Content uploaded by R. L. Bras
Author content
All content in this area was uploaded by R. L. Bras on Feb 12, 2014
Content may be subject to copyright.
1
Chapter 12
The Channel-Hillslope Integrated Landscape
Development Model (CHILD)
Gregory E. Tucker1, Stephen T. Lancaster2, Nicole M. Gasparini, and Rafael L.
Bras
Department of Civil and Environmental Engineering, Massachusetts Institute of Technology,
Cambridge, MA USA
1. INTRODUCTION
Numerical models of complex Earth systems serve two important
purposes. First, they embody quantitative hypotheses about those systems
and thus help researchers develop insight and generate testable predictions.
Second, in a more pragmatic context, numerical models are often called
upon as quantitative decision-support tools. In geomorphology,
mathematical and numerical models provide a crucial link between small-
scale, measurable processes and their long-term geomorphic implications. In
recent years, several models have been developed that simulate the structure
and evolution of three-dimensional fluvial terrain as a consequence of
different process “laws” (e.g., Willgoose et al., 1991a; Beaumont et al.,
1992; Chase, 1992; Anderson, 1994; Howard, 1994; Tucker and Slingerland,
1994; Moglen and Bras, 1995). By providing the much-needed connection
between measurable processes and the dynamics of long-term landscape
evolution that these processes drive, mathematical landscape models have
posed challenging new hypotheses and have provided the guiding impetus
1 Now at: School of Geography and the Environment, University of Oxford, Mansfield Road,
Oxford OX1 3TB United Kingdom, email: greg.tucker@geog.ox.ac.uk
2 Now at: Department of Geosciences, Oregon State University, Corvallis, OR
2 Chapter 12
behind new quantitative field studies and DEM-based analyses of terrain
(e.g., Snyder et al., 2000). The current generation of models, however, shares
a number of important limitations. Most models rely on a highly simplified
representation of drainage basin hydrology, treating climate through a simple
“perpetual runoff” formulation. The role of sediment sorting and size-
dependent transport dynamics has been ignored in most studies of drainage
basin development, despite its importance for understanding the interaction
between terrain erosion and sedimentary basin deposition (e.g., Gasparini et
al., 1999; Robinson and Slingerland, 1998). Furthermore, with the exception
of the pioneering work of Braun and Sambridge (1997), the present
generation of models is inherently two-dimensional, describing the dynamics
of surface evolution solely in terms of vertical movements without regard to
lateral displacement by tectonic or erosional processes.
Our aim in this paper is to present an overview of the Channel-Hillslope
Integrated Landscape Development (CHILD) model, a new geomorphic
modeling system that overcomes many of the limitations of the previous
generation of models and provides a general and extensible computational
framework for exploring research questions related to landscape evolution.
We focus here on reviewing the underlying theory and illustrating the
capabilities of the model through a series of examples. Discussion of the
technical details of implementation is given by Tucker et al. (1999, 2000)
and Lancaster (1998). We begin by briefly reviewing previous work in
landscape evolution modeling. We then discuss the theory and capabilities of
the modeling system, and present a series of examples that highlight those
capabilities and yield some useful insights into landscape dynamics.
2. BACKGROUND
The first quantitative geomorphic process models began to appear in the
1960s, stimulated by the combination of an intellectual shift toward
investigating the mechanics of erosion and sedimentation processes, and the
appearance of digital computers. The earliest models were one-dimensional
slope simulations developed to explore basic concepts in hillslope profile
development (e.g., Culling, 1960; Scheidegger, 1961; Ahnert, 1970; Kirkby,
1971; Luke, 1972; Gossman, 1976). These studies helped to quantify and
formalize some of the concepts of hillslope process and form enunciated by
early workers such as Gilbert (1877) and Penck (1921). Similar one-
dimensional (1D) approaches have more recently been used to examine the
evolution of stream profiles (e.g., Snow and Slingerland, 1987) and alluvial
deposystems (e.g., Paola et al., 1992; Robinson and Slingerland, 1998).
There are clear limitations of the 1D approach for understanding terrain
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
3
morphology, however, and these limitations prompted early efforts to extend
erosion models to two dimensions, though still with a focus primarily on
hillslope morphology (Ahnert, 1976; Armstrong, 1976; Kirkby, 1986).
Driven in part by technological advances, there has been a flowering of
landscape evolution models during the past decade. Many of these models
have focused on watershed evolution and dynamics (Willgoose et al., 1991a;
Howard, 1994; Moglen and Bras, 1995; Coulthard et al., 1997; Tucker and
Slingerland, 1997; Densmore et al., 1998). Although spatial scale is often not
specified, these modeling studies have generally focused on the formation of
hillslope-valley topography within small- to moderate-sized drainage basins
(on the order of several square kilometers or smaller). In parallel with these
developments in watershed geomorphology, a number of researchers have
attempted to model the evolution of terrain on the scale of a mountain range
or larger (e.g., Koons, 1989; Beaumont et al., 1992; Lifton and Chase, 1992;
Anderson, 1994; Tucker and Slingerland, 1994, 1996; Braun and Sambridge,
1997). In these applications, computational limitations dictate the use of a
coarse spatial discretization in which individual grid cells are much larger
than the scale of an individual hillslope, making it impossible to address
explicitly the role of hillslope dynamics, and raising the issue of “upscaling”
as a need in large-scale geomorphic models (Howard et al., 1994). A third
category of models includes cellular statistical-physical models that employ
simple rule sets to address the origin and nature of scaling properties
observed in river networks and terrain (e.g., Chase, 1992; Rigon et al., 1994;
Rodriguez-Iturbe and Rinaldo, 1997). Finally, a number of two-dimensional
models of hillslope-scale soil erosion and rill development have been
developed to study and predict patterns of slope erosion and drainage pattern
initiation (e.g., Smith and Merchant, 1995; Favis-Mortlock, 1998; Mitas and
Mitasova, 1998).
Despite significant progress in theory and model development over the
past decade, the current generation of physically based models suffers from
several limitations: (1) temporal variability in rainfall and runoff has been
largely ignored (cf. Tucker and Bras, 2000); (2) with a few exceptions (e.g.,
Ijjasz-Vasquez et al., 1992; Tucker and Bras, 1998), runoff is usually treated
as spatially uniform (Hortonian) across the landscape, despite the well
known importance of variable source-area runoff generation in humid
regions; (3) lateral erosion by channels has been ignored in the context of
drainage basin evolution; (4) most models use a fixed and uniform spatial
discretization in which only vertical movements of the terrain surface are
allowed (for an exception, see Braun and Sambridge, 1997); (5) the role of
heterogeneous sediment and sorting dynamics is usually ignored for
simplicity, despite their potential impacts on stream profile shape (e.g.,
4 Chapter 12
Snow and Slingerland, 1987; Sinha and Parker, 1996; Robinson and
Slingerland, 1998) and drainage basin structure (Gasparini et al., 1999); and
(6) few efforts have been made to examine the coupling between erosional
and depositional systems (e.g., Johnson and Beaumont, 1995; Tucker and
Slingerland, 1996; Densmore et al., 1998).
3. MODEL FORMULATION
3.1 Overview
The CHILD model simulates the evolution of a topographic surface and
its subjacent stratigraphy under a set of driving erosion and sedimentation
processes and with a prescribed set of initial and boundary conditions (Fig.
1). Designed to serve as a computational framework for investigating a wide
range of problems in catchment geomorphology, CHILD is both a model, in
the sense that it comprises a set of hypotheses about how nature works, and a
software tool, in the sense that it provides a simulation environment for
exploring the consequences of different hypotheses, parameters, and
boundary conditions. Here we will use the term “model” to refer collectively
to the software and the assumptions and hypotheses embedded within it.
The process modules in CHILD are summarized graphically in Figure 2.
Processes incorporated in the model include: (1) climate forcing via a
sequence of discrete storm events with durations, intensities, and inter-
arrival times that may be either random or constant; (2) generation of runoff
by infiltration-excess or saturation-excess mechanisms; (3) downslope
routing of water and sediment using a steepest-descent method; (4)
detachment (erosion) of sediment or bedrock by channelized surface runoff
in rills or stream channels; (5) water-borne downslope transport of detached
sediment; (6) transport of sediment by soil creep and related processes on
hillslopes; (7) meandering of large stream channels; (8) overbank
sedimentation on floodplains; and (9) tectonic deformation. Note that not all
of these processes need to be, or even should be, considered in any particular
application. The point of including a number of different processes is to
allow one to investigate different types of geomorphic system under
different space and time scales, using a common modeling framework that
handles the basic spatial and temporal simulation framework.
In addition to these process modules, CHILD includes capabilities for
managing the spatial simulation framework. The use of an adaptive, irregular
spatial discretization adds several useful capabilities (Braun and Sambridge,
1997; Tucker et al., 2000), including the ability to vary spatial resolution and
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
5
to incorporate the horizontal components of erosion processes (e.g., stream
channel migration) and tectonic motions (e.g., strike-slip displacement). In
addition, the model can simulate depositional history and stratigraphy by
tracking and updating “layers” of deposited material underlying each point
in the landscape, thereby making it possible to model coupled erosion-
deposition systems such as mountain drainage basins and their associated
alluvial fans (e.g., Ellis et al., 1999).
3.2 Continuity of Mass and Topographic Change
Changes in ground surface height, z(x,y), are described by the continuity
of mass equation for a terrain surface, which is expressed in terms of the
divergence of the sediment flux qs (dimensions of bulk volume rate per unit
width):
),,( tyxU
t
z+−∇=
∂
∂s
q
where z is surface height, t is time, and U is a source term that represents
baselevel change or tectonic uplift. The first term on the right-hand side
embodies several different sediment transport and erosion terms and can take
on a number of different forms depending on the assumptions made about
process mechanics. The formulations of the transport and erosion terms and
the numerical solution to (1) are described below.
