Emergence of Characteristic Landscape Scales Through Hillslope-Channel Interactions
Analysis of high-resolution topography reveals that many landscapes contain characteristic spatial scales. These scales appear to have emerged from the erosion and sediment transport processes that have shaped the landscape. In fluvially dissected terrain, for example, the trend in surface slope with increasing drainage area often reaches a maximum at a characteristic drainage area that corresponds approximately to the transition from topographically divergent hillslopes to topographically convergent valleys. Another frequently observed scale is a uniform spacing of first-order valleys that creates a dominant topographic "wavelength." We present a theoretical framework for predicting these two scales in landscapes that are shaped by a combination of soil creep and stream channel incision, and we test this framework by comparing the scales predicted from theory with those measured from high-resolution laser altimetry data. From equations of mass conservation and sediment transport, we derive a characteristic length scale at which the time scales for soil creep and stream incision are equal. Steady-state solutions to a numerical landscape evolution model reveal that this length scale is approximately equal to the square root of the characteristic drainage area and directly proportional to the ridge-valley wavelength. This theoretical result is consistent with the measured slope-area maximum and ridge-valley wavelength at several field sites. The agreement suggests that the characteristic scales observed in drainage networks provide an easily observable record of how geologic and climatic factors regulate long-term sediment transport and erosion rates.
Numerical models of catchment scale sediment transfer: progress, problems and potential
Assessing Sediment Fluxes Within Evolving Rill Networks in an Experimental Landscape
In an effort to better understand the spatial and temporal variation in sediment fluxes within actively evolving rill
drainage networks, high-resolution, photogrammetrically-derived ortho-rectified photographs and digital
elevation models of an experimental landscape were analyzed and compared to basin-outlet sediment
signatures. Within a large 7.1 x 2.4 M flume, a sandy-loam soil bed was prepared at a 5% slope with a broad
central concavity, to represent a natural drainage basin. This soil bed was subjected to repeated simulated
rainstorms of known intensity. During these storms, rill erosion was initiated by a period of intense headcut
development and migration and very high sediment yield in response to baselevel lowering at the flume outlet.
After this initial period of adjustment, a number of sequential waves of degradation propagated throughout the
rill network, scouring pre-existing channels and extending the upstream extent of network tributaries. These
bursts of activity were often interspersed by periods of relative quiescence. While these secondary waves were
clearly visible and quite pronounced with regard to their effect on rill network morphology, they appeared to
have little effect on basin-outlet sediment signatures. Rill networks quickly attained a steady-state sediment
yield at the basin outlet regardless of changes in internal drainage network morphology and sediment fluxes.
Only in the presence of a secondary drop in baselevel was the equilibrium of the steady state rill network upset
- to systematically adjust once again and attain another steady-state condition. These
data provide a unique opportunity to examine the initiation and exploitation of landscapes by rill erosion under
controlled conditions employing extremely precise datasets with a very high temporal and spatial resolution.
These findings have important implications for understanding the relative magnitude of internal and external
controls on drainage network development and landscape evolution.
Numerical Modelling of Experimental Gully Development and Headcut Erosion
Numerical landscape evolution models are typically applied as exploratory tools, helping to understand processes and process interactions, often in simplified or idealized landscapes. In this sense, the models are evaluated qualitatively, based on their ability to broadly reproduce known features of landscape evolution. Quantitative validation of models is far less common. Landscape evolution occurs over large spatial and temporal domains, e.g. 1000's km2 and 100's to 10000's of years, and there is a paucity of adequate data for defining both the initial conditions of the simulations (i.e. initial topography) and the temporal boundary conditions (e.g. climatic or tectonic forcing). However, there is nothing which inherently prohibits the application of numerical landscape evolution models to much smaller spatial and temporal domains, as used in laboratory experiments. Such physical experiments can provide detailed data to which numerical models can be compared quantitatively. Here, the CAESAR landscape evolution model is used to simulate gully development and headcut erosion as observed in a set of physical experiments. The paper analyses the similarities and differences between the physical and numerical simulations, in terms of spatial erosion patterns and in terms of total sediment yield. The paper concludes with a more general discussion on the potential of physical experiments for the validation of numerical landscape evolution models.
