cond-mat0502213/jgr.tex
1: \documentclass[jgrga]{agu2001}
2: 
3: \usepackage{amsmath}
4: \usepackage{amssymb}
5: \ifx\pdftexversion\undefined
6: \usepackage[dvips]{graphicx}
7: \else
8: \usepackage[pdftex]{graphicx}
9: \fi
10: \usepackage{psfrag}
11: \newcommand{\fig}{Fig.\hspace{0.05in}}
12: 
13: \authorrunninghead{LOBKOVSKY ET AL.}
14: \titlerunninghead{THRESHOLD PHENOMENA}
15: 
16: \authoraddr{Alexander E.~Lobkovsky, Department of Earth, Atmospheric
17:   and Planetary Sciences, \\ MIT, Cambridge, MA 02139, USA,
18:   \texttt{<leapfrog@mit.edu>}.}
19: 
20: 
21: \begin{document}
22: 
23: \articleid{1}{10}
24: \paperid{2004JF000172}
25: \journalid{109}{2004/12/24}
26: %\ccc{CODE}
27: 
28: \author{Alexander E.~Lobkovsky}
29: \affil{Department of Earth, Atmospheric and Planetary Sciences, \\
30:   Massachusetts Institute of Technology, Cambridge, MA 02139, USA}
31: 
32: \author{Bill Jensen and Arshad Kudrolli} 
33: \affil{Department of Physics, Clark University, Worcester, MA 01610,
34:   USA}
35: 
36: \author{Daniel H.~Rothman}
37: \affil{Department of Earth, Atmospheric and Planetary Sciences, \\
38:   Massachusetts Institute of Technology, Cambridge, MA 02139, USA}
39: 
40: \vspace{0.2in}
41: 
42: \title{Threshold phenomena in erosion driven by subsurface flow}
43: 
44: \begin{abstract}
45:   We study channelization and slope destabilization driven by
46:   subsurface (groundwater) flow in a laboratory experiment.  The
47:   pressure of the water entering the sandpile from below as well as
48:   the slope of the sandpile are varied.  We present quantitative
49:   understanding of the three modes of sediment mobilization in this
50:   experiment: surface erosion, fluidization, and slumping.  The onset
51:   of erosion is controlled not only by shear stresses caused by
52:   surfical flows, but also hydrodynamic stresses deriving from
53:   subsurface flows.  These additional forces require modification of
54:   the critical Shields criterion.  Whereas surface flows alone can
55:   mobilize surface grains only when the water flux exceeds a
56:   threshold, subsurface flows cause this threshold to vanish at slopes
57:   steeper than a critical angle substantially smaller than the maximum
58:   angle of stability.  Slopes above this critical angle are unstable
59:   to channelization by any amount of fluid reaching the surface.
60: \end{abstract}
61: 
62: \begin{article}
63: 
64: \section{Introduction}
65: \label{sec:intro}
66: 
67: Unlike water, a layer of sand will not flow unless its surface is
68: inclined beyond a characteristic angle, known as the maximum angle of
69: stability \citep{duran99:_sands}.  This simple fact translates into a
70: host of threshold phenomena wherever granular material is found.  Many
71: such phenomena play a crucial role in the erosion of Earth's surface,
72: and very likely manifest themselves in the richness of the patterns
73: exhibited by drainage networks.
74:  
75: Depending on geological, hydrological, and climatological properties,
76: erosion by water is mainly driven either by overland flow or
77: subsurface flow.  The former case occurs when the shear stress imposed
78: by a sheet flow exceeds a threshold
79: \citep{horton45,loewenherz91,dietrich92,montgomery92,howard94}.
80: Erosion in the latter case---known as seepage erosion, or
81: sapping---occurs when a subsurface flow emerges on the surface.  Here
82: the eroding stresses derive not only from the resulting sheet flow but
83: also the process of seepage itself \citep{iverson86:_ground}.  The
84: onset of erosion for both overland flow and seepage is
85: threshold-dependent, but the additional source of stress in the case
86: of seepage has the potential to create significantly different erosive
87: dynamics.
88:  
89: Here we study the seepage case.  Whereas the case of Horton overland
90: flow has been extensively studied
91: \citep{smith72,dietrich92,montgomery94,dunne95,izumi95,izumi00:_linear},
92: seepage erosion has received less attention.  \cite{dunne80,dunne90}
93: suggests that erosive stresses due to seepage are more widespread in
94: typical environments than commonly assumed.  He also provides a
95: detailed description of seepage erosion in the field, together with a
96: discussion of the various factors that influence its occurrence.
97: Another focus of attention has been the controversial possibility that
98: many erosive features on Mars appear to have resulted from subsurface
99: flows
100: \citep{baker82:_mars,higgins82,laity85:_sapping,baker90:_spring,aharonson02}.
101: Although the importance of seepage stresses in erosion have been
102: realized by \cite{howard88b} and \cite{howard88}, comprehensive
103: quantitative understanding is difficult to obtain.  The complexity
104: arises from the interdependent motion of the sediment and fluid---the
105: ``two-phase phenomenon'' \citep{yalin77:_mechanics}--- which, of
106: course, is common to {\em all} problems of erosion.
107:  
108: To further understand seepage erosion, we proceed from experiments
109: \citep{schumm87}.  Questions concerning the origin of ancient Martian
110: channels have motivated considerable experimental work in the past
111: \citep{kochel85,kochel86:_hawaii,howard88,kochel88}.  The process of
112: seepage erosion has also been studied as an example of drainage
113: network development \citep{gomez92}.  Our experiments, following those
114: of \cite{howard88,owoputi01:_erodability} and others, are designed to
115: enable us to construct a predictive, quantitative theory.
116: Consequently, they stress simplicity and completeness of information.
117: Although our setup greatly simplifies much of Nature's complexity, we
118: expect that at least some of our conclusions will improve general
119: understanding, and therefore be relevant to real, field-scale
120: problems.
121:  
122: A previous paper by \cite{schorghofer04:_spont} provided a qualitative
123: overview of the phenomenology in our experiment.  It described the
124: main modes of sediment mobilization: channelization, slumping, and
125: fluidization.  Here we provide quantitative understanding of the onset
126: and transitions between these modes.  
127: 
128: Our emphasis is on the threshold phenomena associated with the onset
129: of erosion, which we will ultimately characterize in the same way that
130: others \citep{duran99:_sands} have characterized the onset of dry
131: granular flow beyond the maximum angle of stability.  This involves a
132: construction of a generalized Shields criterion \citep{howard88} valid
133: in the presence of seepage through an inclined surface.  A major
134: conclusion is that the onset of erosion driven by seepage is
135: significantly different from the onset of erosion driven by overland
136: flow.  We find that there is a critical slope $s_\mathrm{c}$,
137: significantly smaller than the maximum angle of stability, above which
138: the threshold disappears.  Therefore any slope greater than
139: $s_\mathrm{c}$ is unstable to erosion if there is seepage through it.
140: This result is similar to well-known conclusions for the stability to
141: frictional failure of slopes with uniform seepage
142: \citep{taylor65:_fundamentals,iverson86:_ground,howard88}.  An
143: important distinction in our work, however, concerns the mode of
144: sediment mobilization and its local nature.  The existence of the
145: critical slope for seepage erosion may provide a useful quantitative
146: complement to the qualitative distinctions between seepage and
147: overland flow that have already been identified
148: \citep{laity85:_sapping}.
149: 
150: The remaining modes of sediment mobilization, fluidization and
151: slumping, are modeled using well established ideas
152: \citep{iverson86:_ground}.  The result of applying these ideas
153: together with the generalized Shields criterion provides a theoretical
154: prediction of the outcomes of the experiment, i.e., a phase diagram.
155: Agreement between theory and experiment is qualitative rather than
156: quantitative.  We nevertheless believe that our theoretical approach
157: is fundamentally sound and that better agreement would follow from
158: improved experimental procedures.
