Fluid Mechanics II (LMECA2322) — Geophysical and environmental flows

Practical exercises – Tentative date: 12 October 2023

I. Advection, diffusion and reactions

1. [Advection, diffusion and first-order decay] Consider a solute (or tracer) exhibiting a
first-order decay process that is one of the constituents of a fluid mixture flowing in time-
independent domain Ω, which is isolated from its environment. The concentration of this
tracer (defined as a mass fraction) is C(t,x) , where t and x denote the time and the position
vector, respectively. Under the Boussinesq approximation, the solute concentration obeys
partial differential equation (e.g., Deleersnijder 20231)
∂C
= − λ C − ∇ i ( Cv − κ t ∇C ) (1.1)
∂t
where positive constant λ denotes the rate of decay, v(t,x) and κ t (t,x) (>0) are the fluid
mixture velocity (∇ i v = 0) and the eddy diffusivity, respectively. Could the tracer under
consideration be a radionuclide? Determine its mean life and the half-life, and evaluate m(t) ,
the mass of solute contained in Ω at any time t (> 0) , given that [ m(t)]t=0 = m 0 .

The first term in the right-hand side of (1.1) accounts for the first-order decay with constant
decay rate λ (> 0) 2 of the tracer under study. Radionuclides decay as a first-order process
with a constant decay rate. Thus, the tracer under study could be a radionuclide. It is
noteworthy, however, that first-order decay may be used to represent the destruction of
constituents other than radionuclides. The decay rate (e.g., the mortality rate of phyto- or zoo-
plankton in ecological modelling) cannot always be assumed to be a constant and is
sometimes estimated as a function of several environmental parameters.
At time t = 0 , consider a set of N 0 (>>1) particles of the same nature that will
progressively decay at time progresses. Each particle is identified by subscript “n”, with
1 ≤ n ≤ N 0 . Assume that θ n is the lifespan of the n-th particle: this particle exists as long as
t < θ n and transforms into another constituent at t = θ n . This may be taken into account by
means of characteristic function ϖ n (t) , which is such that
⎧1 , t < θ n
ϖ n (t) = ⎨ (S1.1)
⎩0 , θ n < t
Accordingly, the lifespan of the n-th particle identically satisfies
∞
θn = ∫ ϖ n (t) dt (S1.2)
0

whilst the number of particles under study at time t is

1
Deleersnijder E., 2023, Reactive transport equations for estuarine, coastal and shelf sea modelling, Working
note, Université catholique de Louvain, Louvain-la-Neuve, Belgium, 39 pages, available on the web at URL
http://hdl.handle.net/2078.1/277721
2
The physical dimension of λ is time −1 .

1 of 19
 N0
N(t) = ∑ ϖ n (t) (S1.3)
n=1
Needless to say, the number of particles cannot increase as time progresses.
The mean life, θ , is usually defined as the arithmetic mean of the lifespans. By having
recourse (S1.2)-(S1.3), one obtains
∞
N0 ∞ ∞
N0
1 ⌠N
0
1 1 1
θ = 0
N
∑ θn = 0
N
∑ ∫ n
ϖ (t) dt = ⎮ ∑
N 0 ⌡ n=1
ϖ n (t) dt =
N0 ∫ N(t) dt (S1.4)
n=1 n=1 0 0
0
On another note, the half-life, t1/2 , is the time required for the number of particles to reduce to
half of its initial value, leading to
N0
N(t1/2 ) = (S1.5)
2

0.8

0.6
N(t)/N 0

0.4

0.2

0
0 1 2 3 4 5
λt

Figure S1.1. Illustration of formula (S1.7). With normalised variables λ t and

N(t) / N 0 , all first-order decay processes (with a constant decay rate) are taken
into account.

If the particles under study have a constant decay rate (λ ) , then their number is governed
by ordinary differential problem
dN
= − λ N , [ N(t)]t=0 = N 0 >> 1 (S1.6)
dt
the solution of which reads (Figure S1.1)
N(t) = N 0 e− λt (S1.7)
Substituting (S1.7) into (S1.4) and (S1.5) yields the mean life
∞ ∞
1 1 1
θ =
N0 ∫ N(t) dt = N0 ∫ N 0 e− λt dt =
λ
(S1.8)
0 0
and the half-life

2 of 19
 N0 log 2
N(t1/2 ) = N 0 e− λt1/2 = ⇒ t1/2 = = (log 2) × θ ≈ 0.7 × θ (S1.9)
2 λ
The number of nuclei of a radioelement behaves in accordance with (S1.7). For instance,
the mean lives of plutonium-239, carbon-14, caesium-137, strontium-90, iodine-131 and
sodium-24 are about 35,000 years, 8,300 years, 43 years, 40 years, 12 days and 22 hours,
respectively.
Equation (1.1) is to be solved under the initial condition
[C(t,x)]t=0 = C 0 (x) (S1.10)
and impermeability boundary condition
[(Cv − κ t ∇C) i n]x∈Γ = 0 (S1.11)3
where Γ is the boundary of domain of interest Ω , whilst n is its outward unit normal vector.
Multiplying (1.1) by reference density ρi and integrating each term of (1.1) over Ω leads to
=m(t )
!# "# $
⌠ρ ∂C d d
⎮ i
⌡ ∂t
dΩ = ∫
dt Ω
ρ iC d Ω =
dt
m(t) (S1.12)
Ω ####
% #&##### '
Ω is time-independent

∫ ( − ρiλC ) d Ω = −λ ∫ ρ iC d Ω = − λ m(t) (S1.13)

Ω Ω
!#"#
$
=m(t )
=0 , see (S1.11)
!### "###$
∫ ∇ i (Cv − κ t ∇C ) d Ω = ∫ (Cv − κ t ∇C ) i n d Γ = 0 (S1.14)
Ω
%######## Γ
#&#########
'
by using the divergence theorem
where m(t) is the mass of tracer present in Ω at time t. Combining (S1.12)-(S1.14) yields
ordinary differential equation
d
m(t) = − λ m(t) (S1.15)
dt
whose solution reads
m(t) = m 0 e− λt (S1.16)
where m0 denotes the initial mass, i.e.,
m 0 = [ m(t)]t=0 = ∫ ρi C(0,x) d Ω (S1.17)
Ω
The mass of the tracer decreases exponentially at constant rate λ . This is impacted by
neither advection nor diffusion. This is because decay takes place at the same rate at any time
and location, and the domain of interest is isolated from its environment, implying that no
tracer particle crosses the domain boundary.

3
Relation (S1.11) means that the total (advective + diffusive) flux of the tracer under study across the domain
boundary is zero. However, it is seen in Section (3.4) of Deleersnijder (2023, op. cit.) that the advective and
diffusive tracer fluxes across the impermeable domain boundary are both zero.

3 of 19
2. [Water and mass budgets of a nearly enclosed sea]4 The net input of freshwater to the
Black Sea due to “precipitation – evaporation” and river run off is of the order of
300 km 3 / year . The Black Sea and the Mediterranean exchange water through the Bosporus
(see figure below5), in which a two-layer flow regime takes place. In the top (resp. bottom)
layer, the salinity is about 17 × 10 −3 (resp. 34 × 10 −3 )6. Assuming that, over a period of time
of the order of a year, the Black Sea is at a steady state, determine the direction of the flow in
the top layer and the bottom one, and estimate the volumetric flow rates ( km 3 / year ) and the
salt fluxes (kg/year) in each layer. Draw a schematic 7 aimed at illustrating the above-
mentioned processes.

Hint: the Boussinesq approximation is assumed to hold valid.

Consider a constituent of seawater whose concentration (defined as a mass fraction) is

denoted C. The mass of this constituent that crosses motionless surface Γ in the direction of
unit normal vector n reads
Φ = ∫ (ρCv − ρ K i ∇C) i n d Γ (S2.1)
Γ
where ρ , v and K represent the seawater density, the seawater velocity and the eddy
diffusivity tensor, respectively; Φ is measured in kg/s . Under the Boussinesq approximation,
(S2.1) simplifies to
Φ = ρi ∫ (Cv − K i ∇C) i n d Γ (S2.2)
Γ

4
This problem is derived from an exam question from ULiège Professor Jean-Marie Beckers.
5
Source: http://dx.doi.org/10.3390/rs9090892, last viewed on 25 March 2021.
6
The density of seawater in an increasing frunction of salinity, which is the salt concentration, defined as a mass
fraction.
7
A hand-made drawing would be acceptable.

4 of 19
This formula can be used in theoretical developments or in numerical models offering a
sufficiently high space resolution. However, for applications relying on scarce field data, a
simplified approach must be resorted to.
Assume that the only pieces of information available are a single value of concentration C
(which is believed to be sufficiently relevant) and an estimate of the volumetric flow rate, φ
(m 3 / s) . Under these circumstances, (S2.2) reduces to
Φ ≈ ρ iC ∫ v i n d Γ ≈ ρ i φ C (S2.3)
Γ
!#"#
$
≈φ

which obviously is a huge and, yet, not that unrealistic simplification of (S2.2). However, it is
also instructive to realise that it provides the exact mass flux in a channel in which the
concentration and the volumetric flow rate (i.e., the product of the channel cross-sectional
area and the section-averaged longitudinal velocity) are constant. Clearly, for the problem at
hand, given the paucity of the data available, it is formula (S2.3) that will have to be resorted
to.
The Black Sea (see figure above) is connected to the Mediterranean Sea by the Turkish
Straits, i.e., the Bosporus and Dardanelles, which are located at the opposite ends of the Sea
of Marmara.

Bosporus
Mediterranean

!∙ ! ! ! ! !!
! !"
!∙ ! ! ! ! !!
Black Sea

vertical cross-section !∙ ! ! ! ! ≈ 10!" !kg/year

salt fluxes: !
!∙ ! ! ! ! ≈ 10!" !kg/year

! ! ≈ 600!km! /year
(volumetric) water fluxes: ! ! ! ≈ 300!km! /year
! !" ≈ 300!km! /year

Figure S2.1. Schematic representation of the salt and water fluxes entering or
leaving the Black Sea. The figure is not to scale.

Let φ in ≈ 300 km 3 / year , φ s and φ b denote the net input of freshwater due to
“precipitation - evaporation” and river run off, the outgoing water flux ( km 3 / year ) in the
surface layer of the Bosporus and the incoming water flux ( km 3 / year ) in the bottom layer of
this strait, respectively. The steady-state assumption implies that the water mass budget must

5 of 19
be balanced, yielding
ρi φ sC s ≈ ρi φ bC b + ρi φ inC in (S2.4a)
By “water” is meant the fluid mixture under consideration. Hence, the corresponding
concentrations in the above equation are equal to unity, i.e., C s = C b = C in = 1 , so that
(S2.4a) simplifies to
φ s ≈ φ b + φ in (S2.4b)
The incoming freshwater flux makes no contribution to the salt budget of the Black Sea
since, by definition, the salinity of freshwater is zero. Therefore, only the salt fluxes in the top
and bottom layers of the Bosporus are to be taken into account. The salinities in the surface
and bottom layer of the Bosporus are S s ≈ 17 × 10 −3 and S b ≈ 34 × 10 −3 , respectively. The
outgoing salt (mass) flux, ρiφ s S s , must be equal to the incoming one, ρiφ b S b , so that
ρiφ s S s ≈ ρiφ b S b (S2.5)
Combining (S2.4b) and (S2.5) leads to
Ss 17 × 10 −3
φb ≈ φ in ≈ φ in ≈ φ in ≈ 300 km 3 / year (S2.6)
Sb − S s 34 × 10 −3 − 17 × 10 −3
Then, substituting this result into (S2.4b) yields the outgoing water flux through the
Bosporus, i.e.,
φ s ≈ φ b + φ in ≈ 2 φ in ≈ 600 km 3 / year (S2.7)
The associated salt (mass) fluxes can now be evaluated:
ρiφ s S s ≈ ρiφ b S b ≈ 1013 kg/year (S2.8)
All of the abovementioned fluxes are illustrated in Figure S2.1.
It is worth underscoring that the above methodology has been resorted to in a number
studies of nearly closed domains. For instance, Bethoux and Gentili (19998) applied it to the
Mediterranean Sea.

