1

Environmental Fluid Dynamics:

A Brief Introduction
1.1 Introduction .................................................................................................................................3
1.2 Examples .......................................................................................................................................5
1.3 Principles.......................................................................................................................................7
Equations of Motion and Dimensionless Variables • Dimensionless Numbers
and Scales • Waves • Instabilities • Turbulence
1.4 Methods of Analysis ....................................................................................................................9
Numerical Prediction • Process Modeling • Dimensional Arguments and Similarity
Theories • Field Studies
1.5 Illustrations.................................................................................................................................13
Harindra Joseph Environmental Simulations • Process Modeling
Shermal Fernando 1.6 Major Challenges .......................................................................................................................14
University of Notre Dame References...............................................................................................................................................16

1.1 Introduction the solid earth, containing sediments, rocks, and soil), biosphere
(regions that sustain life, covering the lithosphere, portions of
The planet Earth is just one of the 1022 planets in the universe, the hydrosphere, and the lower atmosphere), and cryosphere (ice,
yet it is singularly unique, as it is the only planet known to snow, glaciers, and permafrost; it holds ∼2% of the water sup-
maintain the precise conditions to sustain life. First, the earth is ply, covering 5.7% of the earth’s surface). Common constituents,
covered by an atmosphere, which is held in place by sufficiently transformations, interactions, and feedbacks intertwine these
large gravitational attraction due to its large mass (6 × 1024 kg) spheres. For example, melting of the cryosphere would raise
and radius (average 6371 km). Second, the atmosphere con- the sea level by tens of meters, shut off thermohaline circula-
tains essential gases such as oxygen (∼21% by volume), nitrogen tion of oceans, and drastically change weather patterns and heat
(∼78%), and carbon dioxide (∼0.03%), in requisite amounts for distribution in the atmosphere. Oceans, the major component
breathing, biogeochemical cycling, and for maintaining a nar- of the hydrosphere, contain ∼97% of the world’s water supply.
row temperature range conducive for life. Third, the average dis- They cover 75% of the earth surface with an average depth of
tance from the earth to the sun (1.49 × 1011 m) is such that the 3.7 km, maximum depth of 11 km (Mariana trench), and a mass
temperatures on the earth are neither extremely hot nor cold so of 1.4 × 1021 kg (Gill 1982). Oceans are also the dominant reser-
that the water can be in liquid state, and the nearly circular (vari- voirs of carbon and hence are central to the understanding of
ation <3.5%) orbit of the earth around the sun ensures smaller climate variability (Sarmiento and Gruber 2002). The intersec-
seasonal temperature swings. Fourth, the rate of rotation and tion of the biosphere with other major climatic spheres, with
thermal and gravitational forcing of the earth are just right for lithosphere as the platform, is called the ecosphere or simply
the atmosphere, oceans, and the earth’s interior to be in a quasi put—our Environment (Figure 1.1). The physical and biological
stable or an equilibrium state. This state of the planet, known as (humans, plants, animals, and microorganisms) aspects of the
the climate, prima facie is in a delicate balance dictated by the ecosphere are studied in Ecology (Odum 1971).
ultimate forces of universe. In 2008, the world crossed an important milestone, in that
Five major components are believed to determine the climate: most people now live in urban rather than rural areas (Science
the atmosphere (∼100 km blanket of gasses around the earth), 2011). Concentrated human (anthropogenic) activities exert
hydrosphere (encompassing all water bodies from aquifers to stresses on the environment to the extent of causing apprecia-
clouds, e.g., lakes, streams, and oceans), lithosphere (upper part of ble change of climate, and humans have now entered the epoch

Handbook of Environmental Fluid Dynamics, Volume One, edited by Harindra Joseph Shermal Fernando. © 2013 CRC Press/Taylor & Francis Group, LLC.
ISBN: 978-1-4398-1669-1.

3
4 Handbook of Environmental Fluid Dynamics, Volume One

Atmosphere

Ecosphere
LH

Environment
(ecosystem)

Biosphere LV
Hydrosphere

Su Ice Cryosphere
bs
u rf a
ce fl o w

Lithosphere

FIGURE 1.1 The components of the earth’s climate system and spatial scales of an ecosystem. The influence of the polisphere propagates upscale
via flow and dispersion, which, in turn, penetrates down the scales. Quantification of urban influence on itself and global scales remains a challenge.

of anthropocene. An additional climatic sphere, the polisphere, TABLE 1.2 Oceanic Scales
therefore needs to be introduced to represent anthropogenic Motion Scale Space Scale LH Timescale T
stressors of urban systems (Fernando 2008). In environmental
studies, appropriate spatial and temporal scales are selected, Climate/global 103–104 km Centuries/months
Regional 100–1000 km Months to years
based on a patch of ecological interest (ecosystem), thus sim-
Basin 1000 km Weeks to years
plifying the task of delving into the vast global environment in
Meso ∼10 km Days to weeks
piecemeal manner. Fluid motions on the earth, a communica-
Fine 1–100 m Minutes to hours
tion medium between all spheres, span over 10 decades of scales,
Small <1 m Seconds to minutes
with strong interactions among the scales.
Kolmogorov 1 mm Seconds
An ecosystem with horizontal and vertical spatial scales of
LH and LV, respectively, is illustrated in Figure 1.1. Although a
based on equations of motion broadly falls within the area
given spatial scale is associated with a myriad of time scales, it
of geophysical fluid dynamics (GFD), which encompasses the
is customary to link it with a dominant energy-bearing time
realms of (physical) oceanography, limnology, meteorology,
scale T. Tables 1.1 and 1.2 attempt to identify such scales for
and solid-earth geophysics. Following the pioneering work
oceanic and atmospheric flows. The study of natural flows
of Laplace on tidal forcing, GFD has traditionally dealt with
TABLE 1.1 Atmospheric Scales motions directly affected by earth’s rotation, and hence with the
low Rossby number regime. Meso- and larger scales are in this
Motion Scale Space Scale LH Timescale T
domain, with strong nonlinear interactions among themselves
Global/climate ∼104 km Centuries/months through processes such as barotropic and baroclinic instabili-
Synoptic (regional) 103–104 km Days to months ties and with smaller scales via shear and convective instabili-
Meso ∼102 km Hours to days ties. An example is the link between global climate change and
Urban 10–100 km Minutes to days the reduction of the diurnal temperature range (DTR) of the
Neighborhood 1–10 km Hours to days
atmospheric boundary layer (Karl et al. 1993). Conversely,
City (CBD) 1 km Minutes to hours
changes to the local boundary layer, for example, due to land
Atmospheric boundary layer 1–100 m Minutes to hours
use change, can have climatic impacts (Fernando et al. 2012a).
Street Canyon scale 10–100 m Seconds to hours
Fluid motions akin to anthropogenic activities are studied in
Personal cloud ∼1 m Seconds to minutes
environmental fluid dynamics (EFD), which are usually (but not
Kolmogorov ∼1 mm Seconds
always) unaffected by earth’s rotation. In other words, penetration
Sub Kolmogorov 100 nm–1 mm <Seconds
of human influence to environmental motions is at the nub of EFD
Environmental Fluid Dynamics 5

and thus, many subtopics have emerged within EFD: urban

fluid dynamics (Fernando et al. 2001), geological fluid dynamics
(Huppert 2000), and ecological fluid dynamics (e.g., Goodman
and Robinson 2008; Hondzo and Wuest 2009), to name a few.
A conspicuous application of EFD is flow and pollution disper-
sion in cities (Fernando et al. 2001, 2010). While external pres-
sure gradients and thermal circulation driven by diurnal solar
forcing are responsible for the approaching background meso-
scale flow, which is studied in meteorology, flow adjustments and
enhanced turbulence within building canyons, flow introduced
by differential heating of building surfaces, heat and turbulence
in automobile wakes, effects of HVAC systems, influence of irri-
gation and plant canopies, dust entrainment, and distribution
of pollution from industrial sources, all of which have anthro-
pogenic bent, are in the EFD domain. On the other hand, such
topics as stable boundary layer cut across meteorology, oceanog-
FIGURE 1.2 A natural ecosystem, characterized by clean air, water, raphy, and EFD (Fernando and Weil 2010). In all, the divisions
terrain, flora, and fauna. They are increasingly intruded by factitious are more subjective, and vast overlaps of GFD and its off springs
elements. The challenge is to maintain ecosystem services with mini- are evident—which reflects from the coverage of this volume.
mum impacts on the environment—known as sustainable development.