3.3 Spatial Framework
In order to avoid the limitations associated with grid-based models, the
terrain surface may be discretized as a set of points (nodes) in any arbitrary
configuration. These nodes are connected to form a triangulated irregular
mesh (Figs. 1, 3) (Braun and Sambridge, 1997; Tucker et al., 2000). The
mesh is constructed using the Delaunay triangulation, which is the
(generally) unique set of triangles having the property that a circle passing
through the three nodes of any triangle will contain no other nodes (e.g., Du,
1996). The use of an irregular spatial framework offers several significant
advantages: (1) the model resolution can vary in space in order to represent
certain landscape features, such as floodplains or regions of complex terrain,
at a locally high level of detail (e.g., Fig. 1); (2) adaptive remeshing can be
used to adjust spatial resolution dynamically in response to changes in the
nature or rates of processes occurring at a particular location (e.g., Braun and
Sambridge, 1997; Tucker et al., 2001; and examples below); (3) nodes can
(1)
6 Chapter 12
be moved horizontally as well as vertically, making it possible to simulate
lateral and surface-normal, as opposed to purely vertical, erosion (as, for
example, in the cases of meandering channels and cliff retreat); (4) nodes
can be added to simulate lateral accretion of, for example, point bars in
meandering streams or accretionary wedges at active margins; and (5) the
terrain can be coupled with 3D kinematic or dynamic models of tectonic
deformation in order to simulate interactions between crustal deformation
(e.g., shortening, fold growth) and topographic change. The data structures
used to implement the triangular mesh are described by Tucker et al. (2001).
The Delaunay framework lends itself to a numerical solution of the
continuity equation (Eq 1) using finite-volume methods. Each node (vertex)
in the triangulation, Ni, is associated with a Voronoi (or Thiessen) polygon
of surface area Λi (Fig. 3), in which the polygon edges are perpendicular
bisectors of the edges connecting the node to its neighbors (e.g., Du, 1996;
Guibas and Stolfi, 1986). Thus, the Delaunay triangulation defines the
connectivity between adjacent nodes, while the associated Voronoi diagram
defines the surface area associated with each node as well as the width of the
interface between each pair of adjacent nodes (Fig. 3B). In CHILD, each
Voronoi polygon is treated as a finite-volume cell. Continuity of mass for
each node is written using an ordinary differential equation:
∑
=
Λ
=i
M
j
Sji
i
iQ
dt
dz
1
1,
where zi is the average surface height of node i, Mi is the number of neighbor
nodes connected to node i, and QSji is the total bulk volumetric sediment flux
from node j to node i (negative if the net flux is from i to j). Note that by this
method it is only possible to describe the average rate of erosion or
deposition within a given Voronoi polygon. As described below, the method
used to integrate the flux terms depends on whether the flux is two-
dimensional (e.g., for diffusive sediment transport or kinematic-wave
overland flow routing) or one-dimensional (for streamwise water and
sediment routing). For discussion of the implementation, application, and
advantages of irregular discretization in landscape models, the reader is
referred to Braun and Sambridge (1997) and Tucker et al. (2001).
3.4 Temporal Framework
One of the challenges in modeling terrain evolution lies in addressing the
great disparity between the time scales of topographic change (e.g., years to
geologic epochs) and the time scales of storms and floods (e.g., minutes to
days). Most previous models of drainage basin evolution have dealt with this
(2)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
7
disparity by simply assuming a constant average climatic input (e.g., a
steady rainfall rate or a “geomorphically effective” runoff coefficient). This
approach, while computationally efficient, has three drawbacks: (1) it
ignores the influence of intrinsic climate variability on rates of erosion and
sedimentation (e.g., Tucker and Bras, 2000); (2) it fails to account for the
stochastic dynamics that arise when a spectrum of events of varying
magnitude and frequency acts in the presence of geomorphic or hydrologic
thresholds; and (3) the approach typically relies on a poorly calibrated “cli-
mate coefficient” that cannot be directly related to measured climate data.
In order to surmount these limitations, and to address the role of event
magnitude and frequency in drainage basin evolution, CHILD uses a
stochastic method to represent rainfall variability. The method is described
in detail by Tucker and Bras (2000), and is only briefly outlined here. In
solving the continuity equation, the model iterates through a series of
alternating storms and interstorm periods, based on the Poisson rainfall
model developed by Eagleson (1978). Each storm event is associated with a
constant rainfall intensity, P, a duration, Tr, and an inter-arrival “waiting
time”, Tb (Fig. 4). For each storm, these three attributes are chosen at
random from exponential probability distributions, the parameters for which
can be readily derived from hourly rainfall data (Eagleson, 1978; Hawk,
1992). Alternatively, storm intensity, duration, and frequency may be kept
constant, in which case the approach reduces to the “effective rainfall
intensity” approximation (Tucker and Slingerland, 1997). In either case,
storms are approximated as having constant intensity throughout their
duration, and the same assumption is also applied to the resulting
hydrographs. Runoff-driven transport and erosion processes (described
below) are computed only during storm events. Other processes, including
diffusive creep transport and tectonic deformation, are assumed to occur
continuously, and are updated at the end of each interstorm period (Fig. 5).
Note that the model imposes no special restrictions on time scale, aside
from the fact that it is designed for periods longer than the duration of a
single storm. For simulations involving terrain evolution over thousands to
millions of years (e.g., Tucker and Slingerland, 1997), however, it becomes
computationally intractable to simulate individual storms. For many
applications this problem can be overcome by simply amplifying the storm
and interstorm durations. As long as the ratio Tr/Tb remains the same, the
underlying frequency distributions are preserved. Perturbations in climate
can also be simulated by changing the parameters of the three frequency
distributions (Fig. 4).
8 Chapter 12
3.4.1 Stochastic Rainfall: Example
Figure 6 illustrates the behavior of the model under stochastic rainfall
forcing, in a case where a high erosion threshold (see below) lends the
system a high sensitivity to extreme events. The initial condition consists of
a 30 degree slope upon which are superimposed small random perturbations
in the elevation of each node. Erosion of the slope in response to a random
series of rainfall and runoff events (Fig. 6A) is highly episodic (Fig. 6B). In
this example, a gully forms early on in response to a series of large-
magnitude and relatively long-duration storms (Figs. 6C and E). The gully
develops in an area where the topography of the initial surface leads to local
flow convergence. The reduction in gradient along the gully effectively
stabilizes the system, so that later events have little or no impact. Subsequent
mass movement by slope-driven diffusive creep (see below) leads to gradual
healing of the scar (Figs. 6D and F).
3.5 Surface Hydrology and Runoff Generation
Surface runoff collected at each node on the mesh is routed downslope
toward one of its adjacent neighbor nodes, following the edge that has the
steepest downhill slope (Fig. 3). If a closed depression occurs on the mesh,
water can either be assumed to evaporate at that point, or alternatively a
lake-filling algorithm can be invoked to find an outlet for the closed
depression (Tucker et al., 2000).
The local contribution to runoff at a node is equal to the effective runoff
rate (defined below) multiplied by the node’s Voronoi area, Λ. The drainage
area, A, at a node is the sum of the area of all Voronoi cells that contribute
flow to that node. Total surface discharge can be computed from drainage
area using one of three methods. The first two assume that runoff generation
is spatially uniform, while the third represents variable-source area runoff
generation.
3.5.1 Hortonian (infiltration-excess) runoff
Runoff production (rainfall rate minus infiltration rate) is assumed to be
uniform across the landscape. Assuming steady-state flow, the surface
discharge at any point is equal to
cc IPAIPQ >−= ,)( ,
where Ic is infiltration capacity (L/T).
(4)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
9
3.5.2 Excess storage capacity runoff
Under this approach, the soil, canopy, and surface are collectively
assumed to have a finite and spatially uniform capacity to absorb rainfall.
Any rainfall exceeding this storage depth will contribute to runoff according
to
srr
r
srr DPT,
T
DPT
R>
−
=
where R is local runoff rate (L/T), Dsr is the soil-canopy-surface retention
depth (L), and the resulting discharge at any point is Q = RA. Note that
equation (4) describes a runoff rate that is constant throughout a storm and
equal to the total volume of excess rainfall divided by the storm duration.
Note also that R = 0 if Dsr > TrP.
3.5.3 Saturation-excess runoff
With this option, a modified form of O’Loughlin’s (1986)
topographically based method is used to partition rainfall between overland
and shallow subsurface flow. The capacity for shallow subsurface flow per
unit contour length (qsub) is assumed to depend on local slope (S) and soil
transmissivity (T, dimensions of L2/T),
TS
w
Q
q== sub
sub
where contour length is represented by the width of adjoining Voronoi cell
edges, w. The surface flow component is equal to the total discharge minus
the amount that travels in the subsurface,
TSwPATSwPAQ >−= , .
Here, Q represents surface discharge resulting from a combination of
saturation-excess overland flow and return flow. Note that this method
assumes hydrologic steady state for both surface and subsurface flows, and
thus is most applicable to prolonged storm events and/or highly permeable
shallow soils.
(5)
(5)
(6)
10 Chapter 12
3.5.4 Example
The mechanism of runoff production can impact both terrain morphology
and dynamic responses to changing climate, land-use, or tectonism. For
example, theoretical studies have shown that the mode of runoff production
can have a significant impact on terrain morphology, drainage density, and
the scaling of drainage density with relief and climate (Kirkby, 1987; Ijjasz-
Vasquez et al., 1992; Tucker and Bras, 1998). Figure 7 compares two
simulated drainage basins formed under Hortonian and saturation-excess
runoff production, respectively. All other parameters in the two simulations
are identical. In the saturation case, runoff is rarely generated on hillslopes.
