side is the generation of TKE by buoyancy and repre-
sents thermal forcing. The second term is the production by wind shear and represents dynamic forcing. The
third and fourth terms are TKE productions by turbulence and pressure transport, respectively. These two
terms redistribute TKE vertically, but they do not generate TKE because their vertical integration in the CBL
is 0 [Lenschow et al., 1980; Lin, 2000]. The last term is TKE dissipation.
0
The budget of TKE generation and four associated flux terms (θ w 0 , u0 w 0 , w 0 E, w 0 p0 ) were estimated in Figure 10.
0 0 0 0 0 0 0
The analysis was carried out by obtaining the flux terms (θ w , u w , w E, w p ) at each point and then taking the
horizontal average over the land portion of the domain. In the canopy layer, only the interconnected air space
(grid points outside building blocks) was used to calculate the horizontal average. Figure 10a shows that the
TKE production by buoyancy maximized in the middle of the CBL (z = ~300 m), where the streaky structures
of updrafts were strongest (Figure 9). Buoyancy declined to near 0 at the surface and CBL top. This indicates
that the buoyancy contributed greatly to the formation of streaks in the CBL. To show more detail of the
0
buoyancy, the associated turbulent heat flux (θ w 0 ) was decomposed into four quadrant events [Sullivan
0 0 0 0 0
et al., 1998]: warm air rising (Q1: θþ w þ ), cool air rising (Q2: θ w þ ), cool air sinking (Q3: θ w 0 ), and warm
0
air sinking (Q4: θ w 0 ). Figure 10e shows that the events of Q1 (0.013 K·m/s) and Q3 (0.012 K·m/s) dominated
in the CBL, and they produce most of the net heat flux (0.022 K·m/s) [e.g., Nakao et al., 2016]. They are asso-
ciated with the thermally driven updrafts/downdrafts of streaks. In the canopy layer (z = ~80 m), the sinking
cool air (Q3 = 0.16 K m/s) explains most of the heat flux, likely because of the cavity eddy at windward walls

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5308


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

[Inagaki et al., 2012]. It is partly compensated by the rising cool air (Q2 = 0.08 K m/s) of the return branch of the
cavity eddy (figure not shown).
Figure 10b shows that TKE production by shear was strong in a narrow layer near the rooftop (z = ~80 m). This
high production rate was clearly caused by the urban-induced strong wind shear ( ∂u= ∂z = ~0.028 s1) with
obvious momentum flux (Figure 10f). Like the heat flux, the turbulent momentum flux was also divided into
0 0 0
four quadrant events [Raupach, 1981]: outward interaction (Q1: uþ w þ ), ejection (Q2: u0 w þ ), inward interaction
0
(Q3: u0 w 0 ), and sweep (Q4: uþ w 0 ). Figure 10f shows that the net momentum flux was mainly due to Q2 and
Q4, which contributed downward transport in both the urban canopy layer and CBL [Foster et al., 2006]. The
largest momentum fluxes were in the middle of the CBL (for both Q2 and Q4), which are associated with the
streaks. Besides, the ejection (Q2 = 0.019 K·m/s) was slightly larger than sweep (Q4 = 0.018 K·m/s), which
was consistent with previous studies over urban surface [e.g., Kanda et al., 2004; Park et al., 2015]. Given this
profile of momentum flux and an increasing wind shear ( ∂u= ∂z = ~0.012 s1) with height near the CBL top
[Park and Baik, 2014], another high production rate of TKE by shear occurred in the upper portion of the CBL
(z = ~450 m).
Figure 10c shows that TKE production by turbulent transport was mainly positive (negative) in the upper
(middle and lower) part of the CBL. This profile is similar to that reported in Chen et al. [2015b], except for
an additional local fluctuation near the rooftop. As indicated in equation (4), a local change of TKE can induce
the convergence or divergence in vertical transport of TKE. Figure 10g shows that the flux of vertical (horizon-
tal) kinetic energy, i.e., w 0 3 (w 0 u0 2 þ w 0 v 0 2 ), was largest at z = ~350 m (~450 m). The vertical gradient of trans-
port produced a local TKE gain (loss) in the upper (lower) part of the CBL. There was a secondary peak of
kinetic energy flux near the rooftop (Figure 10g), likely owing to urban-induced ejection and sweep
(Figure 10f). This explains the local disturbance of TKE near the rooftop (Figure 10c). However, Figure 10d
shows that the production term by pressure transport results in a TKE gain (loss) near the surface and in
the entrainment zone (middle of the CBL). This is related to the vertical configuration of pressure perturbation
and vertical motion (Figure 10h).
Figures 10a and 10b show that integration of the shear term throughout the CBL was comparable to that of
the buoyancy term, indicating that both shear and buoyancy effects are important to TKE generation. The
buoyancy term accounted for most TKE in the middle CBL, whereas the shear term dominated its upper
and lower portions. It is recognized that thermal forcing is essential for the occurrence of convection
[Atkinson and Zhang, 1996], whereas shear forcing may have a key role in regulating its shape
[Weckwerth et al., 1997]. The shear effect may help to concentrate warm air at the lowest level into low-
speed streaks [Kanda et al., 2004]. Figures 10c and 10d also show that both transport terms were positive
at the CBL top or entrainment zone. Their combined effect, despite an offset of negative buoyancy, had
considerable magnitude to supply the turbulent energy at the CBL top [Lenschow et al., 1980; Lin, 2000].
Such a vertical transport of TKE may be responsible for the upward growth of turbulent flow [Chen
et al., 2015b].