8
Bethoux J.P. and B. Gentili, 1999, Functioning of the Mediterranean Sea: past and present changes related to
freshwater input and climate changes, Journal of Marine Systems, 20, 33-47,
https://www.sciencedirect.com/science/article/pii/S0924796398000694

6 of 19
3. [Penetration of CFC into the World Ocean] CFC-11 (trichlorofluoro-methane, also
called freon-11) is a gas, whose chemical formula is C Cl3 F . It has been widely used as a
refrigerant. Its release into the Earth's atmosphere has largely contributed to stratospheric
ozone depletion, a phenomenon often referred to as the “ozone hole” by the general public.
Some of the atmospheric CFC-11 has been dissolved into the World Ocean, where it behaves
as a passive constituent. Based on the vertical inventory of CFC-11 in the mid-nineties, which
is displayed in the map below9, estimate the total mass of CFC-11 contained in ocean waters.
It is remarkable that the CFC-11 vertical inventory is largest in the North Atlantic, and is also
significant in the Southern Ocean. Explain this by having recourse to the concept of “great
ocean conveyor”.10

The figure above suggests that the average over the World Ocean of the vertical inventory of
CFC-11 is of the order of I ≈ 1.5 × 10 −6 mole m −2 . The surface area of the World Ocean
represents about 70% of the surface area of Earth. Assuming that the latter is a sphere whose
radius is 6.4 × 10 6 m , then the surface area of the domain of interest is
S ≈ 0.7 × 4 × π × (6.4 × 10 6 )2 m 2 ≈ 3.6 × 1014 m 2 (S3.1)
!### #"#### $
≈ surface area of Earth

There are thus about

I S ≈ (1.5 × 10 −6 mole m −2 ) × ( 3.6 × 1014 m 2 ) ≈ 5.4 × 10 8 mole (S3.2)
of CFC-11 in the World Ocean.
From the periodic table of the elements (see, for instance, Appendix I of the course
material), the mass of one mole of trichlorofluoromethane ( C Cl3 F ) is readily seen to be

9
Source: http://upload.wikimedia.org/wikipedia/commons/2/20/GLODAP_invt_CFC11_AYool.png (last viewed
on March 26th, 2013).
10
Broecker W.S., 1991, The great ocean conveyor, Oceanography, 4, 79-89

7 of 19
 (12.01× 10 −3 kg) + 3 × (35.45 × 10 −3 kg) + (19 × 10 −3 kg) ≈ 137.36 × 10 −3 kg (S3.3)
!##"##$ !##"##$ !# #"## $
≈ mass of 1 mole ≈ mass of 1 mole of ≈ mass of 1 mole
of C (carbon) Cl (chlorine) of F (fluorine)

Thus, the mass of CFC-11 present in the World Ocean (at the time the figure above was
made) is
m ≈ (5.4 × 10 8 ) × (137.36 × 10 −3 kg)
(a) ≈ 7.4 × 10 7 kg (S3.4)
Using field data (e.g. those depicted in
the figure above), Willey et al. (2004 11 )
estimated that the total oceanic CFC-11
inventory as of 1994 was 5.5 × 10 8 moles ,
or 75,500 metric tons. Clearly, the rough
estimates obtained above are in agreement
with these figures. In addition, the numerical
model results examined by Dutay et al.
(200212) are also worth looking at (see figure
opposite).
That the vertical inventory of CFC-11 is (b)

the largest in the North Atlantic and is also

significant in the neighbourhood of Antarctica is in agreement with the fact that the solubility
of CFC-11 increases as the water temperature decreases and is also a signature of the
thermohaline circulation, especially that of deep water formation in the Labrador and Nordic
Seas.
The motions developing in the global ocean over large space scales (i.e., scales comparable
to the size of an ocean basin) are due to the surface wind stress, tides (which are caused by
gravitational forces from the sun and moon) and surface fluxes of heat and water (impacting
the surface salinity), leading to the thermohaline circulation (THC). According to Rahmstorf
(200613), the thermohaline circulation was first discovered and explained as follows:
In 1751 the captain of an English slave-trading (c) ship made the first recorded
measurement of deep ocean temperatures — he discovered that the water a mile
below his ship Fig.was
1. Temporal
very cold,evolution
despiteofthe
(a) the CFC-11 atmospheric
subtropical location. mole fraction
In 1797 in both hemispheres and (b) s
another
CFC-11 inventory. (c) Global CFC-11 model inventory as a function of the average global CFC-11 inve
Englishman, Benjamin Thompson, correctly explained this discovery by cold
models.
currents from the poles, as part of what later became known as the thermohaline
circulation. mainly in regions of cold winter SST and strong vertical mixing such as the Souther
North Atlantic Ocean and the Northwestern Pacific Ocean. Uptake is also high in th
The THC plays Pacific duerole
a crucial to upwelling of subsurface
in the oceanic poleward waters that are(see
heat transport impoverished in CFC. Uptake
figure below,
adapted from Fasullo and Trenberth, 2008 ), which is of the order of 1 PW = 1015 W . This is
14

equivalent to the electrical power generated by about one million single reactor nuclear power

11
Willey D.A., R.A. Fine, R.E. Sonnerup, J. L Bullister, W.M. Smethie Jr. and M.J. Warner, 2004, Global ocean
chlorofluorocarbon inventory, Geophysical Research Letters, 31, L01303, doi: 10.1029/2003GL018816
12
Dutay J.-C., J.L. Bullister, S.C. Doney et al., 2002, Evaluation of ocean model ventilation with CFC-11:
comparison of 13 global ocean models, Ocean Modelling, 4, 89-120
13
Rahmstorf S., 2006, Thermohaline ocean circulation, in: Encyclopedia of Quaternary Sciences, S.A. Elias
(Ed.), Elsevier, Amsterdam,
14
Fasullo J.T. and K.E. Trenberth, 2008, The annual cycle of the energy budget. Part II: Meridional structures
and poleward transports, Journal of Climate, 21, 2313-2325

8 of 19
 stations! The figure below also shows that the atmosphere transports more heat toward the
poles than the ocean. The total poleward heat flux renders the low latitudes cooler, and the
high latitudes warmer. In other words, it reduces the temperature contrast between the poles
and the equator.