159: 
160: \section{Experiment}
161: \label{sec:expt}
162: 
163: \begin{figure}[htbp]
164:   \centering
165:   \includegraphics[width=3.3in]{2004JF000172R-F01_orig}
166:   \caption{Schematic of the experiment as detailed in
167:     \cite{schorghofer04:_spont} as well as the setup for computing the
168:     bulk and surface water flow.}
169:   \label{fig:expt}
170: \end{figure}
171: 
172: In our experimental setup, first introduced by
173: \cite{schorghofer04:_spont}, a pile of identical cohesionless glass
174: beads $0.5$mm in diameter is saturated with water and compacted to
175: create the densest possible packing.  It is then shaped into a
176: trapezoidal wedge inclined at an angle $\theta$ with slope $s =
177: \tan\theta$ as shown in \fig\ref{fig:expt}.  The downslope length of
178: the wedge is $L = 90$ cm, its width across the slope is $119$ cm, and
179: its height in the middle is approximately $11$ cm.  Water enters the
180: sandpile underneath through a fine metal mesh and exits at the lower
181: end of the pile through the same kind of mesh.  A constant head at the
182: inlet is maintained by keeping a constant water level $H$ in the
183: reservoir behind the sandbox with the help of an outflow pipe.  The
184: slope $s$ of the pile and the water level $H$ are the control
185: parameters of the experiment.  The degree of packing of the granular
186: pile is the variable most difficult to control.
187: 
188: Our particular method of feeding water into the sandpile, similar to
189: that of \cite{owoputi01:_erodability}, can be motivated in three ways.
190: The most important justification is the fact that the amount of water
191: flowing on the surface can be finely controlled in our geometry.  This
192: feature is essential in probing the onset of erosion.  Second, our
193: setup allows us to access heads $H$ larger than the height of the
194: pile, which therefore allows us to explore dynamic regimes unavailable
195: if water enters the pile through a mesh in the back.  Third, a similar
196: seepage water flow geometry can exist in the field wherever water
197: travels beneath an impermeable layer that terminates.
198: 
199: We have performed two types of experiments: steady and non-steady.
200: For a fixed water level and in absence of sediment motion, water flow
201: reaches steady state.  By monitoring the total water flux through the
202: system we estimate the time to reach steady state to be approximately
203: ten minutes.  To explore the onset of sediment motion, we raised the
204: water level $H$ in small increments, waiting each time for steady
205: state to be established.  Due to the particular shape of the bulk flow
206: in our experiment, surface flow exists over a finite region of the
207: surface.  The width of this seepage face and therefore the depth of
208: the surface flow can be tuned by changing $H$.  Because of the finite
209: extent of surface flow, its depth and therefore the viscous shear
210: stress reaches a maximum at a certain location.  Thus, by increasing
211: $H$ we can continuously tune the maximum shear stress experienced by
212: the surface grains.  The maximum shear stress reaches a critical value
213: for the onset of sediment motion in a certain location on the slope.
214: As we show below, we can compute where the maximum shear stress occurs
215: and thus can reliably detect the onset of sediment motion visually
216: because we focus our attention on this location.  Once sediment begins
217: to move, channels form almost immediately.  These channels grow in
218: length, width, and depth.  An example of the evolving channel network
219: is shown in \fig\ref{fig:channels}.  Depending on the slope, as the
220: channels deepen, the pile becomes unstable to fluidization or
221: slumping.  For slopes lower than approximately 0.05, the fluidization
222: threshold is reached before sediment is mobilized on the surface.
223: 
224: \begin{figure}[htbp]
225:   \centering
226:   \includegraphics[width=3.3in]{2004JF000172R-F02_orig}
227:   \caption{Examples of the channel network.  Top view of the slope of
228:     the eroding sandpile.  Branched channel heads migrate up the slope
229:     (upward) while the eroded sediment is deposited in the braided
230:     wash downslope.  The horizontal size of the image is approximately
231:     $1$m.  The slope of the pile is $s = 0.1$ and the water level $H =
232:     16.15$ cm.}
233:   \label{fig:channels}
234: \end{figure}
235: 
236: We also explored the non-steady evolution of the bulk and surface
237: water flow and resulting sediment motion by raising the water level
238: $H$ to some higher value from zero.  In this case one of three things
239: can happen.  The pile can be fluidized within a few seconds or fail by
240: slumping as shown in \fig\ref{fig:slump}.  If this does not occur, the
241: water emerges on the surface just above the inlet.  A sheet of water
242: then washes down the slope of the pile.  During this initial wash,
243: sediment is mobilized and incipient channels form.  These channels
244: grow during subsequent relaxation of the bulk water flow towards
245: steady state.  Because of the initial wash's erosive power, channels
246: are able to form and grow for lower water pressures than in steady
247: experiments.
248: 
249: \begin{figure}[htbp]
250:   \centering
251:   \includegraphics[width=3.3in]{2004JF000172R-F03_orig}
252:   \caption{Example of a slumping sandpile viewed at an approximately
253:     45$^\circ$ angle to the slope after the water flow has been
254:     stopped.  The width of the imaged region is approximately $1$m.
255:     Slumping happens along a convex upward arc which looks darker
256:     because it is deeper and therefore wetter.}
257:   \label{fig:slump}
258: \end{figure}
259: 
260: Outcomes of a large number of non-steady experiments and several
261: steady experiments for varying slope $s$ and the water level $H$ are
262: summarized in the phase diagram in \fig\ref{fig:phase}.  Each symbol
263: in the plot represents one experiment.  The sediment is either
264: immobile (stable seepage), or it is mobilized on the surface where
265: channels form (channelization) or in the bulk (slumping or
266: fluidization).  In several experiments, slumping or fluidization
267: happened after channels formed and grew.  In the following sections we
268: describe the computations that allow us to construct the theoretical
269: boundaries between the three different modes of sediment mobilization
270: in our experiment.
271: 
272: \begin{figure}[htbp]
273:   \centering
274:   \includegraphics[width=3.3in]{2004JF000172R-F04_orig}
275:   \caption{Experimental and theoretical phase diagram in the parameter
276:     space of slope and water level.  Experiments that yielded no
277:     erosion are denoted by $\times$; those that produced channels are
278:     indicated by $\bullet$; and those that produced fluidization
279:     and/or slumping within one hour of the beginning of the experiment
280:     are represented by $\triangledown$.  The straight line and
281:     gray-shaded curves are theoretical predictions for the boundaries
282:     separating the four regions indicated by their labels.  The
283:     thickness of the lines indicates uncertainty in the theory.  The
284:     boundary between the uneroded and channelized states is reasonably
285:     well approximated by our theory.  The theoretical boundaries for
286:     fluidization and slumping, however, appear to overestimate the
287:     critical water level, possibly as a result of inhomogeneities,
288:     dynamic changes in the sandpile's shape, or from the assumption of
289:     a steady state.}
290:   \label{fig:phase}
291: \end{figure}
292: 
293: \section{Calculation of the water flow}
294: \label{sec:flow}
295: 
296: Whereas steady-state flow can be readily characterized quantitatively,
297: non-steady flow characterization requires knowledge of the water-table
298: dynamics.  However, the theory of the water-table dynamics is less
299: well established than that of the flow through the bulk of a porous
300: medium.  Also, our steady-state experiments probe all aspects of
301: sediment dynamics.  We can therefore focus on the quantitative
302: characterization of the steady-state flow.
303: 
304: To study the onset of erosion quantitatively we need to be able to
305: establish a correspondence between the experimentally measurable
306: quantities such as the slope $s$, the water level $H$, the size of the
307: seepage face, and the water fluxes.  The seepage and surface fluxes
308: are the most difficult to measure.  In this section we set up their
309: computation.  The computation is designed to enable us to infer water
310: fluxes indirectly by measuring the size of the seepage face.  In the
311: following sections we will use this computation to quantify the onset
312: of erosion and to compute the slumping and liquefaction boundaries of
313: the channelization phase diagram shown in \fig\ref{fig:phase}.