As a result, hillslopes are steep and highly convex (reflecting the dominance
of diffusive creep-type sediment transport; see below). The difference is
reflected in slope-area plots for the two simulated basins. In the case of the
saturation-dominated basin, the hillslope-channel break is well described by
the line of saturation for the mean-intensity storm (Fig. 7D).
3.6 Hillslope Mass Transport
Sediment transport by “continuous” hillslope processes such as soil creep
is modeled using the well-known hillslope diffusion equation (e.g., Culling,
1960; McKean et al., 1993),
zkzk
t
z
dd
2
creep
)( ∇=∇−−∇=
∂
∂,
where kd is a transport coefficient with dimensions of L2/T. Numerical
solution to equation (7) is obtained using a finite-volume approach (Tucker
et al., 2000). The net mass flux for a node is taken as the sum of the mass
fluxes through each face of its Voronoi polygon (Eq (2)). For each pair of
adjacent nodes, the gradient across their shared Voronoi polygon edge is
approximated as the gradient between the nodes themselves. The total flux
between each pair of nodes is thus equal to the topographic gradient between
them multiplied by the width of their shared Voronoi edge, so that Eq (7) is
approximated numerically as
∑
=
Λ
−=
∂
∂i
M
j
ijij
i
di wS
k
t
z
1
creep
,
(7)
(8)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
11
where Sij = (zi - zj)/λij is the downslope gradient from node i to node j, λij is
the distance between i and j (i.e., the length of the triangle edge connecting
them), and wij is the width of the shared Voronoi polygon face (Fig. 3B).
For steeper gradients, a nonlinear form of Eq (7) is arguably more
appropriate to describe the effects of accelerated creep and intermittent
landsliding (e.g., Anderson and Humphrey, 1990; Roering et al., 1999).
Although this type of nonlinear rate law is not presently coded in CHILD, its
incorporation would be straightforward.
Note also that equation (7) is intended to model creep-type processes
rather than wash erosion. Instead, wash and channel erosion are effectively
lumped together under the same formulation, as described below. This
approach has the obvious disadvantage that wash is effectively treated as a
form of rill erosion in which rills have the same hydraulic geometry (i.e.,
width, depth, and roughness properties) as larger channels, with all the
attendant limitations this implies. On the other hand, lumping rill and
channel erosion in a single “runoff erosion” category has the advantage of
simplicity: no extra parameters are needed to differentiate between hillslopes
and channels (as is the case, for example, in the model of Willgoose et al.,
1991a), which emerge solely as a result of process competition (Kirby, 1994;
Tucker and Bras, 1998). Thus, while we acknowledge a need for a more
rigorous sub-model for wash erosion in the future, the treatment of wash as a
general form of channel erosion seems justified given the aims of the model
and the present uncertainty regarding the dynamics of channel initiation.
3.7 Water Erosion and Sediment Transport
At each node, the local rate of water erosion is equal to the lesser of (1)
the detachment capacity, or (2) the excess sediment transport capacity. Both
of these are represented as power functions of slope and discharge, and they
are assumed to be mutually independent. Deposition occurs where sediment
flux exceeds transport capacity (for example, due to a downstream reduction
in gradient). The maximum detachment capacity depends on local slope and
discharge according to
b
b
b
p
cb
n
m
tbc S
W
Q
kkD
−
=
θ
,
where Dc is the maximum detachment (erosion) capacity (L/T), W is channel
width, θcb is a threshold for particle detachment (e.g., critical shear stress),
and kb, kt, mb, nb, and pb are parameters. Note that with suitably chosen
(9)
12 Chapter 12
parameters, equation (9) can represent either excess shear stress (i.e., τ = bed
shear stress, τcb = critical shear stress for detachment) or excess stream
power (Whipple and Tucker, 1999). The shear stress formulation is similar
to that used in the drainage basin evolution models of Howard (1994) and
Tucker and Slingerland (1997), as well as a number of soil erosion models
(e.g., Foster and Meyer, 1972; Mitas and Mitasova, 1998). If the Manning
equation is used to model roughness, mb=0.6, nb=0.7, and kt=ρgn3/5, where ρ
is water density, g is gravitational acceleration, and n is Manning’s
roughness coefficient. If the Chezy equation is used, mb=2/3, nb=2/3, and
kt=ρgC-1/2, where C is the Chezy roughness coefficient (for derivations, see
Tucker and Slingerland, 1997; Whipple and Tucker, 1999).
Channel width is computed empirically, using the well-known scaling
relationships between channel width and discharge (Leopold and Maddock,
1958; Leopold et al., 1964):
b
s
bwbbb QkWQQWW
ω
ω
== ,)/(
where Wb is bankfull channel width, Qb is a characteristic discharge (such as
bankfull or mean annual), kw is bankfull width per unit scaled discharge, and
ωb and ωs are the downstream and at-a-station scaling exponents,
respectively. Although these laws were developed for alluvial streams, they
appear to be applicable to other fluvial systems (e.g., Ibbitt, 1997) such as
steep mountain channels (Snyder et al., 2000).
The transport capacity for detached sediment material of a single grain
size is based on a generalization of common bedload and total-load sediment
transport formulas, which are typically expressed as a function of excess
shear stress or stream power (e.g., Yang, 1996). For steady, uniform flow in
a wide channel,
f
f
f
p
c
n
m
tfs S
W
Q
kWkC
−
=
θ
,
where Cs is transport capacity (L3/T) and kf, kt, mf, nf, and pf are parameters.
As with equation (9), equation (11) can be expressed in terms of excess
shear stress or stream power using suitably chosen values for kt, mf, and nf.
For transport of multiple sediment size-fractions, an alternative approach
based on the method of Wilcock (1997, 1998) is used (this is described
below).
Three end-member cases can arise from equations (9) and (11):
detachment-limited behavior, transport-limited behavior, and mixed-channel
behavior:
(10)
(11)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
13
1. Detachment-limited: If the sediment transport capacity is everywhere
much larger than the rate of sediment supply, the rate of water erosion is
simply equal to the maximum detachment rate,
c
bD
t
z−=
∂
∂,
where zb represents elevation of the channel bed above a datum within the
underlying rock column. This type of formulation has been used in a number
of studies to represent bedrock channel erosion (or more generally, detach-
ment-limited erosion of cohesive, cemented, or non-granular materials) (e.g.,
Howard and Kerby, 1983; Seidl and Dietrich, 1992; Anderson, 1994;
Howard et al., 1994; Moglen and Bras, 1995; Sinclair and Ball, 1996; Stock
and Montgomery, 1999; Whipple and Tucker, 1999; Snyder et al., 2000). An
important assumption is that the sediment flux has no direct control on the
rate of incision, as long as there is sufficient capacity to transport the eroded
material (cf. Sklar and Dietrich, 1998). Note that this case has the practical
advantage of being efficient to solve numerically. Though widely used,
however, the accuracy of this approximation for long-term stream profile
development remains to be evaluated.
2. Transport-limited: If sufficient sediment is always available for
transport and/or the bed material is easily detached (i.e., high kb), streams are
assumed to be everywhere at their carrying capacity. Under this condition,
continuity of mass gives the local rate of erosion or deposition as
x∂
∂
−
−=
∂
∂WC
t
zsb /
)1( 1
ν
,
where ν is bed sediment porosity (usually absorbed into the transport coeffi-
cient kf) and x is a vector oriented in the direction of flow. Transport-limited
behavior has been assumed in a number of models (e.g., Snow and Slinger-
land, 1987; Willgoose et al., 1991a; Tucker and Bras, 1998; Gasparini et al.,
1999), though its applicability to bedrock streams has been questioned (e.g.,
Howard et al., 1994).
3. Mixed-channel systems: The detachment and transport formulas imply
a third category of behavior that arises under conditions of (1) active erosion
into resistant material (e.g., bedrock) and (2) high sediment supply. Under
these conditions, active detachment of bed material must occur (by
definition), but the sediment supply rates are sufficiently high that the local
rate of incision is limited by the excess transport capacity (e.g., Tucker and
Slingerland, 1996). Stream channels falling into this category might be
(12)
(13)
14 Chapter 12
expected to have (on average) a partial cover of alluvium over bedrock; we
thus refer to streams falling into this category as mixed-channel systems
(Howard, 1998). Under certain conditions, the transition point between one
type of behavior (e.g. detachment-limited) and another (e.g. mixed) can be
computed analytically. Mixed channel behavior is discussed in greater depth
by Whipple and Tucker (in review).
3.7.1 Example
In the special case of a constant rate of surface lowering, equations (12)
and (13) both imply a power-law relationship between channel gradient and
contributing area (Willgoose et al., 1991b; Howard, 1994; Whipple and
Tucker, 1999), which is consistent with river basin data (e.g., Hack, 1957;
Flint, 1974). Figure 8 shows an example of such scaling for two simulated
landscapes. The straight lines indicate the trend that would occur under
purely transport-limited conditions (solid line; e.g., Willgoose et al., 1991b)
and under purely detachment-limited conditions (dashed line; e.g., Howard,
1994). Theoretical considerations suggest that longitudinal profile concavity,
which is indicated by the slope of the lines on Figure 8, should generally be
lower in transport-limited alluvial channels (Howard, 1994). The intersection
of the two lines indicates the point at which the gradient required to transport
eroded sediment becomes equal to the gradient required to detach particles.
Upstream of this point, channel gradient is dictated by detachment capacity;
downstream, the channel falls into the “mixed” category in which active
incision occurs but the gradient is controlled by sediment supply. Under
constant runoff, the transition point is abrupt (Fig. 8A), but in the more
realistic case of variable flows, the transition is gradational and spread over
two or more orders of magnitude in drainage area (Fig. 8B). This result
implies that such detachment-to-transport transitions, even if they do exist,
would be very difficult to identify on the basis of morphology alone (this of
course excludes channel-type transitions that are forced by tectonic or other
controls).
