H43F-01
Seismic Control on Location of Lavakas (Midslope Gullies) in Madagascar
Madagascar's central highlands are deeply weathered, with 1-2 m of laterite overlying 10s of metres of saprolite. These unstable materials sit at altitudes on the order of 1000m in recently uplifted, steep terrain characterised by convex hills with slopes averaging 25 degrees and local ridge-valley elevation changes of 100- 500 m. In many areas these slopes bear numerous erosional features called lavakas, with mapped densities up to 25/k$^{2}$ in some areas. Lavakas are tadpole-shaped gullies with vertical sides and flat floors. They are wide at the headwall and taper to a narrow, deeply incised outfall, which connects to the valley drainage. They are not fed by overland flow, but develop by groundwater sapping and subterranean erosion in the porous and friable saprolite beneath the baked lateritic duracrust. The controls on lavaka formation are poorly understood. Interplay between climate, topography, and bedrock geology is known to be important, and human activity is also implicated. However, there are many areas with appropriate rainfall, hillslope geometry and geology where lavakas do not form. Likewise, lavakas form readily in some areas with little human activity, and do not form in other areas that are heavily used. Our analysis suggests that seismic activity may be a significant regional factor driving lavaka formation. Using GIS analysis, we compared the locations of recent earthquakes (1347 events with magnitudes 3.0-5.4, recorded between 1979 and 1995) with the distribution of zones where lavakas are abundant (as mapped by Henri Besairie in 1957), and found that they are strongly correlated. We cannot say whether individual seismic events are responsible for specific lavakas; but the observation that lavakas are most common in zones that are seismically active implies that earthquakes play a significant role. Repeated mild earth-shaking events might aid in loosening the saprolite and making it more vulnerable to lavaka-forming processes; or individual earthquakes might initiate collapse of slopes already weakened by groundwater sapping. More detailed work into the relationships between individual lavakas and local seismicity is needed to determine which of these mechanisms might be active. Our ongoing GIS analysis will quantify the relative importance of geology, topography, and seismicity in the location of lavaka-prone areas.
H43F-02
Feedback Mechanisms Driving Landscape Heterogeneity and Geometry in a Low-Gradient, Pulsed Flow Peatland
The ridge and slough landscape of the Florida Everglades is a historically heterogeneous environment, in which elongated and elevated ridges of sawgrass divide open-water sloughs in a regular pattern that is aligned with the dominant flow direction. With drainage alteration and flow reductions occurring over the past century, the pattern has degraded, and restoration efforts seek a mechanism under which to regenerate topographic heterogeneity, which supports high biodiversity. Based on numerical simulation and a literature synthesis, it is proposed that two dominant feedback mechanisms govern pattern formation in the Everglades: 1) a differential peat accretion feedback, in which species-specific net rates of peat accretion respond to hydroperiod and nutrient availability, and 2) an anabranching river-type feedback, in which flow, sediment deposition, and entrainment depend upon and regulate the morphologic profile. Numerical simulation using a recently developed mechanistic model called {\it PeatAccrete 1.0} shows that the first feedback mechanism regulates vertical differences in topography and the attainment of an equilibrium height. Previous studies on anabranching rivers indicate that the latter likely controls lateral spacing of ridges, longitudinal expansion of topographic features, and the attainment of an equilibrium width. {\it PeatAccrete 1.0} provides insight into ecohydrological feedbacks and offers possible reasons for landscape degradation and possible restoration strategies. Results show that a reduced hydroperiod, decreased flow velocities, and more reduced redox potentials may all contribute to the topographic flattening that has occurred over the past century. In portions of the landscape with remnant topographic heterogeneity, restoration of flow velocities and hydroperiods will likely restore the landscape to its historical form; however, if increased flow velocities transport higher concentrations of phosphorus and sulfate downstream, sawgrass ridges are more readily initiated on former slough bottom locations and may occupy a greater area than in the historical system. Though the {\it PeatAccrete} simulation applies specifically to the Florida Everglades, insights into the two feedback mechanisms governing this landscape are expected to apply to other low-gradient peatlands with abundant flowing surface water and high seasonality in water level.