Sediment Yield Modeling in a Large Scale Drainage Basin
This paper presents the findings of spatially distributed sediment yield modeling in the upper Indus River basin. Spatial erosion rates calculated by using the Thornes model at 1-kilometre spatial resolution and monthly time scale indicate that 87 % of the annual gross erosion takes place in the three summer months. The model predicts a total annual erosion rate of 868 million tons, which is approximately 4.5 times the long- term observed annual sediment yield of the basin. Sediment delivery ratios (SDR) are hypothesized to be a function of the travel time of surface runoff from catchment cells to the nearest downstream channel. Model results indicate that higher delivery ratios (SDR > 0.6) are found in 18 % of the basin area, mostly located in the high-relief sub-basins and in the areas around the Nanga Parbat Massif. The sediment delivery ratio is lower than 0.2 in 70 % of the basin area, predominantly in the low-relief sub-basins like the Shyok on the Tibetan Plateau. The predicted annual basin sediment yield is 244 million tons which compares reasonably to the measured value of 192.5 million tons. The average annual specific sediment yield in the basin is predicted as 1110 tons per square kilometre. Model evaluation based on accuracy statistics shows very good to satisfactory performance ratings for predicted monthly basin sediment yields and for mean annual sediment yields of 17 sub-basins. This modeling framework mainly requires global datasets, and hence can be used to predict erosion and sediment yield in other ungauged drainage basins.
The Influence of Research Designs in Understanding the Control of Morphological Patterns on Bedload Path Lengths in gravel-bed rivers
There may be a strong association between particle path length and the morphologic scale of prevailing pool- bar channel patterns in gravel-bed rivers. It has been shown that tracers introduced in a pool during channel- forming discharges have downstream path length frequency distributions that are symmetrical, with modes coinciding with pool-bar spacing. Evidence to support this hypothesis comes mostly from flume studies and there is only a limited support based on field data. For the past seven years, we have conducted field experiments in streams with gradients ranging from constricted pool and pool-bar systems to step-pools and cascades in order to link displacement distances of tracing particles to the spacing between bed features. Results showed that clast movements could not be predicted from morphological length scales. The objective of this paper is to define why the control of morphological patterns on bedload path lengths could not be seen from our dataset. We have tested hypotheses for which the limited predictive capacity of the morphological length scale results from 1) the identification of some bed units that may have been ambiguous along the bed profiles or 2) the selection of sampling parameters. The experiment was carried out in seven reaches located in Quebec and in the French Alps. The slopes ranged between 0.011 and 0.43 and the representative particle size (d50) from 42 to 110 mm. Detailed topographic maps of the bed were produced in order to describe the morphological patterns. We used passive transponders inserted into clasts to measure displacement distances. Between 100 and 450 clasts of different sizes were tagged in each reach between 2003 and 2008. The tracking was carried out at low flow using a portable antenna. In identifying morphological entities that represent no ambiguity from both field observations and long profiles, we did not observe significant evidence of a morphological control on the path lengths. The results show that even when selecting well defined pool-bar or pool-step single entities, the morphological length scale is not associated with bedload path length in six of the seven reaches. However, in one of the reaches, a step-pool channel, the clasts did tend to move downstream from one pool to the next pool. The influence of sampling strategies was tested using the size and shape of the tagged clasts, the magnitude of the flood events, the channel characteristics, and the duration of the experiments. For some transport events, the size and the shape of the tagged clasts are both correlated with individual displacement distances. These events are not observed at specific discharge values. In general, however, the correlations are not significant. Moreover, the frequency distributions of path lengths are similar whatever the size of the tagged clasts. The percentage of mobile clasts during an initial event, as well as channel and average hydrodynamic characteristics of the streams do not affect the displacement distances or the mobility of the clasts. The duration of the experiment (from the initial seeding to the last recovery of the clasts) appears to be the only sampling factor to play a role on the significance of the relations between displacement distances and channel width, slope and bed material size. The channel morphology may affect the bedload path length but, to test this effect in the field and eventually generalize the results to different channel systems, the sampling protocol must include duration rules applied to all stream channels. This study outlines the importance standardizing research protocols for long term studies of particle path lengths.