314: 
315: \fig\ref{fig:expt} specifies the key quantities and coordinate systems
316: we use in computing the fluxes.  The flow profile is independent of
317: the $x$-coordinate across the slope of the sandpile except near the
318: side walls of the box.  We therefore treat the box as if it were
319: infinitely wide.  Flow is then two-dimensional and the specific
320: discharge vector $\mathbf{q}$ is in the $y$-$z$ plane.  We will use
321: two coordinate systems.  As shown in \fig\ref{fig:expt}, the $z'$
322: coordinate is measured vertically from the bottom of the box while the
323: $z$ coordinate is the normal distance away from the surface of the
324: pile.  The flow is governed by Darcy's law,
325: \begin{equation}
326:   \label{eq:darcy}
327:   \mathbf{q} = - K \nabla \psi,
328: \end{equation}
329: where $K$ is the scalar hydraulic conductivity, and 
330: \begin{equation}
331:   \label{eq:head}
332:   \psi = \pi + z'  
333: \end{equation}
334: is the total hydraulic head of the pore water.  Both $\mathbf{q}$ and
335: $K$ have units of velocity while the scaled pore pressure $\pi =
336: p/(\rho g)$ has units of length.  Here $\rho$ is the density of water
337: and $g$ is the magnitude of the acceleration of gravity.  We have
338: measured $K$ via a \textsf{U}-tube relaxation experiment.  To do so we
339: created a water level difference $\Delta h$ between the two arms of a
340: transparent \textsf{U}-shaped tube of width $\ell$ partially filled
341: with glass beads.  The rate of change of $\Delta h$ is given by
342: $K\Delta h/\ell$.  By measuring the rate of change of $\Delta h$ we
343: deduced the value of the hydraulic conductivity $K = 3.0 \pm 0.1$ mm/s
344: (\cite{schorghofer04:_spont}).  Hydraulic conductivity is sensitive to
345: the packing of the grains and is the variable most difficult to
346: control in our experiment.
347: 
348: Water incompressibility implies $\nabla \cdot \mathbf{q} = 0$,
349: therefore yielding Laplace's equation,
350: \begin{equation}
351:   \label{eq:laplace}
352:   \nabla^2 \pi = 0.
353: \end{equation}
354: 
355: To compute the pore pressure $\pi$, boundary conditions must be
356: specified.  The walls of the box are impenetrable.  Therefore the
357: discharge vector is parallel to the walls.  In other words, the flux
358: $q_\perp$ in the direction $\mathbf{\hat n}$ normal to the walls
359: vanishes.  Thus $q_\perp \equiv \mathbf{\hat n} \cdot \mathbf{q} = 0$.
360: Because the glass beads in our experiment are small, capillarity is
361: important.  When a tube filled with glass beads is lowered into a
362: reservoir of water, the porous bead-pack fully saturates in a region
363: that extends above the surface of the water by a capillary rise
364: $\pi_\mathrm{c}$.  We measured $\pi_\mathrm{c} = 25$ mm for our
365: material.  The capillary rise is a measure of the average radius of
366: the water menisci at the edge of the fully saturated zone.  The pore
367: pressure at the edge of the fully saturated zone is $-\pi_\mathrm{c}$
368: (without loss of generality we set the atmospheric pressure to zero).
369: Water can rise above the fully saturated zone through the smaller
370: pores and narrower throats.  Thus a partially saturated capillary
371: fringe exists above the fully saturated zone.  However, in this fringe
372: the water is effectively immobile since it is confined to the smaller
373: pores and narrower throats.
374: 
375: Since water flows only in the fully saturated zone, we define the
376: water table to be at its edge.  Thus, the pore pressure at the water
377: table is equal to the negative capillary rise $\pi = -\pi_\mathrm{c}$.
378: In steady state the discharge vector is parallel to the water table.
379: This extra condition allows us to determine the location of the water
380: table in steady state.
381: 
382: We neglect the pressure drop across the inlet mesh.  Therefore, the
383: pore pressure at the inlet mesh is $\pi = H - z'$.  The boundary
384: conditions at the surface of the sandpile and at the outlet mesh are
385: more subtle.  When no water seeps out, i.e., when the discharge vector
386: is parallel to the surface, the curvature of the water menisci between
387: grains can freely adjust so that the pressure $\pi$ can vary between
388: zero and $-\pi_\mathrm{c}$.  Therefore when $-\pi_\mathrm{c} < \pi <
389: 0$, no seepage occurs.  Otherwise, the pore pressure equals the
390: atmospheric pressure $\pi = 0$ (we neglect the pressure exerted by the
391: thin layer of water on the surface), and the discharge vector has a
392: component normal to the surface, i.e., there is either exfiltration or
393: infiltration.
394: 
395: \begin{figure}[htbp]
396:   \centering
397:   \includegraphics[width=3.3in]{2004JF000172R-F05_orig}
398:   \caption{(a) Steady state location of the water table (dashed line)
399:     obtained by solving the Laplace's equation as detailed in the
400:     text.  Note that the water table is almost entirely at the surface
401:     of the sandpile except a for small region at the bottom of the
402:     sandpile.  (b) Pore pressure at the water table and surface flow.
403:     Vertical scale for the pressure at the water table is given by the
404:     negative capillary rise $-\pi_\mathrm{c}$, or the pressure at the
405:     water table when it is below the surface of the pile.  Note that
406:     seepage occurs only where the pore pressure reaches atmospheric
407:     pressure.  Slope is $s = 0.1$, water level $H = 15$ cm.}
408:   \label{fig:table_pressure_overland}
409: \end{figure}
410: 
411: To obtain the steady state location of the water table, we guess its
412: position and solve Laplace's equation~\eqref{eq:laplace} with the $\pi
413: = -\pi_\mathrm{c}$ boundary condition on the water table.  We then
414: move the water table in the direction of the local discharge vector by
415: an amount proportional to its length.  Iteration of this procedure
416: converges to the steady-state position of the water table.  An example
417: is shown in Figure \ref{fig:table_pressure_overland}.
418: 
419: Once the steady flow pattern is known, we can calculate the overland
420: water flux $Q(y)$ by integrating the one-dimensional continuity
421: condition which states that the downslope derivative of the overland
422: flux is equal to the seepage flux:
423: \begin{equation}
424:   \label{eq:continuity}
425:   \frac{dQ}{dy} = q_z \equiv \mathbf{\hat z} \cdot \mathbf{q} = -K
426:     \left.\frac{\partial \psi}{\partial z}\right|_{y=0}.
427: \end{equation}
428: 
429: \section{Onset of erosion and channel incision}
430: \label{sec:incision}
431: 
432: In this section we assume, based on direct observation, that the onset
433: of channel incision coincides with the onset of erosion (i.e., we
434: never observed a homogeneously eroding state).  In other words, when
435: the overland water flux becomes strong enough to carry grains, the
436: flow of sediment becomes immediately unstable to perturbations
437: transverse to the downslope direction and incipient channels form
438: \citep{smith72}.  Using this assumption and the calculation of the
439: overland water flux we can deduce the threshold condition for the
440: onset of erosion.
441: 
442: It is universally assumed after \cite{shields36} that the hydrodynamic
443: stresses exerted on the sandpile by the fluid flowing on its surface
444: determine whether cohesionless granular material is entrained.  In the
445: limit of laminar flow, the dominant hydrodynamic stress is the viscous
446: shear stress $\tau_\mathrm{b}$.  Appropriately scaled this shear
447: stress is termed the Shields number \citep{yalin77:_mechanics},
448: defined by
449: \begin{equation}
450:   \label{eq:shields_conv}
451:   \tau^* = \frac{\tau_\mathrm{b}}{(\rho_\mathrm{s} - \rho)gd},
452: \end{equation}
453: where $\rho_\mathrm{s}$ is the density of the granular material, $d$
454: is the grain diameter ($d = 0.5$ mm in our experiment), and the
455: surface is not inclined.