H43F-03
Modeling Chemical Weathering Rates at the Bedrock-regolith Interface
Many process models exist for transport of regolith on slopes. Much less attention has been given to the rate and the processes by which bedrock is transformed into regolith. Landscape evolution models treat this regolith source term as a black box in which the regolith generation rate depends on regolith thickness. We construct a simple model of chemical weathering rates under a regolith as a step towards modeling regolith production processes. We assume that the weathering rate is set by the degree of saturation of water with respect to the minerals that control loosening of particles for physical transport. In a uniform regolith, the concentration {\it c} will asymptotically approach saturation {\it c$_{sat}$}. For vertical water transport, the time water takes to reach the bedrock interface is set by the effective precipitation rate, {\it P$_{eff}$ = P-E}, the regolith porosity, $\phi$, and the regolith thickness, {\it H}. The greater the thickness of the regolith, the longer the water will take to reach its base, and the more saturated it will be. Conversely, the higher {\it P$_{eff}$}, the shorter will be the time to reach the bedrock, resulting in higher rates of dissolution. This formulation predicts an exponential decline in the rate of dissolution and hence in bedrock lowering rate with regolith thickness, falling off with a length scale of {\it {\it L}=P$_{eff}$/(k$\phi$)}. While some have advocated this exponential form, others have documented that the bare bedrock rate is lower than that beneath significant regolith, and have proposed that the dependence on regolith thickness must be humped, i.e., show a maximum at some finite thickness. The development above neglects the contact time of water with the interface. If water runs off rapidly from the surface, the water will have little time to interact, and, despite its aggressive chemistry, will produce little chemical alteration. We account for this by averaging the reaction rate over a hydrologically appropriate timescale {\it T}. This introduces a factor {\it F} representing the fraction of time the bedrock interface is wetted. We consider an expression for {\it F} with the following features. 1) On bare bedrock, contact time is at a minimum, the duration of precipitation in a given year, and {\it F} becomes {\it f}={\it T}$_{ppt}$/{\it T}$_{yr}$. 2) On regolith-mantled surfaces, the contact time increases, and the fractional time wetted, {\it F}, approaches 1. An expression with these properties is {\it F}={\it f}+(1-{\it f})(1- exp(-{\it H}/{\it H*}), where {\it H*} is the regolith thickness at which the fractional contact time has increased to (1-(1/e)) of the way from {\it f} toward 1. We suspect that {\it H*} is small; even a little regolith on the surface should increase {\it F} dramatically. The full model for reaction rate at the bedrock interface ({\it z}={\it H}) becomes d{\it c}/d{\it t}({\it H})=[{\it f}+(1-{\it f})(1-exp(-{\it H/H*}))][{\it k}c$_{sat}$exp(-{\it k}$\phi${\it H/P}]. On bare bedrock ({\it H}=0) the average reaction rate is set by {\it f}{\it k}c$_{sat}$; it should then increase to a maximum at a regolith depth set by {\it H*} and then decline toward zero with a length scale set by {\it L}. This model explicitly addresses climatic control (through {\it P$_{eff}$} and{\it f}) and lithologic control (through the rate constant {\it k} and saturation, c$_{sat}$) on the rate of chemical alteration at the regolith/bedrock interface, and hence on regolith production rates where they are controlled by chemical processes.
H43F-04
Upscaling Schemes for Hydraulic Functions at the Landscape Scale
Soil hydraulic properties at relatively large scales (e.g., remote sensing footprints, large scale hydro-climate model grids) are critically important for land-atmosphere interaction and general circulation models or other large scale hydrologic applications. This study mainly investigates upscaling relationships for two commonly used soil hydraulic conductivity functions (i.e., the Gardner (G), and van Genuchten (VG) equations). We propose two conceptually new criteria to upscale the Gardner and van Genuchten models and to establish their upscaling relationships for steady state vertical flow processes at the landscape scale. The criteria are based on two important hydrologic processes and require that the upscaled hydraulic properties for both models produce ensemble surface reduced soil moisture and predict ensemble vertical water flux across land- atmosphere boundary. The process-based criteria focus on these two important processes because predicting flux rate across land-atmosphere boundary and soil surface moisture is usually a main concern in most practical soil-vegetation-atmosphere transfer models.
H43F-05
Sensitivity of tributaries to water-level fluctuations along the St-Lawrence corridor, Qu\e'bec, Canada
During the course of the last century, variations in the St-Lawrence water levels, caused by different uses of the river and the Great Lakes, have already had major impacts on riparian habitats and on a number of tributaries. Anticipated changes caused by climate change will only accentuate these impacts. At present, climate change scenarios forecast a decrease of the St-Lawrence water discharge by 20% over the next fifty years, which would correspond to a drop in water level between 0.5 and 1 metre at Montréal. Because the St- Lawrence corridor is in a region of lowlands, such fluctuations in water levels are expected to cause major adjustments in the morphology and longitudinal profiles of the tributaries through the erosion and incision of the river bed. These changes are important because ultimately, the sediment loads delivered to the St-Lawrence river are linked to the erosion processes occurring along its tributaries. However, given the important physical diversity of the tributaries, their sensitivity will differ, making it difficult to predict their individual response to environmental changes, especially given the paucity of data on the present state of the tributaries. Thus, our goal was to obtain data on the variability in morphosedimentology and dynamics of the tributaries in order to infer their sensitivity to fluctuations in baselevels. Bathymetric, hydraulic, and sedimentological surveys were conducted in 2004 and 2005 on five tributaries of the St-Lawrence: the Yamachiche, the St-Maurice, and the Batiscan rivers on the north shore, and the Richelieu and St-François rivers on the south shore. These tributaries cover a wide range of sizes and sedimentological characteristics, from cohesive clays to silts and gravel. Their hydrological regimes are also varied and are in some cases regulated, limiting the sediment capacity of certain tributaries. For example, the Yamachiche and St-François rivers have been under intense agricultural pressure, have fine sediments and have migrated significantly over the last fifty years, highlighting their instability and suggesting that changes in water levels may accentuate these erosion processes. On the other hand, the St-Maurice is highly regulated, has greater bank stability, due to its coarser sediment and intact riparian vegetation, and has remained fairly unmoved. This suggests that this river will respond more slowly to fluctuations in water levels. Investigating the current physical conditions of the tributaries will allow a better understanding of the long-term impacts of sustained periods of low water levels due to environmental change, and will help us develop a sensitivity index of the response of these rivers.
H43F-06
Embedded Scales of Self-Similar Turbulent Flow Structures in Gravel-bed Rivers
In spite of the fact that the fractal structure of the turbulent flow in the inertial sub-range is well known and accepted, there are no theoretical frameworks or empirical measurements that have shown the self-similar organisation of geophysical boundary layer flows at a larger scale. In gravel-bed rivers, the fundamental unit of the turbulent flow structure is the combination of interspersed high- and low-speed wedges which scale with flow depth. These elongated and narrow structures occupy the entire water column and their length scale is as much as six times the flow depth. Lasting only a few second, wedges are small with respect to the scale of the river reach and their relationships with known pulsations of the flow at longer time scales have not been investigated. Here we report that the small wedges lasting a few seconds are embedded within progressively larger self-similar flow structures. This finding was made possible by using a novel approach to analyse high frequency velocity measurements taken with electromagnetic velocimeters. Cumulated streamwise and vertical velocity fluctuations over long records show evidence of pulsations at progressively larger scales. The fractal structure is assessed with a box-counting of subregions of the Poincar\'{e} map of the cumulative signatures of the streamwise and vertical velocity components. Small wedges and large flow structures have the same velocity signatures mostly characterised by a negative correlation between the streamwise and vertical velocity components. These large pulsations may control the frequency and strength of the embedded smaller scale wedges. The larger structures may last several minutes with a mean streamwise velocity that can deviate by as much as 15% from the ensemble average. This finding challenges the idea that the flow is stationary over long periods of time even at a constant flow stage. These large flow structures may either be the result of the self-organization of wedges into larger features as wavelet analysis seems to suggest or of the effects of large scale morphological units of the river on the flow. These embedded scales of self-similar turbulent flow structures may have considerable implications on sediment transport and physical habitats in gravel-bed rivers.
