Flow Systems of The Earth's Viscous Subsurface A Complement Togroundwater Flow Systems

Ain Shams Engineering Journal xxx (xxxx) xxx
Contents lists available at ScienceDirect
Ain Shams Engineering Journal

journal homepage: www.sciencedirect.com
Civil Engineering
Flow systems of the Earth’s viscous subsurface: A complement to

groundwater flow systems
Wouter Zijl a, Mustafa El-Rawy b,c,⇑
a
Dept. of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium
b
Civil Engineering Department, Faculty of Engineering, Minia University, Minia 61111, Egypt
c
Civil Engineering Department, College of Engineering, Shaqra University, Dawadmi 11911, Ar Riyadh, Saudi Arabia
a r t i c l e i n f o a b s t r a c t
Article history: Assessment of the long term possibilities and risks related to geological storage requires insight in the
Received 29 June 2020 deep groundwater flow systems. The objective of this paper is to show the relevance of the deep creep
Revised 28 August 2020 flow of the earth’s viscous upper mantle and crust as a complement to the groundwater flow. The paper
Accepted 31 August 2020
presents an approach based on Fourier decomposition of the topography. The creep flow equations are
Available online xxxx
solved analytically, which results in simple indices like penetration depth and relaxation time character-
izing the gravity-driven creep flow. Thanks to the very high effective viscosity of the Earth’s subsurface a
Keywords:
Darcy-like equation is obtained in which the ‘creep conductivity’ is Fourier mode dependent, which
Creep flow
Crust and upper mantle
allows for simple comparison with the hydraulic conductivity for groundwater flow. Order of magnitude
Darcy’s law calculations indicate that for horizontal length scales of 100–1000 km the subsurface creep velocities are
Deep geological storage 0.3–30 mm/year, respectively, which shows that creep velocities in the deep subsurface are significant
Effective viscosity with respect to deep groundwater velocities.
Groundwater flow Ó 2020 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams
Unsteady Stokes Equation University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).
1. Introduction geological storage of greenhouse gases like carbon dioxide (CO2)

and methane (CH4) in deep aquifers or depleted gas fields is a pre-
Global average atmospheric CO2 concentrations have fluctuated requisite. Capture and storage of carbon dioxide are considered
between 180 and 280 parts per million (ppm) in the last essential for meeting the targets for reducing CO2 emissions
420,000 years, but since the Industrial Revolution CO2 concentra- [2,3]. Leung et al. [2] reviewed various technologies and issues
tions have increased to more than 400 ppm [1]. At present, the related to CO2 capture, separation, transport, storage and monitor-
Earth’s Global Average Temperature (GAT) rises at a rate of ing. Cuéllar-Franca and Azapagic [3] analyzed the environmental
0.015–0.018 °C/year. Realistic scenarios limiting global warming impacts of various CCS and CCU options for capture, storage and/
to 1.5–2.0 °C above the pre-industrial level (before 1850) assume or use of CO2. Ajayi et al. [4] reviewed the storage of CO2 in geolog-
human intervention in the climate system. Carbon Dioxide ical formations with emphasis on modeling, monitoring, and esti-
Removal (CDR) is a well-known form of intervention. CDR has mation of capacity.
two components: Carbon Capture and Utilization (CCU) and Car- Also, subsurface storage of hydrogen, biogas, and natural gas
bon Capture and Storage (CCS). To assure stable CCU operation, has to be considered. In addition, long term geological storage of
nuclear waste may also be necessary for carbon-free energy pro-
duction. Regarding CCS, lessons can be drawn from earlier studies
⇑ Corresponding author at: Civil Engineering Department, Faculty of Engineering, related to site selection for repositories of nuclear waste [5] and,
Minia University, Minia 61111, Egypt regarding the short-term technical questions, from petroleum
E-mail addresses: [email protected] (W. Zijl), [email protected] reservoir engineering [6].
(M. El-Rawy).
To assess the long-term possibilities and risks of deep geological
Peer review under responsibility of Ain Shams University.
storage, insight in gravity-driven regional groundwater flow sys-
tems is necessary [7]. Lessons can be drawn from studies related
to site selection for repositories of nuclear waste [5], from studies
on shallow and deep geothermal energy and, especially, from
Production and hosting by Elsevier
https://doi.org/10.1016/j.asej.2020.08.017
2090-4479/Ó 2020 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: W. Zijl and M. El-Rawy, Flow systems of the Earth’s viscous subsurface: A complement to groundwater flow systems, Ain Shams
Engineering Journal, https://doi.org/10.1016/j.asej.2020.08.017
W. Zijl, M. El-Rawy Ain Shams Engineering Journal xxx (xxxx) xxx
studies related to secondary oil migration with the spin-off to [28]. In this paper, this will be called the dynamic boundary condi-
migration of stored CO2 [8–15]. tion. The position of the undulating water table height was
Regarding the subsurface utilization, a rough distinction can be assumed to be related to the topographic height; this may be con-
made between the shallow, intermediate and deep subsurface sidered as a reasonable approximation for regional flow. Thanks to
this boundary condition, Tóth found that discharging groundwater
- shallow subsurface <500 m: groundwater production, brine is not only flowing into the surface waters but also into the lower
storage part of the basin in the neighborhood of the surface waters [26].
- intermediate subsurface 0.5–2 km: storage of solid waste, Regarding the deeper flow, Tóth rediscovered the emergence of
geothermal energy complex flow patterns found earlier by D’Andrimont [29]. A well-
- deep subsurface >2 km: production of gas, oil, salt, storage of known example of such a flow pattern is shown below in Sec-
CO2 tion 2.1, Fig. 1 [7] and aptly called ‘hierarchically nested flow sys-
tems’ by Engelen and Jones [30]. Relevant developments and
The focus of this paper is mainly on the intermediate and deep applications by others than Tóth have been presented by Freeze
subsurface. In the shallow subsurface, the creep flow rates are neg- and Cherry [31], Batelaan [32] and Zijl and El-Rawy [33].
ligibly small with respect to the groundwater flow rates. Moreover, It is relatively simple to extend this approach to the freely mov-
in situations where the shallow subsurface is moving, the creep ing parts of the water table (the parts that are not fixed by the sur-
flow rate is generally negligibly small with respect to compaction face waters). The time evolution of the initial water table height
due to groundwater depletion, e.g., by over-exploitation of fresh can be determined by a kinematic boundary condition relating to
groundwater, or by dewatering polder areas. the time derivative of the water table height to the groundwater
To obtain insight into the subsurface phenomena, a model- velocity normal to the water table [28,24]. The fact that the thus-
based analysis is indispensable. Modeling is generally based on obtained fluxes may deviate significantly from the measured fluxes
(i) a forecast model, (ii) observations/measurements, and (iii) a fil- at the water table can be remediated by application a conventional
ter to determine a weighted average of the forecasts and the obser- flux-based model and use the results to adapt the groundwater
vations as well as an indication of the uncertainty. The model may flow conductivity (the permeability of the subsurface) to match
be (i) comprehensive (based on multi-million grid blocks and time the pressure-based calculation with the conventional flux-based
steps), (ii) Simple (analytic or semi-analytic like, for instance, calculation, without losing the insights gained from the pressure-
Tóth’s model of nested flow systems) and (iii) Indices, i.e., order- based model [16–18,34–39]. A similar approach has been pre-
of-magnitude estimations for qualitative characterization, like sented by Johnson and Fletcher [40] and by Ranalli [25] to deter-
the indices global average temperature (GAT), body mass index mine the simple index effective creep viscosity to characterize
(BMI), gross domestic product (GDP), etc., as used in different con- the postglacial rebound of the Earth’s mantle-crust system.
texts. The filter and its associated uncertainty estimation may be The creep flow discussed in this paper will be based on a similar
comprehensive, e.g., a Kalman Filter [16–18], or more or less intu- approach to the boundary conditions. The dynamic and kinematic
itive, based on the expert’s knowledge and experience. The uncer- boundary conditions are applied on the topographic level instead
tainty of simple models and indices may be based on Bayesian of the water table. The kinematic condition is the same as in
techniques for hypothesis testing [19–20]. groundwater hydrodynamics, while in the dynamic condition the
Indices play an essential role in deciding whether further component of the creeping fluid’s total stress normal to the topo-
research based on comprehensive or straightforward models has graphic level is equal to the atmospheric stress (i.e., equal to minus
to be recommended. In this paper, the focus is on relatively simple
models leading to indices. Indices characterizing Tóthian ground-
water flow systems are hydraulic conductivity, penetration depth,
relaxation time [7,15,21–24]. Deep creep flow and groundwater flow
Creep flow is the long-term motion of the Earth’s upper mantle
and crust [25]. The objective of this paper is to show that, in a more Analycal model resulng in characterisc
or less similar way as for groundwater flow, indices for creep flow indices
systems like effective creep conductivity, penetration depth and
relaxation time makes sense. Although the relatively thin crust Spaal Fourier decomposion of
has higher effective viscosity and density than the upper mantle topographic height
(the crust is ‘‘floating” on the upper mantle) our indices approach
will be based on the concept of one effective creep viscosity and Thóan flow systems
density for the whole large-scale mantle-crust system. Spatial vari- paradox of moving stagnaon points
ations in viscosity, density and other geo-mechanical phenomena
can be accounted for in more comprehensive (numerical) models.
Creep flow equaons
Also see Section 6.
connuity, unsteady Stokes, kinemac and
The creep flow approach presented in this paper is inspired by
dynamic topography condions
Tóth’s ideas about the analysis of gravity-driven shallow and deep
groundwater flow systems, specifically, Tóth’s approach to the
boundary condition on the water table of a groundwater basin Imposed characterisc parameters
[26,27]. Although in the deeper rock formations the permeability gravitaonal acceleraon, density, viscosity
may differ appreciably from the permeability in shallower more
sandy formations, the assumption of homogeneous effective per- Calculated characterisc results
meability yields interesting insights of how the groundwater viscous-ineral force balance, over-damped
behaves. In contrast to the common boundary condition, in which flow criterium, Darcy-like creep flow, creep
the groundwater’s velocity normal to the water table is specified, vs. groundwater velocity, orders of
Tóth imposed the groundwater head on the water table to be equal magnitude
to the water table height, which is equivalent to imposing the
groundwater pressure to be equal to the atmospheric pressure Fig. 1. Flowchart of steps in the workflow of mathematical-physical presentation.
2
the atmospheric pressure). In this way, creep flow patterns that are neighboring fluid particles flowing together to a stagnation
more or less similar to the hierarchically nested groundwater flow point are separated from each other after arriving at the
systems can be found. bifurcation point; one particle will become part of the dee-
Although the boundary conditions for creep flow resemble per flow system, while it’s very close neighbor will flow into
those for the analysis of groundwater flow systems, the equations the local flow system. This radical separation of streamlines
governing the creep flow of the subsurface differ substantially from exemplifies a general phenomenon: predictions break down
the equations for the groundwater flow (volumetric balance equa- at a ‘tipping point.’ For a thorough analysis of two- and
tion and Darcy’s law). Creep flow is governed by the continuity three-dimensional groundwater flow patterns in the neigh-
equation (volumetric balance for incompressible flow) and the borhood of stagnation points, see [41].
unsteady Stokes equation (the Navier-Stokes equation for small (ii) Gradient lines. In this case, the lines with arrows show the
Reynolds number). These flow field equations complemented with direction of the groundwater head gradient. In the points
boundary conditions result in a balance between the inertial forces where two neighboring flow bodies have different direc-
and the viscous forces while the pressure turns out to be litho- tions, the head gradient is equal to zero. Zero-gradient
static. In contrast, the solutions presented by Ranalli [25] and John- points are generally called stationary points. Stationary
son and Fletcher [40] differ because they base their approach on points play an essential role in the explanation of Artesian
the steady Stokes equation (or, equivalently, on the biharmonic flowing wells [7].
equation for the stream function), which means that they neglect
the inertia of the subsurface mass ab initio, without justification. In groundwater flow, the stagnation points and the stationary
For very viscous flow our solution results in a Darcy-like equation points coincide.
in which the ‘creep conductivity’ has a high value for very
extended, superregional, and deep creep flow systems, while it 2.2. Creep flow
has an extremely small value for the shallower regional and local
creep flow systems. Consequently, the flow velocities of the very Also, in creep flow, the iconic picture of Tóthian flow systems, as
extended and deep systems have relatively high velocities com- shown in Fig. 2, holds. In this case, Fig. 2 shows the cross-section of
pared to the very small groundwater flow velocities at the greater a part of the Earth’s mantle-crust system with two-dimensional
depths. Therefore, studies on the subsurface migration of contam- creep flow. Also, in this case, the lines with arrows have two mean-
inants leaking from sites for deep geological storage have to take ings: (i) streamlines and (ii) gradient lines. However, in creep flow,
the creep flow into account as a complement to groundwater flow. the stagnation points do not coincide with the stationary points.
The workflow representing the mathematical-physical presenta-
tion in this paper is summarized in a flowchart; see Fig. 1.
2.3. Stagnation point paradox
2. Creep flow and groundwater flow systems In general, the flow will time-dependent. This means that the
zero-velocity points are moving with a non-zero velocity, which
2.1. Groundwater flow seems to be a paradox. The paradox is solved in statistical thermo-
dynamics and, more specifically, the kinetic theory of non-uniform
Let us first consider groundwater flow by focusing on the iconic gases and liquids [42,43]. These theories, which form the basis for
picture of Tóthian flow systems shown in Fig. 2 [26,27,7]. the equations governing groundwater flow and creep flow, are
The lines with arrows in Fig. 2 have two meanings: based on a six-dimensional phase space with six independent coor-
dinates ½x; y; z; v x ; v y ; v z , where x; y; z are the spatial coordinates
(i) Streamlines. The lines with arrows show streamlines, i.e.,
and v x ; v y ; v z are the fluid velocity components [44]. The equations
they show the direction of the groundwater velocity. In the
governing groundwater flow or creep flow select a time-dependent
points where two neighboring flow bodies have different
subspace with coordinates ½ðxðtÞ; yðtÞ; zðtÞ; v x ðtÞ; v y ðtÞ; v z ðtÞ in
directions, the velocity is equal to zero. In Fig. 2, the starting
which t is the time. Depending on the boundary conditions, one
point of the line pointing to the text ‘Direction of flow’ is in
or more stagnation points S with phase space coordinates
the neighborhood of a zero-velocity point. Zero-velocity
½ðxS ðtÞ; yS ðtÞ; zS ðtÞ; 0; 0; 0 may occur, from which it follows that
points are generally called stagnation points. Stagnation
the velocity of a stagnation point is ðdxS =dt; dyS =dt; dzS =dtÞ. On
points are bifurcation points or ‘tipping points.’ Two
the other hand, the trajectory of a fluid particle P with coordinates
½ðxP ðtÞ; yP ðtÞ; zP ðtÞ; v P;x ðtÞ; v P;y ðtÞ; v P;z ðtÞ follows from the particle
tracking equation ðdxP =dt; dyP =dt; dzP =dtÞ = ðv P;x ; v P;y ; v P;z Þ. This
approach holds in a similar way for quasi-stagnation points in a
suitably chosen x-z cross section with velocity v x ¼ v z ¼ 0 and rel-
atively small velocity v y normal to the cross section. Because in the
phase space approach the six coordinates are on equal footing (are
independent from each other), the particle tracking equation is not
the definition of fluid velocity but is an additional law of nature.
Carathéodory [45] was the first who developed a mathematically
rigorous continuum theory dealing with statistical continuum
thermodynamics and related continuum theories.
2.4. Introduction to the basic equations
On long time scales, the Earth’s subsurface behaves as a viscous

Fig. 2. Cross-section of a homogeneous and isotropic groundwater basin with two-
fluid [25]. Assuming that its rheology may be considered as New-
dimensional groundwater flow; the flow component normal to the picture is tonian, the flow is governed by the Navier-Stokes equation for
assumed to be negligibly small (updated after Tóth [27]). incompressible fluids [46,47,25,48]. Compared to water, which
3
has a viscosity with the order of magnitude 103 Pas, the effective Although in creep flow the pressure head is equal to zero, it is
viscosity of the mantle-crust system amounts to the extremely shown in Section 3.9 that it makes sense to define the topographic
high value of 1021–1022 Pas [40]. In that case, the Reynolds num- potential h. The topographic potential is a harmonic function (sat-
ber is extremely small, which means that the steady inertial forces isfies Laplace equation (1)), while on reference surface z ¼ 0 the
are negligibly small with respect to the viscous forces (the topographic potential is specified as
unsteady inertial forces have to be taken into account). As a conse-
quence, the nonlinear Navier-Stokes equation simplifies to the lin- hðx; y; 0; tÞ ¼ Hðx; y; tÞ ð3Þ
ear unsteady Stokes equation.
where z ¼ Hðx; y; tÞ represents the topographic surface of the
In his presentation on flow systems of planet Earth, Engelen
earth.
[49] suggested to devote special attention to the flow’s vorticity.
Following his advice, the Stokes equation will be decomposed into
3. Flow
two equations: (i) a Laplace equation for the pressure head and (ii)
a diffusion equation for the vorticity (for the proof cf. Appendix A).
3.1. Fourier decomposition
These two equations are not coupled and can, therefore, be solved
independently. However, the boundary conditions for the two
The topographic level with respect to a horizontal reference
equations are coupled. Therefore, the compatibility of the two
level will be decomposed into a series of Fourier modes. For each
solutions has to be checked by substitution into the Stokes
topographic Fourier mode, the flow solution, called the Fourier
equation.
flow system, can be determined in a relatively simple way. The
full solution is then obtained by summing the Fourier flow
2.5. Pressure head and topographic potential systems.

The undulating topography z ¼ Hðx; y; tÞ is decomposed in
The Laplace equation for a general function f is given by Fourier components
! !
@2f @2f @2f X
x

y

x

y
r2 f þ þ ¼0 ð1Þ H x; y; t ¼ Hm ðtÞsin þ m þ Hm
C0 ðtÞcos þ m ð4Þ
@x2 @x2 @z2 m S0 Lm

x
L
y
Lm

x
L
y
A function f satisfying this equation is called a harmonic func- Decomposition in an integral or series of sinuses and cosines is

tion. The x; y coordinates are global horizontal coordinates, while a powerful tool for solving the linearized flow equations (cf.
z is the vertical depth coordinate. The difference between global Appendix D)
coordinates and the Fourier mode-specific horizontal coordinate
is explained in Section 3.1. Under the assumption of a 3.2. Coordinate transform
depth-independent density it makes sense to define the pressure
head We assign to each Fourier mode m its own xm z cross section
with horizontal length scale
p x; y; z; t patm qgz vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
hp x; y; z; t ¼ ð2Þ u 1
qg L ¼t
m
2 m 2
ð5Þ
ð1=Lm Þ þ ð1=L Þ
(p is the pressure, patm is the atmospheric pressure, q is the den-

x y
sity of the Earth’s mantle-crust system, and g is the gravitational In such a way that the topographic height results in the sinu-
acceleration); the thus-defined pressure head is a harmonic func- soidal Fourier mode
tion (Appendix B). In order to find a solution that is compatible qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 m 2
with the flow solution the boundary condition for the pressure Hm ðx; tÞ ¼ ðHm m m m
S0 Þ þ ðH C0 Þ sinðx=L Þ ¼ H0 sinðx=L Þ ð6Þ
head is hP ¼ 0 on the horizontal reference surface z = 0 (cf. Sec-
tion 3). As a consequence of the Laplace equation (Eq. (1)) this For a derivation of this approach see Appendix F. From now on
boundary condition results in hP ¼ 0 in the whole subsurface under the Fourier mode’s index m will be omitted. The part of the Earth’s
consideration; the pressure turns out to be lithostatic; i.e., the mantle-crust system considered below has horizontal extension
pressure is equal to the overload weight per unit area, patm þ qgz
1
pL x 32 pL and depth 0 z < 1.
R 2
(or patm þ qðzÞgdz if the density is a function of only the depth,
z). This weight is an important control of the subsurface’s perme- 3.3. Creep flow field equations
ability as well of its viscosity. The lithostatic pressure gradient is
a function of rock density and is between 22.7 and 25.0 kPa/m The equations governing creep flow can be written in the fol-
(1.0 and 1.1 psi/ft) in coal-bearing sequences [50]. For the order lowing form (Appendices A and C)
of magnitude estimations presented in Section 4 a lithostatic pres- @v x @v z
sure gradient of qg ¼ 30 kPa/m has been assigned to the upper x¼ ðvorticity in y directionÞ ð7Þ
@z @x
mantle and crust under consideration.
!
Also, in incompressible groundwater flow through a homoge- @x @2x @2x
neous and isotropic basin, the pressure head (the hydraulic head, q ¼l þ 2 ðvorticity diffusion equationÞ ð8Þ
@t @x2 @z
in groundwater terminology) is harmonic. However, in that case
the pressure head hP on the reference surface z = 0 is equal to
@v x @v z
the topographic head H above z ¼ 0, where H represents the undu- þ ¼ 0 ðcontinuity equation for incompressible flowÞ
lating topography of the water table [28]. Consequently, in ground- @x @z
water flow the pressure head hp is equal to the topographic ð9Þ
potential h defined below.
4
3.4. Creep flow boundary conditions 3.5. Time scales
To be able to construct analytical solutions the topographic Eq. (8) contains the vorticity-related time scale T V and Eq. (11)
height z ¼ Hðx; tÞ is projected on the horizontal reference level contains the stress-related time scale T S
z ¼ 0. Such a geometrical linearization is sufficiently accurate for
relatively small topographic slopes jHj << L. qL2
TV ¼ ð12Þ
Due to the retreat of a heavy ice cap the topography is rising or l
falling with a rate @H=@t equal to the upward directed vertical
velocity v z on the topographic level z ¼ H projected on z ¼ 0 2l
TS ¼ ð13Þ
@Hðx; tÞ
qgL
¼ v z ðx; 0; tÞ on z ¼ 0 ð10Þ
@t As will be shown below, the ratio T V =T S plays an important role.
In this kinematic boundary condition, the nonlinear term Although we have assumed that both the effective viscosity, l, and
v x @H=@x is neglected in the left-hand side. The subtle difference the density, q, are constant, Eqs. (12) and (13) show that the result-
between surface z ¼ 0 (not necessarily an exactly flat surface and ing solution may be generalized to situations in which l=q is con-
an exactly flat surface Z ¼ 0 is explained in Appendix G. stant (the crust’s higher effective viscosity with respect to that of
The topographic undulations in the relatively thin zone jzj < jHj the mantle will be more or less compensated by its higher density).
above and below reference surface z ¼ 0 result in a force per unit
area on the reference surface. In the subsurface section where 3.6. Initial condition (at time t = 0)
H > 0 this force pushes the topographic level downward (in þz
direction), while in the subsurface section where H < 0 this force It is assumed that initially, at t = 0, there is sudden change in
pulls the topographic level upward (in z direction). The total nor- topographic height, for instance caused by the relatively sudden
mal stress in the earth, rzz ¼ p þ 2l@ v z =@z, and the total normal disappearance of a land ice cap, to a sinusoidally undulating
stress in the earth in the atmosphere, patm , have to be continuous topography
at the earth-air interface. This means that 2l@ v z =@z ¼ p patm at x
the topographic level z ¼ H. In the zone jzj < jHj the pressure is Hðx; 0Þ ¼ H0 sin att ¼ 0 ð14Þ
L
assumed to be lithostatic, i.e., p ¼ patm þ qgz resulting in
2l@ v z =@z ¼ qgH on z ¼ H. Assuming, in addition, that the verti- For the case in which the change in topographic height is not a
cal velocity at z ¼ H and the vertical velocity at z ¼ 0 hardly differ step function, but a function of time, the solution is presented in
from each other, i.e., assuming that ð@ v z =@zÞðx;H;tÞ ffi ð@ v z =@zÞðx;0;tÞ , Appendix G.
results in the dynamic boundary condition at horizontal reference
level z = 0 3.7. Assumed creep flow evolution
@v z qgH
¼ on z ¼ 0 ð11Þ For t > 0 it is assumed that the creep flow causes a slow expo-
@z 2l nential relaxation a to flat topography in which the relaxation time,
In the argument for this boundary condition use has been made of T, follows from the solution for the velocity (cf. Section 3.9)
the lithostatic pressure assumption p ¼ ðp patm qgzÞ=qg ¼ 0 in x
t
the zone jzj < jHj around reference level z = 0, from which it follows Hðx; tÞ ¼ H0 sin exp ð15Þ
that the pressure is lithostatic in the whole flow field, cf. Section 2; L T
and in Appendix C. it is shown that the flow solution that will be The Laplace equation for topographic potential h (similar to Eq.
derived below from Eqs. (7)–(11) is compatible with the assump- (1) for f ) together with the top boundary condition (Eq. (3)) yields
tion of lithostatic pressure.
x z t
Dynamic boundary condition (Eq. (11)) is based on the normal hðx; z; tÞ ¼ H0 sin exp exp ð16Þ
stress, which influences the flow at large depths. Regarding the L L T
condition of zero tangential shear stress at the earth-atmosphere
Eq. (16) shows that L does not only represent the horizontal
interface the following considerations have to be taken into
length scale of the Fourier component under consideration but is
account. The soil and shallow groundwater-bearing layers just
also the relaxation depth of the topographic potential. Half-
below the earth-atmosphere interface have a rheology that may
wavelength pL is the lateral extension of Fourier mode under
be approximated as a thin ‘boundary layer’ with an appreciably
consideration.
smaller apparent viscosity than the effective viscosity of the deep
Rather than specification of a tangential shear stress at top
mantle-crust system under consideration. The tangential shear
boundary z ¼ 0, the velocities representing the creep flow evolu-
stress at the interface between the small-viscosity boundary layer
tion are represented by the expressions
and the very viscous mantle-crust system is unknown. The bound-
x
ary layer is very thin compared to the much deeper reaching mass z t
of crust and mantle. As a consequence, it may reasonably be v x ðx; z; tÞ ¼ v x0 cos exp exp ð17aÞ
L LZ T
assumed that this unknown tangential shear stress does hardly
influence the flow of the much more massive mantle-crust system x
z t
and may therefore be neglected when considering the deeper creep v z ðx; z; tÞ ¼ þv z0 sin exp exp ð17bÞ
L LZ T
flow; instead a creep flow evolution will be assumed, cf. Section 3.7.
The resulting shear stress at the bottom of the boundary layer will The velocities do not generally depend on depth in the same
be ‘‘absorbed” in the boundary layer in such a way that at the way as the topographic potential. Like the relaxation time, T, also
earth-atmosphere interface the tangential shear stress vanishes; the relaxation depth of the velocities, LZ , has to be determined from
for more details cf. Appendix E. the solution.
5
3.8. Solution: Damping Topographic potential h honors the Laplace equation (Eq. (1) for
f ¼ h), not only for one Fourier mode, but also for a superposition
Inertia number I is defined as of many Fourier modes (cf. Section 3.1). Compressive excess pres-
pffiffiffi sures in stationary zones may play a role comparable to stationary
3 3 TV TV
I¼ ¼ 2:598 ð18Þ zones in groundwater.
2 TS TS
In Appendix C the following damping criteria are derived:
3.11. Creep flow systems
If I > 1 then the flow is under-damped; i.e., the amplitudes of

The well-known groundwater flow patterns with local, regional,
topographic undulations are decaying in an oscillating way
and super-regional flow systems [26,7] also apply to creep flow.
with decreasing amplitude (like water waves)
However, in contrast to groundwater flow systems, the flow lines
If I ¼ 1 then the flow is critically damped; the decay is as
in creep flow do not represent the velocity pattern, but the lines
quickly as possible without oscillations
along which the vertical compressive forces are constant. In anal-
If I < 1 then the flow is over-damped; the decay is without
ogy with artesian flowing groundwater wells, there may be zones
oscillations, in such a way that the smaller the inertia number
with over-compressive vertical forces that might cause blowout
the slower the decay
when drilling a well (especially if well completion is poor).
When superposition of a number of Fourier systems is considered,
3.9. Approximation: Negligibly small vorticity
the flow line patterns of creep flow may be more or less similar to the
well-known groundwater flow patterns presented in the context of
Under the condition that
regional groundwater flow systems [26,7]. In contrast to groundwa-
pffiffiffi
TV 3 ter flow systems, the shallow creep flow systems have relatively large
<< I ¼ 0:192 I << 1 ð19Þ relaxation time and relatively small flow velocity, while the deep flow
2T S 9
systemshave relativelyshort relaxationtime and relatively largeflow
it is found in Appendix C that velocity. For more details cf. Appendix D.2.
T L TV
¼ ¼1 ð20Þ
T S LZ 2T S
4. Orders of magnitude: Creep and groundwater flow
from which it follows that the vorticity is equal to
x z 4.1. Deep creep flow
T V 1 H0 t
x¼ cos exp exp ð21Þ
TS TS L L L TS
We consider a subsurface with horizontal extension the half-
Although the vorticity is not exactly equal to zero, the vorticity wavelength pL in the interval 12 pL x 32 pL and with vertical
is sufficiently small to neglect in the expressions for the velocities. extension 0 z < 1 (Fig. 3). The x axis represents a gravitational
Therefore, as good approximation, the flow honors Darcy-like equilibrium level; that is, moving in the x direction at constant
equations (also cf. Appendix D) level z, the gravitational force is constant (equal to qgz). Because
the earth is curved the thus-defined x axis need not be a straight
@h
v x ffi K ð22aÞ line; it may be the x coordinate of an orthogonal curvilinear coor-
@x
dinate system. However, for order of magnitude estimations the
@h scale factors that take the curvature into account may be neglected
v z ffi K ð22bÞ (cf. Appendix G).
@z
In contrast to studies by Johson and Fletcher [40] and Ranalli
with Fourier mode-dependent ‘creep conductivity’ [25], whose focus was on the shallow topographic phenomena,
qgL2 our estimations are focusing on deep flow. Because the relatively
K¼ ð23Þ
2l
The topographic potential hðx; z; tÞ is introduced by Eq. (2) and
presented by Eq. (16) with relaxation time T ¼ T S given by Eq. z-axis
(13); this is the same expression for the relaxation time as pre-
sented by Johnson and Fletcher [40] and Ranalli [25].
3.10. Compressive forces: Creep force systems

x-axis
In creep flow the pressure is lithostatic (cf. Section 2.5 and 0 (1/2) πL π (3/2) πL 2 πL
Appendix D.1); in the flow field there are no excess pressures that
deviate from the lithostatic pressure. This is in contrast with
groundwater flow, in which subsurface locations with excess pres-
sure greater than hydrostatic pressure may cause Artesian flowing
wells above stationary regions [7]. However, in creep flow there Ha-wavelength
may be compressive excess forces (per unit area) comparable with though
the excess pressures in groundwater flow. Let us define the vertical
Fig. 3. Most important Fourier component of topographic height after (relatively
compressive force pzz ¼ ðrzz þ patm þ qgzÞ. For Darcy-like flow sudden) removal of heavy ice cap above the region around x = (3/2) pL. The
(Eqs. (22) and (23)) we find horizontal axis (x-axis) represents the horizontal distance; the vertical axis (z-axis)
x z represents the height of the topography indicated by the blue sinusoidal line. Creep
t
pzz ¼ qgz ¼ qgH0 sin exp exp ð24Þ flow causes a gradual decrease of amplitude; horizontal line z = 0 represents the
L L T final equilibrium situation.
6
pffiffiffi
short topographic wavelengths have already decayed at relatively L < LZ 3L ffi 1:73 L ð31Þ
shallow depths and do not influence the deep flow (cf. Eqs. (21)–
qffiffiffiffiffiffiffi
(23)), reasonable orders of magnitude can be obtained by paying pffiffiffipffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
attention only to one Fourier mode with long half-wavelength. T
3 L=g ffi 1:32 L=g ð32Þ
qffiffiffiffiffiffiffi
1 pffiffiffipffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
4.2. Choice of the earth’s subsurface creep parameters 0 < v0 3 g=LH0 ffi 0:439 g=LH0 ð33Þ
3
For the order of magnitude estimations, the following values are Below the results at the critical point are shown in the tables
assumed: Gravitational acceleration g = 10 m/s2. Earth mantle- presented below for the case L = 100 km and the case
crust density q = 3000 kg/m3; which amounts to a vertical litho- L = 1000 km, again using H0 =L = 0.03% (i.e.,
pffiffiffiffiffi
static pressure gradient of 30 kPa/m. To be sure, the simple model 0 < v 0 0:132 103 gL)
upon which our estimations are based result in rough indications L = 100 km L = 1000 km
of the orders of magnitude that may be expected. These indications
l
0:342 1012 Pa:s l
10:8 1012 Pa.s
may be helpful in defining the spatial and temporal scales of a
more comprehensive numerical model. 100 km < LZ 173 km 1000 km < LZ 1730 km
Two cases with different horizontal subsurface extension will
be investigated in more detail. T
132 s ¼ 3:67 hT
417 s ¼ 6:96 min
Case L = 100 km: subsurface length = pL 300 km ( distance 0 < v 0 0:132 m=s 0 < v 0 0:417 m=s
Brussels-Paris),
Johnson and Fletcher [40] give 3:5 10 Pa.s l 3:5 105 Pa.
Case L = 1000 km: subsurface length = pL 3000 km ( dis-
tance Brussels-Tromsø). s for Lava, and 2:7 1010 Pa.s l 1:1 1011 Pa.s for Syvite.
These values will result in under-damped (oscillatory) flow for
For these two casesH0 = 0.0003L, i.e., the initial average topo- the two above-presented subsurface extensions. For Cucaracha
graphic slope 2H0 =pL is approximately 0.02%. The choice of the clay shale Johnson and Fletcher [40] give l ¼ 5 1012 Pa.s, for
effective viscosity is discussed below. Bearpaw shale l ¼ 1013 Pa.s and for Clay l ¼ 2:3 1013 Pa.s. For
the subsurface with L = 100 km, these values result in over-
4.3. Effective viscosity damped flow, but for the case L = 1000 km these viscosities are
around the critically damped flow. Although these viscosities are
pffiffiffi
The inertia number can be written as I ¼ 34 3g q2 L3 =l2 (cf. Eq. not representative for the much larger effective viscosities that
(18)), from which it follows that the criterion for damped flow, play a role in postglacial rebound, the above-presented results give
I 1 (Section 3.8), can be written as the inequality insight in the creep flow behavior in much smaller parts of the sub-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi surface. For instance, for the much smaller and, hence, much shal-
1 pffiffiffi ffi pffiffiffiffiffi
l
qL 3 3gL ffi 1:14 qL gL ð25Þ lower subsurface with L = 1 km and H0 =L = 0.03%, the criteria for
2 damped flow are l
0:342 1015 Pa.s, 1 km < LZ 1.73 km, T
Substitution into Eqs. (12) and (13) yields 13.2 s, 0 < v 0 13.2 mm/s, which means that in this latter case
sffiffiffiffiffiffiffiffiffiffi the above-presented effective viscosity result in over-damped
4 pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi flow.
TV pffiffiffi L=g ffi 0:877 L=g ð26aÞ
3 3
4.4. Indices: Orders of magnitude
qffiffiffiffiffiffiffiffiffiffi
pffiffiffipffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
T S
3 3 L=g ffi 2:28 L=g ð26bÞ For the below-presented order of magnitude estimations the
pffiffiffi effective viscosity of the mantle-crust system is taken equal to
At the critical point T V =T S ¼ 29 3, i.e., I ¼ 1 (cf. Eq. (18))
The Reynolds number is (cf. Appendix A) l ¼ 5 1021 Pa.s, which represents the effective viscosity for iso-
static rebound. Johnson and Fletcher [40] present 1021 Pa.s
Re ¼ ðT V =T S ÞðH0 =LÞ ð27Þ
l 1022 Pa.s for isostatic rebound. (Below the SI unit second,
pffiffiffi
From which it follows that at the critical point Re ¼ 29 3ðH0 =LÞ which is assumed in the equations, has been converted to year:
and assuming that H0 =L = 0.03% (see above) the Reynolds number 1 year = 31,536,000 s; 1 s = 3.17 10-8 year.)
at the critical point is approximately 0.0001. This very small Rey- Vorticity-related time scale (Eq. (12))
nolds number justifies the neglect of the nonlinear steady inertia
qL2
term from the Navier-Stokes equation. Moreover, because the Rey- TV ¼ 6 109 s 2 1016 year 6 107 s
l
nolds number is inversely proportional to the square of the effec-
tive viscosity (cf. Eqs. (12)–1), the Reynold number will become 2 1014 year
even smaller for creep flow with higher effective viscosity.
Stress-related time scale (Eq. (13))
In Appendix A.3 the following relations are presented
2l
LZ ¼ L=v ðcf:Eq:A:3:1Þ ð28Þ TS ¼ 0:3 1013 s 100; 000year 0:3 1012 s
qgL
T ¼ vT S ðcf:Eq:A:3:7Þ ð29Þ 10; 000year
Time scale ratios
1 pffiffiffi
3 v < 1 ðcf:Eq:A:3:10Þ ð30Þ
3 T V =T S 2 1021
(At the critical point T V =T ¼ 23; cf. Appendix G on initial conditions).
I ¼ 2:598 T V =T S 5 1021
Using inequality 25 and Eqs. (28)–(30) the following
inequalities are found
7
(inertia number, cf. Eq. (18)) the subsurface, with studies related to exploration and production
The above-presented very large stress-related time scale, the as well as carbon capture and storage (CCS). Thanks to the ‘inven-
extremely small vorticity-related time scale and the extremely tion’ of Tóth’s flow systems analysis, the intermediate and deep
small inertia number hold for very viscous flow in the limit subsurface are no longer the exclusive study domains for mining
T V =T S ! 0. and petroleum geologists, but also for hydrogeologists. Although
The general expressions for relaxation depth, characteristic oil and water are immiscible fluids, the oil- and water ‘‘worlds”
depth and penetration depth are: have started to mix at these depths.
In the shallow subsurface, the creep flow rates are negligibly
At relaxation depth z = L the velocity at z = 0 has decreased by a small with respect to the groundwater flow rates. Moreover, in sit-
factorexpð1Þ ¼ 0:37, uations where the shallow subsurface is moving, the creep flow
At characteristic depth z = pL the velocity at z = 0 has decreased rate is generally negligibly small with respect to compaction due
by a factorexpðpÞ ¼ 0:043, to groundwater depletion, e.g., by over-exploitation of fresh
At penetration depth z = 2pL the velocity at z = 0 has decreased groundwater, or by dewatering polder areas. Consequently, the
by a factor expð2pÞ ¼ 0:0019. above-presented order of magnitude of the creep velocities may
be relevant only in the subsurface’s intermediate and deep parts.
For the particular cases, the results are: Let us, therefore, compare the creep flow velocities with the deep
groundwater flow velocities. For an oil/gas reservoir (limestone,
For L = 100 km—1000 km the relaxation depths are 100 km— dolomite) the intrinsic permeability 0.01 millidarcy, from which
1000 km it follows that the hydraulic conductivity 3 mm/year. Assuming
For L = 100 km—1000 km the characteristic depths are 300 km— that the groundwater flow is incompressible and driven by the
3000 km undulations of the topography (water table topography, like in
For L = 100 km—1000 km the penetration depths are 600 km— Tóthian gravity-driven flow systems), the following values for the
6000 km groundwater velocity with respect to the creeping subsurface are
found:
Similar for relaxation time, characteristic time and decay time
(rather than penetration time) Groundwater Darcy velocity: q 1 mm/year
Groundwater transport velocity: v q/n 10 mm/year
At relaxation time t = T the velocity has decreased by a (n = porosity 0.1)
factorexpð1Þ ¼ 0:37,
These orders of magnitude have to be compared with the
At characteristic time t = T the velocity has decreased by a
above-presented creep velocities ranging in order of magnitude
factorexpðpÞ ¼ 0:034,
from 0.3 mm/year to 30 mm/year. Generalizing, we may conclude
At decay time t = T the velocity has decreased by a factor
that that creep flow complements groundwater flow. The assump-
expð2pÞ ¼ 0:0019.
tion of gravity-driven meteoric groundwater flow is doubtful. At
these depths, the compressibility has to be taken into account. In
For L = 100 km—1000 km the stress-related time scales are 100
post-glacial circumstances, this means that the groundwater flow
000 year—10 000 year, respectively. The larger the horizontal
will be compacting, which results in appreciably smaller ground-
length scale, i.e., the larger the lateral extension of the Fourier flow
water velocities [51]. In conclusion: creep flow complements
system, the faster the relaxation to equilibrium and, consequently,
groundwater flow; hydrogeology transforms into fluid geology.
the faster the subsurface creep velocities. This is just the opposite
of the behavior of groundwater flow systems.
5. Summary and conclusions
For T = 100 000 year—10 000 year the relaxation times are 100
000 year—10 000 year Global average atmospheric CO2 concentrations have fluctuated
For T = 100 000 year—10 000 year the characteristic times are between 180 and 280 parts per million (ppm) in the last
300 000 year—30 000 year 420,000 years, but since the Industrial Revolution, CO2 concentra-
For T = 100 000 year—10 000 year the decay times 600 tions have increased to >400 ppm. The Earth’s global average tem-
000 year—60 000 year perature (GAT) has risen by more than 0.8 °C since 1880. Since the
beginning of the Industrial Revolution, readings of the thermome-
4.5. Velocities: Orders of magnitude ter have risen continuously.
Carbon Capture, Utilization, and Storage (CCUS) are considered
Creep conductivity for L = 100 km: K 1000 mm/year an essential strategy for meeting the targets for reducing CO2 emis-
Creep velocity for L = 100 km: V 0.3 mm/year sions. It’s a process that captures CO2 emissions from sources such
Creep conductivity for L = 1000 km: K 100 000 mm/year as coal-fired power plants and either reuses or stores it so that it
Creep velocity for L = 1000 km: V 30 mm/year does not enter the atmosphere. In order to ensure stable carbon
capture and utilization (CCU) operation, the geological storage of
Regarding utilization of the subsurface, a global distinction can CO2 as well as hydrogen, biogas and natural gas in deep aquifers
be made between the shallow, the intermediate, and the deep sub- or exhausted gas fields is a prerequisite. To assess the long-term
surface. Roughly speaking, the shallow subsurface, i.e., the subsur- consequences of deep subsurface storage it is important to dispose
face at depths smaller than 500 m, is used for groundwater of reliable estimates of the subsurface transport characteristic.
production, brine storage, etc. Hydrogeology is mainly devoted to Although groundwater flow is a significant subsurface transport
the study of this part of the subsurface. The intermediate subsur- mechanism, we have shown that the creeping motion of the sub-
face, i.e., the subsurface at depths between 0.5 km and 2 km is used surface itself has also to be taken into account.
to store solid waste, for deep geothermal energy, etc. Finally, the To that aim we have presented simple indices, similar to the
deep subsurface, i.e., the subsurface at depths greater than 2 km, well-known indices for the characterization of groundwater flow
is used to produce natural gas, oil, salt, and the storage of CO2. systems. The paper presents such indices based on considerations
The upstream petroleum activities are focusing on this part of related to the long-term behavior of subsurface after a sudden
8
change in topography (e.g. postglacial rebound). It turns out that are negligible and Eqs. (A.1.1a) and (A.1.1b) simplify to without the
on long time scales: The Earth’s upper mantle and crust behave unsteady Stokes equation.
as a creeping fluid with very high effective viscosity. In this
approach one single value of the effective viscosity is assigned Appendix B. Creep pressure head
the relevant mantle-crust system, except for a relatively thin
‘‘boundary layer” just below the earth-atmosphere interface, The derivative of Eq. (A.1.1a) with respect to x, plus the deriva-
where the apparent viscosity is appreciably smaller. tive of Eq. (A.1.1b) with respect to z while taking the continuity
The derivations are based on a decomposition of the topo- equation (Eq. (9)) into account, yields the Laplace equation for
graphic undulations in Fourier modes, or Fourier flow systems. the pressure head hP
The fluid creep motion is caused by gravity (spatial differences in
topographic height), effective viscosity damping the gravity- @ 2 hP @ 2 hP
þ 2 ¼0 ðA:2:1Þ
induced acceleration, and inertia trying to overcome the damping. @x2 @z
In this approach the effective viscosity has been introduced as a with boundary condition hP ðx; 0; tÞ ¼ 0, from which it follows that
parameter that lumps different fine-scale mechanisms in the the pressure is isostatic (cf. Section 3.4 and Eq. (11)). Eq. (A.2.1) is
apparently viscous large-scale motion of subsurface rock. Appar- the two-dimensional version of Eq. (1). In Appendix C it is shown
ently it is an effective viscosity that does not only depend on ther- that this boundary condition is compatible with the Stokes
modynamic conditions like in ordinary gases and liquids (the equation.
kinetic theory of viscosity [42,43]), but it also depends on other
phenomena like the occurrence of fractures, voids and other inho- Appendix C. Vorticity and velocity
mogeneities. It is shown that for sufficiently high effective viscosity
and not too massive earth the topography is not oscillating (like C.1. Derivation of general solution
water waves) but shows an underdamped decay to an equilibrium
situation. Although the derivation is relatively lengthy, the result- The derivative of Eq. (A.1.1a) with respect to z, minus the
ing equations are very simple: Darcy-like equations for incom- derivative of Eq. (A.1.1b) with respect to x, while taking the defini-
pressible flow, in which the ‘‘creep conductivities” are Fourier- tion of vorticity (Eq. (7)) into account, yields the vorticity diffusion
mode dependent. equation (Eq. (8)).
The results show that for horizontal length scales of 100 km and To simplify the following presentation, we define the length
1000 km the creep velocities are about 0.3 mm/year to 30 mm/ ratio v is as the ratio between the specified horizontal length L
year, respectively. For these horizontal length scales the penetra- and the vertical relaxation length LZ to be determined from the
tion depths amount to the values 600–6000 km. Because the governing flow field equations (Eqs. (7)–(9)) and boundary condi-
crust’s thickness is at least one order of magnitude smaller, these tions (Eqs. (10) and (11))
penetration depths indicate that the mantle’s creep flow
L
dominates. v¼ ðA:3:1Þ
When considering these large horizontal length scales (hun- LZ
dreds of kilometers), the creep velocities are comparable to the Instead of specifying a tangential shear stress at top the bound-
groundwater velocities, while the relaxation times are in the order ary we have introduced expressions for the evolution of the creep
of centuries. For smaller horizontal scales the creep velocities velocities (Eqs. (17a) and (17b)). Substitution into the continuity
become negligibly small which means that on these scales the equation (Eq. (9)) yields
groundwaters flow dominates. In conclusion, deep creep flow
velocities are complementary to the deep groundwater velocities; v x0 ¼ vv z0 ðA:3:2Þ
hydrogeology has to be generalized to fluid geology. Substitution of the proposed velocity model (Eqs. (17a) and
(17b)) into the definition of the vorticity (Eq. (7)) and using Eqs.
Appendix A (A.3.1) and (A.3.2) results in