5. Impacts of Prefrontal Streaky Turbulent Structures on SBF


In this section, we examined how the prefrontal turbulent flow in the CBL modulated the onshore-
propagating SBF. We focused on the streaky turbulent structures induced by urban surfaces and their
impacts on the SBF three-dimensional structures. Since a number of individual streaks may merge with the
SBF (Figure 3), their composite was constructed to show general features of turbulence-SBF interaction.
Both the locations of SBF and streaks were identified as in sections 3.2 and 4.2. Based on the different dis-
tances between the streaks and SBF, four consecutive stages of the turbulence-SBF collision (approach, con-
tact, merger, and restoration), with an interval of ~2 min, were composited to show the interaction process.
In the approach stage, the prefrontal turbulent flow made contact with the SBF. The streamlines of streaks
appeared as columnar vortices aligned streamwise in both sides of the updrafts (Figure 11a). Such streaky
structures of prefrontal turbulence were detailed in section 4.2 (Figures 9, 12a, and 12e). Before contacting
the streaks, the SBF showed a nearly uniform band of upward motion (Figure 12a). The linear boundaries
of cold air (black solid line) and onshore wind speed (blue lines) were also undisturbed. The slope of the fron-
tal surface was very flat with relatively weak updrafts centered at z = ~200 m (Figure 12e).

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5309


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 11. Composite three-dimensional flows of four continuous stages during interaction between streaky structures
and SBF. Lines are streamlines (direction from yellow to black), and shading represents w (m/s, red color for upward
motions and blue for downward). The origin of new coordinate (x’, y’, z) represents the central location of composited
turbulent flow in approach stage.