latitude

heat flux (1015 W = PW) (positive northward)

BE period zonal mean annual cycle of the meridional energy transport in PW by (a) the atmosphere and ocean as
BE RT, NRA #AE/#t, and GODAS #OE/#t; (b) the atmosphere based on NRA; and (c) by the ocean as implied by
The THC's main feature is an overturning, meridional circulation, in
S and GODAS #OE/#t. Stippling and hatching in (a)–(c) represent regions and times of year in which the standard
which cold, salty
monthly mean valueswaters
among estimates, some of which
intermittently sinkinclude
at highthe latitudes
CERES period
and(see text), exceeds 0.5fill
progressively andthe
1.0 PW,
deep ocean. The return
The median annual mean transport by latitude for the total (gray), atmosphere (red), and ocean (blue) 15 accompanied
flow to the surface is more diffuse. Accordingly, Warren (1981 ) wrote
d !2" range (shaded).
[...] the horizontal circulation in the actual ocean may be thought to be a
al transports are characterized by flows 5. Discussion and conclusions
consequence of localized sinking and generalized upwelling.
mer to winter hemispheres that exceed 4
onths. In the regions of maximum trans-
RG2001 The zonal meanet energy
Kuhlbrodt budget
al.: DRIVERS and
OF THE its annual cycle
AMOC RG2001
ty among the estimates is generally less for the globe and for land and ocean domains have
The uncertainty in the monthly mean been examined and the associated meridional trans-
eatest in the tropics and exceeds 1 PW in ports of energy by the atmosphere and ocean have been
ocean in June and July where the uncer- computed. The primary means of indicating sources of
mospheric transports is also a maximum errors is the reproducibility of the results across various
5 PW. datasets. Physical constraints have been used to either
zonal mean total transport (Fig. 7d) can remove the largest systematic errors or indicate where
m the ERBE and CERES fields directly, results are not viable, and one component of the un-
y terms are relatively small. The peak certainty, namely that associated with sampling, is esti-
of 5.9 PW exists at 35° in both hemi- mated. Removing the systematic error may underesti-
!2" range of 0.5 PW in the Southern mate the total uncertainty in any value but it nonethe-
nd 0.3 PW in the Northern Hemisphere. less provides a better assessment of the state of
he atmosphere peaks at 5.1 PW at 41°N knowledge with regard to energy flows. Note that the
39°S with a !2" range of 0.5 and 0.2 PW systematic error was removed by adjusting the albedo
rn and Southern Hemispheres, respec- and thus ASR, and is therefore not constant in latitude.
lobal ocean transport maxima occur at In contrast, Wunsch (2005) unrealistically treats the es-
and 11°S (1.2 PW) where the associated timated error in the net radiation as random although
e 0.3 and 0.5 PW, respectively. meridional transports are constrained by the global bal-

Figure 1. Strongly simplified sketch of the global overturning circulation system. In the Atlantic, warm
and saline waters flow northward all the way from the Southern Ocean into the Labrador and Nordic
The aforementioned sinking
Seas. By contrast, there processes,
is no deepwater resulting
formation in Pacific,
in the North deep andwater formation,
its surface are essential
waters are fresher. in
Deep waters formed in the Southern Ocean become denser and thus spread in deeper levels than those
from the North Atlantic. Note the small, localized deepwater formation areas in comparison with the
widespread zones of mixing-driven upwelling. Wind-driven upwelling occurs along the Antarctic
Circumpolar Current (ACC). After Rahmstorf [2002].

15
Warren B., 1981, Deep circulation of the world ocean, in: Evolution of Physical Oceanography, B. Warren
and C. Wunsch (Eds.), MIT Press, Cambridge, MA, 6-41
surfaces of equal density (diapycnal mixing). Munk and Ekman transport, an upwelling from depth is induced that is
Wunsch [1998] described this mechanism in detail. The associated with the so-called Drake Passage effect (see
action of winds and tides generates internal waves in the Figure 2). In this view it is the strength of Southern Ocean
oceans. These waves dissipate into small-scale motion that9 of 19
winds rather than the oceanic diapycnal mixing that governs
causes turbulent mixing. This mixing of heat lightens water the strength of the AMOC. Note that in this theory the
masses in the deep ocean and causes them to rise in low winds induce large-scale motion of the water masses in
latitudes. Resulting surface and intermediate waters are then the Southern Ocean, which enter the Atlantic and flow to
driving the THC. This is schematically depicted in the figure above (Kuhlbrodt et al., 200416).
This is an elaborate version of the famous and easy to understand sketch of Broecker (199117),
which is usually referred to as the “great ocean conveyor”. This paradigm has spurred a
number of qualitative and quantitative studies of the global ocean circulation and the
associated transport of heat and freshwater (e.g. Speich et al. 200118, Seidov and Haupt
200519). Nonetheless, it comes under increasing criticism since it does not take into account in
an appropriate manner the impact of the wind forcing, the meso-scale eddies (illustrated in the
movie OceanVariability.mp4) and other processes that are generally regarded as diffusive or
turbulent from the basin-scale standpoint (Lozier 201020). In addition, it deceives some into
believing that water or tracer particles follow simple, ribbon-like trajectories, which is
definitely not the case (Holzer and Primeau 200621, Shah et al. 201722).
Clearly, novel — and markedly different — idealisations of the World Ocean's role in the
climate system need be developed23. A little known attempt in this direction was made by
Mouchet and Deleersnijder (200824) and Mouchet et al. (201225), consisting in representing
the World Ocean as a leaky funnel (i.e. a highly idealised, one-dimensional fluid flow model),
which led to quantitative diagnoses that compared favourably with numerical simulation
results.