456: 
457: The conventional Shields number~\eqref{eq:shields_conv} is the ratio
458: between the horizontal force exerted by the flow and vertical force
459: due to grain's weight.  To generalize the notion of the Shields number
460: to the situation with seepage through an inclined surface, we make two
461: changes in Eq.~\eqref{eq:shields_conv}.  We first add the tangential
462: component of the seepage force density $\mathbf{f} = -\rho g \nabla
463: \psi$ acting over a length $d$ to the numerator
464: of~\eqref{eq:shields_conv}.  The numerator thus becomes
465: $\tau_\mathrm{b} + d (\mathbf{\hat y}\cdot \mathbf{f})$.  Note that we
466: did not include the tangential component of the grain's weight to the
467: numerator.  Defined in this way, the generalized Shields number
468: measures the effect of the fluid: both the bulk as well as the surface
469: flows.
470: 
471: \begin{figure}[htbp]
472:   \centering
473:   \includegraphics[width=2.3in]{2004JF000172R-F06_orig}
474:   \caption{The denominator of the generalized Shields number is the
475:     projection onto the $z$-axis of the resultant of the grain's
476:     weight and the seepage force both scaled by $\pi d^2/6$.}
477:   \label{fig:mod_shields}
478: \end{figure}
479: 
480: Second, we replace the denominator of Eq.~\eqref{eq:shields_conv} by
481: the resultant (vectorial sum) of the seepage force on a grain and its
482: submerged weight, as shown in \fig\ref{fig:mod_shields}, both scaled
483: by $\pi d^2/6$ ($d^2$ to obtain stress as in the numerator and $\pi/6$
484: for agreement with the conventional Shields number), projected onto
485: the $z$-axis.  According to \cite{martin71:_seepage} the grains on the
486: surface of the bed experience a seepage force roughly half as large as
487: the grains several layers deep.  Consequently, we assume that the
488: seepage force is reduced by a factor of $a \approx 0.5$; therefore
489: \begin{equation}
490:   \label{eq:shields}
491:   \tau^* = \frac{\tau_\mathrm{b} - a\rho g d \, (\partial \psi/\partial
492:   y)}{(\rho_\mathrm{s} - \rho)gd\cos\theta + a\rho g d \,
493:     (\partial \psi/\partial z)},
494: \end{equation}
495: where $\theta$ is the inclination angle of the surface.  The
496: importance of the seepage stresses for the criterion for the onset of
497: erosion was previously realized by \cite{howard88}.  It can be shown
498: that their equation (10) expressing marginal stability of a surface
499: grain is equivalent to writing $\tau^* = \tan\phi$, the tangent of the
500: angle of internal friction.
501: 
502: The generalized Shields number Eq.~\eqref{eq:shields} is a measure of the
503: relative importance of the tangential and normal forces acting on a
504: grain at the surface of the sandpile.  Therefore, we expect $\tau^*$
505: to be a control parameter for erosion.  In other words, there exists a
506: critical Shields number $\tau^*_\mathrm{c}$, such that when $\tau^* <
507: \tau^*_\mathrm{c}$, surface grains are immobile, and when $\tau^* >
508: \tau^*_\mathrm{c}$, sediment is mobilized.  Note that
509: (\ref{eq:shields}) reduces to the classical definition of the Shields
510: number for a flat surface without seepage.  Also note that since we
511: did not include the tangential component of grain's weight, the
512: critical Shields number at the onset of sediment motion vanishes when
513: the inclination angle $\theta$ reaches the maximum angle of stability.
514: 
515: Although we obtain the seepage force density $\mathbf{f}$ as a result
516: of computing the pore-water pressure, to calculate the boundary shear
517: stress $\tau_\mathrm{b}$ we must estimate the thickness of the surface
518: water layer $h(y)$.  Since this thickness changes slowly in the
519: downslope direction, we can approximate the surface flow by the steady
520: flow of a uniform layer of viscous fluid.  Also, the surface water
521: flux is small enough for turbulence to be of no importance.  The
522: thickness $h$ of laminar surface flow for a given flux $Q$ is
523: \citep{landau87:_fluid}
524: \begin{equation}
525:   \label{eq:thickness}
526:     h = \left(
527:       \frac{3Q \, \eta}{\rho g \sin\theta}
528:     \right)^{1/3},
529: \end{equation}
530: where $\eta$ is the viscosity, while the viscous shear stress exerted
531: on the sandpile is
532: \begin{equation}
533:   \label{eq:tau}
534:   \tau_\mathrm{b} = \rho g h \sin\theta.
535: \end{equation}
536: The particle Reynolds number can then be calculated using the bottom
537: shear stress~\eqref{eq:tau} and shear velocity $v^* =
538: \sqrt{\tau_\mathrm{b}/\rho}$ as
539: \begin{equation}
540:   \label{eq:Re}
541:   \mathrm{Re} = \frac{v^* d}{\nu},
542: \end{equation}
543: where $\nu =\eta/\rho$ is the kinematic viscosity of water.  We
544: estimate that in our experiments, this particle Reynolds number varies
545: between 5 and 20 depending on the slope $s$ of the pile and the water
546: level $H$.  We verify this estimate of the Reynolds number by a direct
547: measurement of the thickness of the surface flow.  We find that this
548: thickness is several grain diameters.  This justifies the laminar flow
549: assumption used in obtaining Eq.~\eqref{eq:thickness}.
550: 
551: Using~\eqref{eq:tau}, the Shields number~\eqref{eq:shields} can now
552: be conveniently rewritten as
553: \begin{equation}
554:   \label{eq:shields-prefinal}
555:   \tau^* = \frac{(h/d)\sin\theta - a \, (\partial \psi/\partial
556:     y)}{(\rho_\mathrm{s}/\rho - 1) \cos\theta + a \, (\partial
557:     \psi/\partial z)}. 
558: \end{equation}
559: This expression can be further simplified by noting that $\pi = 0$
560: along the seepage face.  Therefore $\partial \psi/\partial y =
561: -\sin\theta$ at the surface wherever there is overland flow.  We
562: arrive at the final expression for the modified Shields number which
563: depends on the surface flow thickness $h$, the normal component
564: $\partial \psi/\partial z$ of the seepage force density at the
565: surface, and the seepage force reduction factor $a$
566: \begin{equation}
567:   \label{eq:shields-prefinal1}
568:   \tau^* = \frac{(h/d + a)\sin\theta}{(\rho_\mathrm{s}/\rho - 1)
569:     \cos\theta + a \, (\partial \psi/\partial z)}.
570: \end{equation}
571: 
572: In our geometry, both the surface and the seepage water fluxes reach a
573: maximum somewhere along the slope.  Therefore the Shields number has a
574: maximum value as well.  Below we calculate this maximum Shields number
575: in steady state for a given slope $s$ and water level $H$.
576: 
577: \section{Critical slope for the onset of seepage erosion}
578: \label{sec:crit-slope-seep}
579: 
580: \begin{figure}[htbp]
581:   \centering
582:   \includegraphics[width=3.3in]{2004JF000172R-F07_orig}
583:   \caption{Computed maximum Shields number versus the water level $H$
584:     for a pile of slope $s = 0.1$.  At $H = H_\mathrm{s}$ water first
585:     seeps through the surface and the Shields number jumps to a
586:     nonzero value.  Afterwards it increases rapidly as $(H -
587:     H_\mathrm{s})^{1/3}$ (solid line).  Inset: corresponding size of
588:     the computed seepage face.}
589:   \label{fig:shields_H}
590: \end{figure}
591: 
592: We now explore the consequences of seepage for the phenomenology of
593: the onset of erosion.  Because of the additional force on the surface
594: grains, seepage flow is more erosive than overland flow.  This notion
595: is reflected quantitatively in the generalized Shields number.  Let us
596: examine how the maximum Shields number varies with the water level $H$
597: in our experiment.  A representative graph of the maximum Shields
598: number versus the water level is shown in \fig\ref{fig:shields_H}.