In the contact and merger stages, the inland-propagating SBF encountered the streaks and dramatically
intensified. Figures 11b and 11c show SBF three-dimensional structures with localized updrafts at the inter-
section zone where the streaks collided with the front. Figure 12g shows that ascending motion was locally
enhanced to ~1.0 m/s over the sea breeze head in the merger stage. As a result, the maximum turbulent
momentum and heat fluxes were estimated at 0.04 m2/s2 and 0.01 K·m/s, respectively, 3–4 times those
(0.01 m2/s2 and 0.003 K·m/s) in the approaching stage. This suggests that the merger of streaks with the
SBF can greatly strengthen turbulent fluxes at the front than those at other stages.
Figures 12b and 12c show that the perturbations of temperature and onshore wind speed were convex at the
intersection zone, and there were two concavities at both sides. Such irregular features are somewhat
analogous to the lobes and clefts that are usually observed at the advancing edge of a gravity current
[e.g., Simpson, 1972]. The wind perturbations can be attributed to the superposition of low-speed streaks
on the SBF. The enhanced onshore flow may produce a local intrusion of cold air behind the SBF, leading
to an elevated sea breeze head like the lobes (Figures 12f and 12g). The temperature gradient increased
from ~0.28 K/100 m in contact stage to ~0.63 K/100 m in merging stage, suggesting local frontogenesis.
As a result, the updrafts were locally enhanced over the sea breeze head as the warm air mass overran
the lobes (Figures 12g and 12k). These lobes/clefts in the frontal line were mainly over the urban surface
(Figures 6b and 6d). While encountering various streaks, the advancing SBF was modified along the front
into complex structures. Such a process may lead to an uneven passage of the SBF across the city
[Thompson et al., 2007].
It is well recognized that local circulations of both streaky structures and the sea breeze head possess hori-
zontal vorticity [Miller et al., 2003]. We further examined the turbulence-SBF interaction from the vorticity per-
spective. Regarding the flow pattern of streaks (Figure 9f), positive vorticity directed northward (ζ y > 0) was
estimated downstream of the updrafts, which is clearly shown by the hatched area in Figure 12e. Another
zone of positive vorticity was associated with overturning circulation at the sea breeze head. Figures 12f
and 12g show that the above two types of vorticity merged to strengthen the horizontal vorticity at the
SBF during the contact and merger stages. This is expected to induce an elevated sea breeze head and
strengthen frontal updrafts.
The streaks were also characterized by two counterrotating circulations at both flanks of the updrafts
(Figure 11a), indicating horizontal vorticity along the x axis as shown by the vortex lines in Figure 13a.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5310


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 12. Composited two-dimensional structures of four continuous stages: (a, e, i) for approach stage, (b, f, j) for contact
stage, (c, g, k) for merger stage, and (d, h, l) for restoration stage. Figures 12a–12d are horizontal planes (z = 201 m): shading
indicates w (m/s), black solid line represents isotherm of 300 K at z = 150 m (indicating leftmost boundaries of cold pool
behind SBF), and blue lines denote u’ (deviations from horizontal means, solid for positive and dashed for negative) at
z = 150 m. Figures 12e–12h are x’-z cross sections (y’ = 0 m): shading represents w (m/s), black lines are isotherms of
299–300 K (0.2 K interval), and blue hatched area denotes horizontal vorticity along y axis (ζ y) with a value larger than
3 1
+2.5 × 10 s . Figures 12i–12l are y’-z cross sections along black dashed line in Figures 12a–12d: shading represents w
(m/s). The origin of new coordinate (x’, y’, z) represents the central location of composited turbulent flow in approach stage.

When the streaks encountered the SBF, the vortex lines were tilted upward because of strong lifting ahead of
the front (Figures 13b and 13c) [Dailey and Fovell, 1999]. The modified vorticity generated local disturbances
in the onshore winds, as shown by the black vectors. Such a vorticity-tilting effect of frontal updrafts on hor-
izontal winds may induce inhomogeneous frontal lines associated with lobes/clefts. At microscale, this pro-
duces strong variability along the front, resulting in three-dimensional features of the SBF (Figures 3k, 3n,
and 3q).
After the interaction, the SBF gradually returned to a quasi two-dimensional feature with a nearly uniform
frontal line (Figure 12d), relatively weak frontal updrafts (Figures 12h and 12l), and less disturbed vorticity
(Figures 11d and 13d). The restoration process of the SBF was complete within a few minutes, as the indivi-
dual streaks had a smaller scale than the front. By interacting with a number of streaks, the SBF was continu-
ously modified over urban surfaces and had persistently changing three-dimensional structures. Although
the SBF was locally enhanced (weakened) by updrafts (downdrafts) of streaky structures at the intersection
zone, its overall intensity weakened after the interaction [Ogawa et al., 2003]. Such a weakening is shown
by the decline of temperature gradient on land in Figure 5 (red line).
Previous studies have pointed out that the deceleration of sea breeze intrusion on land may be associated
with the friction drag of rough surface and the friction effect of convective mixing at frontal zone [Simpson
et al., 1977; Physick, 1980; Sha et al., 1991]. Because of the diurnal cycle of turbulent activity, the SBF
penetration may slow down at midday and regain its speed in late afternoon. The vorticity-related processes
of streaks revealed in this section are thought to strengthen the convective mixing, which may partially
account for the SBF’s deceleration after landfall. Since turbulent activity changed little on land during a

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5311


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Figure 13. Same as Figure 11, but for three-dimensional vortex lines (direction from yellow to black). Black vectors
represent air motions associated with vortex lines.

short simulation period, it may explain a nearly constant speed of the SBF penetration after fcst = 19 min
(green line in Figure 5). Further analysis is needed in the future to clarify a detailed impact of turbulence
on sea breeze penetration.

