Quantitative Studies of Alluvial Architecture
assumed to be constructed by any appropriate
fluvial style, following the hyp othesis that the two
major styles, meandering and braided [the «mobile
channels" of Friend (1983)], both construct complexes of bars and minor channels within relatively
stable meander belts. This is certainly the case for
meandering systems, in which lateral migration is
constrained by the alluvial topography of levees
and the resistance met by channels when they erode
laterally into the fine-grained deposits of abandoned
channels and the floodplain. The assumption of a
confined channel belt may not be so appropriate for
braided systems.
The alluvial stratigraphy models discussed here
are autogenic, in that the external, allogenic variables that control the long-term behavior of the river
system are set at constant background values. Tectonism and sea-level change are not assumed to
affect the system, so that average accumulation rates
and regional paleoslope remain constant throughout a model run. Climate change and its effect on
discharge, fluvial style, and sediment supply are also
not considered. Bridge and Leeder (1979) were
among the first to consider the quantitative effects of
tectonism on fluvial aggradation, by building iri the
ability in the computer model to tilt the depositional
surface in a direction perpendicular to paleoslope.
This aspect of the model is noted in Sect. 11.2.1.
The most important controlling factor in al,ltogenic models of alluvial stratigraphy is the determination of the rate and style of avulsion. As noted
earlier in this chapter, little is known about the detailed avulsion history of most rivers, and so some
generalizing assumptions have to be made. Leeder
(1978) and Bridge and Leeder (1979) assumed in
their models that avulsion is not dependent on the
existing position of the channel within the model,
but occurs sufficiently far upstream from the location of the model that the channel is free to move to
the current lowest position on the floodplain. Avulsion frequency was determined by reference to the
limited information on stratigraphic and historical
data (Sect. 10.3), and the recurrence interval was
assumed to follow a distribution fu nction similar to
that of earthquakes, which in any location of tectonic
stress become more likely to occur as time passes.
Mean avulsion periods ranging from 111 to 1780
years were used by Bridge and Leeder (1979).
Given an avulsion event, the channel is moved to
the lowest point on the floodplain and allowed to
accumulate a channel sand body at a preset accumulation rate, ranging from 5 to 40 m/ka. Floodplain
accretion is accomplished using a simple equation
341
that reduces the aggradation rate across the width of
the model cross section in proportion to the distance
from the channel belt. This reflects the fact that
floodplain sediments are derived primarily from
overbank flooding. Compaction of the accumulated
deposits is accomplished, layer by layer, using preset
compaction factors. This process generates relief
on the floodplain surface because floodplain fines
compact less than channel sands, and the least compaction occurs where channel sands are stacked vertically. The model therefore builds in a feedback
effe ct, in that the site of the next lowest position for
channel occupancy will be determined by the outcome of the preceeding sedimentation-compaction
history. Two examples of Bridge and Leeder's (1979)
model are illustrated here (Figs. 10.33, 10.34). The
most realistic of these is the first, in which sand body
distribution appears to be relatively random after
the initial ten avulsion events. The second model
shows a much too regular stacking of the channel
bodies, an effect that occurs when channel width and
depth are set too large relative to aggradation rate.
The effects of the stacking and compaction of nu�
merous preceding events are downplayed in this
model, and successive channel positions are deter�
mined primarily by the position of the two or three
immediately preceding channels. The result is a dis�
tribution of channel bodies not unlike that of bricks
in a walL Nevertheless, the Bridge and Leeder (1979)
model has been much used, particularly by petroleum geologists, because of the insights it yields into
channel stacking patterns and interconnectedness,
fa ctors of considerable importance in the understanding of reservoir predictability and fluid migra�
tion behavior. The model has recently been
expanded and updated by Bridge and Mackey
(1993a), but the improvements are at a level of refinement that does not need to be addressed here.
The application of these model concepts to the
interpretation of real stratigraphic data can be a
complex problem. Willis (1993b,c) provided extremely detailed architectural descriptions of the
Siwalik Group in a segment of the Himalayan
fo reland basin in Pakistan, including quantitative
studies of a 2-km-thick section keyed to the magnetostratigraphic time scale. Variations in sedimentation rates, sand body thickness and sand body
density cannot readily be interpreted using simple
autogenic models because of lack of knowledge of
the number of rivers in the basin and a lack of
detailed information on local tectonism.