599: Below a water level $H_\mathrm{s}(s)$ that is a function of the slope,
600: no water seeps out to the surface of the pile.  Even though the water
601: table may be at the surface, the pressure at the water table is below
602: atmospheric pressure and capillarity prevents seepage.  When $H =
603: H_\mathrm{s}$, i.e., exactly at the onset of seepage, the pressure
604: reaches $\pi = 0$ at some point on the surface.  Since the seepage
605: flux is still zero, $\partial \psi/\partial z = 0$ along the wet part
606: of the surface.  Therefore, just above the seepage onset, when the
607: water layer thickness and the seepage flux are both infinitesimally
608: small, the maximum Shields number is
609: \begin{equation}
610:   \label{eq:shields_onset}
611:   \tau^*_\mathrm{s}(s) = \frac{as}{\rho_\mathrm{s}/\rho - 1}.
612: \end{equation}
613: In contrast to overland flow, the consequence of seepage is that as
614: soon as the water emerges on the surface, the maximum Shields number
615: is some non-zero value which depends on the slope.  This also implies
616: that there exists a critical slope $s_\mathrm{c}$ such that
617: $\tau_\mathrm{s}(s_\mathrm{c})$ is equal to the critical Shields
618: number $\tau^*_\mathrm{c}$, i.e.,
619: \begin{equation}
620:   \label{eq:crit_slope}
621:   s_\mathrm{c} = \tau^*_\mathrm{c}(\rho_\mathrm{s}/\rho - 1)/a.
622: \end{equation}
623: For slopes greater than $s_\mathrm{c}$ seepage is always erosive.
624: Note that for low-density particles this critical slope can be
625: arbitrarily small.  The expression for the critical slope for seepage
626: erosion in Eq.~\eqref{eq:crit_slope} is analogous to well-known
627: formulas for stability of slopes to Coulomb failure due to uniform
628: seepage \citep{iverson86:_ground}.  Our result applies locally to the
629: point where non-uniform seepage first emerges on the surface.  In this
630: situation, the pile is generally stable to Coulomb failure and the
631: sediment is eroded only locally on the surface.
632: 
633: \begin{figure}[htbp]
634:   \centering
635:   \includegraphics[width=2.3in]{2004JF000172R-F08_orig}
636:   \caption{Schematic of the pore pressure $\pi$ as a function of the
637:     downslope coordinate $y$ at the onset of seepage $H =
638:     H_\mathrm{s}$ and just above.}
639:   \label{fig:increment}
640: \end{figure}
641: 
642: We now show that above $H_\mathrm{s}$, the maximum Shields number
643: increases rapidly as a $1/3$ power of the water level excess $(H -
644: H_\mathrm{s})^{1/3}$.  At the onset of seepage, i.e., when $H =
645: H_\mathrm{s}$, the pressure at the water table reaches atmospheric
646: pressure $\pi = 0$ at some point $P$ located at $z = 0$ and $y =
647: y_\mathrm{s}$ on the surface.  Even though the water table is at the
648: surface, there is no seepage anywhere, i.e., $\partial \psi/\partial z
649: = 0$.  Because the pressure is smooth, it can be approximated by a
650: quadratic function near this point so that $\pi \approx b z^2 - c(y -
651: y_\mathrm{s})^2$, where $b$ and $c$ are constants with appropriate
652: dimensions.  When the water level is raised by a small increment
653: $\delta H = H - H_\mathrm{s}$, the lowest order change in the head at
654: the water table is an increase of $\delta H$ with the exception of the
655: region where this increase would lead to a positive pressure.  As
656: illustrated in \fig\ref{fig:increment}, in this region the pore
657: pressure $\pi$ is set to $0$, and thus this region becomes the seepage
658: face.  The width of the seepage face $W$ scales like the square root
659: of $\delta H$, i.e., $W \sim \delta H^{1/2}$ as seen in the inset of
660: Fig.\hspace{0.05in}\ref{fig:shields_H}.  The seepage flux can be
661: estimated by noting that the hydraulic head is modified by an amount
662: $\delta H$ over a vertical region of order $W$.  Therefore we obtain
663: $\partial \psi/\partial z \sim \delta H /W \sim \delta H^{1/2}$.  The
664: total surface flux therefore scales like the product of the seepage
665: flux and the width $W$ of the seepage face, i.e., $Q \sim W \,\partial
666: \psi/\partial z \sim \delta H$.  The lowest order change in the
667: maximum Shields number is due to the change of the surface flow depth
668: $h \sim Q^{1/3} \sim \delta H^{1/3}$.  Thus as we claimed above, just
669: above the water level $H_\mathrm{s}$ for the onset of seepage,
670: \begin{equation}
671:   \label{eq:shields_seepage}
672:   \tau^* \approx \tau^*_\mathrm{s}(s) + \alpha(s) (H - H_\mathrm{s})^{1/3},
673: \end{equation}
674: where the constant $\alpha$ is a function of the slope.  Variation
675: with water level of the computed maximum Shields number shown in
676: \fig\ref{fig:shields_H} is consistent with
677: Eq.~\eqref{eq:shields_seepage}.
678: 
679: \section{Measurement of the critical Shields number}
680: \label{sec:meas-crit-shields}
681: 
682: In the previous sections we have detailed the way of calculating the
683: bulk and surface water fluxes in our experiment and the resulting
684: maximum generalized Shields number.  In this section we use this
685: calculation to examine the onset of the sediment flow and
686: channelization.  Our first goal is to measure the threshold or
687: critical Shields number required for the mobilization of sediment.  We
688: then use this measured value of the critical Shields number to predict
689: the outcome of steady-state experiments for various values of the
690: slope and the water level and thus compute the channelization boundary
691: in the phase diagram in \fig\ref{fig:phase}.
692: 
693: The actual maximum Shields number in the experiment differs from the
694: quantity calculated in Eq.~(\ref{eq:shields-prefinal1}).  In addition
695: to random errors in the measurements of the pile dimensions and water
696: level, there are several sources of systematic error.  For example,
697: the pressure drop across the inlet mesh results in a lower effective
698: hydraulic head.  Also, our measurement of the capillary rise
699: $\pi_\mathrm{c}$ is dependent on a visual estimate of the fully
700: saturated zone and thus can be a source of systematic error.  We
701: indeed find that the size of the seepage face calculated for a
702: particular water level $H$ is greater than measured in the experiment.
703: However, the size of the seepage face translates directly into the
704: surface water flux and therefore the maximum Shields number.
705: 
706: \begin{figure}[htbp]
707:   \centering
708:   \includegraphics[width=3.3in]{2004JF000172R-F09_orig}
709:   \caption{Computed maximum Shields number versus the seepage face
710:     size for three different slopes.  We use this calculation to infer
711:     the Shields number from the experimentally measured size of the
712:     seepage face.}
713:   \label{fig:shields_seep}
714: \end{figure}
715: 
716: The inset of Figure \ref{fig:shields_H} shows the typical dependence
717: of the size of the seepage face on the water level.  The variation of
718: the maximum Shields number with the size of the seepage face is shown
719: in \fig\ref{fig:shields_seep} for three different slopes.  We use
720: this computed correspondence between the size of the seepage face and
721: the maximum Shields number to infer the maximum Shields number in the
722: experiment by measuring the size of the seepage face.
723: 
724: To measure the critical Shields number we raise the water level $H$ in
725: increments of a few millimeters at a time.  Each time the water level
726: is increased, the seepage flow is allowed to reach a steady state.  In
727: each of these steady states we measure the seepage face size and infer
728: the corresponding maximum Shields number.  Eventually, sediment is
729: mobilized and we record the size of the seepage face and compute the
730: corresponding maximum Shields number.  This number is an upper bound
731: on the critical Shields number for our granular material at that
732: particular slope.  The lower bound on the critical Shields number is
733: obtained from the largest seepage face at which no sediment is moving
734: or sediment motion is only transient.  Averaging over several
735: experiments with slope $s = 0.1$ we estimate the critical Shields
736: number to be
737: \begin{equation}
738:   \label{eq:Shields_crit}
739:   \tau^*_\mathrm{c} = 0.085 \pm 0.005.