16
Kuhlbrodt T., A Griesel, M. Montoya, A. Levermann, M. Hofmann and S. Rahmstorf, 2007, On the driving
processes of the Atlantic meridional overturning circulation, Reviews of Geophysics, 45, RG2001,
doi: 10.1029/2004RG000166
17
Broecker W.S., 1991, The great ocean conveyor, Oceanography, 4, 79-89
18
Speich S., B. Blanke and G. Madec, 2001, Warm and cold water routes of an O.G.C.M. thermohaline
conveyor belt, Geophysical Research Letters, 28, 311-314
19
Seidov D. and B.J. Haupt, 2005, How to run a minimalist's global ocean conveyor, Geophysical Research
Letters, 32, L07610, doi: 10.1029/2005GL022559
20
Lozier M.S., 2010, Deconstructing the conveyor belt, Science, 328, 1507-1511
21
Holzer M. and F. Primeau, The diffusive ocean conveyor, Geophysical Research Letters, 33, L14618, doi:
10.1029/2006GL026232
22
Shah S.H.A.M., F.W. Primeau, E. Deleersnijder and A.W. Heemink, 2017, Tracing the ventilation pathways of
the Deep North Pacific Ocean using Lagrangian particles and Eulerian tracers, Journal of Physical
Oceanography, 47, 1261-1280
23
This issue is a controversial one. Some believe that high performance computing has rendered idealised
models obsolete, whilst others think the opposite is true. Today's numerical models produce a huge amount of
real numbers. Making sense of them is no trivial task. Specific interpretation methods are needed and, in this
respect, idealised models may be of use. For instance, see http://hdl.handle.net/2078.1/154174
24
Mouchet A. and E. Deleersnijder, 2008, The leaky funnel model, a metaphor of the ventilation of the World
Ocean as simulated in an OGCM, Tellus, 60A, 761-774
25
Mouchet A., E. Deleersnijder and F. Primeau, 2012, The leaky funnel model revisited, Tellus, 64A, 1931, doi:
10.3402/tellusa.v64i0.19131

10 of 19
4. [Tracer in a water column model] Studying the penetration into the ocean of substances
originating from the atmosphere, be they naturally occurring (e.g., radiocarbon unrelated to
tests of nuclear weapons) or human-made (e.g., chlorofluorocarbons), is ripe with lessons to
be learned about transport processes taking place in the ocean and helps developing
mathematical models thereof.
The simplest model that can be conceived is a “water column model”. The latter is based
on several assumptions, the most important of which is horizontal homogeneity. Accordingly,
if t and z denote the time and a vertical coordinate (with z = −h at the ocean bottom and z = 0
at the water-air interface), the concentration C(t, z) of a constituent (or tracer) dissolved in the
water that is originating from the atmosphere obeys
⎧ ∂C ∂2 C
⎪ ∂t = − λ C + K ∂z 2
⎪
⎨ (4.1)
⎪C(0, z) = 0 , C(t,0) = C s , ⎡ −K ∂C ⎤ =0
⎪⎩ ⎢⎣ ∂z ⎥⎦ z=−h
where positive constants λ , K and C s denote the rate of decay of the tracer under
consideration, the vertical diffusivity and the surface concentration, respectively. Determine
the mean life and the half-life of the tracer? In this model, is the ocean bottom assumed to be
impermeable? Justify your answers!
The solution of differential problem (4.1) reads
∞
C(t, z) cosh(σ ε )
C s
=
cosh( ε )
+ ∑ cn exp ⎡⎣ −(ε + kn2 )τ ⎤⎦ cos(knσ ) (4.2)
n=0
with σ = (z + h) / h , kn = (n + 1 / 2)π , cn = 2(−1)n+1 kn / (ε + kn2 ) , and
t time
τ = 2 = (4.3)
h /K diffusion timescale
λh2 h2 / K diffusion timescale
ε= = = (4.4)
K 1/ λ decay timescale
Show that
C(t, z) cosh(σ ε )
lim = (4.5)
t→∞ C s cosh( ε )
and provide a rough estimate of the time needed for the concentration to approach steady-state
profile (4.5). Draw a sketch representing this steady-state profile for a passive tracer ( λ = 0 ),
for a decaying tracer, and for a tracer decaying faster than the previous one.

As was seen in Problem 1, the mean life and the half-life of a constituent (decaying according
to a first-order law with a constant decay rate) is θ = 1 / λ and t1/2 = (log 2) × θ , respectively.
Let constant S represent the (horizontal) cross-sectional area of the water column under
study. Then, assuming that concentration C(t, z) is defined as a mass fraction (implying that
this concentration is dimensionless), the mass flux (expressed in kg/s and positive upwards)
crossing at time t the section of the water column located at altitude z is

11 of 19
 ∂
C(t, z) Φ (t, z) = − ρi SK (S4.1)
∂z
where constant ρi is the reference density (Boussinesq approximation). By virtue of the
pieces of information provided in the problem statement, this flux is identically zero at the
lower boundary of the domain of interest (Figure 4.1):
⎡ ∂ ⎤
Φ (t,−h) = ⎢ − ρi SK C(t, z) ⎥ = 0 (S4.2)
⎣ ∂z ⎦ z=−h
This means that, in the present model, the ocean bottom is impermeable.

water-air z = 0 Dirichlet
interface σ = 1 boundary conditions
domain of interest : – h ≤ z ≤ 0

Figure 4.1. Illustration of the geometry

K : diffusivity (>0) and boundary conditions of the water
column model under consideration.

water column z=–h zero-flux

bottom σ=0 boundary conditions

Solution (4.2) consists of a steady-state part, which is the first term in the right-hand side
of (4.2), and a sum of an infinite number of modes decreasing exponentially as time
progresses: the larger the order of the mode (i.e., n), the faster the decay. As a consequence,
the typical time for the concentration to approach the steady-state limit is related to the mode
exhibiting the slowest rate of decay, i.e., the n = 0 mode. For the latter, the exponential term
may be transformed as follows:
⎛ t⎞
exp ⎡⎣ −(ε + k02 )τ ⎤⎦ = exp ⎜ − ⎟ (S4.3)
⎝ T⎠
where T denotes the sought-after adjustment timescale (Figure S4.2), i.e.,
h2 / K
T = (S4.4)
ε +π2 / 4
Dimensionless parameter ε , which is often referred to as the Damköhler number, is small
if the decay process is much slower than vertical mixing. In this case, adjustment timescale
(S2.10) admits asymptotic expression

12 of 19
 4h 2 16h 2
− T ~ε , ε →0 (S4.5)
π 2K π 4K
and, hence, the adjustment timescale is essentially related to vertical diffusion. If, on the other
hand, decay is much faster than vertical mixing, the adjustment timescale is chiefly associated
with the first-order decay process:
h2 1
T ~ ~ , ε →∞ (S4.6)
Kε λ

0.8
π2 KT / (4h 2 )

0.6

0.4

0.2

0
0 5 10 15
2
Damköhler number: ϵ = λh / K

Figure S4.2. Representation of adjustment timescale (S4.4) as a function of the

Damköhler number.