(Figure 1.2), thus demarcating it from GFD, albeit nebulously. The

1.2 Examples
pragmatic connotations of EFD to humans and ecosystems per-
meate into quality of life (e.g., safety, health, and comfort), engi- Some applications of EFD are illustrated in Figure 1.3 and dis-
neering designs, environmental planning and management, and cussed in the following. In the urban context, historically cities
hence policymaking. EFD problems tend to be multidisciplinary, have emerged surrounding water resources and hence are mostly

Downstream regional transport

Regional flow aloft (synoptic ~1000 km)

(Pollutants from e
upstream) Fumigation
slop )
ment Up p e km
Entrain Pollen slo 00
w n e ~1
er

ion biogenics o
R iv

Convect D cal
emissions os
es
(m
Surface flow Industry

pe
UHI
n

Slo
Heat + moisture tio
r llu ort
ive Po nsp
R tr a
S ea Brown cloud ay
bree Red tide (photochemical) ew
ze
e
/fr

Vegetation
ds
Roa

Up/dow
n

ilit aves
Heat trapping valle y w
inds

ies
ins nal w
, w xing

tab
er
ds

)
ges

Plankton
ave

Int
i
lan
(w ace m

Mixed land use

sur

(zooplankton, Paved
Wet

surf aces
ind
rm

food chain)
rf
Su
Sto

Construction Dust Unpaved

Op Freeway:
pro tical Dust
.L.

pert heat, pollution

Currents/ ies
ic B
g

noise, turbulence
xin

eddies
nth
Mi

Be

Agriculture/farming
Fr
on
ts

Sed
ime
nt dar y
Airport: Intrusions Boun g
pollution, mixin
noise,
turbulence

FIGURE 1.3 (See color insert.) A schematic of anthropogenically modified environs, and pertinent f luid motions. Coasts, water
bodies,  cities, and terrain add Sisyphean complexity to modeling and prediction. (Adapted from Fernando, H.J.S. et al., Phys. Fluids,
22, 051301, 2010.)
6 Handbook of Environmental Fluid Dynamics, Volume One

located in complex terrain (Fernando 2010). Water and air qual- dispersion from industries (e.g., stacks, metalworking, and min-
ity modeling for complex topography is challenging, which ing), transportation (e.g., roadways, airports, boating), and
become even more problematic for coastal areas because of agricultural and construction sites. On larger scales, cities are
ocean–land–atmosphere interactions. More than 50% of urban pollution islands (area sources), contributing to background
areas are coastal, and currently only about 15% of the U.S. coast- concentrations of air pollutants and greenhouse gases (Crutzen
lines have been developed compared with 5% along the interior 2004). Rising convective plumes loft pollutants to regional flow,
waterways; the former is expected to grow to 25% by 2025. Tides, which transports pollutants downstream. Vertical mixing and
salinity intrusions, wetlands, mangroves, estuaries, bays, beach downward diffusion of pollutants (known as fumigation) occur
erosion, coral cover, scour, shoaling, waves, storm surges, tsu- when the regional flow is perturbed by turbulence induced by
namis on the ocean side, and sea/land breezes and urban heat the surface roughness and UHI of an urban area located down-
island (UHI) on the atmosphere side are some overarching flow stream. The turbulence in a city is related to its morphometry
issues that impact the livability of coastal cities. These features (Britter and Hanna 2003). Pollution plumes of megacities are
are tied to economic activities such as mineral and oil explora- a topic of considerable international interest, alongside climate
tion, effluent outfalls, harbors, navigation, fisheries, and recre- change (Molina et al. 2010). Aviation, vehicles, and industry gen-
ation as well as urban security (Fernando et al. 2005; Paola et al. erate noise pollution, which is strongly dependent on the local
2011). Engineering of rivers with levees, canals, and diversions mean flow, shear, and stratification (Ovenden et al. 2009).
has bearing on natural processes such as river meandering, sedi- The change of hydrologic cycle due to global change impacts
mentation, wetlands, wildlife, and land formation. water availability and quality. Thus, sustainable water manage-
Lakes provide water for recreation, agriculture, and human ment techniques are being actively pursued, most notably novel
consumption, and quality of water therein depends on physical designs for hydraulic dams, aquifer recharge, water treatment,
and biogeochemical processes (Imberger and Patterson 1990). and flood control. Similarly, understanding of circulation,
Flow processes in limnology include seiching in lake basins, eddies, turbulence, instabilities, mixing, stratification, nutri-
strait flows, small-scale mixing in the presence of weak turbu- ents, and biological production (Klein and Lapeyre 2008) are
lence (where molecular diffusion becomes important), sinking processes linked to ecosystem management. Excessive primary
and trapping of microorganisms, and selective withdrawal. production (eutrophication) that leads to anoxia (oxygen defi-
The quality and optical properties of coastal waters and lakes ciency) and release of toxins that affects aquatic life (Smith 1998)
are determined by boundary and thermocline mixing, waves remain crucial issues.
and turbulence, air–sea interactions, run off, subsurface pro- Embedded in urban areas are scales of considerable practical
cesses, vegetation, and nutrients. Fundamental studies of flow– interest. The neighborhood scales (Table 1.1) coincide with eco-
structure interaction, biological and nutrient cycling, two-phase logical patches of heightened air pollutants, which are selected as
flows, gravity and turbidity currents, debris flow, sediment the grid size of mesoscale air quality computations. The central
transport, morphodynamics, and turbulence cut across afore- business district (CBD) scale is of interest to security agencies
mentioned application areas. and fast responders, as CBDs are a target of choice for chemi-
The nonlinear response of anthropogenic forcing on the envi- cal, biological, and radiological (CBR) releases. Enhanced spa-
ronment leads to the so-called emergent phenomena; examples tial inhomogeneity of CBDs requires the use of high-resolution
include UHI in the atmosphere and dead zones in water bod- computational fluid dynamics (CFD) models for predictions.
ies. The former arises due to the high heat capacity of urban Street canyons, courtyards, buildings, parks, vehicles, tress,
built elements, which retain heat absorbed during the day for and heat sources all contribute to this heterogeneity. A myriad
a period longer than that of (rural) surroundings. The result is of processes, from mesoscale approach flow to transport of sub-
a warmer city at night compared to the suburbs. The UHI is Kolmogorov scale aerosols to human lungs, are of interest in epi-
abetted by the lack of moisture within the built environment, demiological studies.
whence incoming solar radiation is mostly used to raise the Air, water, and noise pollution are the main considerations
air temperature rather than consuming as latent heat. If irri- in setting environmental protection regulations. The U.S.
gation is present, evaporation may cause the city to be cooler Environmental Protection Agency (EPA) has set National
during daytime (urban oasis island). The UHI increases the Ambient Air Quality Standards (NAAQS) for pollutants con-
evaporation, and this moisture can be transported downstream sidered harmful to public health (primary standards) as well as
to trigger thunderstorms and precipitation. The heat budget of public welfare (secondary standard) such as those damaging to
the upper urban lithosphere is not well established and is an animals and plants, visibility, and buildings. Six principal pol-
area of active research. Studies of natural hazards (e.g., hurri- lutants identified by the EPA, called criteria pollutants, include
canes, storm surges, tsunamis, and pyroclastic flows) as well as ozone, carbon monoxide, lead, nitrogen and sulfur dioxides,
their urban impacts are also emerging areas that underpin risk and particulate matter (of aerodynamic diameter less than
assessment. 10 μm (PM10) and 2.5 μm (PM2.5)). The PM10 is contributed by
The built environment and anthropogenic forcing strongly the turbulent entrainment of dust, whereas combustion sources
influence the urban atmospheric boundary layer (Fernando are mainly responsible for PM2.5. In addition, currently a group
et al. 2010). Processes therein determine the local air pollutant of 188 air pollutants have been identified, which are suspected
Environmental Fluid Dynamics 7

of being health hazards when present even in small amounts The buoyancy conservation becomes
(e.g., benzene). EPA has recommended water quality criteria
for the protection of aquatic life and human health in surface ⎛ b0 ⎞ ∂b ⎛ b0 ⎞ ∂b ⎛ b0 ⎞ ∂b
water, considering the impacts of about 150 pollutants. Spatial ⎜⎝ u N 2T ⎟⎠ ∂t + ⎜⎝ L N 2 ⎟⎠ uβ ∂x + ⎜⎝ L N 2 ⎟⎠ w ∂z + w
V b V β V
and temporal distribution of pollutants in environmental flows,
⎛ ν ⎞ ⎛ b0 ⎞ ⎛ κ ⎞ ⎡ ∂2b ⎛ LH ⎞ ∂2b ⎤
2
their chemical transformation, and human exposure form the
=⎜ ⎢ + ⎥,
basis of environmental epidemiology and toxicology. ⎝ uH LH ⎟⎠ ⎜⎝ LV N 2 ⎟⎠ ⎜⎝ ν ⎟⎠ ⎢ ∂xβ∂xβ ⎜⎝ LV ⎟⎠ ∂z 2 ⎥
⎣ ⎦