6. Conclusions
The detailed dynamics and thermodynamics of the interaction between SBF and turbulent flow over land,
especially urban-like surfaces, have not been well understood. In the present study, an idealized simulation
using a building-resolving CFD model was performed to explicitly characterize urban-induced turbulent flow
and the fine structures of the SBF. Both the statistics of turbulent flow over land and its impacts on the SBF
were analyzed.
The turbulent flow over urban-like surfaces was adequately developed in the first 15 min by the CFD model
with perturbed initial conditions. The urban-induced streaky turbulent structures in the mixed layer elon-
gated to form low- (high-) speed streaks associated with rising warm (sinking cool) air, indicating strong
vertical transport of heat and momentum flux. The corresponding secondary circulation tended to absorb
the heat from the urban-like surfaces to sustain its own development. The TKE budget shows that the shear
term was large in rooftop level because of strong wind shear with the presence of the block array. The
buoyancy term dominated TKE production in the mixed layer because of warm air rising and cool air sinking.
These results highlight the crucial role of the combination of thermal and dynamic forcing in driving and
sustaining the streaky structures.
The numerical simulation also revealed a strong interaction between the streaks and sea breeze intrusion.
Before reaching land, the SBF had a quasi two-dimensional structure with low sea breeze head. Upon making
landfall, the SBF intensified, with an elevated sea breeze head. KHBs were likely to form in the upper bound-
ary of sea breeze head, owing to strengthened updrafts. Correspondingly, the penetration speed was slowed
from 4.5 to 2.5 m/s, because of frictional drag of the land surface and urban-induced turbulence. When
advancing on land, the SBF interacted with the streaks and developed strong three-dimensional structures.
The merging of streak- and SBF-related vorticity (spanwise) produced an elevated sea breeze head and
strengthened frontal updrafts. The tilting of streak-related vorticity (streamwise) by frontal updrafts led to
a local disturbance in the onshore winds, associated with lobes/clefts. In this manner, the SBF was unevenly
modified, causing it to exhibit irregular features. After the interaction, overall intensity of the SBF gradually
weakened, although its penetration speed appeared undisturbed.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5312


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

The results of this work are informative for understanding coastal urban weather during typical sea breeze
conditions. They focus on an idealized simulation of a daytime stratification with moderately convective con-
dition in the presence of a regular block array. Thus, more experiments dealing with various stratifications
[Antonelli and Rotunno, 2007] and geometries [Roth, 2000; Crosman and Horel, 2010] are required to identify
the basic mechanisms of complex surfaces effects on SBF evolution. For instance, the contrast in stability over
land and sea surfaces might impact the strength and structures of boundary layer convection [e.g.,
Weckwerth et al., 1997], which should lead to different features of turbulence-SBF interaction. Another experi-
ment with a neutral stability condition shows that the turbulent flow over urban surface was much weaker,
the frontal line of SBF was less altered, and SBF’s three-dimensional structures were less obvious. It seems that
the larger the land-sea contrast in stability occurs, the stronger the turbulence-SBF interaction becomes. In
addition, the combination of a reliable mesoscale model and CFD model gives the potential to simulate a rea-
listic case of large coastal cities with tall buildings [e.g., Chen et al., 2015a, 2015b]. Such studies are warranted
to shed further light on the fine-scale dynamics of land-sea interaction and its impacts on local weather.