v z0 x x t
A.1. Continuity and Stokes equation x¼ 1 v2 cos exp exp ðA:3:3Þ
L L LZ T
The Navier-Stokes equation for flow of a Newtonian fluid in two Substitution of Eq. (A.3.3) into the vorticity diffusion equation
dimensions can be written in the following form [52,46–48] (Eq. (8)) and using the definition of the vorticity-related time scale
T V (Eq. (12)) yields
@v x @ v2 @x
q þ xv z ¼ qg hP þ þl ðA:1:1aÞ TV
@t @x 2g @z ¼ 1 v2 ðA:3:4Þ
T

@v z @ v2 @x Substitution of the expression for v z into the stress boundary
q xv x ¼ qg hP þ l ðA:1:1bÞ
@t @z 2g @x condition (Eq. (11)) while using Eq. (A.3.2) and the definition of
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the stress-related time scale (Eq. (13)) yields
where v ¼ v 2x þ v 2z . Reynolds number Re is defined as the ratio
1 H0
between the magnitudes of the steady inertial force (qxv x , qxv z ) v z0 ¼ ðA:3:5Þ
and the viscous force (l@ x=@x; l@ x=@z) (forces per unit area). That
v TS
0
is, Re ¼ ðq=lÞXV=X , where X is the amplitude of the vorticity and Using the kinematic boundary condition (Eq. (10)) with the
0
X is the amplitude of the derivatives of the vorticity with respect to expression for the topographic height (Eq. (15)) and for velocity
x and z. From the velocity-vorticity solution (Eqs. (21)–(23)) it fol- v z (Eq. (17)) results in
lows that Re ¼ ðT V =T S ÞðH0 =LÞ; the ratio between the amplitude of H0
v 2 =2g and that of hP is then 14 Re. For the creep flow considered here v z0 ¼ ðA:3:6Þ
T
we find Re << 1; i.e., the nonlinear inertia terms xv x , xv z and 12 qv 2
9
Combination of Eq. (A.3.5) with Eq. (A.3.6) yields @v z @x

q ¼ l
@t @x
T ¼ vT S ðA:3:7Þ x
T V qgH0 1 z 1 t
¼ sin exp v exp ðA:4:1bÞ
Substitution of Eq. (A.3.7) into Eq. (A.3.4) yields an algebraic T S 2L v 2 L L v TS
equation of third degree
Substitution of the above-presented solutions into the Stokes
v3 v þ 2r ¼ 0 ðA:3:8Þ equation (Eqs. (A.1.1a) and (A.1.1b)) results in
in which r is defined as ð@hP =@x; @hP =@zÞ ¼ ð0; 0Þ; i.e., the pressure is lithostatic:
p ¼ patm þ qgz. This solution of the Stokes equation represents
TV the balance between (i) viscous forces countering the gravity-
r¼ ðA:3:9Þ
2T S driven falling/rising of the rock mass and (ii) the inertial forces try-
ing to continue the falling/rising of the mass. In reality there may
This cubic equation has three solutions. If r2 > 27
1
Eq. (A.3.8) has
be deviations from lithostatic pressure. However, they play a neg-
one real root and a pair of complex conjugate roots; if r2 ¼ 27 1
all
ligible role for the explanation of the essential mechanisms that
roots are real and at least two are equal, and if r < 27 all roots
2 1
govern the motion.
pffiffiffi
are real. In the interval 0 r 19 3 the relevant solution is [53]
pffiffiffi 1 pffiffiffi 1 pffiffiffi D.2. Zero vorticity approximation
1
v ¼ cos arcsinð3 3rÞ 3sin arcsinð3 3rÞ ðA:3:10Þ
3 3 3 pffiffiffi
For r << 19 3 solution (A.12)–(A.15) simplifies to v ¼ 1 r,
For all values of r in this interval the topographic surface is from which Eqs. (21)–(23) follow as a good approximation.
monotonously falling in subsurface section 12 pL x < pL and Although the vorticity is not equal to zero, the vorticity is suffi-
monotonously rising in subsurface section pL < x 32 pL, without ciently small to determine the creep velocity as were the flow
oscillations; see the critically damped and over-damped cases in vorticity-free (irrotational). Only in that case the solution for the
pffiffiffi
Section 3.8. For r << 19 3 this solution simplifies to v ¼ 1 r; this velocity can be approximated by a Darcy-like approximation with
result can also be found by substitution into Eq. (A.3.8) of the series Fourier mode-dependent ‘creep conductivity.’
pffiffiffi In vorticity-free flow (irrotational flow) the flow velocities can
expansion ð1 rÞ3 ¼ 1 3r þ ::: for small r. For r ¼ 19 3 Eq.
pffiffiffi be described by ðv x ; v y ; v z Þ ¼ ð@@xu ; @@yu ; @@zuÞ in which u ¼ Kh is a
(A.3.10) yields v ¼ r 1=3 ¼ 13 3; this can also be found by substitu-
pffiffiffi velocity potential [46–48]. Only in that case the solution for the
tion into Eq. (A.3.8). For r > 19 3 the solution yields complex num-
velocity can be approximated by the Darcy-like approximation
bers, which means that the topographic surface is falling and rising
with Fourier mode-dependent ‘creep conductivity,’ cf. Eqs. (21)–
in an oscillatory way, like waver waves: see the under-damped
(23). This Darcy-like approach is very useful for comparisons with
case in Section 3.8.
groundwater flow (cf. Section 4).
When the superposition of a number of Fourier systems is con-
C.1. General solution for vorticity and velocity
sidered, the velocity potential is equal to
!
The general solution for vorticity and velocity in the interval
X x

y

pffiffiffi u x; y; z; t ¼ um ðz; tÞsin þ m
0 r 19 3 is m S0 Lm Ly
x
x !
T V 1 H0 1 z 1 t
x¼ cos exp v exp ðA:3:11Þ þ um ðz;
x
þ
y
ðA:4:2Þ
TS TS L v 2 L L v TS C0 tÞcos
Lm
Lm

x y
x
v x ðx; z; tÞ ¼ HT 0
S
cos L
exp v Lz exp v1 t
TS
ðA:3:12aÞ where the ‘apparent topographic height’ is given by

z t
x umS0 ðz; tÞ ¼ K m HmS0 exp exp ðA:4:3aÞ
v z ðx; z; tÞ ¼ HT
S
0
v sin L exp v L exp v
1 z 1 t
TS
ðA:3:12bÞ Lm Tm
S

The back transformation to the global x; y; z coordinate system

umC0 ðz; tÞ ¼ K m HmC0 exp Lzm exp Ttm (A.4.3b)
S
transforms the components ð0; x; 0Þ of the vorticity vector in
cf. Section 3.2 and Eq. (4). (Note that coordinates x and y repre-

the Fourier mode-dependent coordinate system to the components sent the global coordinates.) Similar expressions hold for ground-
ðxx ; xy ; xz Þ, in which xz ¼ 0 thanks to the geometric water flow in homogeneous and isotropic porous media.
linearization. However, in the case of groundwater, the conductivity is not Four-
ier flow system dependent; cf. Section 3.11.
Appendix D. Compatibility and the zero vorticity approximation
Appendix E. Shear stress boundary layer
D.1. Compatibility with the Stokes equation
In this appendix an argument is presented to show that the
Solutions (A.3.11)–(A.3.12) can be used to determine the com- shallow non-zero shear stress in the crust is reduced to zero at
ponents of the Stokes equation (Eqs. (A.1.1a) and (A.1.1b) without the earth-atmosphere interface within a relatively thin boundary
the steady inertia terms qxv z and qxv x ) layer 0 < z < D in which the earth has a relatively small apparent
@v x @x viscosity compared to the deeper Earth’s effective viscosity.
q ¼l Regarding the condition of zero tangential shear stress at the
@t @z
x
T V qgH0 1 z 1 t earth-atmosphere interface the following considerations have to
¼þ cos exp v exp ðA:4:1aÞ
T S 2L v L L v TS be taken into account. In the shallow subsurface just below the
atmosphere we may find surface waters, soils and somewhat
10

deeper shallow layers (among which aquifers). Due to its diversity, @v z
rxz ðx; DÞ ¼ 2l ðA:5:5Þ
this shallow subsurface has a relatively complex rheology; on rel- @x ðx;DÞ
atively short time scales (days—years) reversible (elastic) and irre-
versible (plastic) deformations (displacement caused by stress) Continuity of vertical velocity v z ðx; DÞ ¼ v z ðx; DÞ and of shear
play a role, for instance in case of subsidence. These deformations stress rxz ðx; DÞ ¼ rxz ðx; DÞ yields
are caused by relatively small stresses compared to the stresses x
t
that play a role in the deep subsurface. On long time scales (dec- rxz ðx; DÞ ¼ qgHD cos exp ðA:5:6Þ
L TS
ades—centuries) and averaged over large areas these deformations
may appear as creep phenomena (velocity caused by stress). As a where HD ¼ H0 expðD=LÞ; cf. Eq. (13) and Eq. (A.3.12) for v ¼ 1.
consequence, with respect to the much greater depths upon which Substitution of Eq. (A.5.6) into Eq. (A.5.3) while using
we are focusing, the apparently creeping earth-atmosphere inter- x l1 rxz results in the general solution
face may be seen as a relative thin ‘boundary layer’ of which its
qgHD zD zD D

rheology may be simplified by creep flow with a smaller apparent xD ðzÞ ¼ exp þ exp ðA:5:7Þ
viscosity than the effective viscosity of the deep crust-mantle sys- l L D L
tem under consideration. The tangential shear stress at the inter- This solution honors the stress conditions (formulated as vortic-
face between the small-viscosity boundary layer and the very ity conditions) xD ðDÞ ¼ qgHD = l and xD ð0Þ ¼ 0 on z ¼ D, and
viscous mantle-crust system is unknown. Therefore, it may reason- under the condition jDj << L Eq. (A.5.7) simplifies to
ably be assumed that the tangential shear stress in the boundary
layer hardly influences the flow of the much more massive qgH0 zD
xD ðzÞ ¼ 1þ ðA:5:8Þ
mantle-crust system and may therefore be neglected when consid- l D
ering the deeper creep flow. More details about soil and shallow
aquifers can be found in Refs. [57–71]. with HD ¼ H0 expðD=LÞ ¼ H0 . Integration using @ v x =@z ¼ xD ðzÞ
The argument is based on the above-presented relatively simple yields the velocity
" !#
model of a thin boundary layer with small thickness D compared to qgH0 L qgH0 D z D ðz DÞ2 x t
v x ðx; z; tÞ ¼ þ þ cos exp
subsurface depth extension L. Since the boundary later is very thin, 2l l D 2D2 L TS
it is reasonable to assume that the horizontal variations in the flow
ðA:5:9Þ
field are relatively small; i.e., @ v x =@x 0, @ v z =@x 0 and, as a con-
sequence of the continuity equation, also @ v z =@x ¼ @ v x =@z 0. where it has been assumed that at z ¼ D the horizontal velocity in
(The tilde, ~, above the symbols denotes quantities within the the boundary layer, v x ðx; D; tÞ, is equal to the horizontal velocity
boundary layer.) In addition it is assumed that at the bottom of in the deeper mantle-crust system.
the boundary layer, z ¼ D, the velocities in the deeper Earth’s In reality there is not a sharp interface between crust and
mantle-crust system and in the boundary layer are continuous; boundary layer; the transition between boundary layer and crust
i.e., v x ¼ v x and v z ¼ v z at z ¼ D. Under these assumptions the vor- is gradual. Therefore, it may be expected that the layer velocities
ticity in the boundary layer is x @ v x =@z and the shear stress is and the crust velocities have the same order of magnitude, i.e.,
rxz l @ v x =@z l x. D= l L=2l (cf. Eq. (A.5.9)), or D ¼ ðl =2lÞ L. And from the
The vorticity diffusion equation is assumption D << L upon which the above-presented approach is
based, it follows that l << 2l.
q @ x @2 x @2 x The Stokes equations represent only a rough approximation of
¼ þ ðA:5:1Þ
l @t @x2 @z2 the rheology of the shallow subsurface just below the earth-
atmosphere interface. Probably the rheology could be better
where the relatively small term @ 2 x =@x2 is still taken into account described by a visco-elasto-plastic model including compaction,
in order to investigate in more details the above-presented approx-
fracturing, etc. Nevertheless, whatever the rheology in the bound-
imations. Substitution of xðx; z; tÞ ¼ xD ðzÞcosðx=LÞexpðt=T S Þ with ary layer may be, it is reasonable to assume that the boundary
T S ¼ 2l=qgL (cf. Eq. (A.3.12) for v ¼ 1) yields layer is ‘absorbing’ the shear stress developed in the crust to
reduce it to zero at the earth-atmosphere interface. Since the
@ 2 xD 1 q qgL
¼ xD ðA:5:2Þ boundary layer is relatively thin (and has therefore hardly any
@z2 L2 l 2l
mass), it does not significantly influence the massive large-scale
Although apparent viscosity l in the boundary layer may be creep flow of deep reaching mantle-crust.
appreciably smaller than the deeper Earth’s effective viscosity l,
it is assumed that l is sufficiently large to assume Appendix F. Global and local Fourier flow systems
ðq L2 = lÞðqgL=2lÞ << 1. In that case the vorticity diffusion equa-
tion simplifies to F.1. Introduction to Fourier decomposition
@ xD xD
2

¼ 2 ðA:5:3Þ In the global x; y coordinate system a Fourier transform decom-
@z2 L
poses the topographic height Hðx; yÞ into a series, or integral, of
Before presenting a solution, first the conditions on the crust- Fourier components presented by Eq. (4), which will here be writ-
layer interface z ¼ D are presented. The crust’s shear stress just ten as
below z ¼ D is X
Hðx; yÞ ¼ Hm
m
S sinðl x x þl y yÞ þ HC cosðl x x þl y yÞ
m m m

m
ðA:6:1Þ
@v x @v z m
rxz ðx; DÞ ¼ l þ ðA:5:4Þ
@z @x ðx;DÞ where lm m
¼ 1=L and lm m
¼ 1=L . After separation of variables,
x x y y

and since the deeper mantle-crust flow just below z ¼ D is irrota- Hðx; yÞ ¼ Hx ðxÞHy ðyÞ [54,55], the Fourier components Hm m
S and HC
tional, @ v x =@z @ v z =@x ¼ 0, Eq. (A.5.4) results in are determined by inversion integrals for the Fourier series or Four-
11
ier integrals [54,55]. To make the presentation more transparent G.2. Succession of initial conditions: Ice cap retreat
one particular Fourier mode m is considered omitting index m.
In Section 3 it was assumed that at time t ¼ 0 height Hðx; 0Þ
suddenly jumps to H0 sinðx=LÞ. The time evolution for time
F.2. Rotation
0 t < s is then given by Eqs. (21)–(23). The jump in height
Hðx; tÞ may be caused by the sudden retreat of an ice cap.
First a rotation is considered. For that purpose the positive wave
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi At time t ¼ s height Hðx; tÞ may again be changed suddenly
number of the Fourier mode is defined as l ¼ l2x þ l2y ¼ 1=L (cf. from H0 ðx; s eÞsinðx=LÞ to H1 ðx; sÞsinðx=LÞ, where e is an infinites-
Eq. (5)). Substitution of cosh ¼ lx =l and sinh ¼ ly =l into A.6.1 yields imally small positive number. Then the flow solution for t
s has
to be modified by replacing the time-dependent term H0 expðt=T S Þ
*
Hðx; yÞ ¼ HS sin½lðx cosh þ y sinhÞ þ HC cos½lðx cosh
*
in Eqs. (21)–(23) with the term H1 exp½ðt sÞ=T S .
Such an approach is also applicable to infinitesimal time inter-
þ y sinhÞ ðA:6:2Þ vals. In each infinitesimal time interval sn t < snþ1 where
0 0 snþ1 sn ¼ dsn ! 0, time t hardly changes, which means that
A rotated coordinate system with coordinates x and y is found
exp½ðt sn Þ=T S ffi 1. As a consequence, the gradual removal of
by applying the rotational coordinate transforms
0 0 material (e.g. melting ice) @HR ðx; sÞ=@ s from the subsurface over
x ¼ þ x cosh þ y sinh and y ¼ x sinh þ y cosh for rotation over an many small intervals sn t < snþ1 with snþ1 sn ¼ dt ! 0 for
angle h. For our purpose the simplest rotated coordinate system n ¼ 1; 2; :::; N ! 1 can be described by replacing Eq. (10) with
0
is the system in which y ¼ 0, i.e., tanh ¼ y = x. Substitution into
@Hðx; tÞ @HR ðx; tÞ
the above-presented expression yields ¼ v z ðx; 0; tÞ on z ¼ 0 ðA:7:1Þ
@t @t
0 0 0
Hðx Þ ¼ HS sinðlx Þ þ HC cosðlx Þ ðA:6:3Þ without taking care of the initial condition for Eqs. (A.1.1) and
(A.1.2).
F.3. Translation
G.3. General initial conditions
Now a translation is introduced. For that purpose the amplitude
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The other extreme with respect to very viscous flow is critically
pffiffiffi pffiffiffi
of the Fourier mode is defined as H0 ¼ H2S þ H2C (cf. Eq. (6)). In damped flow, for which r ¼ 19 3 and v ¼ 13 3 (cf. Appendix C); i.e.,
pffiffiffi pffiffiffi
addition, we define cosw ¼ HS =H0 and sinw ¼ HC =H0 . Substitution T V =T S ¼ 29 3 (inertia number I ¼ 32 3T V =T S ¼ 1, Eq. (18)) and
into the above-presented expression yields pffiffiffi
v ¼ T=T S ¼ L=LZ ¼ 13 3 (cf. Appendix C). In that case T V =T ¼ 23; this
0 0 0 means that the initial conditions influence the flow considerably.
Hðx Þ ¼ H0 ½sinðlx Þcosw þ cosðlx Þsinw ðA:6:4Þ
However, over-damped flow with an three times higher effective
Substitution of the goniometric relation viscosity results already in T V =T 0.05 and for an effective viscos-
0 0 0
sinðlx Þcosw þ cosðlx Þsinw ¼ sinðlx þ wÞ into (A.6.4) yields ity equal to or greater than three times the critical effective viscos-
ity the flow may be described by Eqs. (A.3.11) and (A.3.12) with
0
H ¼ H0 sinðlx þ wÞ ðA:6:5Þ negligible influence of initial conditions.
0 0
and introduction of the translated coordinate x ¼ x þ w=l ¼ x þ wL
Appendix H. Generalized geometric linearization
yields
HðxÞ ¼ H0 sinðlxÞ ¼ H0 sinðx=LÞ (A.6.6)
In the previous discussion it was assumed that initially at time
cf. Eq. (6).
t ¼ 0, before the retreat of the ice cap, the topography is in equilib-
rium at level z ¼ 0. From the solution for flow after retreat of the
Appendix G. Initial conditions ice cap it follows that for t ! 1 the topography is again in equilib-
rium, now at the deeper level z ¼ H0 because a great amount of
G.1. Initial conditions for very viscous flow ice mass has been removed. Although the levels z ¼ 0 and
z ¼ H0 may suggest that the equilibrium topography is horizon-
In the derivation of Eqs. (A.3.11) and (A.3.12) the initial condi- tal, this is not necessarily the case. When considering the x; y; z
tion for the Stokes equation (Eqs. (A.1.1a) and (A.1.1b)) or the vor- coordinate system as an approximation of a curvilinear orthogonal
ticity diffusion equation (Eq. (8)) was tacitly assumed to be equal coordinate system, the equilibrium topography of the Earth’s
to the velocity and vorticity at the time just after onset of motion. mantle-crust systems may have, at different positions, different
However, in many cases different initial conditions may result in depths and heights with respect to an exactly horizontal reference
different flow patterns after onset of motion. These differences play surface (a level on which at each point the gravitational accelera-
a role during the time-scale of vorticity diffusion time T V , while the tion has the same value).
solution of interest has time-scale T as order of magnitude. Only if A curvilinear orthogonal coordinate system is based on the
T V << T the initial conditions play a negligible role in the larger three orthogonal coordinates n1 ; n2 ; n3 accompanied by three scale
time-scale solution. factors h1 ðn1 ; n2 ; n3 Þ; h2 ðn1 ; n2 ; n3 Þ; h3 ðn1 ; n2 ; n3 Þ: To apply such a
In very viscous flow, where T V =T ¼ T V =T S = coordinate system the equations of motion have first to be written
ðqL2 =lÞðqgL=2lÞ << 1 (cf. Appendix A.3) initial conditions do not in their coordinate-free formulation using differential operators
play a role; they are honored almost immediately. For practical working on vectors [54,55].
applications very viscous flow may, therefore, be considered as The most relevant differential operators are the gradient r
quasi-steady (for theoretical reasons it cannot be exactly quasi (Darcy’s law v ¼ K rh), the curl or rotational r (vorticity
steady; cf. Appendix A.4). More importantly, since initial condi- x ¼ r v ) and the divergence r (continuity equation
tions do not matter, ‘initial conditions’ may be imposed at any time r v ¼ 0). The Stokes equation (Eq. (A.1)) is then written as
t. q @@tv þ rðp qgzÞ ¼ lr ðr v Þ ¼ lr2 v and the vorticity dif-
12
fusion equation (Eq. (8)) in vector form is, q @@tx ¼ lr ðr xÞ ¼ Proc. Int. Symp. on Intermontane Basins: Geology and Resources, Chiang Mai,
Thailand; 1989: 513 pp.
lr2 x [52,56]. [11] Tóth J. Hydraulic continuity in large sedimentary basins. In: Proc. Int. Conf. on
The above-presented vector equations with differentials opera- Groundwater in Large Sedimentary Basins, Perth, Austr; 1990.
[12] Verwey JM. Hydrocarbon migration systems analysis. Amsterdam: Elsevier Sc
tors can be replaced with a differential formulation in the
Publ B.V; 1993. p. 276.
curvilinear orthogonal coordinate system [54,55]. For instance, in [13] Tóth J. Hydraulic continuity in large sedimentary basins. Hydrogeol J 1995;3
a curvilinear orthogonal coordinate system the continuity equation (4):4–16.
[14] Meekes JAC. Seismic techniques for hydro-geological applications. A multi-
(Eq. (9)) is written as
disciplinary study, TNO-IAG report OS-95–25B, Delft; 1995.
" # [15] Engelen GB, Kloosterman FH. Hydrological systems analysis: methods and
1 @ðh2 h3 v 1 Þ @ðh3 h1 v 2 Þ @ðh1 h2 v 3 Þ
rv ¼ þ þ ¼0 8:1Þ applications. Kluwer; 1996.
h1 h2 h2 @n1 @n2 @n3 [16] El-Rawy MA. Calibration of hydraulic conductivities in groundwater flow
models using the double constraint method and the Kalman filter PhD
thesis. Brussels, Belgium: Vrije Universiteit Brussel; 2013.
The scale factors are related to a Cartesian X; Y; Z coordinate [17] El-Rawy MA, Batelaan O, Zijl W. Simple hydraulic conductivity estimation by
system (with straight, not curved axes) by the Kalman filtered double constraint method. Groundwater 53 (3): 401-413.
2 2 2 [18] Zijl W, De Smedt F, El-Rawy M, Batelaan O. The Double estimation by the
2 @X @Y @Z Kalman filtered double constraint method, Groundwater 2015; 53(3):
hi ¼¼ þ þ Constraint Inversion Methodology Equations and Applications in Forward
@ni @ni @ni and Inverse Modeling of Groundwater Flow. Springer; 2017. https://doi.org/
" 2 2 #1 10.1007/978-3-319-71342-7.
2
@ni @ni @ni [19] Enemark T, Peeters LJ, Mallants D, Batelaan O, Valentine AP, Sambridge M.
¼ þ þ 8:2Þ
@X @Y @Z Hydrogeological Bayesian Hypothesis Testing through Trans-Dimensional
Sampling of a Stochastic Water Balance Model. Water 2019; 11:1463.
doi:10.3390/w11071463, www.mdpi.com/journal/water
Considering the retreating ice cap, coordinate n1 is the horizon-
[20] Enemark T, Peeters LJ, Mallants D, Batelaan O. Hydrogeological conceptual
tal coordinate axis while coordinate n3 represents the vertical coor- model building and testing. J Hydrol 2019;569:310–29.
dinate axis, where the flow is in the n1 -n3 cross section and the [21] El-Rawy M, De Smedt F, Batelaan O, Schneidewind U, Huysmans M, Zijl W.
vorticity vector is in the n2 direction. Moving along the n1 -axis Hydrodynamics of porous formations: Simple indices for calibration and
identification of spatio-temporal scales. Mar Pet Geol 2016;78:690–700.
the horizontal Cartesian coordinate changes at a rate @X=@n1 while [22] Stolwijk NHM, Zijl W, Boekelman R. The penetration depth in modelling
the vertical Cartesian coordinate changes at a rate @Z=@n1 . If line natural groundwater flow. J Hydrol 1996;185(1-4):171–82.
element @n1 were a straight (not curved) length, @X=@n1 ¼ cosa [23] Zijl W. Scale aspects of groundwater flow and transport systems. Hydrogeol J
1999;7(1):139–50.
and @Z=@n1 ¼ sina, where a is the angle of the X -axis with respect [24] Zijl W, Nawalany M. Natural groundwater flow. Boca Raton: Lewis Publishers;
to the n1 -axis. From Eq. (A.8.2) is follows then that h1 ¼ 1 (in two- 1993.
[25] Ranalli G. Rheology of the Earth. London: Chapman & Hall; 1995.
dimensional flow the term @ni =@Y is equal to zero). In a similar way [26] Tóth J. A theory of groundwater motion in small drainage basins in central
it can be shown that h3 ¼ 1 too. In this case the n1 -n3 coordinate Alberta, Canada. J Geophys Res 1962;67(11):4375–88.
[27] Tóth J. A theoretical analysis of groundwater flow in small drainage basins. J
system has straight (non-curved) coordinate axes. In reality line Geophys Res 1963;68(16):4795–812.
element @n1 may be curved, and in that case h1 and h3 deviate from [28] De Smedt F, Zijl W. Two- and three-dimensional flow of groundwater. LLC,
Boca Raton, Florida, USA: CRC Press Taylor and Francis Group; 2018.
1. However, if the radius of curvature is large with respect to the [29] D’Andrimont R. La Science Hydrologique (The Science of Hydrology) Revue
thickness of the viscous subsurface under consideration, as may Universelle de Mines, Liège, Bérange; 1906.
be expected from the radius of the earth’s surfaces of equal gravity, [30] Engelen GB, Jones GP. Developments in the Analysis of Groundwater Flow
Systems, IAHS publication 163. Wallingford: International Association of
the deviations from h1 ¼ 1 and h3 ¼ 1 are sufficiently small to be Hydrological Sciences; 1986.
neglected. In that case the curvilinear orthogonal coordinate sys- [31] Freeze RA, Cherry JA. Groundwater. NJ, Prentice-Hall: Englewood Cliffs; 1979.
[32] Batelaan O. Phreatology - Characterizing Groundwater Recharge and Discharge
tem n1 ; n2 ; n3 may be approximated by the curvilinear orthogonal
Using Remote Sensing, GIS, Ecology, Hydrochemistry and Groundwater
coordinate system x; y; z in which the small deviations from Modeling Ph.D. thesis. Belgium: Vrije Universiteit Brussel; 2006.
h1 ¼ 1 and h3 ¼ 1 may be neglected. [33] Zijl W, El-Rawy M. The evolution from an unsteady to a steady mixing zone
between two groundwater flow systems with different concentrations.
Alexandria Eng J 2019;58(2):725–31.
References [34] El-Rawy M, Zijl W, Batelaan O, Mohammed GA. Application of the double
constraint method combined with MODFLOW. In: Proceedings of the Valencia
IAHR international groundwater symposium. Valencia, Spain 22–24
[1] Jr Bruant RG, Guswa AJ, Celia MA, Peters CA. Safe storage of CO2 in deep saline
September; 2010.
aquifers. Environ Sci Techno 2002;36(11):240A–5A. doi: https://doi.org/
[35] El-Rawy M, Mohammed GA, Zijl W, Batelaan O, De Smedt F. Inverse modeling
10.1021/es0223325.
combined with Kalman filtering applied to a groundwater catchment. In:
[2] Leung DYC, Caramanna G, Maroto-Valer MM. An overview of current status of
Proceedings of the MODFLOW and more conference, Golden, USA 6–8 June;
carbon dioxide capture and storage technologies. Renew Sustain Energy Rev
2011.
2014;39:426–43.
[36] El-Rawy MA, Batelaan O, Zijl W. A simple method to apply measured flux and
[3] Cuéllar-Franca RM, Azapagic A. Carbon capture, storage and utilisation
head data for the estimation of regional hydraulic conductivities. In: First
technologies: a critical analysis and comparison of their life cycle
EAGE/TNO Workshop. European Association of Geoscientists & Engineers;
environmental impacts. J CO2 Util 2015;9:82–102.
2015. p. cp-450.
[4] Ajayi T, Gomes JS, Bera A. A review of CO2 storage in geological formations
[37] El-Rawy M, De Smedt F, Zijl W. Zone-integrated double-constraint
emphasizing modeling, monitoring and capacity estimation approaches. Pet
methodology for calibration of hydraulic conductivities in grid cell clusters
Sci 2019;16(5):1028–63.
of groundwater flow models. Transp Porous Med 2018;122(3):633–45.
[5] Maul PR, Metcalfe R, Pearce J, Savage D, West JM. Performance assessment for
[38] Zijl W, El-Rawy M, Batelaan O, Nawalany M. Numerical flow systems analysis
the geological storage of carbon dioxide: Learning from the radioactive waste
based on two water table conditions: Tóth’s head condition and the
disposal experience. Int J Greenhouse Gas Contr 2007;1(4):444–55.
conventional recharge condition. Regional Groundwater Flow: Theory, Appl
[6] Benisch K, Graupner B, Bauer S. The coupled OpenGeoSys-eclipse simulator for
Future Develop 2013;147.
simulation of CO2 storage – code comparison for fluid flow and geomechanical
[39] De Smedt F, Zijl W, El-Rawy M. Double constraint method for pumping test
processes. Energy Procedia 2013;37:3663–71.
analysis. J Hydrol Eng 2018;23(7):06018003. doi: https://doi.org/10.1061/
[7] Tóth J. Gravitational systems of groundwater flow: theory,
(ASCE)HE.1943-5584.0001677.
evaluation. Cambridge UK: Utilization. Cambridge University Press; 2009. p.
[40] Johnson AM, Fletcher C. Folding of viscous layers: mechanical analysis and
188–206.
interpretation of structures in deformed rock. New York: Columbia University
[8] Tóth J. Gravity-induced cross-formation flow of formation fluids, Red Earth
Press; 1994.
region, Alberta, Canada: analysis, patterns and evolution. Water Resour Res
[41] Bresciani E, Kang PK, Lee S. Theoretical analysis of groundwater flow patterns
1978;14(5):805–43.
near stagnation points. Water Resour Res 2019;55(2):1624–50.
[9] Tóth J, Millar RF. Possible effects of erosional changes of the topographic relief
[42] Chapman S, Cowling TG. The mathematical theory of non-uniform gases: an
on pore pressures at depth. Water Resour Res 1983;19(6):1585–97.
account of the kinetic theory of viscosity. Thermal conduction and diffusion in
[10] Tóth J, Otto CJ. Hydrogeology and oil deposits at Pechelbronn-Soultz, Upper
gases. Cambridge Mathematical Library; 1953.
Rhine Graben: Ramification for exploration in intermontane basins. 59–76, In:
13
[43] Hirschfelder JO, Curtiss CF, Bird RB. Molecular theory of gases and liquids. C. F. [60] El-Rawy M, Zlotnik VA, Al-Raggad M, Al-Maktoumi A, Kacimov A, Abdalla O.
and R. B. Bird. New York: Wiley; 1954. xxvi + 1219 pp. Conjunctive use of groundwater and surface water resources with aquifer
[44] Penrose R. The road to reality. New York: Alfred A. Knopf; 2005. recharge by treated wastewater: evaluation of management scenarios in the
[45] Carathéodory C, Variationsrechnung, Frank-von Mises, editor. Die Differential Zarqa River Basin, Jordan. Environ Earth Sci 2016;75(15). doi: https://doi.org/
und Integralgleichungen der Mechanik und Physik. Braunschweig; 1925. 10.1007/s12665-016-5946-1.
[46] Batchelor GK. An introduction to fluid dynamics. Cambridge University Press; [61] Al-Maktoumi A, El-Rawy M, Zekri S. Management options for a multipurpose
1967. coastal aquifer in Oman. Arab J Geosci 2016;9(14). doi: https://doi.org/
[47] Chorin AJ, Marsden JE. A mathemetical introduction to fluid mechanics. New 10.1007/s12517-016-2661-x.
York: Springer-Verlag; 1979. [62] Al-Maktoumi A, Zekri S, El-Rawy M, Abdalla O, Al-Wardy M, Al-Rawas G,
[48] Shinbrod M. Lectures on fluid mechanics. NewYork: Gordon and Breach; 1973. Charabi Y. Assessment of the impact of climate change on coastal aquifers in
[49] Engelen GB. Hierarchically nested energy flow systems of planet earth. In: Oman. Arab J Geosci 2018;11(17). doi: https://doi.org/10.1007/s12517-018-
Proceedings of the international symposium on regional groundwater flow: 3858-y.
theory, applications and future developments, 21-23 June 2013, Xi’an, China; [63] Salem A, Dezso } J, El-Rawy M. Assessment of groundwater recharge,
2013. p. 22–24. evaporation, and runoff in the Drava Basin in Hungary with the WetSpass
[50] McKee CR, Bumb AC, Koenig RA. Stress-dependent permeability and porosity Model. Hydrology 2019; (6): 23.
of coal and other geologic formations. SPE Format Eval 1988;3(01):81–91. [64] Salem A, Dezso } J, El-Rawy M, Lóczy D. Hydrological Modeling to Assess the
[51] Neuzil CE. Groundwater flow in low-permeability environments. Water Efficiency of Groundwater Replenishment through Natural Reservoirs in the
Resour Res 1986;22(8):1163–95. Hungarian Drava River Floodplain. Water 2020; 12 (1): 250.
[52] Truesdell C. The kinematics of vorticity. Bloomington: Indiana University [65] Bourel C, Choquet C, Rosier C, Tsegmid M. Modeling of shallow aquifers in
Press; 1954. interaction with overland water. Appl Math Model 2020;81:727–51.
[53] Abramowitz M, Stegun IA. Handbook of mathematical functions. New [66] El-Rawy M, Batelaan O, Buis K, Anibas C, Mohammed G, Zijl W, and Salem A.
York: Dover Publications. Inc.; 1972. Analytical and Numerical Groundwater Flow Solutions for the FEMME-
[54] Morse PM, Feshbach H. Methods of theoretical physics. New York: McGraw- Modeling Environment. Hydrology 2020; 7(2): 27.
Hill; 1953. [67] Aldaood A. Impact of fine materials on the saturated and unsaturated behavior
[55] Butkov E. Mathematical physics. Reading: Addison-Wesley Publishing of silty sand soil. Ain Shams Eng J 2020;11(3):717–25.
Company; 1973. [68] Mohammadi B, Mehdizadeh S. Modeling daily reference evapotranspiration
[56] Hirasaki GJ, Hellums JD. A general formulation of the boundary conditions on via a novel approach based on support vector regression coupled with whale
the vector potential in three-dimensional hydrodynamics. Quart Appl Math optimization algorithm. Agric Water Manag 2020;237:106145. doi: https://
1968;26(3):331–42. doi.org/10.1016/j.agwat.2020.106145.
[57] Cartwright Keros. Redistribution of geothermal heat by a shallow aquifer. Geol [69] Moazenzadeh R, Mohammadi B. Assessment of bio-inspired metaheuristic
Soc Am Bull 1971;82(11):3197. doi: https://doi.org/10.1130/0016-7606(1971) optimisation algorithms for estimating soil temperature. Geoderma
82[3197:ROGHBA]2.0.CO;2. 2019;353:152–71.
[58] Shan Hsin-yu, Wen-junn Lo. Deformation characteristics of shallow aquifer [70] Mohammadi B, Ahmadi F, Mehdizadeh S, Guan Y, Pham QB, Linh NTT, Tri DQ.
soils. In: Proceedings of the Second International Conference on Unsaturated Developing novel robust models to improve the accuracy of daily streamflow
Soils, vol. 1098. Beijng: International Academic Publishers; 1998. p. 125–130. modeling. Water Resour Manage 2020;34(10):3387–409.
[59] Porter ML, Plampin M, Pawar R, Illangasekare T. CO2 leakage into shallow [71] Mohammadi B, Linh NTT, Pham QB, Ahmed AN, Vojteková J, Guan Y, et al.
aquifers: modeling CO2 gas evolution and accumulation at interfaces of Adaptive neuro-fuzzy inference system coupled with shuffled frog leaping
heterogeneity. Energy Procedia 2014;63:3253–60. algorithm for predicting river streamflow time series. Hydrolog Sci J 2020;65
(10):1738–51.
14

Flow Systems of The Earth's Viscous Subsurface A Complement Togroundwater Flow Systems

Uploaded by

Copyright:

Available Formats

Flow Systems of The Earth's Viscous Subsurface A Complement Togroundwater Flow Systems

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flow Systems of The Earth's Viscous Subsurface A Complement Togroundwater Flow Systems

Uploaded by

Copyright:

Available Formats

Ain Shams Engineering Journal xxx (xxxx) xxx

Contents lists available at ScienceDirect

Ain Shams Engineering Journal

Flow systems of the Earth’s viscous subsurface: A complement to

1. Introduction geological storage of greenhouse gases like carbon dioxide (CO2)

2.4. Introduction to the basic equations

On long time scales, the Earth’s subsurface behaves as a viscous

(p is the pressure, patm is the atmospheric pressure, q is the den-

3.4. Creep flow boundary conditions 3.5. Time scales

If I > 1 then the flow is under-damped; i.e., the amplitudes of

3.10. Compressive forces: Creep force systems

Combination of Eq. (A.3.5) with Eq. (A.3.6) yields @v z @x

qgHD zD zD D

You might also like