1.3 Principles (1.4)n

1.3.1 Equations of Motion and where

Dimensionless Variables N2 = db̄/dz is the background buoyancy (Brünt–Väisälä)
The conservation equations of mass, momentum, energy, and frequency
species form the quantitative basis of EFD. Consider a motion κ is the molecular diffusivity of the stratifying solute
field with typical (horizontal/vertical) velocity (uH, uV), time
(T H, T V), length (LH, LV), pressure (p 0), and buoyancy (b 0)
1.3.2 Dimensionless Numbers and Scales
scales, with a time scale T b for buoyancy fluctuations. The
normalized continuity equation becomes (Fernando and Several familiar nondimensional parameters appear in (1.2)–
Voropayev 2010) (1.4): the Reynolds number; Re = UHLH/v; aspect ratio A = LH/LV;
Rossby number Ro = UH/fLH; Euler number Eu = p0 ρ0 U H2 ; Schmidt
∂uα ∂w ⎛ uV LH ⎞ number Sc = ν/κ; Bulk Richardson number RiB = ∆b0 LV U H2
+ = 0, α = 1,2, (1.1)n (equivalently, an internal Froude number Fri = U H ∆b0 LV with
∂x α ∂z ⎜⎝ uH LV ⎟⎠
RiB = Fri−2 ); and the time scale ratios LH/UHTH and LH/UHTV = UV/
LVTV, which is a Strouhal number St = ωLH/UH, where ω = TH−1 is
where the nondimensional variables are uα* = uα /uH , w* = w/uV, a frequency scale. The RiB in general differs from the gradient
x* = x/LH, and z* = z/LV, z being the vertical coordinate anti- Richardson number Rig  = N2/(∂uα/∂z)2 encountered in hydrody-
parallel to gravitational acceleration −gδi3. The superscript n on namic stability studies, but they can be related. If the Boussinesq
equation number indicates normalized variables, and (*) has approximation is not invoked, the gravity term in (1.3) is associ-
been dropped. When the motions are three dimensional, uVLH/ ated with the Froude number Fr = U H gLV . Each dimensionless
uHLV ∼ 1. Similarly, the momentum equations, to the Boussinesq number signifies the ratio of two forcing terms in the equations
approximation, can be written as of motion.
From (1.2) and (1.3), it appears that at large Re and when
⎛ LH ⎞ ∂ u α ∂uα ∂uα ⎛ fLH ⎞
⎜⎝ u T ⎟⎠ ∂t + uβ ∂x + w ∂z + ⎜⎝ u ⎟⎠ ε αjk ℓ juk LV >> LH Re−1/2 , (1.5)
H H β H

⎛ p ⎞ ∂p ⎛ ν ⎞ ⎡ ∂ 2uα ⎛ LH ⎞ ∂ 2 u α ⎤
2

= − ⎜ 0 2⎟ +⎜ ⎢ + ⎥ the viscous forces can be neglected compared with the iner-

⎝ ρ 0uH ⎠ ∂x α ⎝ LH uH ⎟⎠ ⎢ ∂xβ∂xβ ⎜⎝ LV ⎟⎠ ∂z 2 ⎥ tial forces (i.e., convective acceleration u ⋅ ∇ u ). The scale LV =
⎣ ⎦ ~ ~
LHRe−1/2 is the commonly encountered viscous boundary layer
(1.2)n
scale, and (1.5) is satisfied outside the boundary layer (c.f., the
concept of Reynolds number similarity in Section 1.4). At a dis-
2
⎛ LV ⎞ ⎧⎪ LH ∂w ∂w ∂w ⎪⎫ ⎛ p0 ⎞ ∂p tance sufficiently away from the boundary, the condition LV ∼ LH
⎜⎝ L ⎟⎠ ⎨ u T ∂t + uβ ∂x + w ∂z ⎬ = − ⎜⎝ ρ u 2 ⎟⎠ ∂z is satisfied, whence Re ≫ 1 implies that the viscous effects can be
H ⎩⎪ H V β ⎭⎪ 0 H
neglected compared with the inertial effects. Note that u ⋅ ∇ u is
~ ~
⎛ ν ⎞ ⎡⎛ LV ⎞ ∂2w ∂ 2w ⎤
2
⎛b L ⎞ the only explicit nonlinear term in the equation of motion and
+ ⎜ 0 2 V ⎟ b" + ⎜ ⎟ ⎢⎜ ⎟ + 2 ⎥, (1.3)n
⎝ uH ⎠ ⎝ LH uH ⎠ ⎢⎝ LH ⎠ ∂xβ∂xβ ∂z ⎥ hence key to generating additional (smaller) scales and emergent
⎣ ⎦
phenomena. When Re ≫ 1, the inertial forces need to be bal-
anced by other forces. For example, if the inertial and pressure
where gradient forces balance each other while Eu ∼ O(1), Ro ≫1 and
α, β = 1, 2 RiB ≪ 1, the flow generates three-dimensional turbulence. If the
fℓj is the Coriolis parameter inertial, buoyancy, and pressure force balance is established at
ρ0 is a reference density high Reynolds number, then the flow becomes buoyancy-dom-
b̃ = −g(ρ − ρ0)/ρ0 is the buoyancy, which can be written in inated or stratified turbulence [Eu ∼ O(1), RiB ∼ O(1), Ro ≫ 1].
terms of its mean b̄ and perturbation b values as b̃ = b̄ (z) + b In addition, if Ro ∼ 1 the flow becomes a rotating and stratified
ρ is the density turbulent flow (Riley and Lelong 2000).
8 Handbook of Environmental Fluid Dynamics, Volume One

Turbulent flows consist of a range of length-time scales. 1.3.3 Waves

Although the condition Re = uℓ/v ≫ 1 based on energy bear-
ing scales u and ℓ is satisfied globally, there can be smaller scales When nonlinearities are negligible and a restoring force (e.g.,
where the viscous and inertial effects become of the same order. gravity, buoyancy) is in place countering flow disturbances,
Viscous influence starts appearing at the scale ℓRe−1/2, which is repeating or nearly repeating distortions of physical properties
the well-known Taylor microscale λ of turbulence, as evidenced may occur in space and time that typify wave motions. When
from the earlier discussion. It is generally believed that scales the amplitude A is small, wave motion can be described by a set
much smaller than ℓ are isotropic, and major viscous influence is of linear equations (linear waves). In most cases the fluid (e.g.,
realized at a still smaller length scale ℓK. This, together with cor- viscous) friction is small so as to maintain the repeatability. Any
responding velocity and time scales, uk and tk, respectively, satis- property P (e.g., pressure, surface displacement, etc.) of liner
fies ReK = uKℓK/v = O(1) and tK = ℓk/uk. By hypothesizing that ε (the waves can be described by
kinetic energy dissipation) and v are the governing parameters,
the dissipation (Kolmogorov) scales of turbulence can be written L(P ) = 0, P = Ae iχ , (1.6)
as lK ∼ (v3/ε)1/4, uK ∼ (vε)1/4, and tK = (v/ε)1/2. As such, ε is believed
to occur at Kolmogorov scales, but recent high Reynolds number where χ is the “phase function,” an indicator of the repeatability
observations adumbrate a spectrum roll off at the length scale ∼λ of waves. The wave number k~ and the angular frequency n of
(Figure 1.4). This disparity needs further investigation. waves are defined as k~ = ∇χ = ∂χ/∂xi , n = −∂χ/∂t, whence χ   =
In large-scale geophysical problems, the Coriolis terms ( fUH) k . x − nt. Linear waves do not cause fluid medium to move,
~ ~
dominate the inertial terms U 2H LH or Ro ≪ 1. If the Reynolds when averaged over a single cycle, and thus there is no mass
number is high, a balance may occur between Coriolis and viscous transport. Nevertheless, waves can be efficient transporters of
terms in wall regions, fLH/UH ∼ Re−1(LH/LV)2 or LV ∼ (ν/f)1/2 = LEK, momentum. Waves frequented in natural flows include surface
which holds within the Ekman layer or wall boundary layer. It dif- gravity waves, capillary waves, and internal gravity waves on the
fers from the inertial boundary layer discussed earlier (at Ro ≪ 1) smaller scales, and inertial, inertio-gravity, and planetary waves
( ) on the larger scales (at low Ro). Because of the restoring force,
1/ 2
for which LV ∼ LHRe−1/2. Note that LV LH ~ v fL2H = E1K/2 ,
2
where EK = v fLH is the Ekman number, and at distances larger exchange of energy present in different forms (e.g., potential,
than LH E1K/ 2 from the boundary, the viscous influence can be kinetic, or electromagnetic energy) may occur, thus sustain-
neglected. For the mid-latitude atmosphere, LEK ∼ (0.1/10−4)½ ≈ ing oscillatory motions. For surface waves, this restoring force
30 cm and for oceans LEK ∼ 10 cm based on the molecular viscos- is either the gravity (surface gravity waves) or surface tension
ity. Because of the presence of turbulence, the effective momentum (capillary waves); for internal waves, it is the buoyancy (LH ∼ LV ;
(eddy) diffusivity of natural flows Ke is larger than ν, whence it has LH /U HTV ~ b0 LH U H2 >> 1 or TV ∼ UH/b0); for inertial waves, the
been suggested LEK ≈ π (2Ke/f)1/2 (Kaimal and Finnigan 1994). Coriolis force (TV ∼ f−1); and for magneto-hydrodynamic waves,
the Lorentz force.
10^–1
Linear waves are a mature field of study, but most practical
problems are in the realm of weakly nonlinear and nonlinear
waves (Phillips 1977; Mei 2003). These include wave–current
10^–2 interactions, Langmuir circulation, transitioning of stratified/
rotating  turbulence to internal/inertial waves, differential
f – S(va. 1 m) ((V)2)