0.8
σ = (z+h) / h

0.6

0.4 ϵ=0
ϵ=1
ϵ=10
0.2

0
0 0.2 0.4 0.6 0.8 1
s
C(∞,z) / C

Figure S4.3. Steady-state profile of the concentration for various values of the
Damköhler number ( ε = 0, 1, 10 ). The concentrations at the top of the water
column are equal to each other. The vertical derivative of the concentrations is
zero at the ocean bottom (impermeability condition).

13 of 19
 The steady-state concentration is satisfies
C(t, z) C(∞, z) cosh(σ ε )
lim = = (S4.7)
t→∞ C s C s cosh( ε )
Unsurprisingly, the concentration is constant if the decay rate is zero (passive tracer), i.e.,
C(t, z) = C s . If the constituent actually decays ( ε > 0 ), then the concentration is independent
of the decay rate at the ocean surface, C(∞,0) = C s , and is easily seen to be a decreasing
function of the decay rate at the ocean bottom, i.e.,
∂ ⎡ C(∞,−h) ⎤ sinh( ε )
⎢ ⎥ = − < 0 (S4.8)
∂ε ⎣ C s
⎦ 2 ε cosh 2 ( ε )
In fact, in accordance with elementary physical intuition, the concentration is a decreasing
function of the decay rate at any location below the surface ( σ < 1 ):
∂ ⎡ C(∞, z) ⎤ sinh[(1− σ ) ε ] + (1− σ )sinh(σ ε )cosh( ε )
⎢ ⎥ = − < 0 (S4.9)
∂ε ⎣ C ⎦s 2 ε cosh 2 ( ε )
Figure S4.3 illustrates this discussion.
The surface boundary condition is of a Dirichlet type. This is a simplified version of the
boundary condition usually resorted to for dealing with ocean-air gas exchange. A more
realistic parameterisation of the upward gas flux (across the ocean surface) is
⎡ ∂ ⎤
Φ (t,0) = ⎢ − ρi SK C(t, z) ⎥ = ρi S µ ⎡⎣C(t,0) − C s ⎤⎦ (S4.10)
⎣ ∂z ⎦ z=0
where velocity µ is called the “piston velocity” (e.g., Wanninkhof 1992 26 , Jähne and
Hauβecker 199827) and C s actually is the surface saturation value of the concentration under
consideration. In realistic models, neither the piston velocity nor the saturation concentration
can be taken as constant: they are generally evaluated by means of intricate formulas. If the
piston velocity is large, then (S4.10) is approximately equivalent to Dirichlet boundary
condition 0 = C(t,0) − C s , which does not imply that the tracer flux through the ocean surface
is zero.

26
Wanninkhof R., 1992, Relationship between wind speed and gas exchange over the ocean, Journal of
Geophysical Research, 97 (C5), 7373-7382, http://doi.org/10.1029/92JC00188
27
Jähne B. and H. Hauβecker, 1998, Air-water gas exchange, Annual Review of Fluid Mechanics, 30, 443-468,
http://doi.org/10.1146/annurev.fluid.30.1.443

14 of 19
5. [Diffusion versus growth in a coastal ecosystem] Under certain circumstances (e.g.,
excess of nutrients), the population of algae can grow rapidly in aquatic systems, resulting in
adverse effects (e.g., outcompeting other species, production of neurotoxins, decrease in
oxygenCurrent Status ofSee
concentration). Eutrophication
illustrationsinbelow.
the The left-hand side panel depicts Phaeocystis
Belgian Coastal Zone 28
foam on a Belgian beach in May 1998 , whilst the other is an aerial photograph of a red tide
caused by dinoflagellates off La Jolla, California29.

Build a highly idealised one-dimensional model aimed at identifying the condition under
which diffusive or dispersive processes can thwart such detrimental processes!
Edited by
Hint: inspiration may
Rousseau V., be found
Lancelot inD.Chapter
C. and Cox 15 of the course material.

A relevant one-dimensional model must take into account the competition between growth
and dispersion. The former process tends to cause the algal concentration to increase whereas
the latter has the opposite effect, for dispersion tends to transport algae out of the domain
where conditions for growth are favourable.
In the simplest approach, the key parameters are the growth rate r and the diffusivity K,
which, for the sake of simplicity, are assumed to be positive constants. Accordingly, if t and x
denote the time and a suitable space coordinate, concentration C(t, x) of harmful algae obeys
∂C ∂2 C
= rC + K 2 (S5.1)
∂t ∂x
This equation is to be solved in the domain where the conditions (e.g. the availability of
nutrients and light) are favourable to the growth of the algae under consideration, which is
taken to be x ∈]0, L[ . Then, the simplest boundary conditions read
C(t,0) = 0 = C(t, L) (S5.2)
Clearly, such boundary conditions are not beyond reproach. They are merely meant to
indicate that dispersion transports algae out of the domain, i.e.,
⎡ ∂C ⎤ ⎡ ∂C ⎤
⎢⎣ −K ∂x ⎥⎦ ≤0 , ⎢⎣ −K ∂x ⎥⎦ ≥0 (S5.3)
x=0 x=L
These inequalities stem from the fact that the concentration is zero on the domain boundaries
and is non-negative inside the domain. On the other hand, the initial concentration,

28
Source: Rousseau V., C. Lancelot and D. Cox (Eds), 2006, Current Status of Eutrophication in the Belgian
Coastal Zone, Presses Universitaires de Bruxelles, 122 pages
29
Source: https://commons.wikimedia.org/wiki/File:La-Jolla-Red-Tide.780.jpg, last viewed on 11 March 2021

