Erosion in the Himalayas

One day while having a beer with Dr. Doug Burbank (UC Santa Barbara), he asked me if I wanted to go to Nepal - I said "why not?"; this was my introduction to Himalaya-scale geomorphology. My part of the project is to analyze daily suspended sediment concentrations and chemical fluxes in rivers in the Annapurna region and this has provided me with a nearly inexhaustable supply of data. I was also in charge of training local Nepali to record water levels and collect and filter water samples. A number of different studies have come out of this work.

One of the goals of the project was to determine the spatial distribution of erosion rates and then try to understand the controls on that distribution. Precipitation is often invoked as a first-order control on erosion rates yet we have found that long-term erosion rates in the High Himalayas appear to be independent of climate such that erosion rates are spatially uniform despite an order of magnitude difference in precipitation. Modern-day erosion rates however do seem to be related to precipitation with rapid erosion on the wet and warm southern flank of the Annapurna region and slow erosion on the dry and cold northern region of the Himalayas and the Tibetan Plateau. We reconcile the spatially non-uniform modern rates with the spatially uniform long-term rates by suggesting that, in the north, glacial activity during the Ice Ages drives very rapid erosion that compensates for the slow erosion during the interglacials. This work is in press at Earth and Planetary Science Letters.

From the rainfall and suspended sediment data, I have identified rainfall thresholds that need to be overcome before landslides are triggered. Using a DEM of the field area, I have coupled a simple hillslope hydrology model with a slope stability analysis and have been able to reproduce the observed thresholds. This work was published in Geomorphology 63: 131-143.

Links between the shape of landforms and the climate are always assumed but unequivocal data supporting a connection have been difficult to find. Our field site in the Himalayas presents a great opportunity to look for a climatic signal in the shape of the mountains because of two key characteristics: 1) a steep precipitation gradient exists across the area, and 2) the hillslopes appear to be at their threshold angle, thus negating the influence of potential gradients in the uplift rate. An analysis of a 90m DEM, coupled with a 3-yr record of annual rainfall, suggests a relationship between mean slope angle and mean annual rainfall such that steeper slopes are found in the drier regions (see figure). This work was published in Geology 32: 629-632.

Chemical Weathering

Understanding the controls on chemical weathering of bedrock is key to closing the loop between climate, tectonics, and erosion. I have been analyzing chemical fluxes from 10 watersheds scattered across a steep climatic gradient in the High Himalayas to try to distinguish between climatic and topographic controls on the rates of chemical denudation. The results (soon to be published) indicate a positive relationship between erosion and chemical weathering.

A few years ago, a seminal paper by Raymo and Ruddiman suggested that the uplift of the Himalayas led to the drawdown of atmospheric CO2, thus triggering global cooling. They hypothesized that the exposure of fresh surfaces by rapid erosion accelerated the rates of chemical weathering and subsequent CO2 sequestration. This theory is based on the notion that increases in erosion rate are matched (or exceeded) by increases in weathering rates. To test this idea, I developed a numerical model that couples a power-law distribution of landslides with an equation that represents the common observation that weathering rates decline with time. The results of the model suggest that weathering rates are only proportional to the square root of erosion rate such that large increases in erosion rate are not matched by large in situ increases in weathering rate. This paper is in press at Earth and Planetary Science Letters.

To examine the role of soil properties in controlling the rate of chemical weathering, I created a hillslope flume. The flume holds a slab of synthetic bedrock composed of gypsum and table salt. Materials with different hydraulic conductivities are then poured over the bedrock and water is introduced from the top of the flume to examine how seepage velocity controls chemical weathering rates. I found that the greatest weathering occurs in materials with low hydraulic conductivity but that, once the pore fluid reaches saturation, any further decreases in conductivity are irrelevant. The experiments also demonstrated that the least amount of weathering occurs when there is no soil at all, thus confirming one of GK Gilbert's intuitions. Finally, I derived a numerical model for predicting weathering rates as a function of hydraulic conductivity. This work was published in Geology 34(12): 1065-1068.

Model results

Flume

The Mobilization of Debris Flows from Shallow Landslides

The prevailing theory explaining how rigid landslides transform into debris flows assumes that soils are loose and collapse during failure; as the soil collapses, pore pressures shoot up and debris flow behavior initiates. Simon Mudd (University of Edinburgh) and I tested this theory by analyzing soils from areas that were the sources of debris flows and slumps (ie, failures that did not liquefy). Surprisingly, we found that all of the soils dilated during shear, even the ones that produced debris flows (see figure). This observation, coupled with the observations of others who have suggested that most natural soils are dilational, prompted us to re-examine the standard model of debris flow initiation via a numerical model. We concluded that a soil's potential for liquefaction is independent of porosity (i.e., how loose it is) but sensitive to sand content because of its effect on hydraulic conductivity. This paper was published in Geomorphology 74: 207-218.

Fire and Debris Flows

The relationship between fire and accerated erosion has been well-documented. I took advantage of a proscribed burn near my field site to do a detailed study of post-fire erosional processes. I found that a generally unrecognized process, thin debris flows, was responsible for the greatest amount of sediment transported off the burned hillslopes. These little debris flows begin as slope failures that are only 1-2 cm thick and occur right above a hydrophobic layer. I used the infinite-slope stability analysis to predict the amount of rain that it would take to trigger these after a fire and the results agree very well with observations made by others. This work was published in Earth Surface Processes and Landforms 28(12): 1341-1348.