740: \end{equation}
741: It is not obvious that the generalized critical Shields number for the
742: onset of seepage driven erosion should coincide with the critical
743: Shields number for overland flow.  However our measured value of the
744: critical generalized Shields number is within the scatter of the
745: existing data for overland flow summarized in
746: \cite{buffington97:_systematic}.  Our measurement the critical
747: generalized Shields number is equivalent to measuring the angle of
748: internal friction due to the correspondence of our definition of
749: $\tau^*$ \eqref{eq:shields} and Howard and McLane's equation (10).
750: 
751: Deviations from flatness of the pile's surface result in the
752: fluctuations of the thickness of the surface water film.  As a result,
753: the maximum bottom shear stress in the experiment is systematically
754: greater than that calculated at a given size of the seepage face.
755: Thus the Shields number calculated for a particular size of the
756: seepage face is the lower bound on the actual Shields number in the
757: experiment.
758: 
759: In principle, the critical Shields number should vary with the slope
760: of the pile.  Evidence for this is the fact that at the maximum angle
761: of stability any additional forcing from the water flowing over the
762: bed mobilizes sediment.  Since it is reasonable to assume that the
763: critical Shields number is continuous and monotonic, we arrive at the
764: notion that it decreases monotonically with slope and vanishes at the
765: maximum angle of stability.  For small slopes the critical Shields
766: number is expected to decrease as the cosine of the inclination angle
767: since this is the lowest order change in the stabilizing effect of
768: gravity.  For most slopes in our experiments, $\cos\theta$ is within a
769: few percent of unity and thus we can ignore the variation of the
770: critical Shields number with slope.
771: 
772: This assumption allows us to predict the water level at which erosion
773: and therefore channelization should commence in our experiment.  In
774: Fig.\hspace{0.05in}\ref{fig:phase} a boundary is drawn between regions
775: where sediment is expected to mobilize and remain immobile.  To obtain
776: this line we computed for each slope the water level $H_\mathrm{e}(s)$
777: at which the Shields number is equal to the critical Shields number
778: \eqref{eq:Shields_crit}.  Below this water level, i.e., when $H <
779: H_\mathrm{e}(s)$, the maximum Shields number is below critical and
780: thus sediment is immobile.  Conversely, for $H > H_\mathrm{e}(s)$, the
781: maximum Shields number is above critical and thus sediment is
782: mobilized and channels form.  The channelization boundary is widened
783: because of the uncertainty in the critical Shields number.
784: 
785: Qualitative agreement of the channelization boundary with experiments
786: is perhaps due to the opposite action of two effects.  First,
787: channelization occurs for lower water levels in non-steady
788: experiments.  This happens because in non-steady experiments the
789: maximum Shields number overshoots its steady-state value.  The
790: overshoot is greatest for small slopes.  Second, a pressure drop
791: across the inlet mesh and the compacted region of sand close to it has
792: an opposite effect which increases the water level needed for
793: channelization.  These two effects, though small, could together
794: affect the accuracy of our predictions.  Since these effects act in
795: opposite ways, our the predictions of the calculated channelization
796: water level $H_\mathrm{e}(s)$ agree qualitatively with the
797: experiments.
798: 
799: \section{Fluidization and slumping}
800: \label{sec:fluid}
801: 
802: Having computed the channelization boundary in the phase diagram, we
803: now pursue a quantitative description of the other two modes of
804: sediment mobilization exhibited by our sandpile.  Higher water
805: pressures can cause the sandpile to fail in one of two ways.  First,
806: an upward seepage force can lift sand and result in a fluidization or
807: quicksand instability.  Second, the pile can become unstable to
808: slipping, slumping, or sliding.  Both failure mechanisms have been
809: discussed by a number of studies, e.g., those of
810: \cite{iverson86:_ground} or \cite{martin71:_seepage}.
811: 
812: \begin{figure}[htbp]
813:   \centering
814:   \includegraphics[width=1.5in]{2004JF000172R-F10_orig}
815:   \caption{To predict fluidization, we calculate the total seepage
816:     force on a slice of width $\Delta$ and height $\ell$ and compare
817:     it with its weight.}
818:   \label{fig:fluidize}
819: \end{figure}
820: 
821: Fluidization occurs when at some point $P$ in the sandpile the pore
822: pressure is larger than the total hydrostatic pressure due to the
823: weight of the sand and the water above $P$.  To see this we compute
824: the total seepage force acting on a slice of sand of width $\Delta$
825: between point $P$ and point $O$ on the surface of the pile directly
826: above $P$.  The vertical component of this force is (see
827: \fig\ref{fig:fluidize})
828: \begin{equation}
829:   \label{eq:vert_fluid}
830:   F_\mathrm{v} = \rho g \Delta (\psi_\mathrm{P} -
831:   \psi_\mathrm{O}) = \rho g \Delta (\pi_\mathrm{P} - \pi_\mathrm{O} -
832:   \ell),  
833: \end{equation}
834: where $\ell$ is the height of the slice.  When this force exceeds the
835: submerged weight $\Delta \ell (\rho_\mathrm{t} - \rho) g$ of the
836: slice, the slice is lifted and the bed is fluidized.  Here
837: $\rho_\mathrm{t}$ is the total density of the saturated sand, which
838: for our sand is approximately 2 g/cm$^3$.  Thus fluidization occurs
839: when there exists points $P$ and $O$ on the surface directly above $P$
840: such that
841: \begin{equation}
842:   \label{eq:fluidization}
843:   \pi_\mathrm{P} - \pi_\mathrm{O} > \ell \rho_\mathrm{t}/\rho.
844: \end{equation}
845: For uniform seepage this condition is equivalent to those in
846: \cite{iverson86:_ground} and \cite{martin71:_seepage}.
847: 
848: To construct the fluidization boundary in the phase diagram
849: (\fig\ref{fig:phase}), we find the water level $H_\mathrm{fluid}$
850: above which there exists at least one point in the pile for which
851: condition~\eqref{eq:fluidization} is satisfied.  Below this
852: fluidization water level this condition is not satisfied for any point
853: in the pile.
854: 
855: In addition to fluidization the sandpile can fail by slumping.  This
856: can happen in one of two ways.  Frictional failure can occur in the
857: bulk of the pile due to the seepage stresses.  Alternatively, surface
858: avalanching can occur.  To establish an upper bound on the water level
859: at which the sandpile slumps via either mechanism we use the criterion
860: developed by \cite{iverson86:_ground} for determining when a slope is
861: destabilized by uniform groundwater seepage.  Essentially it requires
862: calculating the vectorial sum of the seepage and gravity forces acting
863: on a small element of soil near the surface.  When the angle between
864: this total force and the downward normal to the surface, which we will
865: call the effective inclination angle, exceeds the maximum angle of
866: stability, the surface grains are destabilized.  We measured the
867: maximum angle of stability to be $\varphi \approx 23^\circ$ for dry
868: glass beads.  The slumping boundary in the phase diagram
869: (\fig\ref{fig:phase}) is constructed by computing the effective
870: inclination angle along the surface of the pile and noting the water
871: level $H_\mathrm{slump}$, at which the effective inclination angle
872: reaches the maximum angle of stability $\varphi$ at some point of the
873: surface.
874: 
875: Figure \ref{fig:phase} shows the critical water level at which
876: fluidization and slumping should occur according to the criteria
877: above.  Failure occurs at systematically lower water levels in the
878: experiment.  There are several effects which can account for this
879: difference.  First, any irregularities in the construction of the pile
880: such as voids or surface height fluctuations make the pile more
881: unstable to fluidization and slumping.  Second, we compute the
882: instability of an uneroded pile, whereas in most experiments, the pile
883: failed after erosion had changed the shape of the pile.  The decrease
884: of pile's height due to erosion increases the head gradient in the
885: bulk and thus makes the pile more prone to slumping and/or
886: fluidization.
