Elsevier

Geomorphology

Volume 53, Issues 1–2, 1 July 2003, Pages 183-196
Geomorphology

Drainage basin evolution in the Rainfall Erosion Facility: dependence on initial conditions

https://doi.org/10.1016/S0169-555X(02)00353-7Get rights and content

Abstract

Four experiments in alluvial drainage basin evolution were carried out in the Rainfall Erosion Facility (REF) at Colorado State University to investigate the dependence of basin evolution on initial topography. Basins were initially undissected. Each experiment began with a unique initial condition representing various end-members of relief and hypsometry. Drainage network development, hillslope processes, basin denudation, and basin response to base-level lowering all depended strongly on the initial topography. No classic model of drainage network evolution was found to be generally applicable. Initially, planar slopes first developed subparallel channels that extended headward dendritically during an early phase of extension. Channel incision occurred first in the interior of the basin where saturation overland flow was greatest, not at the basin outlet as assumed in most classic models of network development. Channels widened over time, initiating lateral migration and drainage capture in the downslope portion of the watershed before transferring lateral migration upslope. Planar basins of larger initial gradient grew headward more quickly and become more deeply entrenched, inhibiting late-stage lateral migration. An experiment with initial relief concentrated at a plateau edge evolved in several unique ways. A high ratio of subsurface-to-surface flow gave rise to mass movements at the plateau edge and outlet channels. Deep channels were quickly cut initially but did not extend far upslope because slope instability undermined channel head migration, leaving the plateau undissected and hence very slow to erode. These results suggest that the distribution of relief within a basin exerts an important control on drainage network pattern and basin denudation. In addition, erosional basins may evolve in several distinct modes characterized by particular combinations of hypsometry, hillslope processes, and mean denudation rate.

Introduction

Flume experiments have added greatly to our understanding of the complex dynamics of the fluvial system (Schumm et al., 1987). Although flume experiments are greatly simplified models of larger-scale drainages, they suggest ways in which elements of the fluvial system couple and give rise to complex dynamics even in the absence of external forcing. Although flume experiments are no substitute for field studies, field work often cannot easily address questions of how landforms within the fluvial system evolve and interact through time, particularly at large scales. Although processes acting in a flume may be different than those of larger basins, the similarity of form between natural basins and model basins in large flumes suggests that the natural and large model basins are at least qualitatively similar. Flume experiments provide null hypotheses against which phenomena associated with tectonic or climatic change can be tested. They also suggest dynamic feedback mechanisms between elements of the fluvial system that may not be apparent in field studies in which landform evolution can only be partially reconstructed. In addition, flume studies may act as reality checks on the numerical simulation models of basin evolution now being widely applied. If a model cannot adequately reproduce the behavior of the controlled conditions of a flume, it is unlikely to be relevant to natural landscapes in which much greater complexity is undoubtedly at work. The behavior of numerical and flume models can be directly and precisely compared because initial conditions can be designed in both cases with a high degree of similarity. In addition, both modeling methods permit measurement of any surface position or boundary flux at nearly any point in the experiment, providing a high potential for the collection of spatially and temporally complete data on model landform evolution.

Perhaps the most influential flume experiments on basin evolution have been performed in the Rainfall Erosion Facility (REF) at Colorado State University by Stan Schumm and his students Parker, 1977, Schumm et al., 1987, Wood et al., 1993, Koss et al., 1994. This work is notable, in part, because the scale of the physical model, 138 m2 of catchment area, is large enough that the model can be considered to be a prototype of a small alluvial basin (Parker, 1977). This size generates sufficient discharge to reproduce realistic landform features, including diffusive (rainsplash-dominated) hillslopes, meandering channels, and fluvial terraces with a wide range of spatial scales (Parker, 1977). Parker (1977) performed two experiments in drainage network evolution with basins of different relief. The early part of his experiments focused on mean basin denudation (sediment yield) while basin response to base-level lowering was the focus of later stages. Parker observed his low-relief basin to evolve by headward growth with simultaneous branching while his higher-relief basin evolved with the rapid formation of a master rill with subsequent side branching. Notably, however, Parker began both of his experiments with two planes intersecting to form a central valley along the length of the flume in order that runoff could be directed to a predetermined outlet point. As such, Parker's networks were biased to grow in the manner he observed. The formation of parallel rills draining to multiple outlet points, for example, would have been impossible in his experimental setup.

Several theoretical models have been introduced for drainage network development. Glock (1931) postulated several phases of development during network evolution including drainage density reduction as overall basin relief decreases during the tectonic quiescence following initial uplift. Horton (1945) introduced a model in which parallel rills developed with a spacing related to the critical distance for overland flow. A network developed in his model by successive rill piracy and cross-grading of the surface as the largest rill downcut faster than adjacent rills and captured drainage. Drainage capture led to enhanced downcutting in a positive feedback process. Howard (1971) proposed that a wave of dissection penetrated the landscape at channel heads. In his model, the drainage network expanded into undissected portions of the basin leaving behind a mature network as it proceeded. Dunne (1980) presented a model based on subsurface flow and seepage erosion. In his model, first-order channels grow headward by seepage erosion. Deflected subsurface flow toward the channel head then enhanced seepage erosion in a positive feedback process. Parker's (1977) results on drainage network evolution have been taken as evidence that Howard's (1971) model is appropriate for basins of low relief while the rapid development of a master rill with subsequent network extension and elaboration is more likely in higher-relief basins. A principal goal of our work was to expand on Parker's study of drainage network evolution by including different initial hypsometries in addition to variable relief between experiments. Also, we wished to determine the effects of a uniform base level along the lower boundary in addition to the concentrated outlet point Parker used.

In addition to the results on drainage network evolution, Parker (1977) and Schumm et al. (1987) introduced the concept of complex response. Complex response refers to the damped, oscillatory variation in sediment yield Parker observed in the phase of his experiments investigating basin response to base-level lowering Parker, 1977, Schumm et al., 1987. Although still not well understood, Parker (1977) introduced a schematic model for complex response involving a coupling between the mainstem channel's longitudinal profile and its cross-section. In Parker's model, base-level lowering creates a concentration of stream power and subsequent incision at the basin outlet that increases sediment yield. Propagation of the knickpoint upslope shifts the locus of incision while the downslope reach widens, providing accommodation space for the storage of sediment from upstream as a fill terrace. This results in a local minimum in sediment yield according to the model. Progradation of deposited sediment partially rejuvenates the channel gradient at the outlet leading to another maximum in sediment yield. Complex response is one possible explanation for cut-and-fill terraces observed in natural drainage networks. Few terraces mapped in the field, however, have ever been attributed to this process (see Ethridge et al., 1998 for a review of those that have). A key research question is how to distinguish terraces formed by autocyclic processes or complex response from those generated by external forcing, such as variations in climate or tectonism. Can bounds be placed on the magnitude or time scale for the creation of terraces by autocyclic mechanisms to help distinguish them from those created by allocyclic processes? Another goal of our experiments was to shed light on complex response and, in particular, determine whether it was robust to variations in initial conditions of the basin.

Many more basic questions of landform evolution remain that can be addressed with physical models. The relationship between basin shape, drainage pattern, and denudation, for example, is poorly known. Although many studies have concluded that sediment yield is most strongly correlated with mean basin relief Ahnert, 1970, Jansen and Painter, 1974, Pinet and Souriau, 1988, Milliman and Syvitski, 1992, Summerfield and Hulton, 1994, sediment yield may be most directly controlled by discharge, grain size, and outlet channel gradient (Dade and Friend, 1998), all of which generally correlate with mean basin relief on a global average but may vary substantially from that average trend. High-elevation plateaus, for example, ought to erode rapidly based on the correlation between sediment yield and elevation. Cosmogenic dating, however, has shown that erosion rates on plateaus are generally very low regardless of climate Bierman, 1994, Bierman and Caffee, 2001, Granger et al., 2001, suggesting that they erode slowly because of their low average hillslope relief and low drainage density. Rainfall erosion experiments can enable the relationship between basin morphology and sediment yield to be precisely determined for an alluvial basin under conditions of constant runoff and erodibility. Physical models can suggest potential controlling factors that would be difficult to assess in natural basins given the large number of variables that often cannot be measured or controlled for.

Flume experiments can also serve as constraints for computer simulation models of landform evolution. Simulation models of landform evolution, although widely applied to a variety of problems, have done a poor job thus far of enhancing our basic understanding of landscape dynamics. Most fluvial landform evolution models currently in use have a number of basic limitations. In addition to other problems, simulation models often use a fixed, uniform grid on which channels cannot migrate laterally or meander. Lateral migration may be a crucial means of basin adjustment in many alluvial environments, particularly those subject to tectonic forcing (Schumm et al., 2000). Channel widths in most models are also assumed to be fixed in time, preventing any cut-and-fill behavior or any mechanism for the generation of fill terraces. In addition, the empirical relationships describing sediment transport in landform evolution models are poorly constrained. Flume experiments may enable a better constraint to be determined by quantitatively relating sediment flux to channel morphology during an experiment. Certainly, landform evolution models will continue to improve and will soon adequately represent a wide range of processes and forms. Careful validation of these models, however, will become even more important as their complexity grows. If a model can reproduce the essential aspects of flume experiments under a variety of initial conditions, we can have more confidence in its ability to predict the evolution of natural landscapes. Thus, an final motivation for our experiments was to provide numerical modelers with a database, building on the work of Stan Schumm and his students, for comparing the processes and results of their models to a prototype alluvial basin under carefully controlled conditions.

Section snippets

Facility description

Four experiments were conducted in the REF, a 15×9.2-m enclosed flume (Fig. 1). Smaller “table top” basins have been used successfully to model landscapes with high relief and landslide-dominated hillslopes Crave et al., 2000, Hasbargen and Paola, 2000, but away from the landslide-dominated regime, they are less likely to reproduce the range of processes and landforms observed in natural fluvial systems.

Although the results we will present in this paper are qualitative observations of the basin

Experimental results

The dominant hillslope process observed in the REF for low-gradient slopes was rainsplash. Larger-gradient, saturated slopes evolved by mass wasting, including slump failure and mudsliding. Channel bank migration was observed to take place by gradual cutting of outer banks in channels of shallow depth and by an episodic process of bank undercutting, slump failure, and reworking of slump material for channels of deeper entrenchment.

We performed four experiments on drainage network and basin

Discussion

None of the theoretical models proposed for the evolution of drainage networks proved to be applicable to our experiments. This is not surprising because none of the models include initial topography or lateral channel migration in any way. Lateral migration was found to be crucial to drainage capture and network integration in our experiments, particularly for basins of low relief. Horton's model of rill formation and subsequent piracy matched some basic elements of the evolution of the

Conclusions

Four experiments in drainage network and alluvial basin evolution were carried out in the Rainfall Erosion Facility (REF) at Colorado State University. The initial conditions were modified from the original experiments of Parker (1977) to enable a uniform base level of erosion to be applied to the basin and multiple outlet channels to compete for drainage and migrate laterally. Basins and their drainage networks evolved in markedly different ways depending on their initial topography. Initially

Acknowledgements

I am indebted to Deborah Bryan for providing courageous assistance and good cheer during these experiments. Frank Ethridge and Stan Schumm generously provided advice, encouragement, support, and logistical help. Members of the Engineering Research Center machine shop provided valuable technical assistance. I also wish to thank two anonymous reviewers for their constructive reviews. This work was supported by the University of Arizona.

References (23)

  • F. Ahnert

    Functional relationships between denudation, relief, and uplift in large mid-latitude drainage basins

    Am. J. Sci.

    (1970)
  • P.R. Bierman

    Using in situ produced cosmogenic isotopes to estimate rates of landscape evolution: a review from the geomorphic perspective

    J. Geophys. Res.

    (1994)
  • P.R. Bierman et al.

    Slow rates of rock surface erosion and sediment production across the Namib Desert and escarpment, Southern Africa

    Am. J. Sci.

    (2001)
  • J.C. Brice

    Erosion and deposition in the loess-mantled Great Plains, Medicine Creek drainage basin, Nebraska

  • W.B. Bull

    Geomorphic Effects of Climatic Change

    (1991)
  • A. Crave et al.

    Analogue modelling of relief dynamics

    Phys. Chem. Earth, Part A

    (2000)
  • W.B. Dade et al.

    Grain-size, sediment-transport regime, and channel slope in alluvial rivers

    J. Geol.

    (1998)
  • T. Dunne

    Formation and control of drainage networks

    Prog. Phys. Geogr.

    (1980)
  • F.G. Ethridge et al.

    Cyclic variables controlling fluvial sequence development: problems and perspectives

  • W.S. Glock

    The development of drainage systems: a synpotic view

    Geogr. Rev.

    (1931)
  • D.E. Granger et al.

    Pliocene–Pleistocene incision of the Green River, Kentucky, determined from radioactive decay of cosmogenic 26Al and 10Be in Mammoth Cave sediments

    Bull. Geol. Soc. Am.

    (2001)
  • Cited by (68)

    • Climate-driven drainage reorganization of small mountainous rivers in Taiwan (East Asia) since the last glaciation: The Zhuoshui River example

      2022, Palaeogeography, Palaeoclimatology, Palaeoecology
      Citation Excerpt :

      The stalagmites δ18O in China indicated that, due to the oscillation of East Asian summer monsoon, during the MIS3 stage and Holocene, precipitation in East Asia (including Taiwan) was abundant but declined in the LGM (Cosford et al., 2008; Cheng et al., 2016, Fig. 4c). Rainfall can extend the river channel headward, increasing the rills density and causing drainage capture by erosion processes (Tucker and Bras, 2000; Pelletier, 2003). Furthermore, local climate also can influence the vegetation coverage in Taiwan.

    • Small catchments evolution on clayey hilly landscapes in Central Apennines and northern Sicily (Italy) since the Late Pleistocene

      2020, Geomorphology
      Citation Excerpt :

      Generally, hard rock type records river changes with topographical signatures and preserves landforms, while in case of soft rocks, the landforms produced can be readily disappear with the erosion of rocks (Stokes et al., 2017; Keen-Zebert et al., 2017). Small catchments and low-order drainage systems, especially when set on clayey bedrock, respond more rapidly to the transformations of physical environment than large ones with regard to the rearrangement of their slope morphometry (Mizutani, 1996; Pelletier, 2003; Della Seta et al., 2007; McGuire and Pelletier, 2016; Brandolini et al., 2018; Harrison et al., 2019; and references therein). Moreover, the initial morphological and morphometrical conditions play a crucial role in driving the evolutionary response of small catchments, thus determining the geomorphological response in terms of sedimentation/incision dynamics (Dotterweich, 2008; Buccolini et al., 2010; Harrison et al., 2019; Pelletier, 2003; Buccolini and Coco, 2013; Tucker, 2015).

    View all citing articles on Scopus
    View full text