15 of 19
C(0, x) = C 0 (x) , will be seen to be unimportant in the present context.
The solution to the abovementioned partial differential problem is
∞
C(t, x) = ∑ a j exp(γ j t)sin(k j x) (S5.4)
j=1

with
L
2 jπ
aj =
L0∫ C 0 (x)sin(k j x) dx , γ j = r − K k 2j , kj =
L
(S5.5)

Growth rate γ j decreases as the order of the mode, j, increases. The higher-order modes
( j >> 1) decrease as time progresses ( γ j < 0 ). However, it is possible that some of the other
modes increase. Accordingly, the condition for an algal bloom to occur is
max γ j = γ 1 > 0 (S5.6)
j
which is equivalent to
1 L2
< 2 (S5.7a)
r π K
Therefore, the first mode increases exponentially if the timescale characterising algal growth,
1 / r , is smaller than that related to dispersion, L2 / (π 2 K ) . In other words, a bloom occurs if
algal growth proceeds faster than dispersion, which transports algae out of the domain where
growing conditions are favourable. Needless to say, no algal bloom will occur if dispersion is
sufficiently efficient, i.e., if
1 L2
> 2 (S5.7b)
r π K
The boundary conditions implemented above are simplistic. It is believed, however, that
having recourse to more realistic boundary conditions would not radically modify the
condition for an algal bloom to develop. For instance, if the x = 0 boundary is impermeable
rather than open (in which case x = 0 may represent the coastline), then (S5.2) transforms to
⎡ ∂C ⎤
⎢⎣ −K ∂x ⎥⎦ = 0 , C(t, L) = 0 (S5.8)
x=0
and the concentration is
∞
C(t, x) = ∑ a j exp(γ j t)cos(k j x) (S5.9)
j=1

with
L
2 ( j − 1 / 2)π
a j = ∫ C 0 (x)cos(k j x) dx , γ j = r − K k 2j , kj = (S5.10)
L0 L
Then, the condition for an algal bloom to take place is γ 1 > 0 , i.e.,
1 4L2
< 2 (S5.11)
r π K
which, as expected, bears similarities with criterion (S5.7a).

16 of 19
 amounts of plastic debris. The highest with a maximum concentration 15 times its ini- observed in the high plastic concentration re-
ncentrations were observed between 22° tial value. This convergence, centered at approx- gion (fig. S4). Although the nonuniform sam-
(Fig. 1 and figs. S1 and S2), where imately 30°N, directly corresponds to the observed pling in this data set cannot resolve short spatial
otal plastic pieces were collected. The high plastic accumulation region (fig. S2). or temporal scale variability, no robust trend was
mple collected in a single 30-min tow Further, the model indicates that the min- observed in the broadest region of plastic accu-
pieces at 24.6°N, 74.0°W in May 1997, imum time for surface tracer (i.e., drifter or plastic) mulation on interannual time scales and longer.
t to 580,000 pieces km–2. The maxi- to reach the collection center from the U.S. east- Although no direct estimates of plastic input
6. [Plastic debris pollution] Nowadays, about eight million tons of mismanaged plastic
mple size reported in Atlantic studies ern30,31 seaboard is less than 60 days, at least half the to the ocean exist, the increase in global produc-
1970s ranged fromwaste 12,000enter
piecesthekm oceans
–2 . The size
time required of plastic
to travel debrisorfound
from Europe Africa.in the
tion marine
of plasticenvironment varyincrease from
materials [fivefold
–2
7,000 pieces km over (16).aComparative-
very wide range The influence
of ordersofofthemagnitude,
Gulf Streamfromis particular-
the size of1976 to 2008
a soda (23)] to
bottle together
muchwithlessthe increase in
astic concentrations were one
than measured in lyThe
micron. evident in some of
transport and the fastest propagation discarded plastic in U.S. municipal solid waste
est to land, such as along the Florida times—40 days from Washington, D.C., and Mi- (MSW) [fourfold increase from 1980 to 2008
transformations of such human-made
Florida Keys, in the Gulf of Maine, ami, Florida, for example—in which tracer trav- (24) (Fig. 3)] suggest that the land-based source
Caribbean islands.debris are still
The average littleeled
plastic known,
along but are now
the coast before entering the gyre of plastic into the ocean increased during the study
tion measured within the Caribbean
the subject of intensive interior.
studiesAlthough
32
. not indicative of the size or period. Ocean-based sources may have decreased
km–2, while
nly 1414 T 112 piecesThere is nothatdoubt,location
however,of land-based sources, or of the age of in response to international regulations prohib-
that their
f of Maine was 1534 T 200 pieces km–2, debris, these estimates demonstrate how quickly iting dumping of plastic at sea (25). Given the
impact on
than an order of magnitude lowermarine
than lifeplasticand, presumably,
entering the ocean near major U.S. pop- measured steady plastic concentration in the west-
ge concentration near 30°N (20,328 T ulation centers could For
human health is a detrimental one. affect an area more than ern North Atlantic, loss terms must exist to off-
es km–2, 29° to 31°N). Although in our 1000 km
instance, a significant percentage of the offshore. set the presumed increase in plastic input to the
on the latitudinal bounds of the highest We observed no strong temporal trends in plas- ocean.
mass of marine animals can consist of
ncentration are well defined, the eastern tic concentration in the 22-year data set (Fig. 3). A change in the type of plastic material en-
plastic because
not yet been determined debris, ofasais illustrated in thevariability
Large interannual figure was observed within tering the ocean could affect the observed amount
rect observations. opposite, which is due thetohigh plastic
Kathy concentration region, and a lin- of floating debris. Density analysis of plastic sam-
Townsend
gion of highest plastic concentration is ear fit to
and was provided to me by Kay Critchell annual mean plastic concentration had ples collected at the sea surface revealed that
sociated with the large-scale subtropical a slope not different from zero (–20 T 217 pieces 99% were less dense than seawater, and ele-
(personal
ce in the surface velocity fieldcommunication,
created June 2017).
driven Ekman currents and geostrophic
n (Fig. 2). This convergence zone, in-
50˚N Fig. 1. Distribution of plastic ma-
rine debris collected in 6136 sur-
y converging streamlines and current
face plankton net tows on annually
less than 2 cm s–1, extends across most
repeated cruise tracks from 1986
btropical North Atlantic basin (20) and to 2008 in the western North At-
with the highest observed plastic con- lantic Ocean and Caribbean Sea.
s. This correspondence not only ex- 40˚N Symbols indicate the location of
plastic distribution but also illustrates each net tow; color indicates the
ng debris acts as a tracer of large-scale measured plastic concentration in
ace currents. pieces km–2. Black stars indicate
the convergence acts to concentrate tows with measured concentration
ebris, the geographic origin of the de- 30˚N greater than 200,000 pieces km–2.
ot be easily determined from current Symbols are layered from low to
plastic pieces
r from the recovered plastic samples per km
2 high concentration.
s. To address this question, we used 200,000
satellite-tracked drifting surface buoys 100,000
21) to examine pathways into and out 20˚N
50,000
ntral region” of high plastic concentra-
20,000
o 34°N, 60° to 70°W). Of 1666 drifters
eployed in the North Atlantic (0° to 10,000