Soil Creep

With the goal of understanding how sediment is transported off of hillslopes, I've investigated a number of soil creep processes. These processes are generally characterized as being slope-dependent, meaning that the sediment flux is a function of slope. On the grassland hillslopes that I studied, the sediment flux by gopher bioturbation was the dominant process. While digging through the soil for plant roots, gophers move a considerable amount of soil. On the basis of simple measurements, I developed an empirically-derived sediment transport equation for gopher bioturbation. This work was published in Earth Surface Processes and Landforms 25: 1419-1428.

The main soil creep process on hillslopes in dry climates that have little ground-level vegetation is dry ravel and this process is particularly important immediately after fires. I derived a slope-dependent sediment transport equation, derived from the momentum equation, to describe this process. I tested the equation with a series of flume experiments and calibrated it with sediment traps set out in the field. This work was published in the Journal of Geophysical Research (PDF).

In the process of writing a review paper on the effects of bioturbation on soil processes, I derived slope-dependent sediment flux equations for tree-throw and root growth-and-decay. This paper was published in the Annual Review of Earth and Planetary Sciences (PDF).

Dry ravel flume

Flume surface

Sediment flux by various bioturbation processes as a function of slope

Shallow Landslides

During the 1997-98 El Nino, record rainfall triggered over 150 shallow landslides with a 10 sq km area near Santa Barbara, California. I studied these failures to understand the mechanics of failure as well and to estimate the amount of sediment evacuated from the hillslopes. Additiionally, many of the hillslopes at the site had been converted from coastal sage scrub to grasslands for grazing animals so I also examined the effect of vegetation conversion on slope stability and long-term changes in soil depths. This work was published in GSA Bulletin 114: 983-990.

Landslides in the sage

Landslide in the grass

Overland Flow

Sediment transport by overland flow is often considered to be a transport-limited process where soil particles are just sitting on the soil surface waiting to be washed downslope. In soils that are high in sticky (i.e., smectitic) clays, the flow itself may be unable detach sediment particles. In this case, detachment of soil particles by raindrop impact provides the material that is transported by overland flow. I did a number of rainfall simulation experiments on a variety of hillslopes with various slope angles, vegetation cover densities, and rainfall intensities to understand the process of sediment detachment by drop impact. With the data from these experiments, I developed an expression, called rainpower, that is derived from a basic consideration of the kinetic energy of raindrops. This work was published in Water Resources Research (PDF).

I collaborated with Dr. Noah Fierer (UC Santa Barbara) to investigate the transport of carbon and nitrogen, in both dissolved and particulate form, in overland flow. We specifically looked at the differences in nutrient transport between grassland hillslopes and sage-scrub hillslopes. We also did some trampling experiments with sawed-off cow hooves ("meat feet") to examine how hoof impact lowers the infiltration capacity of the soil. We found that conversion of sage-scrub to grasses for grazing animals increases the loss of carbon and nitrogen from the hillslopes. This work was published in the Journal of Environmental Quality (PDF).

rainfall simulator

Noah and his meat feet

Coupling the Hillslope and Fluvial Systems

The delivery of sediment to channels is an inherently stochastic process, largely dependent on the vagaries of climate. To explore how climate controls the magnitude and frequency sediment transport on hillslopes, I developed a computer model, driven by rainstorms and fires, that predicts the delivery of sediment. In the model, transport processes such as overland flow and shallow landsliding are explicitly represented. I used the model to examine the effects of climate-induced vegetation change on the amounts and tempo of sediment delivery in a steep, semi-arid landscape. The surprising conclusion from the model was that climatically-driven changes in vegetation, from a doubling of CO2, would have a greater effect on sediment delivery than just changes in the magnitude and frequency of climatic events. This work was published in Water Resources Research (PDF).

I collaborated with Dr. Noah Fierer and Dr. Oliver Chadwick (UC Santa Barbara) to adapt the sediment delivery model to predict the delivery of sediment-bound carbon, nitrogen, and phosphorus in a Mediterranean landscape. This involved detailed measurements of soil carbon, nitrogen, and phosphorus throughout our field site. The results from the model suggest that vegetation controls the tempo of nutrient delivery, which could have important implications for carbon sequestration and the eutrophication of water bodies. This work was published in JGR-Biogeosciences 10.1029/2005JG000032.

With funding from the National Center for Earth Surface Dynamics, Dan Hoffman (PBSJ) and I have conducted flume experiments at St. Anthony Falls Laboratory to investigate how the frequency and magnitude of sediment delivery can affect fluvial processes. The initial results suggest that sediment pulses reduce form drag by filling in the pools and that the decrease in form drag intensifies the rate of bedload transport. In other words, rivers may be able to adjust their transport capacity to accomodate the large, infrequent pulses that typify the sediment delivery regime in steep, mountainous landscapes.

Pre-pulse channel

Sediment pulse moving through.
Note that the pools have filled in

Tidal Channels

Despite highly sinuous planforms and evidence of active bank erosion, tidal channels have relatively slow rates of lateral migration. In an effort to solve this paradox, I closely monitored areas of deposition and erosion in a tidal channel in Marin County, California. This was done through the monthly measurement of 65 erosion pins placed along the length of the channel. Rates of erosion and deposition from the pin measurements were coupled with a bank stability analysis to develop a numerical model for predicting a rate of lateral migration that compared favorably to a rate estimated from a time series of aerial photographs. The paradox was solved through the realization that failed bank material persists for several years, thus protecting the bank from further erosion. This work was published in Estuaries 21(4B): 745-753.