887: 
888: \begin{figure}[htbp]
889:   \centering 
890:   \includegraphics[width=3.3in]{2004JF000172R-F11_orig}
891:   \caption{Effective inclination angle at the surface due to seepage
892:     forces for $s = 0.18$ and three different water levels.  For the
893:     highest water level, a region on the surface has an effective
894:     angle above the maximum angle of stability and thus the slope is
895:     unstable to slumping.  The inset shows the plateau value of the
896:     effective inclination angle as a function of slope.  When the
897:     plateau value reaches the maximum angle of stability, even a small
898:     amount of seepage destabilizes the pile to slumping.}
899:   \label{fig:slump_jump}
900: \end{figure}
901: 
902: At $s \approx 0.2$, a jump in the slumping water level is observed in
903: both the experiment and the model.  This jump is a purely geometric
904: effect.  Slumping occurs when, somewhere along the slope, the
905: effective inclination angle, which includes the effect of the seepage
906: force, exceeds the maximum angle of stability.  As shown in
907: \fig\ref{fig:slump_jump}, for slopes smaller than $0.2$, the
908: effective inclination angle is flat and develops a peak under the
909: water inlet as the water pressure is increased.  When the top of this
910: peak crosses the value of the maximum angle of stability, the pile
911: slumps.  When the slope exceeds $0.2$, however, the value of the
912: plateau in the effective inclination angle is above the maximum angle
913: of stability.  Therefore, for these slopes, the pile will be unstable
914: to slumping as soon as the water emerges on the surface.
915: 
916: 
917: \section{Conclusions}
918: \label{sec:conclusions}
919: 
920: This article reports on our progress in understanding seepage erosion
921: of a simple non-cohesive granular material in a laboratory-scale
922: experiment introduced in \cite{schorghofer04:_spont}.  Our ultimate
923: goal is to construct a quantitative predictive theory of the onset and
924: growth of the channel network observed in this experiment.  This goal
925: requires a complete sediment transport model as well as the
926: calculation of the relevant water fluxes.  Here we obtain the latter
927: and focus on the onset of erosion.
928: 
929: Prediction of the onset of erosion based on the generalized Shields
930: conjecture explains qualitatively the channelization boundary in the
931: experimental phase diagram.  By invoking well established simple ideas
932: we also roughly explain the fluidization and slumping boundaries in
933: the phase diagram.  Greater discrepancy with the experiment for these
934: boundaries indicates that better understanding of the
935: slumping/fluidization mechanisms particular to our experiment is
936: needed.
937: 
938: The central result of our exploration is the introduction of the
939: generalized Shields criterion for seepage erosion.  As a consequence
940: of seepage forces on the surface grains, the threshold for the onset
941: of erosion driven by seepage is slope dependent.  The threshold
942: disappears at a critical slope $s_\mathrm{c}$ determined by the
943: critical Shields number for overland flow and the density contrast
944: between the granular material and water.  In most cases this critical
945: slope is significantly smaller than the maximum angle of stability.
946: We find, therefore, that slopes above this critical slope are unstable
947: to any amount of seepage.  As a consequence, slopes that sustain
948: seepage must be inclined at an angle smaller than the critical angle
949: for seepage erosion.  This behavior contrasts strongly with the
950: threshold phenomena in erosion by overland flow, and therefore
951: provides a mechanistic foundation for distinguishing the two types of
952: erosion.
953: 
954: 
955: \section{Acknowledgements}
956: \label{sec:ack}
957: 
958: This work was supported by a DOE Grants DE-FG02-99ER15004 and
959: DE-FG02-02ER15367.
960: 
961: We thank N.~Sch\"orghofer, W.~Dietrich, and K.~Whipple for helpful
962: discussions.
963: 
964: \bibliographystyle{agu}
965: 
966: \begin{thebibliography}{34}
967: \providecommand{\natexlab}[1]{#1}
968: \expandafter\ifx\csname urlstyle\endcsname\relax
969:   \providecommand{\doi}[1]{doi:\discretionary{}{}{}#1}\else
970:   \providecommand{\doi}{doi:\discretionary{}{}{}\begingroup
971:   \urlstyle{rm}\Url}\fi
972: 
973: \bibitem[{\textit{Aharonson et~al.}(2002)\textit{Aharonson, Zuber, Rothman,
974:   Schorghofer, and Whipple}}]{aharonson02}
975: Aharonson, O., M.~T. Zuber, D.~H. Rothman, N.~Schorghofer, and K.~X. Whipple,
976:   Drainage basins and channel incision on {M}ars, \textit{Proceedings of the
977:   National Academy of Sciences USA}, \textit{99}, 1780--1783, 2002.
978: 
979: \bibitem[{\textit{Baker}(1982)}]{baker82:_mars}
980: Baker, V.~R., \textit{The channels of Mars}, University of Texas Press, Austin,
981:   1982.
982: 
983: \bibitem[{\textit{Baker}(1990)}]{baker90:_spring}
984: Baker, V.~R., Spring sapping and valley network development, in
985:   \textit{Groundwater Geomorphology: The role of subsurface water in
986:   earth-surface processes and landforms}, edited by C.~Higgins and D.~Coates,
987:   chap.~11, Geological Society of America, Boulder, Colorado, 1990.
988: 
989: \bibitem[{\textit{Buffington and Montgomery}(1997)}]{buffington97:_systematic}
990: Buffington, J.~M., and D.~R. Montgomery, A systematic analysis of eight decades
991:   of incipient motion studies, with special reference to gravel-bedded rivers,
992:   \textit{Water Resources Research}, \textit{33}(8), 1993--2029, 1997.
993: 
994: \bibitem[{\textit{Dietrich et~al.}(1992)\textit{Dietrich, Wilson, Montgomery,
995:   McKean, and Bauer}}]{dietrich92}
996: Dietrich, W.~E., C.~J. Wilson, D.~R. Montgomery, J.~McKean, and R.~Bauer,
997:   Erosion thresholds and land surface morphology, \textit{Geology},
998:   \textit{20}, 675--679, 1992.
999: 
1000: \bibitem[{\textit{Dunne}(1980)}]{dunne80}
1001: Dunne, T., Formation and controls of channel networks, \textit{Progress in
1002:   Physical Geography}, \textit{4}, 211--239, 1980.
1003: 
1004: \bibitem[{\textit{Dunne}(1990)}]{dunne90}
1005: Dunne, T., Hydrology, mechanics, and geomorphic implications of erosion by
1006:   subsurface flow, in \textit{Special Paper 252}, pp. 1--28, Geological Society
1007:   of America, Boulder, CO, 1990.
1008: 
1009: \bibitem[{\textit{Dunne et~al.}(1995)\textit{Dunne, Whipple, and
1010:   Aubry}}]{dunne95}
1011: Dunne, T., K.~X. Whipple, and B.~F. Aubry, Microtopography and hillslopes and
1012:   initiation of channels by {H}orton overland flow, in \textit{Natural and
1013:   Anthropogenic Influences in Fluvial Geomorphology}, pp. 27--44, American
1014:   Geophysical Union, 1995.
1015: 
1016: \bibitem[{\textit{Duran}(1999)}]{duran99:_sands}
1017: Duran, J., \textit{Sands, powders, and grains: an introduction to the physics
1018:   of granular materials}, Springer-Verlag, New York, 1999.
1019: 
1020: \bibitem[{\textit{Gomez and Mullen}(1992)}]{gomez92}
1021: Gomez, B., and V.~T. Mullen, An experimental study of sapped drainage network
1022:   development, \textit{Earth Surface Processes and Landforms}, \textit{17},
1023:   465--476, 1992.
1024: 
1025: \bibitem[{\textit{Higgins}(1982)}]{higgins82}
1026: Higgins, C.~G., Drainage systems developed by sapping on {E}arth and {M}ars,
1027:   \textit{Geology}, \textit{10}, 147--152, 1982.
1028: 
1029: \bibitem[{\textit{Horton}(1945)}]{horton45}
1030: Horton, R.~E., Erosional development of streams and their drainage basins;
1031:   hydrophysical approach to quatitative morphology., \textit{Geological Society
1032:   of America Bulletin}, \textit{56}, 275--370, 1945.