3.8 Extension to Multiple Grain Sizes
Size-selective erosion, transport, and deposition are important as agents
that control the texture of alluvial deposits. Textural properties of ancient
strata are important not only for the information they reveal about the
geologic past (e.g., Paola et al., 1992; Robinson and Slingerland, 1998), but
also for their control on the movement and storage of water and hydrocarbon
resources (e.g., Koltermann and Gorelick, 1992). Perhaps surprisingly,
recent work has also shown that the dynamics of size-selective erosion and
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
15
transport can have a significant impact on drainage basin architecture and
evolution (Gasparini, 1998; Gasparini et al., 1999). Size-selective sediment
transport and armoring can also exert important controls on the erosional
history of artificial landforms such as mine tailing heaps, and are therefore
important for engineering applications (Willgoose and Riley, 1998).
To model size-selective erosion and deposition, CHILD uses a two-
fraction (sand and gravel) approach based on the bedload entrainment and
transport functions developed by Wilcock (1997, 1998). The rock or
sediment column underlying each node in the model contains a mixture of
sand and gravel sediment fractions. An active layer of depth Lact defines the
depth over which sediment near the surface is well mixed and accessible for
active erosion and deposition (Gasparini, 1998; Gasparini et al., 1999). The
transport capacities of the two size fractions are given by
54
51
1
1
.
cg
.
gw
sg g)s(
fC
q
τ
τ
−
ρ
τ
−
=
54
51
1
1
.
cs
.
sw
ss g)s(
fC
q
τ
τ
−
ρ
τ
−
=
where qsg and qss are the transport rates of gravel and sand, respectively
(kg/ms), Cw is a dimensionless constant equal to 11.2, fg and fs are the
fractions of gravel and sand in the bed, ρ is water density, s is the ratio of
sediment and water density, g is gravitational acceleration, τ is bed shear
stress, and τcg and τcs are the critical shear stresses needed to entrain gravel
and sand, respectively.
Wilcock (1997; 1998) analyzed the relative mobility of sand and gravel
fractions in gravel-sand mixtures, and found that the initiation of motion
threshold for both fractions approaches a constant (and minimum) value for
mixtures containing more than about 40% sand. The threshold of motion
criterion criteria used in CHILD’s gravel-sand transport module is based on
a piecewise linear fit to the data of Wilcock (1997) (Gasparini et al., 1999).
The active layer represents the depth over which active particle exchange
takes place. For modeling instantaneous transport, the active layer is
typically defined on the basis of grain diameter. For modeling average
transport rates over the duration of one or many floods, this definition is
inappropriate, because local scour and the movement of bars and bedforms
(14)
(15)
16 Chapter 12
allow the flow to access significantly more near-bed sediment than simply
the uppermost one or two grain diameters. Paola and Seal (1995) suggested
that bankfull channel depth might be an appropriate choice for active layer
thickness for calculating long-term average transport rates. However, in the
absence of data on what controls the “effective mixing depth” over a given
time period, we adopt here the simple approach of using an active layer of
constant thickness. Sensitivity experiments by Gasparini (1998), which show
little variation in equilibrium texture patterns with varying active layer
depth, provide some justification for this approach, though we acknowledge
a need for deeper understanding of this issue.
Detachment of cohesive or intact sediment is assumed to be size-
independent and governed by Eq (9). When the multiple grain-size option is
used, detached material is assumed to break down into a user-specified
proportion of gravel and sand, which is then subject to differential
entrainment and transport according to Eqs (14) and (15).
3.9 Deposition and Stratigraphy
There has been an increasing recognition of the importance of coupling
between erosional and depositional systems (e.g., Humphrey and Heller,
1995; Johnson and Beaumont, 1995; Tucker and Slingerland, 1996;
Densmore et al., 1998). An important goal behind developing CHILD has
been to create a system that can be used to investigate these interactions and
their role in shaping the terrestrial sedimentary record. For this reason,
CHILD includes a “layering” module that records depositional stratigraphy.
Each node in the model is underlain by a column of material divided into
a series of layers of variable thickness and properties. Physical attributes
associated with each layer include the relative sand and gravel fractions (if
applicable), the median grain size of each sediment fraction, and the material
detachability coefficient, kb. These properties are assumed to be
homogeneous within a given layer. The time of most recent deposition is
also stored for each layer, so that chronostratigraphy can be simulated.
Finally, each layer also records the amount of time it has spent exposed at
the surface, which is useful for identifying periods of quiescence and may be
applicable to modeling exposure-age patterns in conjunction with
cosmogenic isotope studies.
The active layer depth is fixed in time and space. When material is
eroded from the surface, the active layer is replenished with material from
the layer below. The active layer texture and time of surface exposure are
then updated as a weighted average between the current properties of the
active layer and those of the layer below. Bedrock is never mixed into
sediment layers. When there are no sediment layers below the surface, the
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
17
active layer is depleted until no sediment remains and the channel is on
bedrock. During deposition, material from the active layer is moved into the
layer below before material is deposited into the active layer, so that the
active layer depth remains constant. The layers below the active layer have a
maximum depth; when this depth will be exceeded due to deposition, a new
layer is created.
3.9.1 Example
Fault-bounded mountain ranges and alluvial fans in regions of tectonic
extension are classic examples of close coupling between erosional and
depositional systems (e.g., Leeder and Jackson, 1993; Ellis et al., 1999).
Alluvial fan stratigraphy is shaped by a combination of forces, including
extrinsic factors such as tectonic uplift/subsidence and climate change, and
intrinsic factors related to the dynamics and geometry of sediment erosion,
transport, and deposition. Numerical modeling of these systems can be used
to evaluate the feasibility of conceptual models, to explore their sensitivity to
external controls, and to suggest new hypotheses regarding the stratigraphic
and geomorphic signatures of tectonic and climatic change.
Figure 9 shows a simple example of a simulated mountain range bounded
by an alluvial fan complex. In this example, we have chosen a simple
experimental design in which a block consisting of a cohesionless sand-
gravel mixture rises vertically at a constant rate relative to an adjacent
(fixed) basin surface and its associated baselevel. As one might expect, the
simulation shows a set of alluvial fans that prograde across the basin surface
(Fig. 9). A “wave” of sand-rich sediment progrades ahead of the advancing
fan toes (Figs. 9A, B). Interestingly, size-selective transport occurs not only
within the fan complex but also within the source terrain. Initially, finer
material is removed from the surface of the rising block, leaving behind a
coarsened layer of surface sediment that rims the headward-encroaching
drainages. Thus, the grain-size patterns within the fan complex are
influenced in part by sorting within the source terrain. Whether this effect
occurs in nature must depend on the regolith thickness; while bare rock
slopes offer little opportunity for grain-size fractionation, such fractionation
has been observed to occur on soil-mantled, wash-dominated slopes (e.g.,
Abrahams and Parsons, 1991).
Note that in this example no attempt is made to simulate either
downstream flow branching or sheetflow; rather, flow is effectively spread
across the fan surface through time as channels shift in response to
depositional patterns. A transverse section across the fan complex (Fig. 10)
reveals that the main fan bodies are, perhaps counter-intuitively, slightly
18 Chapter 12
finer than the inter-fan areas. This behavior would have implications for
fluid reservoir modeling, as it implies that in some cases inter-fan areas may
have locally higher hydraulic conductivity.
3.10 Lateral Stream Channel Migration (Meandering)
Owing to the large difference in scale between individual stream
channels and their drainage basins, channels are generally treated as one-
dimensional entities in landscape evolution theory. For many applications,
this choice is entirely appropriate; for others, however, it is problematic
because it neglects the role of floodplains as sediment buffers (e.g., Trimble,
1999). This limitation is particularly severe in analyses of watershed
responses to perturbations (e.g., Tucker and Slingerland, 1997). At the same
time, the morpho-stratigraphic development of floodplains is an important
problem in its own right (e.g., Mackey and Bridge, 1995; Moody et al.,
1999). These issues have motivated the development of a simple “rules
based” model of channel meandering, based on the principle of topographic
steering, which is capable of modeling channel planform evolution on time
scales relevant to valley, floodplain, and stream terrace development
(Lancaster, 1998; Lancaster and Bras, in press).
Lateral channel migration is implemented in CHILD by first identifying
main channel (meandering) nodes on the basis of a drainage area threshold.
Lateral migration of these nodes occurs perpendicular to the downstream
direction, and the rate is proportional to the bank shear stress:
n
ˆ
Eζ
ˆ
weff τ=
where ζ
ˆ
is the migration vector of the outer bank, τw is the bank shear stress
determined by the meandering model of Lancaster and Bras (in press; see
also Lancaster, 1998); n
ˆ
is the unit vector perpendicular to the downstream
direction; and Eeff is the effective bank erodibility, defined by:
0
0
1
0
0
≤=
>
+
−
=
B
B
B
BH
eff
h,E
h,
hH
hP
EE ,
where E0 is the nominal bank erodibility; H is water depth; hB is bank height
above the water surface; and PH is the degree to which the effective bank
erodibility is dependent on bank height, where 10 ≤≤ H
P (Lancaster, 1998).