10^–3 momentum, heat and mass transfer, as well as sediment trans-

port under waves.

10^–4
1.3.4 Instabilities
10^–5 Environmental motions are typically unstable in that perturba-
tions to basic flow states grow in space and time, and after pass-
ing through a series of intermediate transition states, they break
10^–6
down into turbulence. A stability analysis would determine
0.01 0.10 1.00 10.00 100.00 1000.00
whether a perturbation to the basic flow would decay (stable),
Trex L ~14 m f (Hz) λ = 0.035 m λK = 2.8 mm exponentially amplify (unstable), or remain the same (neutral).
If A is sufficiently small, the governing equations can be lin-
FIGURE 1.4 Variance preserving frequency (f ) spectra of turbulent
earized (linear stability analysis), and periodic solutions of the
velocities in the atmosphere. Note the separation between Taylor  (λ)
and Kolmogorov (lK) scales and the roll off of spectra at λ following form (1.6) are sought. If the wave number or frequency becomes
Kolmogorov f−5/3 regime. The full range of scales was obtained via imaginary, the disturbance would grow exponentially, signify-
hot-film anemometry and a limited range using sonic anemometry. ing instability. The flow is stable when all possible disturbances
(Courtesy of Greg Poulos, National Center for Atmospheric Research, do not grow. The locus that demarcates stable and unstable
Boulder, CO.) states is the marginal stability curve on which the disturbances
Environmental Fluid Dynamics 9

are neutral. In the marginal stability state, a small disturbance former (Ellison) scale representing the vertical scale of evolving
may evolve into a periodic state, for example, cellular patterns turbulence and the latter (buoyancy scale) signifying the onset of
observed in Rayleigh Bernard convection. Sometimes the dis- buoyancy effects. The buoyancy influence sets in at a time scale
turbance growth is slower than exponential, simultaneously tN ∼ LH/uH ∼ N−1, typically tN  ≈  (2  − 4)N−1. The Coriolis forces
exhibiting oscillatory behavior with a specific frequency/wave become important when fLH/uH ∼ O(1) or at a time tc ∼ f−1. Since
number, indicating that the restoring forces are greater than the time scale ratio tC/tN is large for oceans (∼10) and atmosphere
what caused the oscillations. Under these conditions, oscilla- (∼102), during the evolution of environmental turbulence, strati-
tions overshoot past the amplitude of previous excursion; this is fication effects come into play first.
referred to as overstability (Turner 1973). At the onset of fossilization, Lb can be recast in terms of the
Linear stability theory is also a mature topic in fluid mechanics, well-known parameterization ε ~ (w ′ 2 )3/2/ LV as Lb ∼ (ε/N3)1/2,
and both linear and weakly nonlinear stability theories are exten- where LR = (ε/N3)1/2 is the Ozmidov length scale. Since ε can be
sively used in EFD (Chandrasekhar 1961; Drazin and Reid 2004). measured in oceans with a higher reliability than w′2 , it is com-
Some examples include instabilities in stratified shear and vortex mon to use LR in oceanographic studies. Because of its space-
flows, two-phase flows, mushy zones, double-diffusive convec- time patchiness, stratified turbulence is intermittent, and hence,
tion, morphological instabilities, and river meandering, many of the flow can be undersampled during measurements. This is
which are addressed throughout this handbook. In addition to particularly acute in oceanic measurements, because repeated
their amenability to elegant mathematical treatment, the aesthetic profiles conducive for averaging cannot be realized by multiple
splendor of instabilities in nature has attracted the attention of casts owing to platform (ship) drift and intermittency of turbu-
scientists and artists alike. lence (Gregg 1987). Although LV is constrained by Lb, there are

1.3.5 Turbulence
no immediate constraints on horizontal growth ⎛ LH ~ uH dt ⎞ ,
⎝ ⎠ ∫
causing (LV/LH) < 1 characterized by large-scale anisotropy at
Three-dimensional active turbulence can be defined as random large times. This anisotropy is felt even at smaller (dissipation)
motions where inertial-vortex forces are balanced by pressure scales when the buoyancy scale Lb ∼ LR ∼ (ε/N3)1/2 becomes
gradient forces, while the influence of body (buoyancy, Coriolis, on the order of the Kolmogorov scale ℓK or when the (Gibson)
and electromagnetic) and viscous forces are negligible (Gibson parameter G = ε/νN2 drops below a critical value ∼1000. As LV/LH
1991). When the body and inertial forces are on the same order, becomes smaller, turbulence is confined to horizontal sheets or
the turbulence (at those scales) is said to be fossilized. In general, thin layers with strong shear in between.
buoyancy influence is first felt at larger length scales l of turbu- Of great practical importance is the efficiency of mixing in
lence, say when l > Lo, where Lo is a threshold. When Lo ≫ LK, stratified flows, defined in terms of the flux Richardson number
turbulence is active in LK < l < Lo and fossil at larger (>Lo) scales,
leading to active-fossil turbulence. When LK ≈ Lo, no active three- −w ′b′
Rf = , (1.8)
dimensional turbulence can be present and the flow is fully fos-
silized. The hydrodynamic state of a stratified turbulent flow is
(
− u′w ′(∂U/ ∂z ) + v ′w ′(∂V / ∂z ) )
dependent on a variety of factors, including the characteristics where (U,V) are the horizontal mean velocities. Numerous exper-
of stratification (N), the length (ℓ) and rms velocity ( ui′2 ) scales imental and theoretical studies exist on Rf (Peltier and Caulfield
of turbulence, as well as the forcing mechanisms present; here 2003; Ivey et al. 2008), given its direct relationship to “eddy dif-
the primed quantities denote turbulent fluctuations. Some fea- fusivities” that are critical for subgrid closure in environmental
tures of stratified turbulence can be illustrated in terms of gov- models. Direct measurements of Rf have been reported in the
erning Equations 1.1 through 1.4 as follows. atmosphere that show disparities with previous laboratory mea-
If turbulence is imposed on a stratified fluid at time t = 0, for surements, which need to be addressed in future studies.
“active” turbulence to exist, the inertial terms should be of the
leading order, and the flow is expected to follow 1.4 Methods of Analysis
b0 1.4.1 Numerical Prediction
LH ~ LV ~ uH t , uH ~ uV , ~ O(1). (1.7)
LV N 2 The prediction of environmental flows is fraught by complexities
of initial and boundary conditions, buoyancy and Coriolis effects,
This initial growth occurs until the buoyancy forces impede the turbulence, anthropogenic forcing, and a vast range of space-time
growth of turbulence, when the conditions b0 LV uH2 ∼ (LV / LH )2 or scales of processes involved that cover more than 10 decades of
N 2 L2V / uV2 ∼ N 2 L2H / uH2 ∼ O(1) are satisfied. Therefore, the vertical scales. Extant computing technologies cannot resolve all natu-
scale of turbulence at the onset of fossilization is LV ∼ uV/N, which ral processes within a single model nor are they being expected
can also be derived using simple energy arguments (Fernando to be resolved in the foreseeable future. Therefore, analyses and
1991). Two length scales of stratified turbulence can be identified predictions are conducted at a given scale or a limited range of
using (1.7): LE = b′ 2 N 2 ~ b0 /N 2 and Lb = w ′ 2 N ~ uV /N , the scales by reducing the governing equations to a manageable
10 Handbook of Environmental Fluid Dynamics, Volume One

which a particular process appears in a grid cell when certain crite-

Solar radiation
ria are satisfied, have been attempted; for example, see the work of
Large et al. (1994). The addition of k values of prevailing processes
has been used so far, and this approach appears to be too simplistic.
The indispensible role of eddy coefficients in modeling has
prompted many fundamental studies, focusing on mixing pro-
cesses and associated fluxes. An example is the parameteriza-
tion of heat (kT) and momentum (km) diffusivities for mesoscale
models. A grid box near the ground, for instance, encompasses
many processes—for example, turbulence due to roughness and
Rainfall instabilities, internal wave generation, radiation and breaking,
Land Ocean
surface
land-air heat fluxes, and evapotranspiration. Biogeochemical
processes are also of interest in integrated models, which sharply
increases model complexity. Examples include anthropogenic
activities of cities, biological transformation in natural waters,
and evolution of particulate matter (PM) in the lower atmosphere
Global model due to entrainment of crustal material and chemical reactions.
Multiscale simulations consist of grid nesting and exchang-
ing information between different scales, both down the scales
FIGURE 1.5 Depiction of an earth systems model. The processes (downscaling; Salathé et al. 2008) and from small to larger scales
(e.g., cities, mountain ranges) of scales smaller than horizontal resolution (upscaling). Downscaling is the most convenient and provides
∼100 km need to be parameterized. The primitive equations of motions an opportunity for large-scale simulations to run independently
with subgrid closure are solved with boundary/initial conditions and of smaller scale models. Upscaling is tedious and couples simula-
global forcing. The atmosphere and oceans are treated separately.
tions of multiple scales. Only limited progress has been made in
mathematical  set. As illustrated in Figure 1.5, the atmosphere- upscaling studies, although their importance is widely appreci-
ocean general circulation models (AOGCM) cover the entire ated. Scale interactions are of particular interest in cases where
earth, and the primitive equations are solved with external inputs slow changes at larger scales may cause amplified response
and bottom and top boundary conditions. Their bane is the low at smaller scales, sometimes leading to emergent phenomena
resolution (∼100 km), requiring mathematical representation (Section 1.2) as well as environmental disasters.
(parameterization) of salient unresolved processes within grid Experimentally capturing physical phenomena across a swath
boxes via eddy coefficients of momentum (km) and species (kc), viz. of space-time scales requires remote sensors or sensor networks of
appropriate resolution and footprint. Dense, fast sensor networks
⎛ 1 ⎞ with cyber infrastructure capable of transferring and process-
ui′u′j = 2km ⎜ eij − θδ ij ⎟ (i ≠ j) (1.9)
⎝ 3 ⎠ ing information are being sought after for decision support and
environmental management. On the global scale, for example,
and the World Meteorological Network’s “Global Weather Watch”
dC receives and archives atmospheric information. The Global Ocean
ui′c′ = kc , (1.10) Observing System (GOOS) under development is its oceanic
dxi
counterpart. A number of networks operate on regional scales, for
where example, AmeriFlux for fluxes of momentum, CO2, and energy.
θ = ∂ui/∂xi is the dilatation There are also networks with extensive spatial coverage with a
_ myriad of observables. The FLUXNET, a global network of micro-
eij is the mean rate of strain
meteorological tower sites that measures fine-scale turbulence
Naturally, the eddy coefficients are sensitively dependent on pro- and mean quantities, is of this ilk so are the multispectral satellite
cesses being modeled. In nested simulations, the output from instruments (e.g., ASTER, MODIS) that capture environmental
AOGCMs is fed into mesoscale (∼1–10 km resolution) models, information at meter scales. In model–data intercomparisons,
which in-turn provides input to CFD models (resolution ∼1–10 m). individual data stations are usually taken as a footprint for a grid
For the latter, variants of Reynolds Averaged Navier Stokes box, but within a given grid box the variability is extensive, and
Simulations (RANS), Large-Eddy Simulation (LES), and Detached methods to handle such heterogeneity are required.
Eddy Simulation (DES) are used in practice. The most scientifically
useful would be the Direct Numerical Simulation  (DNS), where
1.4.2 Process Modeling
all turbulence scales down to Kolmogorov scales are resolved, but
computer demands for such simulations are insurmountable. A Laboratory and numerical models are widely employed to gain fun-
challenge would be to develop eddy diffusivities of general utility damental insights on natural processes, concentrating on a single or
to account for multiple processes. The conditional diffusivities, in at most a few processes. The translation of results from such models
Environmental Fluid Dynamics 11

to improve operational predictability, nevertheless, occurs rather have been useful for identifying new phenomena arising from non-
slowly. This is, in part, due to the facts that only a few research groups linear interactions between different processes. The literature on
are working on operational model development/improvement and process modeling is voluminous, and this topic has been discussed
the laborious nature of model validation required after new imple- extensively in reviews and monographs. A simple example is given
mentations are made. In numerical modeling, the process-based in Section 1.5 on ocean bottom boundary layer, which, in labora-
approach has been extended to develop reduced or parsimonious tory or theoretical studies, is treated as equivalent to flow above an
models that include a selected set of processes, and such models oscillating bottom of a semi-infinite volume of water.
Laboratory modeling has been used in two ways. The first
W 0° E uses traditional model testing based on dynamic similitude,
20° 20° where the matching of geometric and dynamic parameters is
40 attempted to the extent possible between the model and actual
°
40 ° (Boyer and Davies 2000). The most difficult to match are the
aspect ratio (horizontal/vertical scales) and the Reynolds num-
ber, both of which are very large in geophysical flows. Similarity

60°
°
60

and physical arguments permit to reduce the number of rele-

vant dynamic variables to a few and model environmental situ-
ations with some success (e.g., Figure 1.6). On the other hand,
80°
80°

idealized flow configurations can be used without matching of

nondimensional numbers, allowing the development of param-
eterizations of general utility and collection of benchmark data
100°
100°

for model evaluation (Figure 1.7). Care must be taken to ensure

that the model and actual flows are in the same hydrodynamic
regime (e.g., laminar versus turbulent). Recent advances in flow
12
°
120

0°

14
0° 0°
14
° 160
160 180°
°
(a)
(a) (b)
W 0° E
20° 20°

° 40
40 °

(c) (d)
°

60°
60
80°

80°
100°

(e) (f )
100°

E
U
°

12
120

0°

(g) (h)
0° 0°
14
14
FIGURE 1.7 (See color insert.) Laboratory modeling of turbulent
°
160 °
160 mixing in a two-layer stably stratified fluid, with fluid layers separated
(b) 180°
by a sheared density interface. A generic flow configuration was used
here to obtain results of general validity. To (h) represent cases with
FIGURE 1.6 (a) Laboratory modeling of Antarctic Circumpolar different Bulk Richardson number Ri _B (Section 1.3.2) with L_V
Current in a rotating dishpan, with matching of a selected number replaced by the depth of the upper turbulent layer. (a) Ri _B = 1.8,
of dimensionless parameters. (b) Streamlines from the experiments, (b) 3.2, (c) 4.5, (d) 5.5, (e) shortly after (d), (f) 5.8, (g) 9.2 and (h) shortly
which agree well with ocean observations. (From Boyer, D.L. et al., after (d). (From Strang, E.J. and Fernando, H.J.S., J. Fluid Mech., 428,
J. Geophys. Res., 98(C2), 2587, 1993. With permission.) 349, 2001. With permission.)
12 Handbook of Environmental Fluid Dynamics, Volume One

diagnostic techniques have greatly improved the quality, reso- where ϖ and ϑ are constant coefficients. An example for (1.13)
lution, and volume of laboratory data. would be the neglect of ν in the wake problem described earlier.
When ν is included, u(x)/U = f1(x/D, Re), where Re = UD/ν is
the Reynolds number, and thus for Re → ∞ the dependence on
1.4.3 Dimensional Arguments
it, and hence on ν, can be neglected. This procedure is referred
and Similarity Theories
to as the Reynolds number similarity. Although (1.13) and (1.14)
Complexities of natural flows behoove a suite of mathematical are nominally applicable at extreme limits, they do have appli-
tools for understanding and analysis of natural flows. In typi- cations to intermediate stages of a phenomenon, when suitably
cal process studies, basic building blocks of the flow are iden- interpreted. For example, if the wake example discussed earlier
tified and studied in depth. Perhaps the simplest analysis tool is complicated by the presence of stable stratification with buoy-
is the dimensional analysis (Bolster et al. 2011) and its elegant ancy frequency N, then an extra π (=U/ND) term comes into play
extensions such as the generalized dimensional analysis (Long and the self-similarity is lost, viz.,
1963) and similarity theories (Baranblatt 1996). If a dependent
parameter in dimensional domain is a function of n inde- u ⎛x U ⎞
= f2 ⎜ , , (1.15)
pendent parameters (a1,a2,…,an), the dimensional analysis U ⎝ D ND ⎟⎠
axiomatizes that physical laws relating these quantities is inde-
pendent on human-derived units and hence can be written as but it is possible to seek similarity solutions in the limit x/D → ∞ as
a functional relationship between dimensionless parameters.
The  Buckingham π Theorem states that the number of nondi- m
mensional (π) numbers possible from the total dimensional set u ⎛ x⎞ ⎛ x ⎞ u ⎛ x⎞ (1.16)
= f3 ⎜ or = f3 ⎜ ⎟ ,
is (n − m + 1), m being the number of basic units involved, viz., U ⎜⎝ D ⎟⎠ ⎝ D(U/ND)n ⎟⎠ u0 ⎝ ℓ0 ⎠

π = π(π1 , π 2 ,…, πn − m ), (1.11) where (1.16) looks as if the flow has a new similarity veloc-
ity u 0  = U(x/M)m and length ℓ0 = D(U/ND)n scales, with
where, generally in deriving π and π i ’s, ai ’s are involved at least diminished significance of D. Obviously, these new scales
once. The actual functional form of π (π i) needs to be derived are valid when x/D → ∞ but not exactly at those extremes
using alternative methods, for example, laboratory/numerical (where the velocity tends to be either 0 or ∞, depending on m).
experiments or physical arguments. Pragmatically, x/D → ∞ solution is valid for x ≫ D, that is at
While classical dimensional analysis is a powerful tool, more (intermediate) values of x much greater than D and less than
detailed results are possible using extensions of it. Similarity x → ∞. At these distances, the influence of D is decadent, lead-
solutions are one of them, where for evolving flows the results ing to a new length scale. This exemplifies an intermediate
are expressed in terms of two dimensionless parameters, π = asymptotic solution.
π (π 1). Similarity theory allows obtaining of spatial distribution The scaling analysis presented in Section 1.3.1 is also help-
of a time evolving solution via similarity transformations. A ful in delineating dominant dynamical balances and hence for
simple example would be the turbulent wake of a cylinder of experimental data analysis and obtaining exponents for (1.14).
diameter D, where any property such as the velocity u(x) at a Such is a more rigorous form of dimensional analysis, allowing
downstream distance x along the centerline is dependent on the one to simplify the number of dimensional parameters based on
free-stream velocity U, D, and x, where the kinematic viscosity physical insight and anisotropic scaling (Long 1963). The final
ν has been neglected (see the following). The ensuing similarity solution, however, involves dimensionless (sometimes universal)
solution is simple, u(x)/U = f(x/D), where f, f1, … are functions. constants, which need to be determined using numerical and
When two or more independent π numbers are involved, experimental means. A combination of all these techniques is a
similarity solutions are not guaranteed. In certain cases, how- powerful resource for quantification of processes while illumi-
ever, it is possible to seek self-similar solutions at the extremes of nating underlying physics.
π i values. Consider

π = π(π1 , π 2 ), (1.12) 1.4.4 Field Studies

The prediction of environmental motions are predicated by diffi-
which may be converted to a self-similar form of a single inde- culties with subgrid closure, complexity of initial and boundary
pendent parameter in the limit of π 2 → 0 or ∞ as conditions, numerical issues, and sensitivities of numerical solu-
tion to boundary conditions (Lorenz 1976). As such, model vali-
(1.13) dations with field data as well as model verifications are essential
π = π(π1 )[self similarity of the first kind]
components of model development (Beven 2002). Field experi-
ments are tedious because of the vast domains involved, spatial
⎛π ⎞
π = π 2ϖ ⎜ ϑ1 ⎟ [self similarity of the second kind], (1.14) and temporal inhomogeneities within the domain, instrument
⎝ π2 ⎠ limitations and complexities, logistics, need for a group of
Environmental Fluid Dynamics 13

skilled personal with diverse expertise, setup and operational Although available photochemical models predict observed air
costs, efforts required for compliance with environmental regu- quality distribution trends satisfactorily, quantitative predictions
lations, as well as natural variability (Acts of God). The advent of pollutant concentrations, their distribution, and the timing of
of remote sensors and satellites has added immensely to envi- pollution episodes exhibit significant errors. Alternatives such as
ronmental research and monitoring, and if their scanning rates stochastic tools (e.g., neural networks) or empirical correlations
are faster than the variability of the processes of interest, useful are used for convenience, which gives area-specific predictions
four-dimensional information is possible, including delineation with acceptable reliability (Fernando et al. 2012b). In all, the cur-
of individual processes and their evolution. rent state of environmental forecasting leaves much to be desired,
calling for improvements in model designs (architecture), numeri-
cal algorithms, and parameterizations. These as well as new para-
1.5 Illustrations digms of modeling ought to be given top priority in developing a
1.5.1 Environmental Simulations vision for improved environmental forecasting (Jakob 2010).

In principle, at present, nested simulations downscale from

global to 1 m scale. A repertoire of models exclusively dedicated 1.5.2 Process Modeling
for different scales are available, and Figure 1.8a and b show the A case of simple process modeling is discussed in the follow-
results of a nested simulation where NCEP’s eta model (synop- ing, by considering a deep water layer subjected to a reversing
tic), MM5 mesoscale model, and a CFD model with k − ε closure square wave shear stress at the bottom z = 0. The aim is to mimic
(Baik et al. 2003) were used to predict flow in a downtown area the bottom boundary layer of a natural water body subjected to
at 2 m resolution. The computational domain is a commercial- reversing tides. With respect to a coordinate system moving with
ized area of Houston with buildings and street canyons. It shows the bottom, the flow aloft is seen as oscillating while exerting a
velocity vectors at a specific point in time, and the model picks reversing shear stress on the bottom. In the theoretical analysis,
up typical flow features such as wakes, flow separation, and con- the problem can be posed as
vergence/divergence zones. Obviously, simulations of this type
need to be validated with well-designed field campaigns. ∂ 2u ∂u
ν = ,
While nested simulations may include meter scale features, ∂z 2 ∂t
typical air quality calculations are based on mesoscale grids of size
with boundary conditions
1–12 km (Lee et al. 2007). Each grid represents a reasonable foot-
print of an urban, rural, or a mixed domain, known as the neighbor- ∂u ⎧⎪ ρ 0u*2 for 2ka < t < (2k + 1)a
hood scale. This is the scale over which health advisories are issued. −µ = T (t ) at z = 0, T (t ) = ⎨ ,
∂z 2
⎩⎪ −ρ 0u* for (2k + 1)a < t < (2k + 2)a
Air quality predictions even on such scales are computationally
daunting, mainly confined to research or for planning purposes. (1.17a,b)

Flow pattern at 1 m agl

600 600

500 500

400 400
Northing (m)

Northing (m)

300 300

200 200

100 100

0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
(a) Easting (m) (b) Easting (m)

0 m <= z < 7 m 7 m <= z < 16 m 16 m <= z < 19 m

19 m <= z < 21 m 21 m <= z < 81 m

FIGURE 1.8 (a) The building geometry in the computational domain. (Courtesy of S.-M. Lee.) (b) Simulated flow field at 1 m above ground level.
Filled polygons represent buildings.
14 Handbook of Environmental Fluid Dynamics, Volume One