Acknowledgments References
This research was supported by the
Special Program for Applied Research Adams, E. (1997), Four ways to win the sea breeze game, Sail. World, March, 44–49.
on Super Computation of the Allen, J. R. L. (1971), Mixing at turbidity current heads, and its geological implications, J. Sediment. Petrol., 41(1), 97–113.
NSFC-Guangdong Joint Fund (second Antonelli, M., and R. Rotunno (2007), Large-eddy simulation of the onset of the sea breeze, J. Atmos. Sci., 64, 4445–4457.
phase). The authors also thank the Atkins, N. T., R. M. Wakimoto, and T. M. Weckwerth (1995), Observations of the sea-breeze front during CaPE. Part II: Dual-Doppler and aircraft
National Supercomputing Center in analysis, Mon. Weather Rev., 123(4), 944–969.
Guangzhou for providing the Atkinson, B. W., and J. W. Zhang (1996), Mesoscale shallow convection in the atmosphere, Rev. Geophys., 34(4), 403–431, doi:10.1029/
supercomputing resource. All data 96RG02623.
provided in this study may be obtained Berry, R. E., and L. R. Taylor (1968), High-altitude migration of aphids in maritime and continental climates, J. Anim. Ecol., 37(3), 713–722.
by contacting the corresponding author Bornstein, R. D., and W. T. Thompson (1981), Effects of frictionally retarded sea breeze and synoptic frontal passages on sulfur dioxide
at chenguixing@mail.sysu.edu.cn. concentrations in New York City, J. Appl. Meteorol., 20(8), 843–858.
Britter, R. E., and S. R. Hanna (2003), Flow and dispersion in urban areas, Annu. Rev. Fluid Mech., 35, 469–496.
Castillo, M. C., A. Inagaki, and M. Kanda (2011), The effects of inner- and outer-layer turbulence in a convective boundary layer on the
near-neutral inertial sublayer over an urban-like surface, Boundary Layer Meteorol., 140(3), 453–469.
Charba, J. (1974), Application of gravity current model to analysis of squall-line gust front, Mon. Weather Rev., 102, 140–156.
Chen, G., X. Zhu, W. Sha, T. Iwasaki, H. Seko, K. Saito, H. Iwai, and S. Ishii (2015a), Toward improved forecasts of sea-breeze horizontal
convective rolls at super high resolutions. Part I: Configuration and verification of a down-scaling simulation system (DS3), Mon. Weather
Rev., 143(5), 1849–1872.
Chen, G., X. Zhu, W. Sha, T. Iwasaki, H. Seko, K. Saito, H. Iwai, and S. Ishii (2015b), Toward improved forecasts of sea-breeze horizontal
convective rolls at super high resolutions. Part II: The impacts of land use and buildings, Mon. Weather Rev., 143(5), 1873–1894.
Collier, C. G. (2006), The impact of urban areas on weather, Q. J. R. Meteorol. Soc., 132(614), 1–25.
Crosman, E. T., and J. D. Horel (2010), Sea and lake breezes: A review of numerical studies, Boundary Layer Meteorol., 137, 1–29.
Dailey, P. S., and R. G. Fovell (1999), Numerical simulation of the interaction between the sea-breeze front and horizontal convective rolls.
Part I: Offshore ambient flow, Mon. Weather Rev., 127(5), 858–878.
Dailey, P. S., and R. G. Fovell (2001), Numerical simulation of the interaction between the sea-breeze front and horizontal convective rolls.
Part II: Alongshore ambient flow, Mon. Weather Rev., 129(8), 2057–2072.
Dias, M. A. F. S., and A. J. Machado (1997), The role of local circulations in summertime convective development and nocturnal fog in Sâo
Paulo, Brazil, Boundary Layer Meteorol., 82(2), 135–157.
Eliasson, I., B. Offerle, C. S. B. Grimmond, and S. Lindqvist (2006), Wind fields and turbulence statistics in an urban street canyon, Atmos.
Environ., 40, 1–16.
Finkele, K. (1998), Inland and offshore propagation speeds of a sea breeze from simulations and measurements, Boundary Layer Meteorol.,
87(2), 307–329.
Finkele, K., J. M. Hacker, R. A. D. Byron-Scott, and H. Kraus (1995), A complete sea-breeze circulation cell derived from aircraft observations,
Boundary Layer Meteorol., 73(3), 299–317.
Finnigan, J. J., R. H. Shaw, and E. G. Patton (2009), Turbulent structure above a vegetation canopy, J. Fluid Mech., 637, 387–424.
Foster, P. C., F. Vianey, P. Drobinski, and P. Carlotti (2006), Near-surface coherent structures and the vertical momentum flux in a large-eddy
simulation of the neutrally-stratified boundary layer, Boundary Layer Meteorol., 120(2), 229–255.
Giometto, M. G., A. Christen, C. Meneveau, J. Fang, M. Krafczyk, and M. B. Parlange (2016), Spatial characteristics of roughness sublayer mean
flow and turbulence over a realistic urban surface, Boundary Layer Meteorol., 160(3), 425–452.
Inagaki, A., and M. Kanda (2010), Organized structure of active turbulence over an array of cubes within the logarithmic layer of atmospheric
flow, Boundary Layer Meteorol., 135(2), 209–228.
Inagaki, A., M. C. L. Castillo, Y. Yamashita, M. Kanda, and H. Takimoto (2012), Large-eddy simulation of coherent flow structures within a
cubical canopy, Boundary Layer Meteorol., 142, 207–222.
Iwai, H., et al. (2008), Dual-Doppler lidar observation of horizontal convective rolls and near-surface streaks, Geophys. Res. Lett., 35, L14808,
doi:10.1029/2008GL034571.
Jurelionis, A., and D. Bouris (2016), Impact of urban morphology on infiltration-induced building energy consumption, Energies, 9(3), 177.
Kanda, M. (2006), Large-eddy simulations on the effects of surface geometry of building arrays on turbulent organized structures, Boundary
Layer Meteorol., 118(1), 151–168.
Kanda, M., R. Moriwaki, and F. Kasamatsu (2004), Large eddy simulation of turbulent organized structure within and above explicitly resolved
cubic arrays, Boundary Layer Meteorol., 112, 343–368.
Kataoka, H., and M. Mizuno (2002), Numerical flow computation around aeroelastic 3D square cylinder using inflow turbulence, Wind Struct.,
5, 379–392.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5313


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Kawai, T., M. Kanda, K. Narita, and A. Hagishima (2007), Validation of a numerical model for urban energy-exchange using outdoor
scale-model measurements, Int. J. Climatol., 27, 1931–1942.
Lapworth, A. (2000), Observation of atmospheric density currents using a tethered balloon-borne turbulence probe system, Q. J. R. Meteorol.
Soc., 126(529), 2811–2850.
LeMone, M. A. (1973), The structure and dynamics of horizontal roll vortices in the planetary boundary layer, J. Atmos. Sci., 30, 1077–1091.
Lenschow, D. H., J. C. Wyngaard, and W. T. Pennell (1980), Mean-field and second-moment budgets in a baroclinic, convective boundary
layer, J. Atmos. Sci., 37, 1313–1326.
Leonard, B. P. (1979), A stable and accurate convection modeling procedure based on quadratic upstream interpolation, Comput. Methods
Appl. Mech. Eng., 19(1), 59–98.
Lilly, D. K. (1962), On the numerical simulation of buoyant convection, Tellus, Ser. A Dyn. Meteorol. Oceanogr., 2, 148–172.
Lin, C. L. (2000), Local pressure-transport structure in a convective atmospheric boundary layer, Phys. Fluids, 12, 1112–1128.
Liu, A. Q., G. W. K. Moore, K. Tsuboki, and I. A. Renfrew (2004), A high-resolution simulation of convective roll clouds during a cold-air
outbreak, Geophys. Res. Lett., 31, L03101, doi:10.1029/2003GL018530.
Liu, H., and J. Sang (2011), Numerical simulation of roll vortices in the convective boundary layer, Adv. Atmos. Sci., 28, 477–482.
Lund, T. S., X. Wu, and K. D. Squires (1998), Generation of turbulent inflow data for spatially-developing boundary layer simulations,
J. Comput. Phys., 140, 233–258.
Lyons, W. A., R. A. Pielke, C. J. Tremback, R. L. Walko, D. A. Moon, and C. S. Keen (1995), Modeling impacts of mesoscale vertical motions upon
coastal zone air pollution dispersion, Atmos. Environ., 29(2), 283–301.
Macdonald, R. W., R. F. Griffiths, and D. J. Hall (1998), An improved method for the estimation of surface roughness of obstacle arrays, Atmos.
Environ., 32(11), 1857–1864.
Masselink, G., and C. B. Pattiaratchi (1998), The effect of sea breeze on beach morphology, surf zone hydrodynamics and sediment
resuspension, Mar. Geol., 146, 115–135.
Miao, S., and F. Chen (2008), Formation of horizontal convective rolls in urban areas, Atmos. Res., 89(3), 298–304.
Miller, S. T. K., B. D. Keim, R. W. Talbot, and H. Mao (2003), Sea breeze: Structure, forecasting, and impacts, Rev. Geophys., 41(3), 1011,
doi:10.1029/2003RG000124.
Mirocha, J. D., J. K. Lundquist, and B. Kosović (2010), Implementation of a nonlinear subfilter turbulence stress model for large-eddy
simulation in the advanced research WRF model, Mon. Weather Rev., 138(11), 4212–4228.
Mitsumoto, S., H. Ueda, and H. Ozoe (1983), A laboratory experiment on the dynamics of the land and sea breezes, J. Atmos. Sci., 40,
1228–1240.
Moeng, C.-H., and P. P. Sullivan (1994), A comparison of shear- and buoyancy-driven planetary boundary layer flows, J. Atmos. Sci., 51,
999–1022.
Muñoz-Esparza, D., B. Kosovi, J. Mirocha, and J. V. Beeck (2014), Bridging the transition from mesoscale to microscale turbulence in numerical
weather prediction models, Boundary Layer Meteorol., 163(3), 409–440.
Nakane, H., and Y. Sasano (1986), Structure of a sea-breeze front revealed by scanning lidar observation, J. Meteorol. Soc. Jpn., 64(5), 787–792.
Nakao, K., Y. Hattori, and H. Suto (2016), Numerical investigation of a spatially developing turbulent natural convection boundary layer along
a vertical heated plate, Int. J. Heat Fluid Flow, doi:10.1016/j.ijheatfluidflow.2016.09.006.
Nielsen, J. W. (1992), In situ observations of Kelvin–Helmholtz waves along a frontal inversion, J. Atmos. Sci., 49, 369–386.
Ogawa, S., W. Sha, T. Iwasaki, and Z. Wang (2003), A numerical study on the interaction of a sea-breeze front with convective cells in the
daytime boundary layer, J. Meteorol. Soc. Jpn., 81(4), 635–651.
Ookuchi, Y., M. Uryu, and R. Sawada (1978), A numerical study on the effects of mountain on the land and sea breezes, J. Meteorol. Soc. Jpn.,
56, 368–385.
Papanastasiou, D. K., D. Melas, and I. Lissaridis (2010), Study of wind field under sea breeze conditions; an application of WRF model, Atmos.
Res., 98, 102–117.
Park, S.-B., and J.-J. Baik (2013), A large-eddy simulation study of thermal effects on turbulence coherent structures in and above a building
array, J. Appl. Meteorol. Climatol., 52(6), 1348–1365.
Park, S.-B., and J.-J. Baik (2014), Large-eddy simulations of convective boundary layers over flat and urbanlike surfaces, J. Atmos. Sci., 71(5),
1880–1892.
Park, S.-B., J.-J. Baik, and B.-S. Han (2015), Large-eddy simulation of turbulent flow in a densely built-up urban area, Environ. Fluid Mech., 15,
235–250.
Patankar, S. V. (1980), Numerical Heat Transfer and Fluid Flow, pp. 80–100, Hemisphere Corp., Washington, D. C.
Physick, W. L. (1980), Numerical experiments on the inland penetration of the sea breeze, Q. J. R. Meteorol. Soc., 106, 735–746.
Raasch, S., and G. Harbusch (2001), An analysis of secondary circulations and their effects caused by small-scale surface inhomogeneities
using large-eddy simulation, Boundary Layer Meteorol., 101, 31–59.
Rao, P. A., H. E. Fuelberg, and K. K. Droegemeir (1999), High resolution modeling of the Cape Canaveral area land–water circulations and
associated features, Mon. Weather Rev., 127, 1808–1821.
Raupach, M. R. (1981), Conditional statistics of Reynolds stress in rough-wall and smooth-wall turbulent boundary layers, J. Fluid Mech., 108,
363–382.
Raynor, G. S. (1974), Mesoscale transport and dispersion of airborne pollens, J. Appl. Meteorol., 13(1), 87–95.
Reible, D. D., J. E. Simpson, and P. F. Linden (1993), The sea breeze and gravity-current frontogenesis, Q. J. R. Meteorol. Soc., 119(509), 1–16.
Roth, M. (2000), Review of atmospheric turbulence over cities, Q. J. R. Meteorol. Soc., 126(564), 941–990.
Ryu, Y. H., J. Smith, M. L. Baeck, and E. Bouzeid (2016), The influence of land surface heterogeneities on heavy convective rainfall in the
Baltimore–Washington metropolitan area, Mon. Weather Rev., 144(2), 553–573.
Salesky, S. T., M. Chamecki, and E. Bou-Zeid (2017), On the nature of the transition between roll and cellular organization in the convective
boundary layer, Boundary Layer Meteorol., 163, 41–68.
Sha, W. (2002), Design of the dynamics core for a new-generation numerical model of the local meteorology [in Japanese], Kaiyo Mon., 2,
107–112.
Sha, W. (2008), Local meteorological model based on LES over the Cartesian coordinate and complex surface [in Japanese], in Meteorological
Research Note, vol. 219, edited by E. Y. Fujiyoshi, pp. 21–26, Meteorol. Soc. of Jpn. Press, Tokyo.
Sha, W., T. Kawamura, and H. Ueda (1991), A numerical study on sea land breezes as a gravity current: Kelvin-Helmholtz billows and inland
penetration of the sea-breeze front, J. Atmos. Sci., 48(14), 1649–1665.
Simpson, J. E. (1969), A comparison between laboratory and atmospheric density currents, Q. J. R. Meteorol. Soc., 95(406), 758–765.
Simpson, J. E. (1972), Effects of the lower boundary on the head of a gravity current, J. Fluid Mech., 53, 759–768.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5314


Journal of Geophysical Research: Atmospheres 10.1002/2016JD026247

Simpson, J. E. (1982), Gravity currents in the laboratory, atmosphere, and ocean, Annu. Rev. Fluid Mech., 14, 213–234.
Simpson, J. E. (1994), Sea Breeze and Local Winds, pp. 7–19, Cambridge Univ. Press, Cambridge, U. K.
Simpson, J. E., and R. E. Britter (1980), A laboratory model of an atmospheric mesofront, Q. J. R. Meteorol. Soc., 106, 485–500.
Simpson, J. E., D. A. Mansfield, and J. R. Milford (1977), Inland penetration of sea-breeze fronts, Q. J. R. Meteorol. Soc., 103, 47–76.
Smagorinsky, J. (1963), General circulation experiments with the primitive equations, Mon. Weather Rev., 91, 99–164.
Stephan, K., H. Kraus, C. Evenz, and J. M. Hacker (1999), Sea-breeze front variations in space and time, Meteorol. Atmos. Phys., 70, 81–95.
Stull, R. B. (1988), An Introduction to Boundary Layer Meteorology, pp. 587–620, Kluwer Acad., Boston, Mass.
Sullivan, P. P., C. H. Moeng, B. Stevens, D. H. Lenschow, and S. D. Mayor (1998), Structure of the entrainment zone capping the convective
atmospheric boundary layer, J. Atmos. Sci., 55(19), 3042–3064.
Sweeney, J. K., J. M. Chagnon, and S. L. Gray (2014), A case study of sea breeze blocking regulated by sea surface temperature along the
English south coast, Atmos. Chem. Phys., 14, 4409–4418.
Thompson, W. T., T. Holt, and J. Pullen (2007), Investigation of a sea breeze front in an urban environment, Q. J. R. Meteorol. Soc., 133, 579–594.
Von Kármán, T. (1940), The engineer grapples with nonlinear problems, Bull. Am. Meteorol. Soc., 46, 615–683.
Wakimoto, R. M., and N. T. Atkins (1994), Observations of the sea-breeze front during CaPE. Part I: Single-Doppler, satellite and cloud
photogrammetry analysis, Mon. Weather Rev., 122(6), 1092–1114.
Weckwerth, T. M., J. W. Wilson, R. M. Wakimoto, and N. A. Crook (1997), Horizontal convective rolls: Determining the environmental condi-
tions supporting their existence and characteristics, Mon. Weather Rev., 125(4), 505–526.

JIANG ET AL. INTERACTION OF TURBULENCE AND SEA BREEZE 5315

You might also like