This bank height dependence directly couples topography and migration
(16)
(17)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
19
rate. Each channel node in the model actually has both right and left bank
erodibilities, and these values are determined from a weighted average of Eeff
values calculated for neighboring nodes falling on either side of the line
perpendicular to the downstream direction (Fig. 11). We write the effective
erodibility at node i of the bank on the e
ˆ
-side as
()
21
1221
dd
dEdE
Ei,effi,eff
e
ˆ
i,eff +
+
=
where e
ˆ
is the unit vector in the direction of either the left ( n
ˆ
) or right
(n
ˆ
−) bank; Eeff,i1 and Eeff,i2 are the effective erodibilities of the bank nodes
with respect to node i; and d1 and d2 are the distances of the bank nodes from
the line parallel to the unit vector, e
ˆ
(Fig. 11).
We use the meandering model of Lancaster (1998) to find τw in (16) as a
function of channel curvature upstream. Movement of a channel node
indicates that the channel centerline has moved, i.e., that one bank has been
eroded while deposition has occurred at the other. As the channel migrates,
existing nodes are deleted from the moving channel’s path, and new nodes
are added in the moving channel’s wake. Node movement and addition
require re-determination of node stratigraphy.
A flow chart in Figure 12 illustrates the implementation of meandering
within the CHILD model. The discretization of meandering channel reaches
is dependent on channel width and is, in general, different from the
discretization of the surrounding landscape. This procedure is described in
more detail in Lancaster (1998).
3.10.1 Examples
An example simulation incorporating the stream meander model is
shown in Figure 13. Here the model is configured to represent an idealized
segment of floodplain, with a large stream (point source of discharge)
entering at the top of the mesh and exiting at the bottom. The hydrology and
initial topography are patterned after Wildcat Creek, a 190 km2 drainage
basin in north-central Kansas. In this example, the mainstream elevation is
forced with a series of cut-fill cycles (representing millennial-scale climate
impacts), while the stream planform is free to migrate laterally. Each point
along the main channel is moveable. Dynamic remeshing is used to ensure
that the mainstream is adequately resolved. Whenever a moving channel
point comes very close to a fixed “bank” point, the latter is removed from
the mesh. To ensure an adequate level of spatial resolution within the
floodplain, a new point is added in the “wake” of a moving channel point
(18)
20 Chapter 12
whenever the moving point has migrated a given distance away from a
previously stored earlier location (which is then updated). The net result is
that the floodplain is modeled at a locally high resolution relative to the
surrounding uplands (Fig. 13).
A similar approach can be used to investigate the development of incised
meanders in bedrock such as those of the Colorado Plateau (e.g., Gardner,
1975) and the Ozark Mountains. Lancaster (1998) modeled the development
of terrain under active uplift, incision, and stream meandering, and found
that coupling between bank height and the rate of cut-bank erosion exerts an
important influence on the resultant topography and channel planforms.
3.11 Floodplains: Overbank Sedimentation
Valley-fill sediments often contain an important record of paleoclimate,
paleo-geomorphology, and prehistory (e.g., Johnson and Logan, 1990). Most
studies of the formation and dynamics of river basins have treated streams as
essentially one-dimensional conduits of mass and energy. Yet valley-fill
sediments are inherently three-dimensional features, and to model their
stratigraphy properly requires an alternative approach. The one-dimensional
approach cannot, for example, resolve important aspects of alluvial
stratigraphy such as the distribution of channel and overbank deposits (e.g.,
Mackey and Bridge, 1995). Motivated by this limitation, CHILD includes
the capability to model overbank sedimentation using a modified form of
Howard’s (1992) floodplain diffusion model. Under this approach, the rate
of overbank sedimentation during a flood varies as a function of distance
from a primary channel and local floodplain topography. Average rates of
floodplain sedimentation are known to decay with distance from the source
channel due to diffusion of turbulent energy. The local rate of sedimentation
is also presumed to depend on the height of the floodplain relative to water
surface height. During a given storm event, the rate of overbank
sedimentation at a given point is
()( )
λ−µ−η= /dzDOB exp
where DOB is the vertical deposition rate (dimensions of L/T), z is local
elevation, d is the distance between the point in question and the nearest
point on the main channel, η is the water surface height at the nearest point
on the main channel, µ is a deposition rate constant (T-1), and λ is a distance-
decay constant. “Main channel” is defined on the basis of a drainage area
threshold; typically, the model would be configured with a large channel fed
in as a boundary condition for this type of application, so that there would be
no ambiguity about what constitutes a primary channel (e.g., Fig. 13). Water
(19)
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
21
surface height is computed as the sum of bed elevation, z, and water depth,
H, using a simple empirical hydraulic geometry approach for H:
b
s
bwbbb QkH,)Q/Q(HH δ
δ==
where Hb is bankfull channel depth, Qb is a characteristic discharge (such as
bankfull or mean annual), kh is bankfull depth per unit scaled discharge, and
δb and δs are the downstream and at-a-station scaling exponents, respectively
(Leopold et al., 1964). Equation (19) is only applied for events in which H >
Hb.
3.11.1 Example
Combining channel meandering and overbank deposition makes it
possible to simulate the development of three-dimensional alluvial
stratigraphic architecture, which has been the goal of a number of different
models (e.g., Howard, 1992; Mackey and Bridge, 1995; Teles et al., 1998).
The fill terraces depicted in Figure 13 are formed during times of rising
baselevel along the main channel. Lateral channel migration etches out the
fills during intervals of cutting. Inset terraces are formed during subsequent
fill episodes (Fig. 13). Among other things, this type of stratigraphic
simulation can provide a basis for developing and testing improved
geostatistical methods for modeling 3D subsurface architecture (e.g., de
Marsily et al., 1998).
4. DISCUSSION: APPLICATIONS AND
LIMITATIONS
All models involve a tradeoff between simplicity and realism. What
makes the CHILD model unique is its ability to examine interactions among
a wide range of processes, in scenarios that range from simple to complex.
The examples herein use simple, idealized scenarios to illustrate these
processes. The model’s design reflects the fundamental recognition that the
characteristics of one part of a river basin are determined in large part by the
characteristics of the basin upstream and, to a lesser degree, downstream.
The inclusion of many process modules and alternative parameterizations
in one model (Fig. 2) is intended to enhance the researcher’s ability to
address very simple and well-posed questions by carefully selecting a subset
of process equations and configuring these with appropriate initial and
boundary conditions. By comparing model behavior under varying levels of
(20)
22 Chapter 12
complexity and/or different process models, the validity and robustness of
different simplifying assumptions can be tested. One can examine, for
example, the consequences of relaxing the common assumption of
homogeneous sediment size (Gasparini et al., 1999), or assess the
appropriateness of using a “characteristic storm” parameter as a surrogate for
time-varying rainfall and flood discharge (Tucker and Bras, 2000).
CHILD has been developed as a framework for modeling changes in
drainage basin terrain over a range of space and time scales. Although there
are no explicit limits to spatial scale, the assumption of hydrologic steady
state during storm events is most valid for relatively small watersheds (less
than perhaps 100km2), in which the time of concentration is shorter than the
duration of a typical storm. Similarly, the assumption of spatially uniform
precipitation rate, infiltration capacity and soil transmissivity is only
appropriate for small watersheds (although one might also wish to make
similar assumptions in simple “what if” studies of large-scale landscape
evolution). At the lower end of spatial scales, the approximation of steady,
uniform unidirectional flow loses validity for the length scales on which
momentum and backwater effects become important (on the order of
decimeters to meters). The assumption of steady rainfall and runoff during
storms also implies that the model is most applicable to time periods much
longer than the duration of a single storm. The upper limit to time scale is
dictated only by performance considerations, and in fact for certain
applications it is feasible to magnify storm and interstorm durations to
enhance computational speed.
Distributed models necessarily involve a tradeoff between speed and
resolution. The CHILD model’s TIN-based framework offers an advantage
in this regard, because it makes it possible to vary spatial resolution as a
function of dominant process or landscape position (Figs. 1 and 13). On the
other hand, the use of variable spatial resolution complicates the inclusion of
“scale-dependent physics” (i.e., equations whose rate constants depend on
spatial scale). This may be a blessing in disguise, for although it makes the
problem of calibration in engineering applications more difficult it also
provides a disincentive to scale-dependent “tuning” of parameters. Use of a
variable-resolution numerical mesh, if handled properly, may also help
resolve certain scaling issues that arise as a result of averaging terrain
properties over an arbitrary and fixed discretization scale. For example, with
an irregular discretization method it becomes possible (at least in principle)
to construct a discretized terrain surface that uses the minimum necessary
number of computational points to accurately represent hillslope gradient at
all points in a region of complex terrain. The TIN framework also opens the
door to bridging the two fundamental and disparate scales in watershed
hydrology, that of the channel and that of the basin as a whole.
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
23
Although it is intended to serve a wide range of applications, the CHILD
model’s roots lie in large-scale drainage basin morphology and evolution.
The form of many of the equations used in CHILD reflects this emphasis.
Thus, the sediment transport equations are based on formulas commonly
used to predict bedload transport rates, and the model at this stage includes
no explicit treatment of suspended or wash load (which are presumably of
lesser importance in controlling stream gradients). Similarly, the model at
present includes no expressions for landsliding or for eolian transport. The
emphasis on physical rather than chemical process renders CHILD
inapplicable in solution-dominated environments (e.g., karst terrain). It
should be emphasized, however, that CHILD is designed with extensibility
in mind, and the modular design of the software reflects this (Tucker et al.,
2001). Recent efforts to adapt CHILD for applications in forestry (Lancaster
et al., 1999) and flood hydrology (Rybarczyk et al., 2000) demonstrate the
utility of constructing modular and extensible numerical modeling systems.
There is no simple answer to the question of how to test and validate a
model such as CHILD because it is in essence not one model but many, each
with different assumptions, aims, and requirements. Ultimately, the basis for
validation or rejection of a model should depend on the nature of the
problem addressed. Nonetheless, it is worth noting that several methods for
evaluating the predictions of landscape evolution theory have been advanced
recently. Statistical approaches have been widely used to examine drainage
network properties (e.g., Rodriguez-Iturbe and Rinaldo, 1997), although
some network statistics suffer from a lack of discriminant ability (e.g.,
Kirchner, 1993). Experimental approaches have also been used (Hancock
and Willgoose, in review). Arguably the most promising tests of landscape
evolution theory come from settings in which paleo-topography (e.g., Stock
and Montgomery, 1998) and driving factors such as uplift rate (e.g., Merritts
and Vincent, 1989; Snyder et al., 2000) are independently known. Much can
be learned by testing the morphologic predictions of landscape evolution
models against observed terrain in regions where some type of equilibrium is
believed to exist (e.g., Willgoose, 1994). The CHILD model, as with other
models based on similar fluvial erosion formulations, can successfully
reproduce theoretically predicted slope-area scaling under conditions of
spatially uniform erosion rate (a form of equilibrium). This simply reflects
the fact that the exponent terms in the fluvial transport and erosion terms can
be chosen such that the model-predicted scaling under either equilibrium
(uplift-erosion balance) or transient decline agrees with observed values
(e.g., Willgoose et al., 1991a; Howard, 1994; Willgoose, 1994; Whipple and
Tucker, 1999; Tucker and Whipple, in review). However, one disadvantage
of testing models on the basis of equilibrium states, aside from the difficulty
24 Chapter 12
in establishing the existence of such states in the first place, is the potential
for equifinality (i.e., different processes may lead to the same outcome, as in
the case of slope-area scaling discussed by Tucker and Whipple, in review).
Landscapes characterized by a transient response to a known perturbation
contain useful information about process dynamics that is often lost in
equilibrium states (Tucker and Whipple, in review; Whipple and Tucker, in
review). Hence, one of the key research needs is to identify transient
landscapes in which knowledge of the nature and timing of the causative
external perturbation, whether of tectonic, climatic, geomorphic, or human
origin, can be obtained. For short-term phenomena such as gully
development, there is a need for detailed monitoring to establish time
sequences of landform development.
5. SUMMARY AND CONCLUSIONS
CHILD is a new computer model of drainage basin evolution that
integrates a wide variety of processes, many of which have not been
included in previous models of drainage basin evolution. The model is
designed to serve as a general-purpose framework for investigating a range
of issues in drainage basin geomorphology, with an emphasis on morpho-
logical development. Some components of the model, such as the treatment
of channel and hillslope erosion, use an approach similar to that of existing
models. The model also includes a number of new features and capabilities
that are designed to foster the development of theoretical geomorphology by
making it possible to investigate in greater detail the feedbacks between
hillslope/channel hydrology and landscape evolution, and to examine
coupling between erosional and depositional systems. The incorporation of
(1) meandering and (2) floodplain deposition, which have not before been
included in models of drainage basin evolution, makes it possible to
investigate the development of alluvial stratigraphy in the drainage basin
context. Other capabilities, which are unique in their combination, include
(3) stochastic storm variability, with an explicit link to climate data; (4) both
detachment- and transport-limited fluvial erosion, with transport of either
single- or dual-size sediment; (5) explicit tracking of subsurface stratigraphy,
including time of deposition, textural properties, and deposit exposure ages;
(6) variable, triangulated discretization and adaptive remeshing, which allow
detailed resolution of particular features and representation of horizontal
surface motion; and (7) infiltration-, storage- and saturation-excess runoff
mechanisms, the last of which provides a direct link between topography and
hydrology. Other capabilities, including a dynamic vegetation component
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
25
(Tucker et al., 1999) and kinematic thrust-fault propagation, are under
development and will be described elsewhere.
To implement these capabilities, CHILD includes several process
“modules.” In some cases, these represent alternative models for the same
process (e.g., Hortonian versus saturation-excess runoff generation).
Although the number of parameters in the model is potentially quite large,
the many different capabilities and process equations are in fact developed
with simplicity and flexibility in mind. CHILD’s extensible design facilitates
the process of comparing alternative process models and conducting
sensitivity experiments that address the basic (and important) questions of
“what matters and when.”
Although developed with an emphasis on research applications, CHILD’s
more detailed treatment of hydrology also makes it well suited to potential
applications in land management and erosion prediction. Most soil erosion
models, such as USLE and WEPP, assume a one-dimensional and
unchanging topography. These limitations, though appropriate for estimating
soil loss under rill and interrill erosion, are poorly suited to modeling gully
and channel incision, phenomena in which dynamic modification of
landforms and flow aggregation play a central role.
Acknowledgments
This research was supported by the U.S. Army Corps of Engineers
Construction Engineering Research Laboratories (USACERL), the U.S.
Army Research Office, and the Italian National Research Council. We are
grateful to Lainie Levick for a helpful review, and to William Doe and
Russell Harmon for convening the workshop and GSA Special Session that
led to this volume.
REFERENCES
Abrahams, A.D., and Parsons, A.J., 1991, Relation between sediment yield and gradient on
debris-covered hillslopes, Walnut Gulch, Arizona: Geological Society of America
Bulletin, v. 103, p. 1109-1113.
Ahnert, F., 1970, A comparison of theoretical slope models with slopes in the field:
Zeitschrift fur Geomorphologie, v. 9, p. 88-101.
Ahnert, F., 1976, Brief description of a comprehensive three-dimensional process-response
model of landform development: Zeitschrift fur Geomorphologie, v. 25, 29-49.
Anderson, R.S., 1994, Evolution of the Santa Cruz Mountains, California, through tectonic
growth and geomorphic decay: Journal of Geophysical Research, v. 99, p. 20,161-20,179.
26 Chapter 12
Anderson, R.S., and Humphrey, N.F., 1990, Interaction of weathering and transport processes
in the evolution of arid landscapes, in Cross, T., ed., Quantitative dynamic stratigraphy:
Englewood Cliffs, N.J., Prentice-Hall p. 349-361.
Armstrong, A.C., 1976, A three-dimensional simulation of slope forms: in Quantitative Slope
Models, ed. by F. Ahnert, Zeitschrift fur Geomorphologie, Supplement Band 25, p. 20-28.
Beaumont, C., Fullsack, P., and Hamilton, J., 1992, Erosional control of active compressional
orogens, in McClay, K.R., ed., Thrust tectonics: New York, Chapman and Hall, p. 1-18.
Braun, J., and Sambridge, M., 1997, Modelling landscape evolution on geological time scales:
a new method based on irregular spatial discretization: Basin Research, v. 9, p. 27-52.
Chase, C.G., 1992, Fluvial landsculpting and the fractal dimension of topography:
Geomorphology, v. 5, p. 39-57.
Coulthard, T.J., Kirkby, M.J., and Macklin, Mark G., 1997, Modelling the impacts of
Holocene environmental change in an upland river catchment, using a cellular automaton
approach, in Brown, A.G., and Quine, T.A., eds., Fluvial processes and environmental
change, British Geomorphological Research Group symposia series, p. 31-46.
Culling, W.E.H., 1960, Analytical theory of erosion: Journal of Geology, v. 68, p. 336-344.
Densmore, A.L., Ellis, M.A., and Anderson, R.S., 1998, Landsliding and the evolution of
normal-fault-bounded mountains: Journal of Geophysical Research, v. 103, p. 15,203-
15,219.
Du, C., 1996, An algorithm for automatic Delaunay triangulation of arbitrary planar domains:
Advances in Engineering Software, v. 27, p. 21-26.
Eagleson, P.S., 1978, Climate, soil, and vegetation: 2. the distribution of annual precipitation
derived from observed storm sequences: Water Resources Research, v. 14, p. 713-721.
Ellis, M.A., Densmore, A.L., and Anderson, R.S., 1999, Development of mountainous
topography in the Basin Ranges, USA: Basin Research, v. 11, p. 21-41.
Favis-Mortlock, D., 1998, A self-organizing dynamic systems approach to the simulation of
rill initiation and development of hillslopes: Computers & Geosciences, 24 (4), p. 353-
372.
Flint, J.J., 1974, Stream gradient as a function of order, magnitude, and discharge: Water
Resources Research, v. 10, p. 969-973.
Foster, G.R., and Meyer, L.D., 1972, A closed-form erosion equation for upland areas, in
Shen, H.W., ed., Sedimentation: Symposium to Honor Professor H. A. Einstein, p. 12.1-
12.19, Colorado State University, Fort Collins, CO.
Gardner, T.W., 1975, The history of part of the Colorado River and its tributaries; an
experimental study, Field Symposium - Guidebook of the Four Corners Geological
Society (8), p. 87-95.
Gasparini, N.M., 1998, Erosion and Deposition of Multiple Grain Sizes in a Landscape
Evolution Model, unpublished M.Sc. thesis, Massachusetts Institute of Technology.
Gasparini, N.M., Tucker, G.E., and Bras, R.L., 1999, Downstream fining through selective
particle sorting in an equilibrium drainage network: Geology, v. 27, no. 12, p. 1079-1082.
Gilbert, G.K., 1877, Report on the geology of the Henry Mountains: U.S. Geographical and
Geological Survey of the Rocky Mountain Region, Washington, D.C., U.S. Government
Printing
Gossman, H., 1976, Slope modelling with changing boundary conditions---effects of climate
and lithology: Zeitschrift fur Geomorphologie, v. 25, p. 72-88.
Guibas, L., and Stolfi, J., 1985, Primitives for the manipulation of general subdivisions and
the computation of Voronoi diagrams: ACM Transactions on Graphics, v. 4, no. 2, p. 74-
123.
Hack, J.T., 1957, Studies of longitudinal stream profiles in Virginia and Maryland: U.S.
Geological Survey Professional Paper no. 294-B, 97 pp.
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
27
Hancock, G., and Willgoose, G., in review, The use of a landscape simulator in the validation
of the SIBERIA catchment evolution model: declining equilibrium landforms: Water
Resources Research, submitted.
Hawk, K.L., 1992, Climatology of station storm rainfall in the continental United States:
parameters of the Bartlett-Lewis and Poisson rectangular pulses models, unpublished M.S.
thesis, Department of Civil and Environmental Engineering, Massachusetts Institute of
Technology, 330pp.
Howard, A.D., 1992, Modeling channel and floodplain sedimentation in meandering streams,
in Lowland Floodplain Rivers: Geomorphological Perspectives, John Wiley & Sons,
Chichester, United Kingdom, p. 1-41.
Howard, A.D., 1994, A detachment-limited model of drainage basin evolution: Water
Resources Research, v. 30, p. 2261-2285.
Howard, A.D., 1998, Long profile development of bedrock channels; interaction of
weathering, mass wasting, bed erosion, and sediment transport, in Tinkler, Keith J., and
Wohl, Ellen E. (editors), Rivers over rock; fluvial processes in bedrock channels,
Geophysical Monograph, 107, p. 297-319
Howard, A.D., Dietrich, W.E., and Seidl, M.A., 1994, Modeling fluvial erosion on regional to
continental scales: Journal of Geophysical Research, v. 99, p. 13,971-13,986.
Howard, A.D., and Kerby, G., 1983, Channel changes in badlands: Geological Society of
America Bulletin, v. 94, p. 739-752.
Humphrey, N.F., and Heller, P.L., 1995, Natural oscillations in coupled geomorphic systems:
an alternative origin for cyclic sedimentation: Geology, v. 23, p. 499-502.
Ibbitt, R.P., Evaluation of optimal channel network and river basin heterogeneity concepts
using measured flow and channel properties, J. Hydrol., 196, 119-138, 1997.
Ijjasz-Vasquez, E.J., Bras, R.L., and Moglen, G.E., 1992, Sensitivity of a basin evolution
model to the nature of runoff production and to initial conditions: Water Resources
Research, v. 28, p. 2733-2741.
Johnson, D.D., and Beaumont, C., 1995, Preliminary results from a planform kinematic model
of orogen evolution, surface processes and the development of clastic foreland basin
stratigraphy, in Dorobek, S.L., and Ross, G.M., eds., Stratigraphic Evolution of Foreland
Basins, SEPM Special Publication 52, p. 3-24.
Johnson, W.C., and Logan, 1990, Geoarchaeology of the Kansas River Basin, central Great
Plains, in Lasca, N.P., and Donohue, J., eds., Archaeological Geology of North America:
Geological Society of America, Decade of North American Geology Centennial Special
Volume 4, p. 267-299.
Kirchner, J.W. 1993, Statistical inevitability of Horton’s laws and the apparent randomness of
stream channel networks: Geology (Boulder), 21 (7), p. 591-594.
Kirkby, M.J., 1971, Hillslope process-response models based on the continuity equation: in
Slopes: form and process, Institute of British Geographers Special Publication 3, p. 15-30.
Kirkby, M.J., 1986, A two-dimensional simulation model for slope and stream evolution, in
Abrahams, A.D., ed., Hillslope processes: Winchester, Mass., Allen and Unwin, p. 203-
222.
Kirkby, M.J., 1987, Modelling some influences of soil erosion, landslides and valley gradient
on drainage density and hollow development: Catena Supplement, v. 10, p. 1-14.
Koltermann, C.E., Gorelick, S.M., 1992, Paleoclimatic signature in terrestrial flood deposits,
Science, 256 (5065), p. 1775-1782.
Koons, P.O., 1989, The topographic evolution of collisional mountain belts: a numerical look
at the Southern Alps, New Zealand, American Journal of Science, v. 289, p. 1041-1069.
28 Chapter 12
Lancaster, S.L., 1998, A nonlinear river meander model and its incorporation in a landscape
evolution model, Ph.D. thesis, Massachusetts Institute of Technology.
Lancaster, S.L., and Bras, R.L., 2000, A simple model of river meandering and its comparison
to natural channels: Hydrologic Processes, in review.
Lancaster, S.L., Hayes, S.K., and Grant, G.E., 1999, The interaction between trees and the
landscape through debris flows: Eos, Transactions AGU, v. 80 (46), p. F425.
Leeder, M.R., Jackson, J.A., 1993, The interaction between normal faulting and drainage in
active extensional basins, with examples from the western United States and central
Greece: Basin Research, 5 (2), p. 79-102
Leopold, L. and Maddock, T., 1953, The hydraulic geometry of stream channels and some
physiographic implications: Professional Paper 252, United States Geological Survey.
Leopold, L.B., Wolman, M.G., and Miller, J.P., 1964, Fluvial processes in geomorphology:
New York, W.H. Freeman, 522 pp.
Lifton, N.A., and Chase, C.G., 1992, Tectonic, climatic and lithologic influences on landscape
fractal dimension and hypsometry: implications for landscape evolution in the San Gabriel
Mountains, California: Geomorphology, v. 5, p. 77-114.
Luke, J.C., 1972, Mathematical models of landform evolution: Journal of Geophysical
Research, v. 77, p. 2460-2464.
Mackey, S.D., and Bridge, J.S., 1995, Three-dimensional model of alluvial stratigraphy;
theory and applications, Journal of Sedimentary Research, Section B: Stratigraphy and
Global Studies, 65 (1), p. 7-31.
Marsily, G. de, 1998, Some current methods to represent the heterogeneity of natural media in
hydrogeology: Hydrogeology Journal, v. 6, p. 115-130.
McKean, J.A., Dietrich, W.E., Finkel, R.C., Southon, J.R., and Caffee, M.W., 1993,
Quantification of soil production and downslope creep rates from cosmogenic 10Be
accumulations on a hillslope profile: Geology, v. 21, p. 343-346.
Merritts, D., and Vincent, K.R., 1989, Geomorphic response of coastal streams to low,
intermediate, and high rates of uplift, Mendocino junction region, northern California:
Geological Society of America Bulletin, v. 101, p. 1373-1388.
Mitas, L., and Mitasova, H., 1998, Distributed soil erosion simulation for effective erosion
prevention: Water Resources Research, v. 34, p. 505-516.
Moglen, G.E., and R.L. Bras, 1995, The Effect of Spatial Heterogeneities on Geomorphic
Expression in a Model of Basin Evolution, Water Resources Research v. 31, no. 10, p.
2613-23.
Moody, J.A., Pizzuto, J.E., and Meade, R.H., 1999, Ontogeny of a flood plain, Geological
Society of America Bulletin, 111 (2), p. 291-303.
O’Loughlin, E.M., 1986, Prediction of surface saturation zones in natural catchments: Water
Resources Research, v. 22, p. 794-804.
Penck, W., 1921, Morphological analysis of land forms: a contribution to physical geography:
translated by Czech, H., and Boswell, K.C., London, Macmillan, 1953, 429 pp.
Paola, C., Heller, P.L., and Angevine, C.L., 1992, The large-scale dynamics of grain-size
variation in alluvial basins, 1: theory: Basin Research, v. 4, p. 73-90.
Paola, C., and Seal, R., 1995, Grain size patchiness as a cause of selective deposition and
downstream fining: Water Resources Research, v. 31, p. 1395-1407.
Rigon, R., Rinaldo, A., and Rodriguez-Iturbe, I., 1994, On landscape self-organization:
Journal of Geophysical Research, v. 99, p. 11,971-11,993.
Robinson, Ruth A. J., Slingerland, Rudy L., 1998, Origin of fluvial grain-size trends in a
foreland basin; the Pocono Formation on the central Appalachian Basin, Journal of
Sedimentary Research, 68 (3), p. 473-486.
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
29
Rodriguez Iturbe, I., and Rinaldo, A., 1997, Fractal river basins; chance and self-organization,
547 pp.
Roering, J.J., Kirchner, J.W., and Dietrich, W.E., 1999, Evidence for nonlinear, diffusive
sediment transport on hillslopes and implications for landscape morphology: Water
Resources Research, v. 35, no. 3, p. 853-870.
Rybarczyk, S.M., Ivanov, V.Y., Bras, R.L., and Tucker, G.E., 2000, Representing complex
topography in distributed rainfall/runoff modeling: a TIN based approach, Paper presented
at the International Francqui Chair Workshop on The Future of Distributed Hydrological
Modelling, April 13th, Leuven, Belgium.
Scheidegger, A.E., 1961, Mathematical models of slope development: Geological Society of
America Bulletin, v. 72, p. 37-50.
Seidl, M.A., and Dietrich, W.E., 1992, The problem of channel erosion into bedrock: Catena
Supplement 23, p. 101-124.
Sinclair, K., and Ball, R.C., 1996, Mechanism for global optimization of river networks for
local erosion rules: Physical Review Letters, v. 76, p. 3360.
Sinha, S.K., and Parker, G., 1996, Causes of concavity in longitudinal profiles of rivers:
Water Resources Research, v. 32, p. 1417-1428.
Sklar, L., and Dietrich, W.E., 1998, River longitudinal profiles and bedrock incision models;
stream power and the influence of sediment supply, in Wohl, E., and Tinkler, K., Rivers
Over Rock: Fluvial Processes in Bedrock Channels, AGU, Geophysical Monograph 107,
p. 237-260.
Smith, Terence R., Merchant, George E., 1995, Conservation principles and the initiation of
channelized surface flows, in Costa, John E. (editor), Miller, Andrew J. (editor), Potter,
Kenneth W. (editor), Wilcock, Peter R. (editor), Natural and anthropogenic influences in
fluvial geomorphology; the Wolman Volume, Geophysical Monograph, 89, p. 1-25.