where The plausible solution for (1.21a,b)would be u′ = u′(z′,t′) or

k > 0 is an integer
u is the x-direction velocity u ⎛t z ⎞
= f5 ⎜ , , (1.22)
μ(=νρ0) is the dynamic viscosity 2
u* a/ν ⎝ a aν ⎟⎠
ρ0 is the density
z is directed upward which contains one less variable than (1.20). Still the problem is
not self-similar, but the generalized dimensional analysis helped
The other initial and boundary conditions are u(z → ∞, t) → 0, and simplify the functional form.
u(z,0) = 0. The analytical solution for the velocity field becomes, Natural flows are turbulent, and hence Reynolds number sim-
using Laplace transforms, ilarity can be invoked for the interior of the flow, away from the
boundaries. This means that, as z aν → ∞, the self-similarity
4u*2
∞
⎧⎪ ⎡ (2k + 1)π ⎤1/2 ⎫⎪
∑
a 1 of the second kind can be used for (1.22), viz.,
u(z , t ) = exp ⎨− ⎢ z⎬
π3/2 ν (2k + 1)3/2 2aν ⎥⎦
k =0 ⎩⎪ ⎣ ⎭⎪ −1
⎛ z ⎞ ⎛t⎞
⎧⎪ (2k + 1)πt f5 = ⎜ f6 ⎜ ⎟ , (1.23)
(2k + 1)πz 2 3π ⎫⎪ ⎝ aν ⎟⎠ ⎝ a⎠
× cos ⎨ − − ⎬
⎪⎩ a 2aν 4⎪
⎭
∞ and hence
u2 ⎡ 1 ⎛ aξ ⎞ ⎛ ξ⎞
1/ 2
⎤
+ *
π ν ∫ ⎢ 3/2 tanh ⎜ ⎟ exp(−ξt )cos ⎜ ⎟ z ⎥ dξ.
⎢ξ
0 ⎣
⎝ 2⎠ ⎝ ν⎠ ⎥⎦
2
u a ⎛t⎞
u = * f6 ⎜ ⎟ , (1.24)
z ⎝ a⎠
(1.18)
The bottom of the water column therefore should oscillate as which is a self-similar solution for the velocity field. A similar
analysis can be performed for the boundary layer thickness and
∞
4u*2 ⎧ (2k + 1)πt 3π ⎫ rms turbulent velocities. While this idealized case can be read-
∑ (2k + 1)
a 1
u(0, t ) = cos ⎨ − ⎬. (1.19)
π3/2 ν 3/2
⎩ a 4⎭ ily verified in the laboratory and provide useful guidance for
k =0
data analysis, in natural flows the shear stress variation is more
Although an elegant closed-form solution is possible, in nature irregular and additional effects such as background currents and
the flow is turbulent and hence (1.19) is hardly valid. Coriolis forces complicate the flow.
As an alternative, consider conventional dimensional analysis Figure 1.9a shows a typical oceanic case where a rotating tidal
with governing dimensional variables a, ν, z, t, and u*. With fun- current is evident, which considerably differs from that discussed
damental dimensions being length and time, three independent previously. Nonetheless, expressions of the form (1.24) are of
dimensionless parameters are possible: some utility in presenting data of the oceanic bottom boundary
layer. Figure 1.9b shows vertical velocity profiles taken in the
u ⎛ t z u2 a ⎞ same area. Lacking an alternative, the profiles near the bottom
= f4 ⎜ , , * ⎟. (1.20) have been approximated by log-linear power laws convention-
u* ⎝ a aν ν ⎠
ally used for steady boundary layers. The interior of the fluid
Note that all dimensionless parameters in (1.20) also appear in sometimes shows a decrease in velocity with height, as in (1.24),
(1.18), but in (1.20) the functional form f2 is unknown and no but the variability from one profile to another is significant. The
similarity solutions are viable. Nevertheless, simplifications are presence of background currents and other factors needed to be
possible using the generalized dimensionless analysis. By intro- considered in interpreting field observations. Idealized labora-
ducing the dimensionless variables u′ = u/u 0, t′ = t/t0, and z′ = tory simulations and analysis of fundamental processes, however,
z/z0, where u0 = u*2 a/ν , t0 = a, and z 0 = aν obtained using the are immensely useful in interpreting field observations as well as
scaling analysis of (1.17a,b), we get inferring prevailing processes at a given field site.

∂ 2u′ ∂u′
= , 1.6 Major Challenges
∂z ′2 ∂t ′
Improved understanding and quantification of the dynamics of
with
environmental (air and water) flows and their interaction with
∂u ′ anthropogenic, biological, cryological, chemical, and geologi-
− = T ′(t ′) at z ′ = 0; and cal components of the earth system are at the heart of EFD.
∂z ′
⎪⎧ 1 for 2k < t ′ < (2k + 1) In a broader perspective, EFD parses the challenges of environ-
T ′(t ) = ⎨ . mental prediction, management, and planning on a firm scien-
⎩⎪ −1 for (2k + 1) < t ′ < (2k + 2) tific footing while developing tools to successfully address them.
(1.21a,b) EFD tends to focus on the small-scale end of environmental
Environmental Fluid Dynamics 15

0.5

0.4

0.3 35 19:20 19:50 20:20 20:50 21:20 21:50

0.2 17 16
18 4 3 15
5 30

Height above the sea floor (m)

2
0.1 19 6 1
7 14 25
v (m/s)

0 20 0
20
–0.1
21 9 13
–0.2 15
10 12
22 24
–0.3 11 10
23
–0.4 5

–0.5
0
–0.5 –0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 0.5 0.2 0.25 0.3 0.35 0.4 0.45 0.5
(a) u (m/s) (b) |U| (m/s)

FIGURE 1.9 (a) A hodograph of rotating tidal flow measured at a mean water depth of 35.5 m (total water depth 38 m) on the shelf of the East China
Sea (Yellow Sea); the numbers at the arrows are times of 1 h averaged current vectors after 15:00 of March 26, 2005. (From Lozovatsky, I. et al., Cont.
Shelf Res., 28, 318, 2008. With permission.) (b) Profiles of velocity magnitude at the location of (a). The oft-used log-layer region is indicated by the dotted
lines. While (1.24) is applicable for an idealized oscillating tidal flows, the presence of a mix of processes in nature precludes its direct usage in the field.

motions, particularly the meso- and smaller scales, the latter Conversely, on the atmospheric side, issues of engineering inter-
being unresolved in most predictive models. The EFD research est are the plume stacks, urban designs, indoor–outdoor air
employs a suite of approaches—analytical, numerical, labora- exchange, HVAC systems, and sound pollution.
tory experimental, and field observational—with the aim of The emphasis on nonfossil energy resources (solar, hydro,
achieving better forecasts for environmental motions. Studies wind, and nuclear power) is expected to continue. Small hydro-
focusing on one or a few processes have had the most success, kinetic and wind turbines are becoming popular, and ways of
as evident from such discoveries as double diffusion, lee waves, locating them for optimal efficiency and minimum environ-
solitary waves, and a host of instabilities via process studies. mental impacts are being sought out by commercial developers.
The progress in tackling nonlinear phenomena has been slow, Nuclear power plant operations have a strong EFD presence,
although some important new (emergent) behaviors resulting including cooling water recirculation and emergency planning
from nonlinear processes have been delineated (e.g., lee waves). based on transport and diffusion.
To this end, a reductionist approach is used, incorporating a On the technological front, smart buildings, transportation
selected number of processes. infrastructure, aircraft, ship and spacecraft operations, as well
Some grand challenges loom large for EFD. The burgeoning as emergency hazard alert systems rely on acquisition, process-
world’s population has led to large (mega) cities and aggravated ing, and mining of real-time environmental data as well as uti-
anthropogenic stresses on the environment. Urban areas are the lization of historic data in conjunction with predictive models.
sources of heat emissions, air pollution, drag (due to buildings), To this end, the deployment of cyberinfrastructure is a frontier
greenhouse gases, and hence anthropogenic climate change. issue of environmental research. Benefits of environmental
Cities in turn bear the brunt of climate change, and their very resources (ecosystem services) sensitively depend on water avail-
sustainability may depend on local response to global change. ability, air and water quality, local microclimate, and health of
Key elements for planning and management of green, smart, ecosystems.
and secure cities are the local meteorology, air and water quality, In all, environmental fluid dynamics suffuses a host of grand
human comfort, and water availability, which are common top- challenges of modern times, unfolding exciting research oppor-
ics in EFD research. Further, risk assessment tools used in envi- tunities ranging from molecular mixing in reservoirs to pedes-
ronmental-hazard mitigation studies hinge on flow predictions. trian comfort in urban canyons to cyberinfrastructured (smart)
EFD also plays a key role in hydraulic engineering, for cities with hazard warning capabilities. With reverberating
example, in the design of dams, wells, waterways, cooling societal calls for environmental sustainability, understanding
systems, storm water and flood control systems, irrigation and prediction of environmental motions are becoming ever
canals, and desalinization plants. Some specific issues of con- so more important. This yearning combined with emergence of
cern are dispersion, turbulent mixing, outfalls, wetlands, sedi- powerful analytical, laboratory, field experimental, remote sens-
ment transport, remediation, chemical reactions, dead zones ing, numerical, and computing tools promise a bright future for
(e.g., eutrophication and hypoxia), and underwater acoustics. environmental fluid dynamics.
16 Handbook of Environmental Fluid Dynamics, Volume One