Quite different approachs to modeling have
been taken by other researchers. Several modeling
assumed to be constructed by any appropriate
fluvial style, following the hyp othesis that the two
major styles, meandering and braided [the «mobile
channels" of Friend (1983)], both construct complexes of bars and minor channels within relatively
stable meander belts. This is certainly the case for
meandering systems, in which lateral migration is
constrained by the alluvial topography of levees
and the resistance met by channels when they erode
laterally into the fine-grained deposits of abandoned
channels and the floodplain. The assumption of a
confined channel belt may not be so appropriate for
braided systems.
The alluvial stratigraphy models discussed here
are autogenic, in that the external, allogenic variables that control the long-term behavior of the river
system are set at constant background values. Tectonism and sea-level change are not assumed to
affect the system, so that average accumulation rates
and regional paleoslope remain constant throughout a model run. Climate change and its effect on
discharge, fluvial style, and sediment supply are also
not considered. Bridge and Leeder (1979) were
among the first to consider the quantitative effects of
tectonism on fluvial aggradation, by building iri the
ability in the computer model to tilt the depositional
surface in a direction perpendicular to paleoslope.
This aspect of the model is noted in Sect. 11.2.1.
The most important controlling factor in al,ltogenic models of alluvial stratigraphy is the determination of the rate and style of avulsion. As noted
earlier in this chapter, little is known about the detailed avulsion history of most rivers, and so some
generalizing assumptions have to be made. Leeder
(1978) and Bridge and Leeder (1979) assumed in
their models that avulsion is not dependent on the
existing position of the channel within the model,
but occurs sufficiently far upstream from the location of the model that the channel is free to move to
the current lowest position on the floodplain. Avulsion frequency was determined by reference to the
limited information on stratigraphic and historical
data (Sect. 10.3), and the recurrence interval was
assumed to follow a distribution fu nction similar to
that of earthquakes, which in any location of tectonic
stress become more likely to occur as time passes.
Mean avulsion periods ranging from 111 to 1780
years were used by Bridge and Leeder (1979).
Given an avulsion event, the channel is moved to
the lowest point on the floodplain and allowed to
accumulate a channel sand body at a preset accumulation rate, ranging from 5 to 40 m/ka. Floodplain
accretion is accomplished using a simple equation
341
that reduces the aggradation rate across the width of
the model cross section in proportion to the distance
from the channel belt. This reflects the fact that
floodplain sediments are derived primarily from
overbank flooding. Compaction of the accumulated
deposits is accomplished, layer by layer, using preset
compaction factors. This process generates relief
on the floodplain surface because floodplain fines
compact less than channel sands, and the least compaction occurs where channel sands are stacked vertically. The model therefore builds in a feedback
effe ct, in that the site of the next lowest position for
channel occupancy will be determined by the outcome of the preceeding sedimentation-compaction
history. Two examples of Bridge and Leeder's (1979)
model are illustrated here (Figs. 10.33, 10.34). The
most realistic of these is the first, in which sand body
distribution appears to be relatively random after
the initial ten avulsion events. The second model
shows a much too regular stacking of the channel
bodies, an effect that occurs when channel width and
depth are set too large relative to aggradation rate.
The effects of the stacking and compaction of nu�
merous preceding events are downplayed in this
model, and successive channel positions are deter�
mined primarily by the position of the two or three
immediately preceding channels. The result is a dis�
tribution of channel bodies not unlike that of bricks
in a walL Nevertheless, the Bridge and Leeder (1979)
model has been much used, particularly by petroleum geologists, because of the insights it yields into
channel stacking patterns and interconnectedness,
fa ctors of considerable importance in the understanding of reservoir predictability and fluid migra�
tion behavior. The model has recently been
expanded and updated by Bridge and Mackey
(1993a), but the improvements are at a level of refinement that does not need to be addressed here.
The application of these model concepts to the
interpretation of real stratigraphic data can be a
complex problem. Willis (1993b,c) provided extremely detailed architectural descriptions of the
Siwalik Group in a segment of the Himalayan
fo reland basin in Pakistan, including quantitative
studies of a 2-km-thick section keyed to the magnetostratigraphic time scale. Variations in sedimentation rates, sand body thickness and sand body
density cannot readily be interpreted using simple
autogenic models because of lack of knowledge of
the number of rivers in the basin and a lack of
detailed information on local tectonism.
Quite different approachs to modeling have
been taken by other researchers. Several modeling