H43F-07
Methods for Determining Streambank Critical Shear Stress and Erodibility: Implications for Erosion Rate Predictions
According to the US EPA, excess sediment is a significant cause of water quality impairment for rivers. The goal of this study was to compare different methods of determining two parameters used to estimate streambank erosion, soil critical shear stress (Ċc) and erodibility (kd), and to determine the impact of those differences on streambank erosion predictions. At twenty-five field sites, bank erosion tests were conducted using a submerged jet test device to measure critical shear stress and erodibility for streambanks composed of fine grained soils. Additionally, soil samples were collected and analyzed for particle size distribution, Atterberg limits, bulk density, and root density. Critical shear stress was estimated using Shield¡¦s diagram (SD), and three empirical equations based on percent clay (Pc), plasticity index (Iw), and median particle size (D50). Additionally, using a single set of Ċc values, the kd measured by the jet test was compared to two empirical kd relationships. Using these parameter values, streambank erosion rates were predicted for a local stream. Study results showed the measured Ċc estimates were as much as four orders of magnitude greater than the SD, Pc, and D50 estimates, indicating the SD and empirical methods underestimate Ċc for fine grained soils. The two empirical kd equations produced similar values. Both empirical parameter estimates were generally two orders of magnitude less than the kd values measured in situ. Erosion predictions followed the same trend as the kd data; the measured parameter values produced higher erosion predictions than the two empirical methods, although all three methods resulted in unrealistic erosion estimates. Field validation of these methods over a wide range of soil types is recommended to further develop methods of estimating kd and Ċc for streambank soils. Additionally, the effects of riparian vegetation on near bank shear stresses should be explored further.
H43F-08
Riparian Vegetation Influence on Stream Channel Dimensions: Key Driving Mechanisms and Their Timescales
Combined results from field-based investigations and flume experiments demonstrated key mechanisms driving channel widening following the reforestation of riparian zones in small streams. Riparian reforestation is a common occurrence either due to restoration efforts, intended to improve water quality, temperature regimes, and in-stream physical habitat or due to passive reforestation that is common when agricultural land uses decline. Previous studies have documented the influence of riparian vegetation on channel size, but driving mechanisms and the timescales at which they operate have not been evaluated. Field-based investigations were conducted in the Sleepers River basin in northeastern Vermont to revisit streams that were previously surveyed in the 1960s. We measured channel dimensions, large woody debris (LWD), and steam velocities in reaches with non-forested and forested riparian vegetation, in reaches currently in transition between vegetation types, and reaches with no change in riparian vegetation over the last 40 years. Flume experiments were performed with a 1:5 scale, fixed-bed model of a tributary to Sleepers River. Two types of riparian vegetation scenarios were simulated: 1) forested, with rigid, wooden dowels; and 2) non-forested, with synthetic grass carpeting. Three-dimensional velocities were measured during flume runs to determine turbulent kinetic energy (TKE) during overbank flows. Results showed that stream reaches with recently reforested vegetation have widened since the mid 1960s, but are not as wide as reaches with older riparian forests. LWD was more abundant in reaches with older riparian forests than in reaches with younger forests; however, scour around LWD did not appear to be a significant driving mechanism for channel widening. Velocity and TKE measurements from the prototype stream and the flume model indicate that TKE was significantly elevated in reforested reaches. Given that bed and bank erosion can be amplified in flows with high TKE, channel widening may be driven by increased turbulence generation in reforested reaches and may operate on a much shorter timescale than previously thought. Understanding the driving mechanisms and the timing of this channel widening phenomenon is important to predict geomorphic change due to riparian reforestation efforts, inform stream restoration designs, and evaluate the ultimate impact on aquatic ecosystems.
H43F-09
Sediment Sources and Storage in the Chesapeake Bay Watershed
Physically and chemically, sediment is a pollutant of concern in many waterbodies. In the Chesapeake Bay, sediment is having an adverse effect on the living resources and habitat of the Chesapeake Bay and its watershed. Identifying significant sources of watershed sediment is important in reducing sediment loads. In the Chesapeake Bay, several approaches were used to understand the sources, transport, and storage of watershed-derived sediment. From 1985 through 2001, the U.S. Geological Survey collected suspended sediment at 35 stations draining portions of the 103,000 km$^{2}$ Chesapeake Bay. Of the 35 sites, 4 of the 6 highest sediment yields were in the Conestoga River Basin, Pennsylvania, which drains to the Susquehanna River. In the Susquehanna River Basin (43,600 km$^{2}$), erosion rates were determined using atmospheric $^{10}$Be at 92 river outlets and confirmed that the highest rates of erosion were in the Conestoga River Basin. In three small watersheds draining to the Chesapeake Bay -- the Pocomoke River (157 km$^{2}$), Little Conestoga Creek (68.1 km$^{2}$), and Mattawoman Creek (92.8 km$^{2}$) -- sediment sources were identified using a sediment-source identification approach. In this approach, the sources of fine-grained suspended sediment in transport can be established by comparing physical and chemical properties of the suspended sediment to potential sources. In this study, suspended sediment ($<$ 0.062 mm) collected during storm runoff was compared to upland sediment sources (cropland, construction sites, and forest) and channel corridor sources (channel banks and bed) using radionuclides ($^{210}$Pb, $^{137}$Cs), stable isotopes ($^{13}$C, $^{15}$N), and total C, N, and P. Preliminary results are available for two of the three watersheds. In the Pocomoke River watershed, which drains the Coastal Plain physiographic province, ditch beds which were dug to drain cropland were a significant source of sediment. In the Little Conestoga Creek watershed, which drains the Piedmont physiographic province, river banks and cropland were significant sources.
H43F-10
Using Tripod Mounted Lidar to Determine Riverbank Erosion Rates Along the South River, Virginia.
To determine the rate of supply of contaminated sediments introduced into the river from erosion, a program of bank surveys is underway along South River using ground LIDAR combined with more traditional methods of historical aerial photo analysis. During a pilot study, we surveyed two sites in January 2006. Each site consisted of a reach several hundred meters in length. Both banks of the river at each site were surveyed simultaneously. Each survey was completed in about 4 hours. Setting up equipment and around 10 targets at each site consumed most of the survey time. Three GPS (Global Positioning System) units accurate to within 0.01 m were used to georeference the LIDAR data. Over 2x106 data points were obtained at each site. Root- mean-square survey errors were all less than 0.007 m. The data provide an exceptional database for quantifying bank morphology. The survey instrument was setup so that points are spaced at scales of 0.01 m at distances of 100 m. At this scale very detailed morphological features are imaged, including overhangs, small protrusions, individual roots, and details of the riparian vegetation (including trunks and branch canopy structure). These data can be used to determine bank morphology and roughness caused by vegetation and bank irregularities. Repeat surveys will determine detailed spatial patterns of erosion along each reach. Ground LIDAR technology, therefore, has the potential to provide new insights to better understand bank morphology and related processes of erosion and deposition.
H43F-11
Impacts of the Variability of Discharges on the Long-Term Geomorphic Evolution of a Watershed
Over long periods, watersheds respond to fluvial processes driven by a wide range of discharges. Low discharges have little ability to modify the land surface, but this contribution is magnified because such flow rates are common. High discharges have a much greater ability to modify the land surface but rarely occur. The importance of a given discharge to the erosion of a basin can be calculated by multiplying the discharge's frequency of occurrence and the erosion rate produced by the discharge. The discharge that contributes the most geomorphic work is called the geomorphic effective event (GEE). In this analysis, we aim to understand the behavior of the GEE through analytical and numerical approaches. A generic stream power model with a threshold is used to describe either the detachment or transport of sediment by flowing water. The exponential and log-normal distributions are used to describe the variability of discharges. The analytical results suggest that the return period of the GEE depends primarily on the threshold value when the exponent on discharge is less than two. Otherwise, it depends primarily on the exponent. The GEE usually cannot be substituted for the probability density function of discharge because it produces a different long-term erosion rate, thus we propose a new definition for the GEE, which produces the same long-term erosion rate. Furthermore, the return period of the GEE can vary spatially in a basin. For example, the return period can be different between locations where the fluvial process is dominant and sub-dominant if the threshold is non-zero. For a detachment-limited model, the return period of the GEE is different upstream and downstream of knickpoints, and for a transport-limited model, the return period is different along channel profiles even at steady state. Spatial variation in streamflow generation also produces spatial variations in the return period of the GEE.
H43F-12
Self-similarity in the Classification of Channel Network Planforms
The geometry of channel network planforms can vary significantly between regions depending on the local lithologic, tectonic, and climatic conditions. This tendency has led to the classification of channel networks as dendritic, parallel, rectangular, pinnate, trellis, etc. Classification of channel networks has been done using both qualitative and quantitative methods. Qualitative methods rely on visual inspection and can therefore produce subjective results. Available quantitative methods produce more objective and reproducible classifications, but they require calculation of a large number of network attributes and do not always agree with visual classifications. In this study, a simpler classification method is proposed based on deviations from self-similarity. It is well known that the planform geometry of dendritic networks exhibits approximate self- similarity. Thus, it is our hypothesis that several of the traditional classifications correspond to distinct deviations from self-similarity. To identify such deviations, three measures of channel networks are used: the accumulation of drainage area along channels, the irregularity of channel courses, and the angles formed by merging tributaries. The accumulation of drainage area captures the way in which tributaries add drainage area to a channel as one moves downstream. The irregularity of the channel course measures horizontal deviations of the channel course from a straight course. The junction angles consider the angles at which tributaries merge. In each case, the measures are rescaled in a way that allows deviations from self-similarity to be easily recognized. The three measurements were calculated for fifteen samples of five network classifications: dendritic, parallel, rectangular, pinnate, and trellis. The sample networks were previously classified in the literature or are located in regions where a given network type is known to predominate. In our analysis, dendritic networks tend to exhibit approximate self-similarity as expected from the literature. The remaining network types tend to exhibit deviations from self-similarity in some or all of the three measures. The forms of the deviations varied for the different network types. For example, parallel and pinnate networks appear to exhibit distinct types of self-affinity over a range of scales, while rectangular and trellis networks appear to deviate from any type of scaling invariance. These deviations can be used to quantitatively and objectively classify the planforms of drainage networks.
H43F-13
Order and disorder in starved sandy bedforms
Bed forms are among the most common features associated with the transport of sediment by air or by water. Understanding the fluid flow and sediment transport conditions that create specific bed form morphologies is a critical element of interpreting the stratigraphic record and is also a key element in predicting rates of sediment transport in active river channels. The characteristics and dynamics of bed forms that exist in uniform or well-mixed bed sediment conditions are well understood; less is known about bed forms that develop as fine sediment is transported over a bed of coarse immobile grains. We present data from large- scale laboratory experiments that describe bed form morphology, grain size, and distribution for suspended sediment transport over a coarse-grained immobile bed that is partially covered with fine sediment. Sand transported in suspension over coarse immobile grains resulted in a consistent sequence of bed configurations that included flow parallel sand stripes and flow transverse barchan dunes. The bed forms nearly buried the immobile grains while sand bed elevations between the bed forms were of order 0.5 times the immobile grain height (rb) or less. Suppressed entrainment for zs less than about 0.5 rb and accelerated entrainment for zs between 0.5 and 1.0 rb was observed in related experiments. This distribution of entrainment rates is consistent with previously reported distributions of shear stress over sand stripes and provides a mechanism for the organization of the bed into sand patches. Accelerated entrainment that occurs as immobile grains become exposed result in rapid scour of interstitial sand. Thus, an initially flat bed with sand elevation of order 1 rb quickly reorganizes into a bed of sand patches (stripes or barchans) and non- patches that have lower rates of sand entrainment. The transition from these isolated bedforms to a full sediment bed can be predicted using a modified bed state stability diagram. The axes of the stability diagram are the boundary Reynolds number and the volume of the mobile sediment, expressed by the ratio of zs to rb. In general, increasing the amount of available sediment results in a transition from sand stripes to barchans to fully developed dunes. However, because the near-bed hydraulic regime and degree of supply limitation are important in determining the bed state, other trajectories are possible. For example, if grain size and/or bed stress increases as supply decreases, one may observe a transition from sand stripes to barchans, a behavior not predictable with previous stability diagrams.