References Fernando, H.J.S., Zajic, D., Di Sabatino, S., Dimitrova, R.,

Hedquist, B., and Dallman, A. 2010. Flow, turbulence and
Baik, J.J., Kim, J.J., and Fernando, H.J.S. 2003. A CFD model for pollutant dispersion in urban atmospheres. Phys. Fluids 22:
simulating urban flow and dispersion. J. Appl. Meteorol. 051301–051319.
42(11): 1636–1648. Gibson, C.H. 1991. Laboratory, numerical, and oceanic fossil tur-
Baranblatt, G.I. 1996. Scaling, Self Similarity and Intermediate bulence in rotating and stratified flows. J. Geophys. Res. 96:
Asymptotics, Cambridge University Press, Cambridge, U.K. 12549–12566.
Beven, K. 2002. Towards a coherent philosophy for modelling the Gill, A.E. 1982. Atmosphere-Ocean Dynamics, Academic Press,
environment. Proc. R. Soc. Lond. 458(2026): 2465–2484. New York.
Bolster, D., Hersberger, R.E., and Donnelley, R.J. 2011. Dynamics Goodman, L. and Robinson, A. 2008. On the theory of advec-
similarity, the dimensionless science. Phys. Today 64(9): tive effects on biological dynamics in the sea. III. The role of
42–47. turbulence in biological–physical interactions. Proc. R. Soc.
Boyer, D.L., Chen, R.-R., and Tao, L. 1993. Physical model of A 464: 555–572.
bathymetric effects on the Antarctic circumpolar current. Gregg, M.C. 1987. Diapycnal mixing in the thermocline: A review.
J. Geophys. Res. 98(C2): 2587–2608. J. Geophys. Res. 92(C5): 5249–5286.
Boyer, D.L. and Davies, P.A. 2000. Laboratory studies of oro- Hondzo, M. and Wüest, A. 2009. Do microscopic organisms feel
graphic effects in rotating and stratified flows. Annu. Rev. turbulent flows? Environ. Sci. Technol. 43(3): 764–768.
Fluid Mech. 32(1): 165–202. Huppert, H.E. 2000. Geological fluid mechanics. In: G.K. Batchelor,
Britter, S. and Hanna, S. 2003. Flow and dispersion in urban areas. H.K. Moffatt, and M.G. Worster (Eds.) Perspectives in Fluid
Annu. Rev. Fluid Mech. 35: 469–496. Dynamics: A Collective Introduction to Current Research,
Chandrasekhar, S. 1961. Hydrodynamic and Hydromagnetic Cambridge University Press, Cambridge, U.K., pp. 447–506.
Stability, Courier Dover Publications, Mineola, NY. Imberger, J. and Patterson, J.C. 1990. Physical limnology. Adv.
Crutzen, P.J. 2004. New directions: The growing urban heat and Appl. Mech. 27: 303–475.
pollution “island” effect—Impact on chemistry and climate. Ivey, G.N., Winters, K.B., and Koseff, J.R. 2008. Density stratifica-
Atmos. Environ. 38: 3539. tion, turbulence, but how much mixing? Annu. Rev. Fluid
Drazin, P.G. and Reid, W.H. 2004. Hydrodynamic Stability, Mech. 40: 169–184.
Cambridge University Press, Cambridge, U.K. Jakob, C. 2010. Accelerating progress in global atmospheric
Fernando, H.J.S. 1991. Turbulent mixing in stratified fluids. Annu. model development through improved parameterizations—
Rev. Fluid Mech. 23: 455–493. Challenges, opportunities and strategies. Bull. Am. Meteorol.
Fernando, H.J.S. 2008. Polimetrics: The quantitative study of Soc. 91: 869–875.
urban systems (and its applications to atmospheric and Kaimal, J.C. and Finnigan, J.J. 1994. Atmospheric Boundary Layer
hydro environments). J. Environ. Fluid Mech. 8(5–6): Flows: Their Structure and Measurement, Academic Press,
397–409. New York, 289pp.
Fernando, H.J.S. 2010. Fluid dynamics of urban atmospheres in Karl, T.R., Jones, P.D., Knight, R.W., Kukla, G., Plummer, N.,
complex terrain. Annu. Rev. Fluid Mech. 42: 365–389. Razvayev, V., Gallo, K.P., Lindseay, J., Charlson, R.J., and
Fernando, H.J.S. and Voropayev, S.A. 2010. Shallow flows in Peterson, T.C. 1993. Asymmetric trends in surface tempera-
the atmosphere and oceans: Geophysical and engineering ture. Bull. Am. Meteorol. Soc. 74: 1007–1023.
applications. J. Hydro-Environ. Res. 4: 83–87. Klein, P. and Lapeyre, G. 2008. The oceanic vertical pump
Fernando, H.J.S. and Weil, J. 2010. An essay: Whither the induced by mesoscale and submesoscale turbulence. Annu.
SBL?—A shift in the research agenda. Bull. Am. Meteorol. Rev. Mar. Sci. 1: 351–375.
Soc. (BAMS), 91(11): 1475–1484. Large, W., McWilliams, J.C., and Doney, S.C. 1994. Oceanic verti-
Fernando, H.J.S., Klaic, Z., and McCulley, J. (Eds.). 2012a. National cal mixing: A review and a model with a nonlocal boundary
Security and Human Health Implications of Climate Change. layer parameterization. Rev. Geophys. 32(4): 363–403.
NATO ARW Series, Springer Publishing, New York. Lee, S.-M., Fernando, H.J.S., and Clarke, S. 2007. Modeling
Fernando, H.J.S., Lee, S.-M., Anderson, J., Princevac, M., of ozone distribution in the State of Arizona in support
Pardyjak, E., and Grossman-Clarke, S. 2001. Urban fluid of 8-hour non-attainment area boundary designations.
mechanics: Air circulation and contaminant dispersion in Environ. Model. Assess. 12: 63–74.
cities. J. Environ. Fluid Mech. 1: 107–164. Long, R.R. 1963. The use of the governing equations in dimen-
Fernando, H.J.S., Mammarella, M.C., Grandoni, G., Fedele, P., sional analysis. J. Atmos. Sci. 20(3): 209–212.
DiMarco, R., Dimitrova, R., and Hyde, P. 2012b. Forecasting Lorenz, E.N. 1976. Nondeterministic theories of climatic change.
PM10 in metropolitan areas: Efficacy of neural networks. Quat. Res. 6(4): 495–506.
Environ. Pollut. 163: 62–67. Lozovatsky, I., Liu, Z., Wei, H., and Fernando, H.J.S. 2008. Tides
Fernando, H.J.S., Mendis, G., McCulley, J., and Perera, K. 2005. and mixing in the northwestern East China Sea Part I:
Coral poaching worsens tsunami destruction. EOS 86: Rotating and reversing tidal flows. Cont. Shelf Res. 28:
301–304. 318–337.
Environmental Fluid Dynamics 17

Mei, C.C. 2003. The Applied Dynamics of Ocean Surface Waves, Riley, J.J. and Lelong, M.-P. 2000. Fluid motions in the presence
World Scientific, Singapore. of strong stable stratification. Annu. Rev. Fluid Mech. 32:
Molina, L.T. et al. 2010. An overview of the MILAGRO 2006 cam- 613–657.
paign: Mexico City emissions and their transport and trans- Salathé, E.P., Jr., Steed, R., Mass, C.F., and Zhan, P.H. 2008. A high-
formation. Atmos. Chem. Phys. Discuss. 10: 7819–7983. resolution climate model for the U.S. Pacific Northwest:
Odum, E.P. 1971. Fundamentals of Ecology, 3rd edn., Saunders, Mesoscale feedbacks and local responses to climate change.
New York. J. Climate 21: 5708–5726.
Ovenden, N., Shaffer, S., and Fernando, H.J.S. 2009. Determining Sarmiento, J.L. and Gruber, N.L. 2002. Sinks for anthropogenic
the impact of meteorological conditions on noise propa- carbon. Phys. Today 55(8): 30–36.
gation away from freeway corridors. J. Acoust. Soc. Am. Science. 2011. Doom or vroom? Special section on population.
126(1): 25–35. Science 333: 540–594.
Paola, C., Twilley, R.R., Edmonds, D.A., Kim, W., Mohrig, D., Smith, V.H. 1998. Cultural eutrophication of inland, estuarine
Parker, G., Viparelli, E., and Voller, V.R. 2011. Natural pro- and coastal waters. In: M.L. Groffman and P.M. Pace (Eds.)
cesses in delta restoration: Application to the Mississippi Success, Limitations, and Frontiers in Ecosystems Sciences,
delta. Annu. Rev. Mar. Sci. 3: 67–91. Springer-Verlag, New York.
Peltier, W.R. and Caulfield, C.P. 2003. Mixing efficiency in strati- Strang, E.J. and Fernando, H.J.S. 2001. Turbulence and mixing in
fied shear flows. Annu. Rev. Fluid Mech. 35: 135–167. stratified shear flows. J. Fluid Mech. 428: 349–386.
Phillips, O.M. 1977. Dynamics of Upper Ocean, Cambridge Turner, J.S. 1973. Buoyancy Effects in Fluids, Cambridge University
University Press, Cambridge, U.K. Press, London, U.K.