Snow, R.S., and Slingerland, R.L., 1987, Mathematical modeling of graded river profiles:
Journal of Geology, v. 95, p. 15-33.
Snyder, N.P., Whipple, K.X., Tucker, G.E., and Merritts, D.J., 2000, Landscape response to
tectonic forcing: DEM analysis of stream profiles in the Mendocino triple junction region,
northern California: Geological Society of America Bulletin, in press.
Stock, J., and Montgomery, D.R., 1999, Geologic constraints on bedrock river incision using
the stream power law: J. Geophys. Res., v. 104, p. 4983-4993.
Teles, V., de Marsily, G., Perrier, E., 1998, Sur une nouvelle approche de modelisation de la
mise en place des sediments dans une plaine alluviale pour en representer l’heterogeneite:
C. R. Acad. Sci. Paris, Sciences de la Terre et des Planetes, v. 327, p. 597-606.
Trimble, S.W., 1999, Decreased rates of alluvial sediment storage in the Coon Creek basin,
Science, 285 (5431), p. 1244-1246.
Tucker, G.E., and Bras, R.L., 1998, Hillslope processes, drainage density, and landscape
morphology: Water Resources Research, v. 34, p. 2751-2764.
Tucker, G.E., and Bras, R.L., 2000, A stochastic approach to modeling the role of rainfall
variability in drainage basin evolution: Water Resources Research, Water Resources
Research, v. 36(7), pp. 1953-1964..
Tucker, G.E., Gasparini, N.M, Bras, R.L., and Lancaster, S.L.,1999, A 3D Computer
Simulation Model of Drainage Basin and Floodplain Evolution: Theory and Applications,
Technical report prepared for U.S. Army Corps of Engineers Construction Engineering
Research Laboratory.
Tucker, G.E., Lancaster, S.T., Gasparini, N.M., Bras, R.L., and Rybarczyk, S.M., 2000, An
object-oriented framework for distributed hydrologic and geomorphic modeling using
triangulated irregular networks: Computers and Geosciences, in press.
30 Chapter 12
Tucker, G.E., and Slingerland, R.L., 1994, Erosional dynamics, flexural isostasy, and long-
lived escarpments: a numerical modeling study: Journal of Geophysical Research, v. 99, p.
12,229-12,243.
Tucker, G.E., and Slingerland, R.L., 1996, Predicting sediment flux from fold and thrust belts:
Basin Research, v. 8, p. 329-349.
Tucker, G.E., and Slingerland, R.L., 1997, Drainage basin responses to climate change: Water
Resources Research, v. 33, no. 8, p. 2031-2047.
Whipple, K.X., and Tucker, G.E., 1999, Dynamics of the stream power river incision model:
implications for height limits of mountain ranges, landscape response timescales and
research needs: J. Geophys. Res., v. 104, p. 17,661-17,674.
Wilcock, P., 1997, A method for predicting sediment transport in gravel-bed rivers: Technical
report for the U.S. Forest Service Rocky Mountain Forest and Range Experiment Station,
59 pp.
Wilcock, P., 1998, Two-fraction model of initial sediment motion in gravel-bed rivers:
Science, v. 280, p. 410-412.
Willgoose, G.R., 1994, A statistic for testing the elevation characteristics of landscape
simulation models: Journal of Geophysical Research, v. 99, p. 13,987-13,996.
Willgoose, G.R., Bras, R.L., and Rodriguez-Iturbe, I., 1991a, A physically based coupled
network growth and hillslope evolution model, 1, theory: Water Resources Research, v.
27, p. 1671-1684.
Willgoose, G.R., Bras, R.L., and Rodriguez-Iturbe, I., 1991b, A physical explanation of an
observed link area-slope relationship: Water Resources Research, v. 27, p. 1697-1702.
Willgoose, G.R., and Riley, S., 1998, The long-term stability of engineered landforms of the
Ranger Uranium Mine, Northern Territory, Australia: Application of a catchment
evolution model: Earth Surface Processes and Landforms, v. 23, p. 237-259.
Yang, C.T., 1996, Sediment transport: theory and practice: New York, McGraw Hill, 396pp.
FIGURE CAPTIONS
Figure 1. Example terrain simulations produced by CHILD. Thin solid lines are contours
and heavy lines indicate drainage pathways. (A) simulation of gullying on an actual watershed
(unnamed watershed on Fort Carson in the foothills of the Colorado Front Range near
Colorado Springs). Inset shows a segment of the triangular mesh. (B) hypothetical fault-
bounded mountain range. (C) valley and floodplain simulation, illustrating meandering stream
with variable-resolution mesh. (D) rising mountain block with alluvial fans. Scales in (B),
(D), and (D) are nominally 10 km, 1 km, and 2.5 km, respectively.
Figure 2. Components of the CHILD model.
Figure 3. Elements of the irregular computational mesh, showing nodes (solid circles),
triangle edges (black lines), and Voronoi polygons (gray lines). Each Voronoi polygon acts as
a finite volume cell. (A) streamflow is routed downslope from node to node along triangle
edges, following the route of steepest descent. (B) two-dimensional diffusive exchange of
sediment between node N and its neighbors. The diffusive mass flux per unit width between
any two nodes is computed using the gradient between them. Multiplying unit flux by the
width of their shared Voronoi polygon edge gives the total mass exchange rate.
12. The Channel-Hillslope Integrated Landscape Development
Model (CHILD)
31
Figure 4. Schematic illustration of Poisson rectangular pulse rainfall model (after Eagleson,
1978).
Figure 5. Flow chart illustrating the sequence of computations in CHILD.
Figure 6. Example of simulated gully erosion and healing in response to stochastic
variations in rainfall intensity and duration. Here, a gully system forms and begins to heal on
a planar slope (30 degrees, 100 by 100 meters) that is subjected to a series of random storm
events. The landscape is highly sensitive to extreme events, owing to a large threshold for
runoff erosion (τc) and a high soil erodibility coefficient (kb). (A) time series of rainfall events
(durations not shown). (B) mean elevation of the surface through time, highlighting the
episodic nature of denudation. Arrows indicate the times corresponding to plots C and D. (C)
perspective plot of slope immediately after the last gully-forming episode. (D) perspective
plot at the end of the simulation. (E) and (F) show contour plots at these two time slices.
Equation set used in this run is –dz/dt = kb(τ-τc) (kb=1.6 x 10-5 m
2 s kg-1), with τ =
kt(Q/W)2/3S2/3 (Pa) (kt=6.2 x 104 Pa s2/3 m-4/3), τc = 20 Pa, and W = 0.001 Q0.5 with Q in m3/s.
Rainfall parameters are 640.P = mm/hr, 32=
r
T hr, and 148=
b
T hr; hillslope diffusivity
(kd) is 0.01 m2/yr.
Figure 7. The influence of runoff-production mechanism on terrain morphology. (A)
simulated drainage basin under infiltration-excess (Hortonian) runoff production (Eq (3)). (B)
simulated basin under saturation-excess runoff production, using the O’Loughlin (1986)
model (Eq (6)). (C), (D) plots of surface slope versus contributing area for the two cases. The
line in (D) represents the line of saturation for the mean-intensity storm. In these examples
runoff erosion is modeled as SQdt/dz .50
−∝ . Parameters are 92.P = mm/hr, 5=
r
T yr,
95=
b
T yr, kd = 0.01 m2/yr, U = 0.1 mm/yr, and in (B) T=105 m2/yr.
Figure 8. Slope-area plots from two simulations illustrating a downstream transition from
detachment-limited to transport-limited behavior under (A) constant runoff and (B) variable
(stochastic) runoff. Both simulations are in equilibrium with a constant and spatially uniform
rate of baselevel fall. The transport and erosion coefficients are adjusted so that the theoretical
transition point occurs at the same drainage area in both cases. Although fluvial erosion
theory predicts that such a transition should occur in many rivers, the result shown in (B)
implies that transitions may be so smooth as to be undetectable in data.
Figure 9. Example of a simulated mountain-fan system, showing progradation of a set of
alluvial fans in response to block uplift along a vertical fault. The substrate is treated as a
cohesionless sediment pile containing a mixture of sand and gravel sediment fractions.
Shading indicates the relative proportion of sand in the uppermost (active) sediment layer,
with lighter shades indicating higher sand fraction. (A) 20,000 years after onset of uplift; (B)
40,000 years; (C) 100,000 years. Inset in (C) shows the location of cross-section in Figure 10.
Uplift rate is 1 mm/yr, diffusivity is 0.01 m2/yr, and rainfall parameters are 11.P = mm/hr,
3=
r
T yr, and 97=
b
T yr.
Figure 10. Stratigraphic cross-section through the fan complex in Figure 9. Section is taken
parallel to the strike of the range through the center of the fan complex (indicated by a-a’ in
Figure 9C). Black shading indicates less than 60% sand content.
32 Chapter 12
Figure 11. Illustration of right bank ( n
ˆ
e
ˆ
−= ) erodibility determination for node i. Eeff,i1 and
Eeff,i2 are effective erodibilities with respect to node i at adjacent nodes that are distances d1
and d2, respectively, from the line parallel to the unit vector, n
ˆ
−. In the coordinate system
shown, the s-direction is parallel to the flow edge, and the n-direction is perpendicular to the
flow edge. Delaunay triangulation is in thin solid lines, Voronoi diagram is in dashed lines,
and flow edges are in heavy black.
Figure 12. Flow chart showing the implementation of meandering.
Figure 13. Simulation of channel meandering and floodplain development. (A) perspective
view of simulated topography, highlighting stream pattern and development of terraces
(elevations are interpolated to regular grid for plotting purposes). (B) view of triangulated
mesh, showing densification in the area of the floodplain (see text for details).