o 90°W) from 1989 to 2009, 24 drifters 10˚N 5000

oyed in the central region and 116 others 0
rough this region. The trajectories of
ntral region” drifters were strongly con- 90˚W 80˚W 70˚W 60˚W 50˚W

As marine plastic 2010

3 SEPTEMBER debrisVOL
mostly
329 originate
SCIENCE from land and rivers, their maximum
www.sciencemag.org
concentrations should be observed in the immediate vicinity of the coasts if their
concentration obey the well-known properties of the solution of an advection-diffusion

30
Jambeck J.R., R. Geyer, W. Wilcox, T.R. Siegler, M. Perryman, A.Andrady, R. Narayan and L.L. Law, 2015,
Plastic waste inputs from land into the ocean, Science, 347, 768-771
31
Countless initiatives are aimed at addressing this regrettable situation, which has gone on for too long. See, for
instance, Tangaroa Blue (www.tangaroablue.org).
32
https://topios.org/

17 of 19
equation.
Small plastic debris typically millimetres in size located at the ocean surface or very near
to it can be collected by plankton net tows. The figure above (which is Figure 1 of Law et al.
201033) shows that such debris tend to accumulate in the North Atlantic Subtropical Gyre, i.e.,
hundreds of kilometres away from the nearest shores.
How can one reconcile these field data with the results of the theory of classical advective
and diffusive transport?
Hint: Inspiration may be found in Deleersnijder (201734).

Let t and x h = (x, y) represent the time and the horizontal part of the position-vector, where x
and y are horizontal coordinates. The concentration n(t,x h ) of plastic debris located at or
very near to the ocean surface (number of debris per unit area) is likely to obey the following
two-dimensional advection-diffusion equation
∂n
= − λ n − ∇ h i (nu − κ h∇ h n) (S6.1)
∂t
where ∇ h is the horizontal part of the del operator; u is the sum of the surface velocity of the
water and an appropriate velocity representing the impact of the wind (e.g. Critchell et al.
2015 35 ); κ h is the relevant horizontal (turbulent) diffusivity. In the equation above,
degradation and sinking processes are taken into account by means of a first-order law,
including non-negative parameter λ , which is unlikely to be a constant. Clearly, (S6.1) may
be rewritten as follows
∂n
= − µn − (u − ∇ hκ h ) i ∇ h n + κ h∇ 2h n (S6.2)
∂t
with
µ = λ + ∇h i u (S6.3)
Assume that a local concentration maximum is located at a point whose position-vector is
r(t) . Then, the value of the debris concentration at this point obeys
d ⎡ ∂n dr ⎤
n[t,r(t)] = ⎢ + i ∇hn ⎥ (S6.4)
dt ⎣ ∂t dt ⎦ x=r(t )
Next, combining (S6.2) and (S6.4) yields
d ⎡ ⎛ dr ⎞ ⎤
n[t,r(t)] = ⎢ − µ n − ⎜ u − ∇ hκ h − ⎟ i ∇ h n + κ h∇ 2h n ⎥ (S6.5)
dt ⎣ ⎝ dt ⎠ ⎦ x=r(t )
At a local maximum, ∇ h n = 0 , allowing one to simplify (S6.5) as follows

33
Law K.L., S. Morét-Ferguson, N.A. Maximenko, G. Proskurowski, E.E. Peacock, J. Hafner and C.M. Reddy,
2010, Plastic accumulation in the North Atlantic Subtropical Gyre, Science, 329, 1185-1188
34
Deleersnijder E., 2017, Analytical developments regarding plastic debris pollution of the marine environment,
Working Note, Université catholique de Louvain, Louvain-la-Neuve; Belgium, 10 pages, available at URL
http://hdl.handle.net/2078.1/190114
35
Critchell K., A. Grech, J. Schlaefer, F.P. Andutta, J. Lambrechts, E. Wolanski and M. Hamann, 2015,
Modelling the fate of marine debris along a complex shoreline: Lessons from the Great Barrier Reef, Estuarine,
Coastal and Shelf Science, 167, 414-426, https://doi.org/10.1016/j.ecss.2015.10.018

18 of 19
 d
n[t,r(t)] = ⎡⎣ − µ n + κ h∇ 2h n ⎤⎦ (S6.6)
dt x=r(t )

Laplacian ∇ 2h n is non positive. However, if ∇ h i u is negative (convergent surface velocity)

and its order of magnitude is sufficiently large that µ be negative, then it is possible that
dn[t,r(t)] / dt be positive, implying that the concentration would increase. In the long run, the
maximum value of the concentration would occur inside the domain of interest. If, on the
other hand, parameter µ is positive (which is the case if ∇ h i u > 0 ), then the maximum value
of the concentration will decrease. The maximum would remain located on the boundary of
the domain, probably at the mouth of one of the estuaries injecting plastic debris into the
ocean.
The key difference between the equation governing the concentration of plastic debris and
that obeyed by a dissolved constituent (exhibiting a first order decay process) is that in the
latter the (three-dimensional) velocity field is divergence-free, implying that a local
concentration maximum exhibits a non-increasing concentration value. By contrast, for plastic
debris, velocity u is unlikely to be divergence-free, which may cause the maximum value of
the concentration to be located inside the domain of interest.

19 of 19