1033: 
1034: \bibitem[{\textit{Howard}(1988)}]{howard88b}
1035: Howard, A.~D., Groundwater sapping experiments and modelling, in
1036:   \textit{Sapping Features of the Colorado Plateau, a Comparative Planetary
1037:   Geology Field Guide}, edited by A.~D. Howard, R.~C. Kochel, and H.~E. Holt,
1038:   pp. 71--83, NASA Scientific and Technical Information Division, Washington,
1039:   D.C., 1988.
1040: 
1041: \bibitem[{\textit{Howard}(1994)}]{howard94}
1042: Howard, A.~D., A detachment-limited model of drainage basin evolution,
1043:   \textit{Water Resources Research}, \textit{30}, 2261--2285, 1994.
1044: 
1045: \bibitem[{\textit{Howard and McLane}(1988)}]{howard88}
1046: Howard, A.~D., and C.~F. McLane, Erosion of cohesionless sediment by
1047:   groundwater seepage, \textit{Water Resources Research}, \textit{24},
1048:   1659--1674, 1988.
1049: 
1050: \bibitem[{\textit{Iverson and Major}(1986)}]{iverson86:_ground}
1051: Iverson, R.~M., and J.~J. Major, Groundwater seepage vectors and the potential
1052:   for hillslope failure and debris flow mobilization, \textit{Water Res.
1053:   Research}, \textit{22}(11), 1543--8, 1986.
1054: 
1055: \bibitem[{\textit{Izumi and Parker}(1995)}]{izumi95}
1056: Izumi, N., and G.~Parker, Inception of channelization and drainage basin
1057:   formation: upstream-driven theory, \textit{J. Fluid Mech.}, \textit{283},
1058:   341--363, 1995.
1059: 
1060: \bibitem[{\textit{Izumi and Parker}(2000)}]{izumi00:_linear}
1061: Izumi, N., and G.~Parker, Linear stability analysis of channel inception:
1062:   downstream-driven theory, \textit{J. Fluid Mech.}, \textit{419}, 203--38,
1063:   2000.
1064: 
1065: \bibitem[{\textit{Kochel and Piper}(1986)}]{kochel86:_hawaii}
1066: Kochel, R.~C., and J.~F. Piper, Morphology of large valleys on {H}awaii -
1067:   evidence for groundwater sapping and comparisons with {M}artian valleys,
1068:   \textit{J. Geophys. Res.}, \textit{91}(B13), E175--92, 1986.
1069: 
1070: \bibitem[{\textit{Kochel et~al.}(1985)\textit{Kochel, Howard, and
1071:   McLane}}]{kochel85}
1072: Kochel, R.~C., A.~D. Howard, and C.~McLane, Channel networks developed by
1073:   groundwater sapping in fine-grained sediments: analogs to some {M}artian
1074:   valleys, in \textit{Models in Geomorphology}, edited by M.~J. Woldenberg, pp.
1075:   313--341, Allen and Unwin, Boston, 1985.
1076: 
1077: \bibitem[{\textit{Kochel et~al.}(1988)\textit{Kochel, Simmons, and
1078:   Piper}}]{kochel88}
1079: Kochel, R.~C., D.~W. Simmons, and J.~F. Piper, Groundwater sapping experiments
1080:   in weakly consolidated layered sediments: a qualitative summary, in
1081:   \textit{Sapping Features of the Colorado Plateau, a Comparative Planetary
1082:   Geology Field Guide}, edited by A.~D. Howard, R.~C. Kochel, and H.~E. Holt,
1083:   pp. 84--93, NASA Scientific and Technical Information Division, Washington,
1084:   D.C., 1988.
1085: 
1086: \bibitem[{\textit{Laity and Malin}(1985)}]{laity85:_sapping}
1087: Laity, J.~E., and M.~C. Malin, Sapping processes and the development of
1088:   theater-headed valley networks on the {C}olorado plateau, \textit{Geol. Soc.
1089:   Am. Bull.}, \textit{96}(2), 203--17, 1985.
1090: 
1091: \bibitem[{\textit{Landau and Lifshitz}(1987)}]{landau87:_fluid}
1092: Landau, L.~D., and E.~M. Lifshitz, \textit{Fluid mechanics}, Pergamon Press,
1093:   Oxford, England; New York, 1987.
1094: 
1095: \bibitem[{\textit{Loewenherz}(1991)}]{loewenherz91}
1096: Loewenherz, D.~S., Stability and the initiation of channelized surface
1097:   drainage: a reassessment of the short wavelength limit, \textit{Journal of
1098:   Geophysical Research}, \textit{96}, 8453--8464, 1991.
1099: 
1100: \bibitem[{\textit{Martin and Aral}(1971)}]{martin71:_seepage}
1101: Martin, C.~S., and M.~M. Aral, Seepage force on interfacial bed particles,
1102:   \textit{J. Hydrol. Div.}, \textit{HY 7}, 1081--1100, 1971.
1103: 
1104: \bibitem[{\textit{Montgomery and Dietrich}(1992)}]{montgomery92}
1105: Montgomery, D.~R., and W.~E. Dietrich, Channel initiation and the problem of
1106:   landscape scale, \textit{Science}, \textit{255}, 826--830, 1992.
1107: 
1108: \bibitem[{\textit{Montgomery and Dietrich}(1994)}]{montgomery94}
1109: Montgomery, D.~R., and W.~E. Dietrich, A physically based model for the
1110:   topographic control on shallow landsliding, \textit{Water Resources
1111:   Research}, \textit{30}, 1153--1171, 1994.
1112: 
1113: \bibitem[{\textit{Owoputi and Stolte}(2001)}]{owoputi01:_erodability}
1114: Owoputi, L.~O., and W.~J. Stolte, The role of seepage in erodability,
1115:   \textit{Hydrol. Process.}, \textit{15}, 13--22, 2001.
1116: 
1117: \bibitem[{\textit{Sch\"orghofer et~al.}(2004)\textit{Sch\"orghofer, Jensen,
1118:   Kudrolli, and Rothman}}]{schorghofer04:_spont}
1119: Sch\"orghofer, N., B.~Jensen, A.~Kudrolli, and D.~H. Rothman, Spontaneous
1120:   channelization in permeable ground: theory, experiment, and observation,
1121:   \textit{J. Fluid Mech.}, \textit{503}, 357--74, 2004.
1122: 
1123: \bibitem[{\textit{Schumm et~al.}(1987)\textit{Schumm, Mosley, and
1124:   Weaver}}]{schumm87}
1125: Schumm, S.~A., M.~P. Mosley, and W.~E. Weaver, \textit{Experimental Fluvial
1126:   Geomorphology}, Wiley, New York, 1987.
1127: 
1128: \bibitem[{\textit{Shields}(1936)}]{shields36}
1129: Shields, A., \textit{Anwendung der {\"A}hnlichkeitsmechanik und der
1130:   {T}urbulenzforschung auf die {G}eschiebebewegung}, Heft 26, Mitteilung der
1131:   Preussischen Versuchsanstalt f\"ur Wasserbau und Schiffbau, Berlin, Germany,
1132:   (In German), 1936.
1133: 
1134: \bibitem[{\textit{Smith and Bretherton}(1972)}]{smith72}
1135: Smith, T.~R., and F.~P. Bretherton, Stability and the conservation of mass in
1136:   drainage basin evolution, \textit{Water Resources Research}, \textit{3},
1137:   1506--28, 1972.
1138: 
1139: \bibitem[{\textit{Taylor}(1965)}]{taylor65:_fundamentals}
1140: Taylor, D.~W., \textit{Fundamental of soil mechanics}, chap.~16, fifteenth ed.,
1141:   John Wiley \& Sons, Inc., New York, 1965.
1142: 
1143: \bibitem[{\textit{Yalin}(1977)}]{yalin77:_mechanics}
1144: Yalin, M.~S., \textit{Mechanics of sediment transport}, 2nd ed., Pergamon
1145:   Press, Oxford, New York, 1977.
1146: 
1147: \end{thebibliography}
1148: 
1149: \end{article}
1150: 
1151: \end{document}
1152: 
1153: %%% Local Variables: 
1154: %%% mode: latex
1155: %%% TeX-master: t
1156: %%% End: 
1157: