You are on page 1of 22

Basin Research (2002) 14, 379400

Quantifying the sequence stratigraphy and drowning


mechanisms of atolls using a new 3-D forward
stratigraphic modelling program (CARBONATE 3D)
G. M. D. Warrlich, D. A. Waltham and D. W. J. Bosence
SedTec Modelling Group, Department of Geology,
Royal Holloway University of London, Egham, Surrey,
TW20 0EX, UK

ABSTRACT
This paper describes a new 3-D forward numerical model (CARBONATE 3D) that simulates the
stratigraphic and sedimentological development of carbonate platforms and mixed
carbonatesiliciclastic shelves by simulating the following sedimentary processes: (1) Carbonate
shallow, open-marine production, dependent on water depth, restriction and sediment input; (2)
Carbonate shallow, restricted-marine production, dependent on water restriction; (3) Pelagic sediment production and deposition; (4) Coarse and fine siliciclastic input; (5) Erosion, transport and
redeposition of sediment, dependent on currents, slope, depth and restriction as well as sediment
grain-size and composition; (6) Dissolution of subaerially exposed carbonate. In this paper the model
is used to investigate the controlling mechanisms on the sequence stratigraphy of isolated carbonate
platforms and atolls and to predict distinctive architectural signatures from different drowning
mechanisms. Investigation of the mechanisms controlling atoll strata shows that although relative
sea-level is the major control, antecedent topography, environmental setting and early diagenesis have
profound influence on what stratigraphic geometries and facies develop. Hence care must be taken if
sea-level curves are interpreted from real stratigraphies. Atoll drowning by fast sea-level rise, by
lowered production and by repeated exposure and fast subsequent sea-level rises are investigated and
different stratigraphic signatures for the respective mechanisms predicted. A fast relative sea-level rise
results in a bucket-shaped morphology developed prior to drowning and a sharp transition from the
platform margin facies to a pelagic cover. Drowning caused by lowered platform margin production is
predicted to result in the development of a dome-shaped, shallow-water shoal over the whole platform
top prior to drowning. Fourth order amplitudes of several tens of metres, typical of `icehouse' settings,
cause atoll drowning at subsidence rates where atolls subject to fourth order amplitude of only a few
metres, typical of `greenhouse' settings, can keep up with the rising sea-level. In the resultant strata,
vertical facies belts are less well developed but horizontally extensive facies bands are more prominent.
High fourth order amplitudes (up to 80 m) without sufficient third order scale subsidence will not lead
to drowning, however, as the platform can recover in each fourth order lowstand. These results
suggest that atolls might be easier to drown in `icehouse' rather than in `greenhouse' conditions but
only in situations with suitably high rates of longer-term relative sea-level rise or sufficient lag times.

INTRODUCTION
In the last 20 years stratigraphic forward modelling of
carbonate platform stratigraphies has contributed to identifying and quantifying mechanisms controlling carbonate
platform evolution as well as unravelling the interplay of
these mechanisms (e.g. Bice, 1988; Scaturo et al., 1989;
Correspondence: G.M.D. Warrlich, Shell Technology, Carbonate Development Team (SEPTAR), Volmerlaan 8, Postbus 60,
Rijswijk 2280 AB, The Netherlands. E-mail: Georg.Warrlich
@shell.com
2002 Blackwell Science Ltd

Bosence & Waltham, 1990). Forward modelling has


further been used to understand the development of specific stratigraphies (e.g. Lawrence et al., 1990; Bosence
et al., 1994; Liu et al., 1998). The current aims are to use
stratigraphic simulations to make predictions about unexposed geometries and facies, to quantify sedimentary
budgets, test sequence stratigraphic models and provide
volumes of facies poorly imaged in the subsurface.
Widely used conceptual models of how carbonate platforms develop are based on qualitative concepts, limited
observation of the sedimentological processes acting in
carbonate depositional environments and observation of
379

G. M. D. Warrlich et al.
preserved strata, deposited over geological time scales.
Without quantitative tests of these conceptual models,
geologists cannot be sure that the processes that form
such large-scale and long-lived structures such as carbonate platforms (i.e. stratigraphic processes) are correctly
described and their interrelations within a stratigraphy
are understood. This limits our understanding of these
systems and the predictive capabilities of such models.
To overcome this problem, a mathematical description
of the stratigraphic processes is required to allow quantitative testing of conceptual stratigraphic models. Since
strata form over geological periods of time, testing cannot
be performed with real-time experiments, the method
commonly used in natural sciences. Numerical experiments with stratigraphic forward computer models offer
an alternative approach. Stratigraphic concepts are formulated as mathematical algorithms, which can be executed
on computers to produce virtual stratigraphies. If the
stratal patterns predicted by conceptual models can be
reproduced by a stratigraphic forward computer model
this suggests that the conceptual models are internally
consistent and the processes adequately understood.
Most stratigraphic forward modelling computer models
have been limited to reproducing 2-D sections of 3-D
stratigraphies. Carbonate stratigraphies can sometimes be
approximated well in 2D, due to the importance of in situ
sediment production. However 2-D programs cannot explore 3-D effects and forcing mechanisms, which have
profound importance in many settings, i.e. when the antecedent topography varies along strike or where out-ofsection transport occurs. Similarly carbonatesiliciclastic
interactions cannot be investigated as siliciclastics typically
derive from a point source.
Here we present a 3-D stratigraphic forward model,
CARBONATE 3D (Warrlich et al., 1999), which simulates the development of pure carbonate and mixed
carbonatesiliciclastic stratigraphies by modelling the
important sedimentary processes with deterministic
algorithms. The program is designed to be applicable to
a variety of carbonate platform settings (e.g. rimmedshelves, isolated and attached platforms, carbonate ramps
and mixed carbonatesiliciclastic systems) and because
sediment transport is an important component of many
carbonate systems (e.g. Burgess et al., 2001), it can simulate sediment transport in all directions. CARBONATE
3D is designed to operate with short run times on a
personal computer so that it can easily and conveniently
be used to test different hypotheses on the development of
real stratigraphies and can help determine the validity of
different hypotheses.
In this paper, we describe the algorithms of CARBONATE 3D and simulate the stratigraphy of an idealized atoll
to demonstrate that the sedimentary and stratigraphic
processes for such a system are simulated adequately, for
if that was not the case we would not get a match. Having
established this consistency between conceptual model
and CARBONATE 3D, the program is then used to
investigate the relative importance of the different controls
380

on atoll stratigraphies. It is found that although relative


sea-level is very important, antecedent topography, environmental setting and early diagenesis as well as an `icehouse' or `greenhouse' sea-level curves have profound
influences as well and need to be considered when interpreting sea-level histories from atoll stratigraphies. Investigations on the drowning of atolls predict that different
drowning mechanisms produce distinctive stratigraphic
signatures.

DESCRIPTION OF CARBONATE 3D
The algorithms of CARBONATE 3D simulate various
sedimentary processes that contribute to the evolution of
a sedimentary surface. Many time steps of a few kilo-years
are used to generate strata over model times of a few
million years. The processes included in the model are
described below.

Carbonate production
Marine carbonate sediments are produced within the
marine environment in what has been described as the
`subtidal carbonate factory' (sensu James & Kendall,
1992) which comprises the shallow marine shelf environment. Here we extend this concept and divide marine
carbonate production into three carbonate factories: benthic shallow, open-marine production, benthic shallow,
restricted-marine production, and pelagic, open-marine
production.
CARBONATE 3D simulates these three different carbonate factories with a respective maximum production
rate M, at which carbonate sediments are produced under
optimal conditions. If conditions deviate from the optimal
configuration, the production rates decrease. Generalizing
the approach of Chappell (1980), who defined four `stress
functions' that reduce reef growth, we define stress functions that model the deviation from optimal conditions for
each carbonate factory. In the most general form, the stress
function S, as a function of the horizontal coordinates
(x, y), is defined by:
Sx; y Px; y=M

Where P is the `stressed' rate of the carbonate factory


considered. The details of how production in each carbonate factory is reduced by the stress functions are explained
below.
Shallow, open-marine production processes
Shallow, open-marine carbonate production results
mainly from marine benthic organisms that are dependent
on photosynthesis and require clear, nutrient poor, welloxygenated, shallow marine waters. Hence the production
rates decrease with decreasing light levels, caused by
increasing water depth (Bosscher & Schlager, 1992), or,
with increasing deviation from fully marine waters (i.e.
increase in salinity, nutrients, temperature or influx of
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning

P0 x; y S0 M0

Where M0 is the maximum production rate for shallow,


open-marine production processes and S0 is the stress
function describing the deviation from M0. S0 is a function
of the stress factors listed above:
S0 S0 z; L; U0

Where z is the water depth, L the transported sediment


load and U0 models the reduction in production rate
caused by restriction, i.e. deviation from fully marine
waters as defined above.
Assuming that the various stress factors act independently of each other, they can be modelled with independent
functions and S0 can be written as:
S0 z; L; U0 D0 zU0 x; yL0 L

Where D0 is the function modelling the depth dependency


of the shallow, open-marine production and L0 is the
function used to simulate the effects of transported sediment on production rate.
Normalized production/depth profile (D0(z)). Almost all
carbonate forward models developed so far use some
type of overall decrease of production rate with water
depth (e.g. Bice, 1988). Two schools of thought for determining the production rate/depth profile exist: (1) Deriving the profile from physical processes like light
attenuation or water temperature (e.g. Lerche et al., 1987)
or (2) using empirically derived profiles (e.g. Bosence &
Waltham, 1990). The first method is more process-based,
but the determination of the variables required is difficult,
and possibly impracticable for the geological past. Also
maximum production rates and depth profiles vary for
different communities, geographical settings and geological time. Hence CARBONATE 3D follows the second
method and uses a production rate-depth relationship
where the rates and shape of the profile can be adjusted
with geologically/palaeontologically constrained data to
the carbonate producing system simulated. To achieve
this, the overall shape of D0(z) is defined as shown in
(Fig. 1). The depth stress factor function D0 together
with a maximum production rate (M0) can be used to
simulate profiles like the one measured over present day
coral reefs (e.g. Hubbard et al., 1990). In addition, production profiles for settings that may have smaller differences
2002 Blackwell Science Ltd, Basin Research, 14, 379400

between shallow and deep-water production (e.g. carbonate ramps) are also reproducible (e.g. Boylan et al., 2002).
Stress on shallow, open-marine carbonate production due
to deviation from fully marine waters (restriction) (U0
(x, y)). Deviation from shallow, open-marine waters (i.
e. restriction) may be caused by an increase in salinity,
temperature, nutrients or influx of fresh water. Increase in
restriction reduces the production rates of the organisms
that are mainly responsible for shallow, open-marine production rate as they only thrive under fully marine conditions. A simple way to quantify all these restriction effects
is to determine for each point on a platform top its distance
to open marine waters, i.e. the platform margin.
A mathematically simple and computationally fast way
to `measure' this distance to open water is to smooth the
water depth z by convolving it with a smoothing filter:
zs x; y zx; y  Gx; y

Where * denotes convolution, z is the un-smoothed depth,


zs the smoothed one (Fig. 2a) and G is a suitable filter. In
this paper we use:

3
r
Gx; y 2 exp 1
where r  L
6
pL
L
Where r2 x2 y2 and L is the horizontal scale length for
restriction. Hence a location on the platform top, close to
the margin, will have a smoothed depth greater than that of
a location of the same un-smoothed depth but further into
the interior of the platform (Fig. 2a). In this way the
smoothed depth indicates the distance of each location to
the platform margin. Increasing the L value smooths the
topography more.
A simple way then to model the effects of restriction on
shallow, open-marine production rate is to set up the
restriction stress factor function U0 as a function of the
smoothed depth zs (Fig. 2b):

Shallow, open-marine production rate (m kyr1)


2.0
4.0

D0 = 0
D0 = z/zt

D0 = 1

M 0D 0

Depth (m)

fresh water e.g. Lerche et al., 1987), or, when wave


energy is too high. Shallow water carbonate production
rates are also generally reduced when the depositional
system is subject to influx of sediment (Hubbard &
Scaturo, 1985), which might also be associated with
decreasing the light levels.
The conditions for this type of carbonate production are
met best at the margins of carbonate platforms. Hence the
products of this production process are labelled `platform
margin' in the computer model output.
The shallow, open-marine production rate P0 is given
by:

zt
zb

zzb
D0 = exp ( )
D

20

Pp

40

Shallow, open-marine
production rate
Pelagic production rate

0.02

0.04

Pelagic production rate (m kyr1)

Fig. 1. A typical production rate depth profile for shallow,


open-marine production (solid line) and pelagic sedimentation
(dotted line).

381

G. M. D. Warrlich et al.
(a)
Depth

Sea-level
Platform morphology
zs
z

Restriction due
to distance from
open water

(b)
Sea-level
1.0

U0

Platform morphology

0.0
Restriction due
to distance from
open water

(c)
Sea-level

1.0

U0

Platform morphology
Sr

0.0

Fig. 2. Illustrations of restriction functions over an idealized isolated carbonate platform (grey area). (a) Depth profile z (black dotted)
over a carbonate platform and the same depth profile smoothed by convolution with a triangular filter, zs (solid). The smoothed depth zs
decreases across the flat platform top and hence can be used to measure the distance from open, fully marine water on the flat platform
top (sea-level dot-dash). (b) Restriction function U0 calculated from the un-smoothed depth (dotted) and from the smoothed depth
(solid). The restriction calculated from the smoothed depth (solid) reduces shallow, open-marine production gradually across the
platform top away from the open water at the margins. (c) Restriction functions caused by distance from open water for shallow,
open-marine (U0 solid black) and shallow, restricted-marine production (Sr dotted).


U0 x; y 1

exp


zs x; y
sR

where sR determines the depth to which restriction is


effective and only non-negative values of zs are allowed.
A location is fully restricted if the smoothed depth is 0.
Stress on shallow, open-marine production by transported sediment (L0). The stresses of sediment influx and turbidity
in the water column on carbonate production rates are
difficult to quantify, but are known to reduce production
rates of, for example, corals (Hubbard & Scaturo, 1985).
CARBONATE 3D uses a simple exponential decrease in
production rate with increasing amount of transported
sediment L, with threshold rates and scaling factors defined by the user:


L Lt
L0 L exp
8
if L > Lt
sL
L0 L 1 if L  Lt

Where sL is the scaling factor and Lt the threshold sediment load for the restriction by transported sediment. The
transported sediment at locations (x, y) is the amount of
entrained sediment moving over that position. Scaturo
et al. (1989) also simulate reduction of carbonate growth
by the influx of siliciclastic sediments with an exponential
function.

waters and also require a low energy hydrodynamic setting. These conditions are best met in the interior of
platforms, where many benthic, photosynthesizing calcifying algae contribute to carbonate production (Neumann
& Land, 1975). This production rate is in general much
lower than that of the shallow, open-marine production
(e.g. Bosence, 1989). Due to the higher wave energy and
more open marine conditions at platform margins, the
shallow, restricted-marine production rate decreases towards the platform margin, where it is replaced by shallow,
open-marine carbonate production.
The shallow, restricted-marine production rate Pr at
every point is defined by a maximum production rate Mr
under optimal conditions for this type of production,
diminished by a stress function Sr.
Pr Mr Sr

The stress function must simulate the stress factors of


increasing hydrodynamic energy, depth and open-marine
conditions. The simplest way to model all these stress
factors in one function is to define Sr as exponentially
decreasing with the spatially smoothed depth zs (Eq. (5)
and Fig. 2a):


zs x; y
Sr exp
11
sR
Using Eq. (7), Eq. (11) can be written as:
Sr 1

Shallow, restricted-marine production processes


Production in restricted-marine water applies to communities that can tolerate deviations from normal marine
382

10

U0

12

From this it follows that Pr is maximal in the platform


interior and decreases towards the platform margin and in
deeper water (Fig. 2c). Therefore the products of this
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning


production process are labelled `platform interior' in the
computer output.
Pelagic production
The pelagic production rate Pp is maximal in the open
ocean and decreases when the water gets shallower (Fig. 1).
Hence the stress function Sp can be modelled as depth
dependent only:
Pp x; y Mp Sp z

13

Again a simple function to model this kind of depth dependent stress on production rate is an exponential increase in production with increasing water depth with a
user defined scale factor sp:


z
Sp 1 exp
14
sp
Bowman & Vail (1999) simulate pelagic productivity in a
similar way. The products are labelled `pelagics' in the
computer output.

Siliciclastic sediment input


Interaction between carbonates and siliciclastics are an
important factor in the development of many carbonate
platforms (e.g. Scaturo et al., 1989). In CARBONATE
3D, siliciclastic sediments are simply supplied to the edge
of the model area as point or line sources. They can be
introduced as two grain-sizes, fine-grained and coarsegrained. Detailed simulation of subaerial and fluvial transport of siliciclastic sediments has not yet been incorporated
into our program.

Sediment dispersal
The redistribution of sediments is of critical importance in
the development of carbonate platforms (e.g. Bosence &
Waltham, 1990). In CARBONATE 3D, sediment dispersal is simulated by separately modelling the generic processes of sediment generation by disintegration, sediment
entrainment, transport and deposition of sediment as
detailed below.
Sediment types
Sediment distribution, i.e. the rates and thresholds for
entrainment and deposition, as well as the mode (bedload or suspended-load) and direction of transport, is
dependent on grain-size and composition. Therefore
these two parameters have to be modelled for all sediments
to be distributed.
In order to maintain fast model execution, the properties of the `virtual' sediments to be distributed are muchsimplified versions of real sediment properties. Grain-size
and composition of real sediments are described here
by two parameters: carbonatesiliciclastic ratio and grain/
2002 Blackwell Science Ltd, Basin Research, 14, 379400

matrix ratio. Following the classification of Dunham


(1962), grain to matrix ratios are used to label the sediment. Grain refers to grain-sizes of medium-grained sand
to granule size, and matrix to mud to fine-grained sand
sizes. Hence every virtual sediment that is to be distributed
(i.e. the products of the algorithms simulating sediment
input) has to be classified in these terms before it enters the
distribution algorithm. In the following, `grain' and
`matrix' are sometimes referred to loosely as `coarsegrained' and `fine-grained', respectively.
The sediments that are distributed on a carbonate platform can be grouped into two classes:
 Siliciclastic sediments brought into the system by rivers
or shorelines;
 Sediments produced by breaking down hard substrate
(disintegration).
These sediments are described with the two parameters
introduced above. Sediments input by rivers are classified
as siliciclastic and grain, or siliciclastic and matrix,
according to their grain-size.
Sediments produced by disintegration can be siliciclastic or carbonate. If they are carbonate, their grain-size is
related to the carbonate factory that produced the sediments. In the simulated atoll shown later in this paper it is
assumed that products of shallow, open-marine production process are taken to disintegrate 70% into `grain' or
coarse-grained sediments and 30% into `matrix' or finegrained sediment. Products of the shallow-restrictedmarine production and of the pelagic production are set
to disintegrate into 100% `matrix' or fine-grained sediments. Bosscher & Southam (1992) make a similar simplification for their 2-D model based on observations on
present day platforms of the Bahamas and Florida where
most of the mud is produced in the platform interior
(Neumann & Land, 1975). However, in CARBONATE
3D, the user can change these grain/matrix ratios for
carbonate depositional environments which show evidence
for different proportions, e.g. early lithification of platform interior muds can lead to disintegration of platform
interior sediments partly into coarse-grained material.
The distribution algorithm is therefore applied to four
different `virtual sediment types': grain carbonate, matrix
carbonate, grain siliciclastic and matrix siliciclastic. When
these virtual sediment types re-enter the sediment distribution algorithm, they keep their grain-size and composition classification. In the following, we describe how the
generic sediment distribution processes are modelled.
Disintegration of rock into sediments
Disintegration is defined as an in situ conversion of hard
substrate (solid rock, solidified sediments and biologically
produced structures) into loose sedimentary grains (sediment load). In the marine realm, many physical, chemical
and biological processes break down hard substrate. The
rate of breakdown B into sedimentary grains depends on
the erosion process, the nature of the substrate being
eroded and the environment it occurs in (Trudgill,
383

G. M. D. Warrlich et al.
1985). The sediments produced by disintegration processes differ in texture (grain-size and shape) and in composition.
CARBONATE 3D only simulates marine environments and hence only disintegration due to wave, current
and biological activity is modelled. This is achieved with a
simple depth and position dependent model based on
concepts of energy distribution in marine carbonate depositional environments. Disintegration rate is linked to
energy, which is highest in the surf zones at platform
margins and decreases with water depth and on the platform top with distance from open water. Making a similar
approach as that for the shallow, open-marine production,
the disintegration rate B can be written as:
B EzUE x; y

15

Where E(z) models the disintegration rate with depth in


the same way as D0(z) models the shallow, open-marine
production (Fig. 1). UE (x, y) is used to measure the
distance to open water in the same way as U0 (x, y) in
Eq. (7), but with independent length scale L and depth
decay length s. However, this process would be modelled
more realistically with a 3-D wave and current model
based on energy in the water. This is however, mathematically very complex and has not been attempted here.
Entrainment of loose sediment
Analysis of the physical forces acting on sediment grains
(e.g. Middleton & Southard, 1984), shows that grains are
entrained if the shear stress t applied to them is greater
than the critical shear stress for entrainment te for that
grain type. Thus in the program, loose carbonate or siliciclastic grains and matrix are entrained if the applied shear
stresses are greater than the te for each sediment type
(cf. Bitzer & Pflug, 1989; Cao & Lerche, 1994). Siliciclastic
sediments brought in by rivers are treated as already entrained when they enter the model area. It is assumed in
the following that the shear stresses produced by currents,
waves or slopes all have similar effects on sediments with
respect to entrainment and sediment transport direction.
Shear stresses acting on the sediment. In an aqueous environment, the total shear stress ttotal acting on sedimentary
grains may be simplified as the combination of the shear
stresses from currents, waves and slopes:
ttotal tcurrent twave tslope

16

The shear stress tensors are further simplified to vectors.


The shear stress due to the slope tslope can be approximated to:
tslope rgD sinaG=G

17

where Dr is the excess density of the submerged sediment,


g is the gravitational acceleration, D the grain-size and a
the maximum slope angle parallel to the gradient vector G.
384

D
D

Fig. 3. Diagram illustrating how Eq. (17) is derived. The view is


parallel to the maximum slope gradient G.

This equation implies that the sediment is modelled as a


fluid of thickness D, as illustrated in Fig. 3.
The effect of the shear stress due to waves (twave) and
currents (tcurrent) is modelled at present in a very simple
way by a unidirectional shear stress vector field where the
shear stress value varies with depth:
twave=current twave=current z

18

The variation with depth is again modelled in the same


way as D0 (Fig. 1) and E (Eq. (15)) with no current above
sea-level. The variables M, zt, zb and s can again be
defined by the user, based on sedimentological data. This
profile can be used as a simple model for wind induced
waves where the shear stress decreases with depth from a
maximum value at the surface, or for a unidirectional
current in an open basin where the maximum value can
be at any depth. It is however, unsatisfactory in many other
situations with more complex current patterns.
The critical shear stress te for entrainment. Much experimental work has been carried out to establish the threshold
shear stress for grain movement. Most of it is based on the
work of Shields (1936), who introduced the dimensionless
Shields variable y, from which the critical shear stress can
be derived for each grain-size D of sediments:
te y rgD

19

Since then more experimental data have been collected


to better constrain y (e.g. Yalin & Karahan, 1979). However, this has only been carried out for unconsolidated
sediments exposed to unidirectional currents, for very
short time spans and under laboratory conditions. In
CARBONATE 3D, y can be used to determine te for the
grain and the matrix grain-sizes, but this is not very satisfactory, since values for y are extracted from experiments
over a time period much shorter than the average time-step
used in the program (15 kyr). Also, published y values are
not very representative for carbonate grains with their
wide range in shape and density.
Due to these difficulties and the lack of data to define y,
the critical shear stress for entrainment te is estimated
from angles of depositional slopes. Equation (17) is used
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning


to convert an angle of repose ac into a threshold shear
stress te:
te rgD sinac

20

Equating Eq. (19) with Eq. (20) leads to the relationship:


y sinac

21

Using published y values for spherical grains, this predicts angles of a few degrees for siliciclastic systems, as
observed in reality. Hence a critical shear stress derived
from the angle of depositional slope is a crude estimate but
in our opinion the best option available.
As a relationship between slope angle and grain-size is
observed in depositional systems (e.g. Kenter, 1990), different threshold angles ac can be used for coarse and fine,
carbonate and siliciclastic sediments, respectively, in
CARBONATE 3D. These ac values can be constrained
from measurements of maximum angles of deposition in
the field. This method will give a minimum for te, since
threshold shear stresses for erosion (under currents) are
found to be higher than those for deposition (e.g.
Hjulstrm, 1935), especially for fine-grained sediments.
Disintegration and entrainment due to oversteepening of
depositional slopes
Sediments are also disintegrated and entrained if they
form slopes greater than a critical angle (cf. Strobel et al.,
1989). This process is independent of water depth. The
critical angle for slope failure as is also dependent on the
composition and grain-size of the slope sediments. Kenter
(1990) has derived a relationship between the slope angle
and sediment texture (grains and matrix) from data of
twenty, mainly ancient, carbonate platforms. From these
data, a critical angle for each grain/matrix ratio is interpolated. In CARBONATE 3D, if the slope angle is above
the critical angle, sediment is disintegrated and removed
until the stable slope angle is re-established.
Subaerial dissolution
Subaerially exposed carbonate sediments are removed out
of the system through dissolution (cf. Bosence & Waltham,
1990). This process is modelled by a user-entered rate of
removal of carbonate sediments that are above sea-level.
The sediment is simply taken out of the simulation and
does not reappear.
Sediment transport
In CARBONATE 3D, sediment transport is simulated by
moving the entrained sedimentary load L along the total
shear stress vector-field ttotal from its position of entrainment until the shear stress is lower than the critical
shear stress for deposition td, then the load is deposited
(Fig. 4). All entrained sediments are transported until they
are redeposited, or are out of the model area. Here the
underlying assumption is that within one time-step of
2002 Blackwell Science Ltd, Basin Research, 14, 379400

the simulation run, entrained sediments are moved from


positions out of equilibrium (i.e. where the shear stress is
above the critical shear stress) to positions in equilibrium
with the shear stress environment.
Transport direction. The transport direction at each point
(x, y) is the direction of the total shear stress at (x, y) (see
Eq. (16) for definition of ttotal). Since tslope is proportional
to D (Eq. (17) whereas tcurrent/wave is not, the transport
direction is dependent on grain-size, i.e. it varies for the
coarse-grained and fine-grained sediments. It follows from
Eq. (17) that the slope affects the transport direction of
coarse-grained material more than the transport direction
of fine-grained sediments. Hence coarse-grained material
tends to follow slopes, whereas fine-grained sediments
tend to follow wave and current directions (Fig. 4c, d). In
the absence of slopes, however, e.g. on shallow water
platform tops, it follows from Eq. (17), that the coarsegrained material follows the current and wave directions
(Fig. 4c, d). This behaviour of coarse-grained material is
typical of bed load, that of fine-grained material typical of
suspended load. Since coarse-grained sediments are transported mainly as bed load and fine-grained sediments
mainly as suspended load, the program simulates these
two different, grain-size dependent modes of transport as
natural systems behave.
Sediment deposition
Critical shear stress for redeposition td. Sediments are
deposited if the shear stress is below a critical value td
(e.g. Hjulstrm, 1935). Similar to sediment entrainment,
td is poorly defined by measurements and stable slope
angles (e.g. from Kenter, 1990) are used to derive
approximate values. For simplicity, in the applications
shown here, the critical shear stress for deposition td is
set at the same value as the critical shear stress for
entrainment te
Deposition rate. Assuming the deposition rate to be proportional to the sediment load L and applying the continuity equation to preserve sediment mass, the load L, at each
point where sedimentation happens along the path S, is
given by:
L L0 e S Sd =X
22
L0 is the initially entrained load, X is the characteristic
transport distance and Sd is the point along the path
S where deposition starts.
A short transport distance X leads to deposition of most
sediments adjacent to the point where shear stress first
becomes low enough for deposition. A large X, however,
results in spreading the redeposited sediment further out,
simulating a more diffusive redeposition pattern. Hence in
CARBONATE 3D, the user can set the value of X for the
four virtual sediment types individually with fine-grained
sediments generally having a large X and coarse-grained
sediments having smaller values for X.
385

G. M. D. Warrlich et al.
Sediment entrainment and redeposition
(b) Fine-grained sediment

(a) Coarse-grained sediment

120
100
80
60
40
20

120
100
80
60
40
20

00

00

50

50

00

40

0
00

20

00
10

00

0 0

1000

000
3000 4
2000

5000

Sediment transport
(c) Coarse-grained sediment

40

00
30

00

00

20

0
00

0 0

1000

2000

3000

4000

5000

(d) Fine-grained sediment

5000 m

5000 m

Point of
entrainment
+
Transport
and
deposition

80m
60m

Transport,
no
deposition

0m
0m

5000 m

0m
0m

120 m

100 m

m
80m

60

m
40

20

40m

5000 m

Fig. 4. Entrainment and re-deposition of sediment in CARBONATE 3D; (a & b) An arbitrary sedimentary surface (5  5 km) with
sea-level at 120 m (no emergent areas) and a dominant wave or current direction towards the bottom left (Fig. 4d). The strength of the
shear stress due to the waves or current decreases with depth (cf. Eq. (18)). In the dark grey areas (in a, b & c), the shear stress is above the
critical shear stress for sediment entrainment, in the light grey areas it is below critical. Making the simplification that the shear stress for
erosion te is equal to that for re-deposition td, loose sediment is entrained in dark areas and deposited in light areas. For fine-grained
carbonate sediments (b), which have a low critical shear stress deposition only occurs in the deepest flat areas, where also the shear stress
due to the waves or currents is lowest. (c & d) Plan view of the surface of a & b, respectively. The total shear stress vector field is shown
for coarse-grained sediments (c mainly down slope, except on the flat platform top where it follows the dominant current direction)
and fine-grained sediments (d follows the wave or current direction and is affected by slope only in deeper waters where the wave or
current strength decreases). In (c), the transport path of a coarse-grained sediment load entrained at point ` ' is also shown. In the dark
grey area, the shear stress is above the critical shear stress for the coarse-grained carbonate sediments and the sediment load is
transported (solid line) along the total shear stress vector field, here governed mainly by the wave or current direction. Off the shelf, the
total shear stress falls below the critical shear stress for deposition (light grey) and re-deposition occurs (dotted line). Here, in the deeper
water, the transport direction is mainly governed by the slopes.

Values for characteristic transport distances can be


obtained from field data; for example by measuring the
distance over which beds change form coarse to fine in a
basinward direction. This will give an estimate over which
distance the main load of coarse sediment is deposited. As
in the case of deriving critical shear stress from the angle of
deposition, this method is crude estimate to get a handle on
parameters that are needed in the modelling, but are very
poorly defined by data. This approach of using field observations to constrain transport distances is discussed
further later in the paper.
386

Virtual facies
Sedimentary processes described above create processbased facies defined in the model as platform margin,
platform interior, pelagic, coarse-grained redeposited carbonate, fine-grained (matrix) redeposited carbonate,
coarse-grained siliciclastic, and fine-grained (matrix) siliciclastic. Since all sedimentary processes are simulated
within each time step, every volume in the virtual stratigraphy is made up of the products of several processes (e.g.
40% redeposited fine-grained carbonate, 30% redeposited
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning


coarse-grained siliciclastic and 10% shallow restrictedmarine in situ, 20% shallow, open-marine in situ). This
combination of process-based facies can be related to a real
rock description in several ways. The simplest way, which
is used in the examples given here, is to describe each
volume with the process-based facies that forms the largest
fraction in the volume (e.g. the volume described above
would be assigned the facies: redeposited fine-grained
carbonate). The facies can also be assigned on the basis of
defined facies as described by Boylan et al. (2002), by
labelling all those sediments with a defined mixture of
different virtual sediment types as one facies (e.g.
50100% platform margin, 030% platform interior,
010% pelagics, 2080% coarse-grained carbonate redeposited, 020% fine-grained redeposited carbonate,
010% coarse-grained siliciclastic redeposited and
010% fine-grained redeposited siliciclastics may be defined as `reefal boundstone').

Simulation of eustatic sea-level and


subsidence
Tectonic subsidence can vary across the modelled area.
Differential subsidence is defined at various points in the
model area and is interpolated for the rest of the area. In
the examples shown here, the subsidence does not vary
across the model area and the sum of subsidence and
eustatic sea-level are displayed as a relative sea-level curve.

APPLICATION OF CARBONATE 3D TO
SIMULATING AN OFFSHORE ATOLL
To illustrate most functions of the program and to demonstrate that CARBONATE 3D simulates the stratigraphic
processes for such a system adequately, we show here the
responses of an offshore atoll to third order sea-level
changes under `greenhouse' conditions. The computer
output is compared to the conceptual sequence stratigraphic model of an atoll of Handford & Loucks (1993)
(Figs 5a and 7a). This atoll model was chosen for comparison because it provides detailed information, synthesized
from various sources, on surface morphology, facies, depositional sequences, systems tract and responses to wind,
third and fourth order sea level changes. We are not aware
of a study of a single, real atoll that provides this level of
information. However, in using this atoll model, we have
had to make some assumptions and give some values for
spatial and temporal scales to the model together with
other input parameters required for the simulation. Elsewhere, the program has also been shown to reproduce the
observed overall geometries and a 79% facies match for a
Late Miocene fringing reef platform at Njar, SE Spain
(Warrlich, 2001).
Subsequently we use the program to investigate different sequence stratigraphic styles of atolls with different
sea-level histories (`icehouse' vs. `greenhouse'), initial
surfaces, energy setting and early diagenesis. Finally we
investigate the different mechanisms for drowning atolls
2002 Blackwell Science Ltd, Basin Research, 14, 379400

and show how they have different sequence stratigraphic


styles.

Modelling a `greenhouse climate' atoll


CARBONATE 3D requires a surface to start the simulation from (initial surface) and the relative changes in
sea-level over the time of simulation in the form of a
relative sea-level curve. Parameters for the sediment
supply and dispersal are also required. These parameters
can be bracketed from measurements of present-day rates
and flume tank experiments, geological fieldwork or can be
deduced from the order of magnitude estimates. The
parameters used in our simulations are constrained by
such rates and these are listed in Table 1 and discussed
below. If the program is found to give realistic results with
parameters bracketed from the lines of evidence mentioned above, this is taken as evidence that the program is
able to simulate the main processes that control the sediment accumulation on atolls. This line of reasoning is
discussed further in the discussion part of the paper.
Initial surface and relative sea-level curve
To compare the simulation with the idealized atoll of
Handford & Loucks (1993) (Figs 5a and 7a) similar initial
surfaces and sea-level curves were chosen (Fig. 5d, e).
Since the above authors give no spatial dimensions of
their atoll, the initial surface used in the simulation is based
on the dimensions of a typical Pacific atoll, Fangataufa
(Buigues, 1997). The simulated area measures 15 km by
15 km with a relief of 300 m (Fig. 5). The slopes do not
exceed 5 , which is typical for the starting surface of many
atolls (Read, 1985).
Handford & Loucks (1993) use a third order sea-level
curve with a superimposed fourth order curve (Fig. 5a)
for their conceptual model. Third order cycles range
from 1 to 10 Myr and have average amplitudes of 50 to
100 m (cf. Goldhammer, 1994). Fourth order cycles are
taken as 100 kyr in duration (cf. Reading & Levell, 1996)
and have typical amplitudes of a few metres during greenhouse periods (cf. Goodwin & Anderson, 1985). The
sea-level curve used in the model consists of a 4 m amplitude 100 kyr sinusoid superimposed on a 40 m amplitude
1 Myr sinusoid to simulate Handford & Loucks' fourth
and third order cycles, respectively (Fig. 5d). The run
time for this simulation (herein called the `standard atoll
run STAR') is about 30 min on a 600-MHz Pentium
processor.
Modelling of windwardleeward effects
In the Handford & Loucks' conceptual model the atoll is
subject to a dominant wind direction, which affects sediment dispersal. This is modelled by setting the constant
wave/current shear stress field parallel to the given wind
direction to simulate wind-induced waves and currents
that control the sediment transport direction in the shallow
387

G. M. D. Warrlich et al.

Wind

(a)
4TH ORDER

DEBRIS
FLOW

HST

EUSTASY

LST

TST

3RD ORDER

LST
TIME

HST

HST
LST

Lowstand
Systems Tract

NC

IDE

BS

TST

LST

THICKNESS

LST

SU

RELATIVE SEA LEVEL

Handford & Loucks (1993)

TIME

CARBONATE 3D
computer model
output
Win
d

(b)

(c)

Platform interior
Platform margin

1km

Pelagics

Coarse-grained
redeposited

A
W

Relative sea level [m]

(d)

in

1km
450
400
350

LST

300

Platform margin

Coarse-grained
redeposited

Platform interior

Fine-grained
redeposited

Time [kyr]

Pelagics

250
0

500

200

400

600

800

1000

Wind

Time line, 100 kyr interval

(e)

400

200

Height [m]

300

100

Sea-level

1km

Fig. 5. Comparison of the initial lowstand conditions of an idealized atoll (a from Handford & Loucks, 1993) with the computer
output of CARBONATE 3D, after 200 kyr runtime (be). This is shown as a 3D computer output with the facies in grey shades (b),
a plan view (c) and a cross section (e) along the line AA' (see c for line of section). Bathymetry (c & e) is measured from a basin floor of 0 m
to an initial surface height of 300 m. Fringing reefs, a karstic plain and erosion of the carbonate sediments into coarse-grained sediments
as well as transport and re-deposition of these sediments onto the slopes are reproduced. The transport direction of the entrained
sediments is only slightly affected by the overall wind direction (e).

water. From simple geometric relationships of wave


heights the size of this wind-induced shear stress is estimated to be a few N m 2 at the surface and in very shallow
depths. It then exponentially decreases with depth over
388

several tens of metres to a negligible value. In this


simulation, the shear stress is held constant for the uppermost 10-m depth and then decreases exponentially with a
scale-depth of 5 m (Table 1).
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning


Table 1. Parameters used in the standard atoll simulation run (STAR).
Parameters used in the standard atoll run (STAR)
Run time
Time step
Cell size

dt
dx

1000.0
2.0
200.0

[kyr]
[kyr]
[m]

Carbonate production
Shallow, open-water maximum production rate
Top depth of interval of maximum production rate
Bottom depth of interval of maximum production rate
Scale depth for shallow water production
Threshold for production restriction by sediment load
Scale for production restriction by sediment load
Shallow, restricted-water maximum production rate
Horizontal scale length for restriction
Scale depth for restriction
Pelagic maximum production rate
Scale depth for pelagic production

Mo
zt
zb
sD
Ft
sF
Mr
sd
sR
Mp
sp

3.0
0.0
5.0
20.0
1.0
6700
0.6
1400.0
30.0
0.05
150.0

[m kyr
[m]
[m]
[m]
[m kyr
[m kyr
[m kyr
[m]
[m]
[m kyr
[m]

Disintegration
Maximum rate of disintegration
Height above sea-level up to which erosion is effective
Bottom depth of interval of maximum disintegration rate
Scale depth for disintegration by shallow water processes
Subaerial dissolution rate

ME
zEs
zEb
sE
A

1.5
2.0
3.0
35.0
0.0

[m kyr 1]
[m]
[m]
[m]
[m kyr 1]

Dg
Dm
Mc
zct
zcb
sc

0.8
0.008
5.0
0.0
10.0
5.0

[mm]
[mm]
[N m 2]
[m]
[m]
[m]

ag

25

[ ]

am

5
40
20

[ ]
[ ]
[ ]

Xg
Xm

1500.0
8000.0

[m]
[m]

Entrainment
Grain size coarse-grained sediment
Grain size fine-grained sediment
Maximum shear stress due to wave and current/wave action
Top depth of interval of maximum shear stress
Bottom depth of interval of maximum shear stress
Scale depth for shear stress due to current/wave action
Critical slope angle for entrainment and deposition for
coarse-grained sediment
Critical slope angle for entrainment and deposition for
fine-grained sediment
Angle of repose for pure coarse-grained slopes
Angle of repose for pure fine-grained slopes
Sediment distribution and deposition
Transport distance, coarse-grained sediment
Transport distance, fine-grained sediment

Carbonate production
The platform margin production rates used are constrained by present-day rates. Maximum reported coral
reef growth rates are up to 15 m kyr 1 (e.g. Schlager,
1981). However considering an average platform margin
production rate (rather than coral reef growth rate),
as required for this simulation program, rates of
14 m kyr 1 are quoted for coral reef rimmed margins
(e.g. Bosence et al., 1994). These production rates also
incorporate carbonate production of other organisms
living on the margin and on the deeper slopes. Here
we use a depth profile shown in Fig. 1, where M0 is
3 m kyr 1 in shallow water and decays with depth over
several tens of metres to a negligible value.
Restriction due to sediment in the water column
(modelled with stress function L0) is difficult to quantify
and the parameters controlling this were adjusted such
2002 Blackwell Science Ltd, Basin Research, 14, 379400

]
]
1
]
1

that shallow, open-marine production was reduced markedly on the leeward side of the atoll during the late transgression and highstand conditions, when sediment supply
is judged to be a maximum with respect to limited accommodation (Sarg, 1988).
The parameters that define restriction due to distance
from open water U0 (depth scale for the restriction and the
smoothing width for the lagoon) can only be quantified on
an order of magnitude scale, based on observation of
lagoon sizes and depths. The smoothing width for the
lagoon for this atoll, sd, is on the scale of 12 km. sR
controls the depth to which restriction is effective (see
Eq. (7)) and is set such that U0 approaches 1 in water
deeper than several tens of metres, i.e. when open water
is approached or when deeper water develops on the platform top. Platform interior production rates are reported
at about 0.5 m kyr 1 (Bosence, 1989).
389

G. M. D. Warrlich et al.

(a)

(b)

Win

1km

A
W

Relative sea-level [m]

(c)

in

1km
450
400

TST

Platform margin

Coarse-grained
redeposited

Platform interior

Fine-grained
redeposited

Pelagics

350
300
Time [kyr]
250
0

500

200

400

600

800

1000

Wind

Time line, 100 kyr interval

(d)
300
200

Sea-level

Height [m]

400

100

1km

Fig. 6. Computer output of CARBONATE 3D of the transgressive systems tract (after 520 kyr runtime). Only the platform margin can
keep up with the sea level rise and a `bucket' morphology develops (d). Eroded platform top sediments are transported preferentially
leeward and are deposited in the lagoon or on the slopes (b).

Pelagic production rates range from 0.0050.08 m


kyr 1 (various authors, data compiled by Enos, 1991).
They decrease to a negligible value in shallow water, i.e.
less than several tens of metres depth (Fig. 1).
Sediment dispersal
Submarine erosion rates reported for reef limestones range
from 0.5 to 4 m kyr 1 (Trudgill, 1985). Since the simulated disintegration is caused by wave action, it decreases
with water depth to a negligible value over the order of
several tens of metres. Erosion due to oversteepening is
simulated independent of water depth. Critical angles of
40 for pure coarse-grained slopes and 20 for pure finegrained slopes are used (cf. Kenter, 1990). In each cell
linear interpolation according to the coarse/fine ratio is
used to derive a critical angle. No subaerial dissolution was
used in these runs, simulating a very arid climate platform
setting. Using subaerial dissolution values of 1.0 m kyr 1,
as would be found in more humid settings (Trudgill, 1985)
leads to thinner strata than the ones shown in Figs 58.
The grain sizes used for the coarse and fine-grained
sediment are coarse sand (0.8 mm) and fine to medium
390

silt (8 mm), respectively, reflecting common types of sediment size produced on carbonate platforms (e.g. bioclastic
sands and lagoonal muds; Ginsburg & Lowenstam, 1958).
Transport distances for fine-grained sediments are taken
to be about five times as far as those of coarse-grained
sediment (Table 1).

Lowstand systems tract (LST)


Conditions for the idealized atoll in a low sea-level stand
are summarized in Fig. 5(a) (Handford & Loucks, 1993),
and the modelling results presented in Figs. 5(b, d, e)
(Note that colour versions of the Figs 58 can be
downloaded from www.gl.rhul.ac.uk/research/nummodel.html).
In the conceptual model, thin, fringing lowstand reefs
develop around the subaerially exposed atoll top. The
steep, shallow margins are subject to slope failure and
mass flows. Windward and leeward effects are limited,
since only small amounts of sediment eroded from the
fringes are transported to the leeward margin.
In the virtual strata produced by CARBONATE 3D
(Figs. 5b, c, e) in situ platform margin growth occurs only
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning

(a)

GRAIN-SHOAL
ISLAND
WIND

AGGRADING
REEF

LE
M EW
AR AR
GI D
N

4TH ORDER

HST

Initial
surface

LST

TST

EUSTASY

3RD ORDER

LST

HST

SU

LST

THICKNESS

TST

BS

IDE

Highstand
Systems Tract

W
IN
M DW
AR A
GI RD
N

TIME
LST

CE

RELATIVE SEA LEVEL


TIME

Handford & Loucks (1993)


CARBONA TE 3D
computer model
output
Win
d

(b)

(c)

1km

A
W

Relative sea-level [m]

(d)

in

1km
450
400

HST

Platform margin

Coarse-grained
redeposited

Platform interior

Fine-grained
redeposited

Pelagics

350
300
Time [kyr]
250
0

500

200

400

600

800

1000

Time line, 100 kyr interval

(e)

Sea-level

Wind A

400

200

Height [m]

300

100

1km

Fig. 7. Comparison of the highstand conditions of an idealized atoll (a from Handford & Loucks, 1993) with the computer output of
CARBONATE 3D, after 820 kyr runtime (be). Decreasing accommodation space during the sea-level highstand leads to more
shedding of sediments, which are eroded on the platform top, onto the slopes. During the early sea-level highstand shallower water
depths on the platform top lead to and an increase in production and export of platform interior, fine-grained material, which is
deposited onto a flatter, deeper leeward slope (e). As the HST develops, periodic exposure of the platform top causes production shut
down which results in export of less sediment on the leeward margin. The windward margin steepens up to 30 and slope failure here
leads to a slight retreat of this margin and slope failure deposits on the windward slope. This is different from the conceptual model (a)
and is discussed in the text.

2002 Blackwell Science Ltd, Basin Research, 14, 379400

391

G. M. D. Warrlich et al.
(a)
500

Greenhouse (standard run)

Time line, 100 kyr interval

Horizontal cut at 300 m

400

Height [m]

1km

Relative sea-level [m]

Sea-level

400

300
250
0

200
100

Time [kyr]
200

400

600

800

1000

1km

500

Icehouse

Relative sea-level [m]

Sea-level

1km

300

350

450
400

(b)

400

Height [m]

Horizontal cut at 300 m

300
200

350
300

100

250

Time line, 100 kyr interval

Time [kyr]
200

400

600

800

1000

1km

500

Height [m]

Lower angle Initial surface


Sea-level
Sea-level curve as in 8A

(c)

400
300
200
100

1km

Time line, 100 kyr interval

Platform margin

Coarse-grained
redeposited

Platform interior

Fine-grained
redeposited

Pelagics

1km

500

Lower erosion rate


Sea-level

1km

(d)

400

Height [m]

Horizontal cut at 300 m

Sea-level curve as in 8A

300
200
100

1km
Time line, 100 kyr interval

500

No dominant
wind direction

1km

Sea-level curve as in 8A
Time line, 100 kyr interval

392

(e)

400

Sea-level

Height [m]

Horizontal cut at 300 m

300
200
100

1km

2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning

(f)
500

Lower angle of
repose

400

Sea-level

1km

Height [m]

Horizontal cut at 300 m

300
200

Sea-level curve as in 10A


100

1km

Time line, 100 kyr interval

(g)

Sea-level

Drowning by rapid
sea-level rise

Time line, 100 kyr interval

Height [m]

400

1300
1200
1100
1000
900
800
700
600
500
400
300

200
100

Time [kyr]
0

Horizontal cut at 300 m

300

1km

(h)

200 400 600 800 1000

500

Drowning by
lowered
production

Sea-level

1km

400

Height [m]

1km

500

at 1300m height

Relative sea-level [m]

Horizontal cut at 300 m

300
200

Sea-level curve as in 8A
100

1km

Time line, 100 kyr interval

(i)

Sea-level

Drowning by rapid rises


and falls in sea-level

1km
Time line, 100 kyr interval

500

at 600m height
Height [m]

400

Relative sea-level [m]

Horizontal cut at 300 m

600

300
200

500

100
400

1km
Time [kyr]

300
0

200 400 600 800 1000

Fig. 8. (af) Comparison of different simulations of atoll strata after 1000 kyr runtime in which the controlling parameters are varied,
one at a time, to compare with the standard atoll run (STAR) in (a) to assess the respective influence on the stratigraphic development.
(gi) Simulations to illustrate the stratigraphic responses of an atoll to drowning by different mechanisms. Each simulation is shown as a
cross-section along the line shown in Fig. 7(c) and a horizontal slice through the simulated stratigraphy at 300 m height (note the black
area around the slice through the strata is background). A represents the standard atoll run taken through the FSST after the full
runtime of 1000 kyr. b shows the effect of a higher amplitude (40 m) in the 100 kyr cyclicity. c is a simulation on an initial surface with
lower angles of slope. df show the effects of lower erosion rates, varying dominant wind direction and early lithification, respectively.
(g) Simulation of drowning caused by a rapid rate of relative sea-level rise produces a bucket-shaped morphology and rapid upward
transition from platform top facies to a pelagic cover. (h) Environmental stress, such as increase or decrease in water temperature
produces a dome-shaped, coarse-grained shoal covering the whole platform prior to drowning. (i) High fourth order sea-level
amplitudes lead to drowning under conditions where atolls subjected to a lower fourth order amplitude can keep up with rising sea-level.

2002 Blackwell Science Ltd, Basin Research, 14, 379400

393

G. M. D. Warrlich et al.
in a narrow band along the flanks of the atoll, and steepens
these flanks to about 12 from initial maximum slope angle
of 5 . Some progradation, limited by the small, shallowwater area available, is reproduced on the windward and
leeward margins and thin lowstand wedges of redeposited
sediment develop (Fig. 5b, e). Hardly any platform interior sediments are produced as the platform top is emergent
during the lowstand, and therefore coarse-grained redeposited sediments produced at the margins dominate the
slopes. The topographic contours (Fig. 5c) show that the
sediment distribution is fairly even around the atoll, although small volumes of sediment eroded on the shallow
tops of the fringing margin are preferentially transported
leeward. Strata deposited during initial falling sea-level are
progressively exposed (Fig. 5e). Pelagic sediments occur in
the deeper areas around the atoll and become dominant at
depths greater than about 150 m.

Transgressive systems tract (TST)


A fast sea-level rise (~1 m kyr 1) leads, at the chosen
parameter setting, to the development of raised rims and
a bucket morphology (cf. Read, 1985 Fig. 6) as observed
in the Quaternary of many atolls (e.g. Purdy & Bertram,
1993). In the simulation, the platform margin `keeps up'
with the rising sea-level, but a deeper lagoon develops
(1020 m) and is filled mainly with fine-grained platform
interior facies and redeposited coarse-grained sands,
eroded at the margins and transported into the lagoon
(Fig. 6a, b, d). The platform margins aggrade and the
slopes steepen up to angles of up to 25 . Windward
leeward effects are more noticeable than in the LST:
thicker deposits occur on the leeward slope, as sediments
eroded on the platform top and margins are preferentially
transported leewards by the simulated current field.
The redeposited slope facies are mainly coarse-grained.
This is because more platform margin strata is eroded and
70% of this facies is assumed to disintegrate into coarsegrained sediments. Only some platform interior fine sediment is exported from the platform top as this is confined
mainly to the deep lagoon. Fine-grained redeposited sediments preferentially occur on the leeward slope as their
distribution is controlled mainly by the dominant current
direction, whereas the coarse-grained sediments have a
wider radial distribution as the slopes as well as the currents control their transport (Fig. 6b). The slope sediments dilute the pelagic contribution on the leeward
margin, so that pelagic facies occur at greater depths here
than on the windward slope.

Highstand systems tract (HST) and falling


stage systems tract (FSST)
Decreasing accommodation during sea-level highstand prevents further aggradation and most sediment produced on
the platform top is shed into deeper water (highstand
shedding) (Droxler & Schlager, 1985) and redeposited
mainly on the leeward slope, causing significant leeward
394

progradation of the platform (Fig. 7b, c, e). Restriction


increases on the platform top due to the shallowing of
water depths, and more platform interior facies are deposited. This is also reflected in the redeposited sediments
on the slope, where more redeposited fine-grained facies
are laid down in deeper waters, on slopes with lower
angles, than coarse-grained redeposited sands (cf. Figs. 6d
and 7e). This increase in redeposited fine, as opposed to
coarse, can be seen in the late TST to HST strata (Fig. 7e).
The dominant wind direction affects the fine-grained
sediments, assumed to be transported in a suspended
mode of transport, more than the coarse-grained ones.
Hence fine-grained sediments form the biggest sediment
fraction only on the leeward slopes, whereas some coarse
sediments fringe the entire platform.
The windward margin steepens up to above 30 and
slope failure leads to a slight retreat of the margin and
redeposition of eroded coarse material on the deeper, lower
angle part of the windward slope (Fig. 7e). These windward effects differ from the conceptual model proposed by
Handford & Loucks (1993; Fig. 7a), where more sediments
are shed off the windward margin than in our simulation,
where most material from the windward crest is transported leewards across the platform as is observed in
some Caribbean banks (Droxler & Schlager, 1985).
The final stage of the run, the FSST (Falling stage
systems tract; Posamentier et al., 1992), is shown as a
cross section in Fig. 8(a). Periodic exposure during the
late stages of the HST and early FSST is caused by the
fourth order cyclicity. This results in thinner slope deposits in the late HST and early FSST than in the early
HST (Fig. 8a).
The standard atoll run (STAR) described above demonstrates that we can reproduce the main stratigraphic
features of a conceptual atoll over third order, a sea-level
cycle with our algorithms. This implies that the translation
of the conceptual ideas of the stratigraphic processes of
carbonate systems into mathematical algorithms is satisfactory and we are simulating the important processes
adequately otherwise we could not reproduce the observed features of the conceptual model (see discussion
later).

INVESTIGATING CONTROLS ON THE


SEQUENCE STRATIGRAPHY OF
ISOLATED PLATFORMS AND ATOLLS
Having gained confidence that the important stratigraphic
processes on an atoll are being simulated adequately, CARBONATE 3D can now be used to investigate the relative
importance of the different processes that control sequence stratigraphic geometries and facies associations.
This is performed by changing one parameter value whilst
leaving the others as set in the standard atoll run or STAR
(Table 1). The results are summarized in Fig. 8 as crosssections along the line AA' (see Fig. 7c) at the end of each
run (after 1 Myr) and horizontal slices through the stratigraphies are cut at 300 m.
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning

`Icehouse' vs. `greenhouse' sea-level curves


Increasing the amplitude of the 100 kyr cyclicity to 40 m
(4 m in STAR, cf. Fig. 8a, b) simulates proposed `icehouse' sea-level curves (cf. Goldhammer, 1994). Compared to the STAR (Fig. 8a) the higher 100 kyr
amplitudes lead to a more aggradational architecture with
slightly steeper windward and leeward slopes; the `bucket
morphology' is more pronounced, as predicted by the
qualitative model of Wright (1992). Less sediment is
transported into the basin and the overall sediment
volume produced is reduced (Fig. 8b). The distinct
lateral zonation of platform interior and platform margin
facies develops on the platform top only during the beginning of the fast rises of the 100 kyr cycles when the
platform top is covered by shallow water. As the sea-level
in each 100 kyr cycle rises at a rate where the platform
top production cannot keep-up, shallow-water production
decreases with increasing water depth. This leads to
the development of a high-energy, coarse-redeposited
facies being deposited across the platform top during
transgression and highstand of every 100 kyr cycle. This
results in a vertically more cyclic arrangement of facies
than in STAR, making across-platform correlation easier
(cf. Lehrmann & Goldhammer, 1999). This run demonstrates a high sensitivity of the atoll system to relative sealevel changes.

Morphology of the initial surface


It has been known that the initial surface, or antecedent
topography, has an important control on atoll development
(Purdy, 1974). The 3-D shape of the initial surface also has
a profound influence on whether progradational, aggradational or retrogradational stratigraphic geometries develop. In Fig. 8(c), the initial surface is altered so that the
maximum angle of slope is reduced from 5 (in STAR) to
about 2 and the effects are: the area in shallow water is
increased, which leads to higher carbonate production
that results in thicker progradational and downstepping
wedges during the initial LST and subsequently increased
progradation of the leeward margin during the TST, HST
and FSST. More production leads to an increase in stratigraphic volume compared to STAR. The windward
margin is asymmetrical as with STAR, but the horizontally layered platform top differentiates between finegrained sediment in sheltered, windward sides and
coarse-reworked sediments towards the leeward margin.
A similar pattern can be seen today on the Great Bahama
Bank (Purdy, 1963).
The increased sediment flux towards the leeward
margin compared to STAR leads to more restriction
there and hence less in situ platform margin production.
However, the leeward margin accumulation keeps up with
the windward margin by deposition of eroded sediments to
compensate the decrease in situ production. This is particularly noticeable during the late TST and HST when
the windward margins are platform margin dominated and
2002 Blackwell Science Ltd, Basin Research, 14, 379400

leeward margins are dominated by redeposited coarsegrained sands.

Environmental setting: energy and dominant


wind direction
The environmental setting also has a profound influence
on the development of the strata. A low energy setting, as
maybe found in an intra-cratonic basin, was simulated with
a reduction in the maximum rate for shallow water erosion
(or `disintegration') from 1.5 to 1.0 m kyr 1 (Fig. 8d).
Compared to STAR the amount of coarse sediment redeposited on the leeward margin is reduced by about 50%,
resulting in no leeward progradation of the atoll and a
much steeper leeward margin (~27 ). Also less coarseredeposited sediment is found in the lagoon. Overall, a
more symmetrical stratigraphy develops with aggradational and concentric facies belts. Hence erosion rate has
a profound, but generally underestimated effect on the
platform morphology and can determine whether platform
shows aggradational or progradational geometries.
A setting with no dominant wind direction was modelled by rotating the current by a few tens of degrees from
time step to time step during the model run (Fig. 8e). This
leads to very symmetric geometries and more aggradational strata than STAR. Such a pattern differs from
symmetric platforms developed in high amplitude, high
frequency periods under a dominant wind direction by the
concentric belt of coarse-redeposited facies, the vertical
continuity of facies belts, absence of platform-top cycles
and less pronounced bucket morphology.

Early lithification
A lower angle of repose for coarse sediments of 25 and for
fine sediments of 15 was used (cf. Kenter, 1990) to simulate the effects of reduced early lithification in platform
slopes (Fig. 8f ). Compared to STAR this results in lower
slope angles (max. angle of 21 ) of the windward margin,
greater areas of shallow water production and transport of
more material onto the leeward slope. This leads to increased leeward progradation. As our algorithms also
translate a lower angle of repose into a lower critical
shear stress for sediment entrainment (Eq. (20)), the
lower angle of repose also result in increased amount of
coarse-redeposited facies in the platform interior.

INVESTIGATING CAUSES OF DROWNING


OF ATOLLS
Factors controlling atoll drowning have been debated extensively in the last 20 years (e.g. Schlager, 1981, 1998;
Zempolich, 1993; Flood, 2001) but have not been investigated quantitatively with respect to stratigraphic aspects.
Here we use CARBONATE 3D to investigate the processes of drowning and the predicted sedimentological and
stratigraphic responses to different proposed drowning
mechanisms.
395

G. M. D. Warrlich et al.
Mechanisms that have been considered (e.g. Schlager,
1981; Erlich et al., 1990) to cause atoll drowning are rapid
sea-level rise, environmental stresses on the carbonate
producing organisms, and repeated drops in relative sealevel followed rapid rises during which carbonate producers cannot re-establish or keep up. These different
drowning mechanisms have been simulated and their respective sedimentary and stratigraphic signals compared
and summarized in Fig. 8(gi).

Drowning by fast sea-level rise


Figure 8(g) shows the results of a run with all parameters
set as in STAR, but the linear rate of subsidence is increased from 0.1 m kyr 1in STAR to 1 m kyr 1, to give
a sea-level of 1300 m at the end of the run. Due to the faster
sea-level rise no lowstand progradation occurs in the LST,
but a narrower, aggradational stratigraphy develops. The
fast sea-level rise leads to a pronounced bucket morphology (bucket to `empty bucket' cf. Schlager, 1989). The
higher producing margins can keep up with the rising sealevel, but the slower producing platform interior cannot
and becomes a depo-centre for coarse, redeposited sediments. No shallow platform interior facies develop because
the water depths are too great. The windward margin is
made up mainly of in situ platform margin facies, which
may be interpreted as reefs, whereas the leeward margin
consists of a mixture of platform margin and coarse redeposited facies, indicating a sand shoal deposit. This asymmetry in windward and leeward margins contrasts with
STAR (cf. Fig. 8a).
Backstepping of both margins occurs as has been predicted by Erlich et al. (1993). The atoll drowns during the
third order TST, when the 1 Myr sinusoidal sea-level
change rate adds to the linear subsidence rate to give a
rate of approximately 1.5 m kyr 1.
During the rest of the run, when the sea-level rises to
1300 m, up to 40 m of pelagic sediments accumulate on top
of the atoll. These thin over the windward margin, where
platform margin production persists longer as it is least
affected by the additional stress caused by suspended
sediment in the water column. Note that the windward
margin is the last point to drown as the production rates are
relatively higher here than on the leeward margin. Similar
drowning asymmetry has been described from drowned
Miocene platforms in SE Asia where windward margins
survive the longest prior to drowning (Erlich et al., 1993).
Here the asymmetry may be further emphasized by synsedimentary tectonic tilting in response to fault-block rotation (Erlich et al., 1993; Bosence et al., 1998).

Drowning by environmental stresses


A reduced platform margin maximum production rate of
1.7 m kyr 1 is used in Fig. 8(h) with platform interior
production unaltered. This is to simulate the stresses
caused by changes in water temperature, nutrients or
suspended siliciclastic sediment in the water column,
396

which might result, for example, in a change from fasterproducing coral-dominated platform margin to an algaldominated community with a slower rate of production.
The reduced platform margin production rate affects the
platform less during the LST and early TST when accommodation is limited. However, during the TST, the atoll
develops into a dome-shaped, coarse-redeposited shoal
covering the whole platform top. Atoll drowning occurs
during the late TST (after 500 kyr) caused by the additional stress of the faster, third order sea-level rise. The
sand shoals backstep significantly prior to drowning,
caused by relatively high erosion on the platform margins.
During sea-level highstand and fall, most of fine sediment
still being produced in the platform top interior is exported
to the leeward slope. A reduced HST is developed
(drowning sequence) and no FSST is recognizable.
Compared to drowning by a fast relative sea-level rise,
this drowning sequence shows more progradation during
the initial LST and a more symmetric morphology develops as the reduced production rate affects the whole
platform top. The atoll drowned by environmental stress
(Fig. 8h) shows very different morphology, architecture
and facies in the TST and HST to that drowned by rapid
sea-level rise. The environmentally stressed atoll does not
backstep to such an extent and maintains a wide, flat
platform top of reworked sands and muds with sloping
margins. No bucket morphology develops and platformtop sediments are shed onto the leeward margin. Responses to the different stresses of changes in temperature,
nutrient supply and community changes cannot be investigated with CARBONATE 3D since these processes are
not specifically modelled.

Drowning by repeated drops and fast


successive rises
This was initially attempted by using the sea-level curve
shown in Fig. 8(b) and increasing the amplitude of the
100 kyr cycles to 80 m (results not shown here). However,
this does not result in drowning of the atoll, as the platform
recovers during each 100 kyr lowstand. The strata produced were similar to that in Fig. 8(b), with increased
amounts of coarse-redeposited facies, loss of the bucket
morphology and formation of a flat topped mounded
morphology. Increasing the linear subsidence from 0.1 to
0.3 m kyr 1 and keeping the 40 m high-frequency amplitude of the `icehouse run' (Fig. 8b), results in permanent
drowning of the platform (Fig. 8i). At this linear subsidence rate and a 4 m amplitude 100 kyr cycle (as used in
STAR), the atoll can keep up and does not drown.
A high fourth order sea-level curve results in less production over the whole third-order cycle as the platform
top is emergent or in deep water for longer periods of time.
Hence drowning of atolls is easier under `icehouse' than
under `greenhouse' conditions, but a sufficiently fast rate
of longer-term relative sea-level rise is required to prevent
recovery of the isolated platform during fourth-order
lowstands. Only if a lag time in the onset of carbonate
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning


production in the order of 100 kyr after platform emergence is assumed, could the platform be drowned by a
fourth order sea-level oscillation for lower rates of longterm relative rise.
Drowning by high-amplitude fourth-order sea-level oscillations produces more laterally continuous facies sheets
on the platform top than in the standard run (Fig. 8a) or
when drowning results from a rapid linear rise alone
(Fig. 8g). As in the `icehouse run' (Fig. 8b) small-scale
stratigraphic cycles are caused by rapid increases in water
depth over the whole platform top when production
cannot keep up with the sea-level rise rates. Compared to
drowning by a steady sea-level rise, the bucket morphology
and the aggradational, concentric facies belts are less well
developed and a more symmetric platform develops. Significant backstepping occurs just prior to drowning as is
the case in the other model runs.

DISCUSSION
Validity of stratigraphic forward modelling
CARBONATE 3D can reproduce the stratal geometries,
broadly assigned facies distributions and systems tracts of
an atoll deposited over a third order sea-level change with a
superimposed fourth order cyclicity. It simulates in situ
platform margin, interior, slope and pelagic facies distributions comparable to those described in Handford &
Loucks's (1993) conceptual model of an atoll. This is also
true for the redistribution of coarse and fine-grained sediments, which are governed by a combination of the slopes
and wind-induced waves and currents.
Windwardleeward effects on facies and geometries are
also reproduced as are depositional sequences generated in
response to sea-level changes.
Minor asymmetries in topography and facies develop
during the run (Figs 5b, c, 6a, b and 7b, c). Although
apparently realistic, these are caused by numerical artifacts
due to the combined effects of sediment transport, sediment deposition and depth-dependent in situ production
(e.g. Waltham, 2000). A further problem with CARBONATE 3D in this application is the oversimplified current/
wave model. This does not take into account the decrease
in shear stress as currents pass across the platform top.
A more sophisticated current model needs to be developed. Despite these criticisms we are able to reproduce
most of the features described in a well-known and cited
conceptual model of an atoll (Handford & Loucks, 1993).
It is best to use well-understood fundamental physical
laws in numerical models. However, many geological processes and parameters are poorly described, poorly understood, and not well constrained by geological data.
Consequently, a more semi-empirical or `rule based'
(cf. Paola, 2000) approach must be used, and parameters
estimated from semi-quantitative field and lab observations. The approach taken in CARBONATE 3D to
model sediment distribution using `transport distances' is
an example of this. Constraining such parameters is
2002 Blackwell Science Ltd, Basin Research, 14, 379400

difficult and has been tackled by different authors in


different ways. Parameters like transport distance can be
bracketed by field measurements, e.g. the offshore transport of shallow water ooids on a carbonate ramp (Aurell
et al., 1995) so observations made at a few locations in the
field are used to model the behaviour of the whole depositional system. This approach must be used with care. For
example, measuring transport distance from an outcrop
example, and then using the model to reproduce the strata
in that outcrop example proves little and may be considered circular reasoning. Applying data measured from
one example to a different example, as has been done here,
is more robust in that it suggests that the measurements are
more generally applicable.
A further line of evidence why forward models using
`rule-based' approaches are not circular is, paradoxically,
their failure to reproduce real world features of depositional systems. For example, a model based on slope
diffusion cannot simulate sediment removal from a flat
platform or sediment transport across it because the sediment transport processes are not adequately represented in
such a model. A truly circular model should always produce a perfect answer.
This discussion has implications for what a match between a numerical and conceptual model actually means.
As discussed above, CARBONATE 3D is based on semiquantitative equations of sedimentological and stratigraphic processes. These are in turn derived from analysis
of deposited strata and sedimentary processes, i.e. conceptual models of how carbonate systems work. Hence a
match between the two might be considered a mere consistency check. However, whereas internal consistency
cannot be demonstrated for a conceptual model, it can be
for numerical models; they are falsifiable. If the wrong
processes are simulated or the processes are not adequately
modelled, then the conceptual model (or a real stratigraphy)
cannot be matched. For example, using a diffusion-based
sediment transport mechanism to simulate sediment transport on a carbonate platform top cannot reproduce highstand shedding of a carbonate platform and the right
amount of progradation (e.g. Hussner et al., 2001).
Hence the failure or success of stratigraphic numerical
models to reproduce real stratigraphies and conceptual
models tells us whether we have understood and described
the relevant processes adequately. The numerical models
can be used to falsify conceptual models, i.e. if they fail to
work adequately then our models are too simple or do not
describe the relevant processes. The corollary is that the
match of a conceptual model with a numerical model gives
us more confidence that we have identified and properly
described the relevant processes. If the conceptual model
is not translated into mathematical language and tested
with a computer model we have no way of knowing if it
works. Only when we have confidence in our models
should we use them to make predictions. However, as
with all models, the assumptions underlying the method
must be considered carefully to avoid attributing undue
significance to the model results.
397

G. M. D. Warrlich et al.

Modelling the sequence stratigraphy of atolls


Having established that we are simulating the important
sedimentological and stratigraphic processes adequately,
CARBONATE 3D is used to investigate the controls on
the formation of isolated platforms and atolls. Whilst sealevel changes have a very strong control on stratigraphic
architecture, antecedent topography, platform setting and
early diagenesis are also found to have major influences on
the stratigraphy of atolls. It is important to recognize these
factors to avoid misinterpretation and assignment of all
changes in architectural style to relative sea level changes.
For example, the progradation caused by a lower angle of
the slope of the underlying surface or smaller amounts of
early lithification might be misinterpreted as resulting
from slower rates of sea-level rises. Here it is shown to
form as a response to an increased area in shallow water
where high carbonate production is possible. This is ultimately caused by the lower angle of slope and less early
lithification, respectively, and not by a slower rate of rising
sea-level, as might be interpreted.
During `icehouse' conditions more symmetric atolls
with a bucket morphology and less well-developed concentric facies belts develop compared to atolls formed with
a `greenhouse' sea-level curve. The `icehouse' atoll has
more laterally continuous facies sheets covering the
whole platform top developed which would be expected
to be cyclically arranged, making across platform correlation easier. `Greenhouse' platform stratigraphies are predicted to be characterized by aggradationally stacked
concentric facies belts.
Comparing the runs shown in Fig. 8(af ), it can be
concluded that the area of shallow water available for carbonate production (i.e. size of the carbonate factory) and
the accommodation space available are the main controls on
the amount of carbonate sediments produced. This comparison also shows that highest angles of slope develop
when the platform margin production is at optimum
values, i.e. when shallow water is available for the platform
margin carbonate factory. Therefore steep angles of slope
develop during the `greenhouse' and `icehouse' runs
(Figs. 8a, b) and in the run with lowered erosion rates
(Fig. 8d). Lower slope angles form during drowning, increased sediment flux (Fig. 8e) as well as in more obvious
situations where underlying slopes are less steep (Fig. 8c) or
when sediments have lower angles of repose (Fig. 8f ).
Characteristic stratigraphic patterns of drowning sequences are predicted for different drowning mechanism
end-member cases. However the influence of different
initial surfaces might also be an overriding control and
combination of the different factors may lead to non-linear
results. Atoll drowning caused by an increased rate of
relative sea-level rise is distinct and is predicted to have a
more bucket-shaped morphology (cf. Wright, 1992) and
a rapid upward-transition from platform margin facies to a
pelagic cover. A markedly asymmetric stratigraphy develops on the atoll in response to increased windward
production rates (cf. Erlich et al., 1993). Atoll drowning
398

due to environmental stress however, results in the development of a dome-shaped, coarse-grained shoal covering
the whole platform prior to drowning. This type of shoal is
observed in the stratigraphy of the drowned Devonian atoll
of Judy Creek (Wendte & Stoakes, 1982).
High-amplitude fourth order sea-level changes do not
cause platform drowning independently of other factors,
since platforms can recover during the lowstands. However, fourth-order changes may facilitate drowning during
a longer-term relative rise as the platform top is in conditions favouring carbonate production for a shorter period
during each fourth-order cycle. This suggests that permanent drowning is controlled as much by long-term
subsidence rate and amplitude of third order cyclicity.
Atolls might be more likely to drown with typical `icehouse' amplitudes and frequencies of sea-level change
rather than in `greenhouse' conditions, but permanent
drowning will still be dependent on the tectonic setting
and the longer-term sea-level curve.

CONCLUSIONS
A stratigraphic forward computer model that can simulate
the evolution of carbonate platforms and mixed, shallowwater, carbonatesiliciclastic systems in 3D has been
developed. It is based on mathematical representation of
stratigraphic and sedimentary processes. Consistency between a published conceptual model of an isolated atoll and
computer simulations is demonstrated. This gives confidence that the processes resulting in atoll formation are
understood and adequately modelled in three dimensions.
Investigating the controls on the development of atoll
stratigraphies shows that although relative sea level is the
major control, antecedent topography, environmental setting and early diagenesis also have profound influence on
what geometries and facies develop. Hence care must be
taken if sea level curves are interpreted from real strata, as
similar stratigraphic geometries may result from different
processes.
Investigating mechanisms of atoll drowning with CARBONATE 3D predicts different stratigraphic signatures
for drowning by a fast sea-level rise, drowning by frequent
exposure and reflooding and drowning by environmental
stresses.
Due to the way the algorithms are constructed, CARBONATE 3D is applicable to many different carbonate
platform settings and can be used to investigate real and
conceptual models of carbonate depositional environments
and stratigraphies as well as carbonatesiliciclastic interactions. The program's fast run times (approximately
30 min on standard PCs) allow it to be used to test different
hypotheses of how actual stratigraphies develop. Simulations of real stratigraphies can be used as an input into
hydrocarbon reservoir models, which are then based on
stratigraphic and sedimentary processes, an improvement
to methods currently used.
2002 Blackwell Science Ltd, Basin Research, 14, 379400

A new 3-D model to quantify stratigraphy and drowning

ACKNOWLEDGEMENTS
The authors would like to thank the EU Commission and
the Thomas Holloway scholarship for sponsoring the PhD
project of G. Warrlich, on which this paper is based, and
TOTALFINAELF for sponsoring postdoctoral research.
Trevor Burchette (BP, Sunbury), Peter Burgess (Shell E &
P, Rijswijk), Bruno Caline, Patrick Henriquel and Enzo
Insalaco (TOTALFINAELF), Yvette Baker & Amy Boylan (Royal Holloway University of London), contributed
valuable discussion. Peter Burgess and Bruce Wilkinson
are thanked for critical and constructive reviewing and
editing.

REFERENCES
Aurell, M., Bosence, D.W.J. & Waltham, D.A. (1995) Carbonate ramp depositional systems from a late Jurassic epeiric
platform (Iberian Basin, Spain): a combined computer modelling and outcrop analysis. Sedimentology, 42, 7594.
Bice, D. (1988) Synthetic stratigraphy of carbonate platform and
basin systems. Geology, 16, 703706.
Bitzer, K. & Pflug, R. (1989) DEPO3D: A three-dimensional
model for simulating clastic sedimentation and isostatic compensation in sedimentary basins. In: Quantitative Dynamic
Stratigraphy (Ed. by T.A. Cross), pp. 335348. Prentice
Hall, Englewood Cliffs, NJ.
Bosence, D.W.J. (1989) Biogenic carbonate production in Florida Bay. Bull. Mar. Sci., 44, 419433.
Bosence, D.W.J., Cross, N. & Hardy, S. (1998) Architecture
and depositional sequences of Tertiary fault-block carbonate
platforms; an analysis from outcrop (Miocene, Gulf of Suez)
and computer modelling. Mar. Petroleum Geol., 15, 203221.
Bosence, D.W.J., Pomar, L., Waltham, D.A. & Lankester,
T.H.G. (1994) Computer modelling a Miocene carbonate
platform, Mallorca, Spain. AAPG Bull., 78, 247266.
Bosence, D.W.J. & Waltham, D.A. (1990) Computer modelling
of the internal architecture of carbonate platforms. Geology,
18, 2630.
Bosscher, H. & Schlager, W. (1992) Computer simulation of
reef growth. Sedimentology, 39, 503512.
Bosscher, H. & Southam, J. (1992) CARBPLAT a computer
model to simulate the development of carbonate platforms.
Geology, 20, 235238.
Bowman, S.A. & Vail, P.R. (1999) Interpreting the stratigraphy
of the Baltimore canyon section, offshore New Jersey with
PHIL, a stratigraphic simulator. In: Numerical Experiments in
Stratigraphy: Recent Advances in Stratigraphic and Sedimentologic Computer Simulations (Ed. by W. Harbaugh, W.L.
Watney, E.C. Rankey, R. Slingerland, R.H. Goldstein &
E.K. Franseen). Spec. Publishers SEPM, 62, 117138.
Boylan, A.L., Waltham, D.A., Bosence, D.W.J., Badenas, B.
& Aurell, M. (2002) Digital rocks: linking forward modelling
to carbonate facies. Basin Res., 14, 401415.
Buigues, D.C. (1997) Geology and hydrogeology of Mururoa and
Fangataufa, French Polynesia. In: Geology and Hydrogeology of
Carbonate Islands (Ed. by H.L. Vacher & T.M. Quinn),
pp. 433452. Elsevier Science B.V., Amsterdam.
Burgess, P.M., Wright, V.P. & Emery, D. (2001) Numerical
forward modelling of peritidal carbonate parasequence development: implications for outcrop interpretation. Basin Res.,
13, 116.
2002 Blackwell Science Ltd, Basin Research, 14, 379400

Cao, S. & Lerche, I. (1994) A quantitative model of dynamical


sediment deposition and erosion in three dimensions. Computers Geosciences, 20, 635663.
Chappell, J. (1980) Coral morphology, diversity and reef
growth. Nature, 286, 249252.
Droxler, A.W. & Schlager, W. (1985) Glacial versus interglacial sedimentation rates and turbidite frequency in the Bahamas. Geology, 13, 235239.
Dunham, R.J. (1962) Classification of carbonate rocks according
to depositional texture. In: Classification of Carbonate Rocks
(Ed. by W.E. Ham). Mem. Am. Ass. petrol. Geol., 1, 108121.
Enos, P. (1991) Sedimentary parameters for computer modelling. In: Sedimentary Modeling: Computer Simulations and
Methods for Improved Parameter Definition (Ed. by E.K.
Franseen, W.L. St. Watney, C.G.C. Kendall & W. Ross)
Bull. Kansas Geol. Surv., 233, 6399.
Erlich, R.N., Barrett, S.F. & Guo, B.J. (1990) Seismic and
geologic characteristics of drowning events on carbonate platforms. AAPG Bull., 74, 15231537.
Erlich, R.N., Longo, A.P. & Hyare, S. (1993) Response of
carbonate platform margins to drowning: Evidence of environmental collapse. In: Carbonate Sequence Stratigraphy (Ed. by
R.G. Loucks & J.S. Sarg), Mem. Am. Ass. Petrol. Geol., 57,
241266.
Flood, P.G. (2001) The Darwin Point' of Pacific Ocean atolls
and guyots: a reappraisal. Palaeogeography, Palaeoclimatology,
Palaeoecology, 175, 147152.
Ginsburg, R.N. & Lowenstam, H.A. (1958) The influence of
marine bottom communities on the depositional environment
of sediments. J. Geol., 66, 310318.
Goldhammer, R.K. (1994) Sequence stratigraphic architecture
and cycle stacking patterns of platform carbonates An evaluation of autocyclic, deterministic and stochastic controls.
Tulsa, AAPG 199495 Distinguished Lecture Tours. AAPG
Bull., 78, 1808.
Goodwin, P.W. & Anderson, E.J. (1985) Punctuated aggradational cycles: a general hypothesis of episodic stratigraphic
accumulation. J. Geol., 93, 515536.
Handford, C.R. & Loucks, R.G. (1993) Carbonate depositional
sequences and systems tracts responses of carbonate platforms to relative sea-level changes. In: Carbonate Sequence
Stratigraphy (Ed. by R.G. Loucks & J.S. Sarg), Mem. Am.
Ass. Petrol. Geol., 57, 341.
Hjulstrm, F. (1935) Studies of the morphological activity of
rivers as illustrated by the river Fyris. Bull. Geol. Inst. Uppsala,
25, 221527.
Hubbard, D.K., Miller, A.I. & Scaturo, D. (1990) Production
and cycling of calcium carbonate in a shelf-edge reef system
(St. Croix, U.S. Virgin Is.): Applications to the nature of reef
systems in the fossil record. J. Sedimentary Petrol., 60,
335360.
Hubbard, D.K. & Scaturo, D. (1985) Growth rates of seven
species of scleractinian corals from Cane Bay and Salt River,
St. Croix, USVI. Bull. Maine Sci., 36, 117125.
Hussner, H., Roessler, J., Betzler, C., Petschick, R. & Peinl,
M. (2001) Testing 3D computer simulation of carbonate
platform growth with REPRO: the Miocene Llucmajor carbonate platform (Mallorca). Palaeogeography, Palaeoclimatology, Palaeoecology, 175, 239247.
James, N.P. & Kendall, A.C. (1992) Introduction to carbonate
and evaporite facies models. In: Facies Models. Response to sealevel Change (Ed. by R.G. Walker & N.P. James). Geol. Ass.
Canada, 265275.

399

G. M. D. Warrlich et al.
Kenter, J.A.M. (1990) Carbonate platform flanks: slope angle
and sediment fabric. Sedimentology, 37, 777794.
Lawrence, D.T., Doyle, M. & Aigner, T. (1990) Stratigraphic
simulation of sedimentary basins: Concepts and calibration.
AAPG Bull., 74, 273295.
Lehrmann, D.J. & Goldhammer, R.K. (1999) Secular variation
in parasequence and facies stacking patterns of platform carbonates; a guide to application of stacking-patterns analysis in
strata of diverse ages and settings. In: Advances in Sequence
Stratigraphy: Application to Reservoirs, Outcrops and Models
(Ed. by P.M. Harris, A.H. Saller & J.A. Simo), Spec. Publishers
SEPM., 63, 187225.
Lerche, I., Dromgoole, E., St. Kendall & C.G.C., Walter,
L.M. & Scaturo, D. (1987) Geometry of carbonate bodies: a
quantitative investigation of factors influencing their evolution. Carbonates Evaporites, 2, 1542.
Liu, K., Pigram, C.J., Paterson, L., St. Kendall & C.G.C.
(1998) Computer simulation of a Cenozoic carbonate platform,
Mariou Plateau, North-East Australia. Spec. Publishers International Ass. Sediment., 35, 145161.
Middleton, G.V. & Southard, J.B. (1984) Mechanics of Sediment Movement, 2nd edn., SEPM Short Course no. 3., chapter
6, 165216.
Neumann, C.A. & Land, L.S. (1975) Lime mud deposition and
calcareous algae in the Bight of Abaco: a budget. J. Sedimentary
Petrol., 40, 763786.
Paola, C. (2000) Quantitative models of sedimentary basin
filling. Sedimentology, 47, 121178.
Posamentier, H.W., Allen, G.P., James, D.P. & Tesson, M.
(1992) Forced regressions in a sequence stratigraphic framework; concepts, examples, and exploration significance. AAPG
Bull., 76, 16871709.
Purdy, E.G. (1963) Recent calcium carbonate sediments of the
Great Bahama Bank. 2 Sedimentary facies. J. Geol., 71,
472497.
Purdy, E.G. (1974) Reef configurations; cause and effect. In: Reefs
in Time and Space; Selected Examples from the Recent and Ancient
(Ed. by L.F. Laporte), Spec. Publishers SEPM, 18, 976.
Purdy, E.G. & Bertram, G.T. (1993) Carbonate concepts from
the Maldives, Indian Ocean. AAPG studies in Geology, Vol. 34,
AAPG, Tulsa, USA.
Read, J.F. (1985) Carbonate platform facies models. AAPG Bull.,
69, 121.
Reading, H.G. & Levell, B.K. (1996) Controls on the sedimentary rock record. In: Sedimentary Environments: Processes, Facies
and Stratigraphy (Ed. by H.G. Reading), pp. 536., Blackwell
Science, London.
Sarg, J.F. (1988) Carbonate Sequence stratigraphy. In: Sea Level
Changes: an Integrated Approach (Ed. by C.K. Wilgus, B.S.
Hastings, C.G. St. Kendall, H.W. Posamentier, C.A. Ross &
J.C. Van Wagoner). Spec. Publishers SEPM., 42, 155181.
Scaturo, D.M., Strobel, J.S., St. Kendall, C.G.C., Wendte,
J.C., Biswas, G., Bezedek, J. & Cannon, R. (1989) Judy

400

Creek: A case study of two-dimensional sediment deposition


simulation. In: Controls on Carbonate Platform Development
(Ed. by P.D. Crevello, J.J. Wilson, J.F. Sarg & J.F. Read),
Spec. Publishers SEPM., 44, 6476.
Schlager, W. (1981) The paradox of drowned reefs and carbonate platforms. Geol Soc. Am. Bull., 92, 197211.
Schlager, W. (1989) Drowning unconformities on carbonate
platforms. In: Controls on Carbonate Platform Development
(Ed. by P.D. Crevello, J.J. Wilson, J.F. Sarg & J.F. Read),
Spec. Publishers SEPM., 44, 1525.
Schlager, W. (1998) Exposure, drowning and sequence boundaries on carbonate platforms. In: Reefs and Carbonate Platforms
in the Pacific and Indian Oceans (Ed. by G.F. Camoin & P.J.
Davies), Spec. Publishers International Ass. Sediment., 25, 321.
Shields, A. (1936) Anwendung der Ahnlichkeitsmechanik und
der Turbulenz Forschung auf die Geschiebebewegung.
Mitteilungen der Preussischen Versuchsanstalt fur Wasserbau
und Schiffbau (Berlin).
Strobel, J., Cannon, R., St. Kendall, C.G.C., Biswas, G. &
Bezdek, J (1989) Interactive (SEDPAK) Simulation of Clastic
and Carbonate sediments in shelf to Basin settings: Computers
& Geosciences, 15, 12791290.
Trudgill, S. (1985) Limestone Geomorphology., Longman, New
York.
Waltham, D. (2000) Computer modelling of complex systems.
31st IGC meeting abstracts.
Warrlich, G.M.D. (2001) Computer modelling of carbonate
platform evolution. Unpublished PhD Thesis, University of
London.
Warrlich, G.M.D., Bosence, D.W.J. & Waltham, D.A.
(1999) CARBONATE 3D: a computer model to simulate
carbonate platform evolution in 3 dimensions. J. Conf. Abstracts, 4, 979.
Wendte, J.C. & Stoakes, F.A. (1982) Evolution and corresponding porosity distribution of the Judy Creek reef complex,
Upper Devonian, central Alberta. In: Canada's Giant Hydrocarbon Reservoirs (Ed. by W.G. Cutler), Canadian Soc. Petrol.
Geol. Core Conference Manual, pp. 6381.
Wright, V.P. (1992) Speculations on the controls on cyclic
peritidal carbonates: ice-house versus greenhouse eustatic controls. Sedimentary Geol., 76, 15.
Yalin, M.S. & Karahan, E. (1979) Inception of sediment transport. J. Hydraulics Division, 105, 14331442.
Zempolich, W.G. (1993) The drowning Succession in the Jurassic Carbonates of the Venetian Alps, Italy: A record of supercontinental breakup, gradual eustatic rise, and eutrophication
of shallow water environments. In: Carbonate Sequence Stratigraphy (Ed. by R.G. Loucks & J.S. Sarg), Mem. Am. Ass.
Petrol. Geol., 57, 63105.
Manuscript received 30 March 2001; Manuscript accepted 13 May
2002.

2002 Blackwell Science Ltd, Basin Research, 14, 379400

You might also like