EARTH SURFACE PROCESSES AND LANDFORMS

Earth Surf. Process. Landforms (2020)

© 2020 John Wiley & Sons, Ltd.
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/esp.4843

The role of soil pipe and pipeflow in headcut

migration processes in loessic soils
Ximeng Xu,1,2,3 Glenn V. Wilson,3 Fenli Zheng1* and Qiuhong Tang2
1
State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Northwest
A&F University, Yangling 712100, Shaanxi China
2
Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources
Research, Chinese Academy of Sciences, Beijing 100101, China
3
USDA-ARS National Sedimentation Laboratory, Oxford, MS 38655, USA

Received 24 September 2019; Revised 18 February 2020; Accepted 19 February 2020

*Correspondence to: Fenli Zheng, Institute of Soil and Water Conservation, Northwest A&F University, No. 26 Xi’nong Road, Yangling, Shaanxi 712100, China. E-mail:
flzh@ms.iswc.ac.cn

ABSTRACT: Headcut formation and migration was sometimes mistaken as the result of overland flow, without realizing that the
headcut was formed and being influenced by flow through soil pipes into the headcut. To determine the effects of the soil pipe
and flow through a soil pipe on headcut migration in loessic soils, laboratory experiments were conducted under free drainage con-
ditions and conditions of a perched water table. Soil beds with a 3-cm deep initial headcut were formed in a flume with a 1.5-cm
diameter soil pipe 15 cm below the bed surface. Overland flow and flow into the soil pipe was applied at a constant rate of
68 and 1 l min 1 at the upper end of the flume. The headcut migration rate and sediment concentrations in both surface (channel)
and subsurface (soil pipe) flows were measured with time. The typical response was the formation of a headcut that extended in
depth until an equilibrium scour hole was established, at which time the headcut migrated upslope. Pipeflow caused erosion inside
the soil pipe at the same time that runoff was causing a scour hole to deepen and migrate. When the headcut extended to the depth of
the soil pipe, surface runoff entering the scour hole interacted with flow from the soil pipe also entering the scour hole. This interac-
tion dramatically altered the headcut processes and greatly accelerated the headcut migration rates and sediment concentrations.
Conditions in which a perched water table provided seepage into the soil pipe, in addition to pipeflow, increased the sediment con-
centration by 42% and the headcut migration rate by 47% compared with pipeflow under free drainage conditions. The time that
overland flow converged with subsurface flow was advanced under seepage conditions by 2.3 and 5.0 min compared with free
drainage conditions. This study confirmed that pipeflow dramatically accelerates headcut migration, especially under conditions
of shallow perched water tables, and highlights the importance of understanding these processes in headcut migration processes.
© 2020 John Wiley & Sons, Ltd.

KEYWORDS: subsurface erosion; seepage; concentrated flow; piping; gully erosion; soil erosion

Introduction ephemeral gully erosion is not well understood (Wilson, 2011).

Soil erosion by water degrades soil and the transport of sedi-
ments into surface waters causes many environmental and so-
cial problems (Poesen, 2018). Water erosion by overland flow
has been researched widely in many aspects, like splash and
sheet erosion (Dunne et al., 2010), rill erosion (Govers
et al., 2007; Sun et al., 2013), gully erosion (Poesen et al., 2003;
Castillo and Gómez, 2016), and so on, while soil erosion and
sediment transport processes associated with subsurface flow
are not as well understood (Wilson et al., 2018). Swanson
et al. (1989) credits Leopold and Miller (1956) as being the first
to ‘recognize subsurface erosion (piping) as a mechanism for
headward extension of existing gullies’. Erosion by piping has
been found to be a significant sediment source worldwide for
all soil types and hydrogeologic conditions, and a critical pro-
cess of landscape evolution dynamics and natural hazard in-
ducement (Bernatek-Jakiel and Poesen, 2018), yet its role in
Rodzik et al. (2009) qualitatively associated piping with gully
erosion for snow melt conditions based on field observations.
However, quantitative assessment of the role of piping on gully
headcut migration under different hydrologic conditions is vir-
tually non-existent.
Subsurface flow contributes to soil erosion by way of water
content/pressure effects on soil properties, seepage contribu-
tions to surface and subsurface preferential flows, as well as
seepage forces on soil particle detachment (Fox and
Felice, 2014; Fox et al., 2014; Wilson et al., 2018). As reviewed
by Wilson and Fox (2013), there are many different subsurface
erosion processes that have been associated with the term ‘pip-
ing’. This paper will apply the distinction made by Wil-
son (2011) and others (Jones, 1987; Chappell, 2010) that
piping is internal erosion as a result of flow through a discrete
conduit or linear voids as described by Bernatek-Jakiel and
Poesen (2018). Pipeflow is such a rapid and non-uniform form
 X. XU ET AL.
of preferential flow (Bryan and Jones, 1997) that shear stress ex-
ceeds the critical shear stress in soil pipes. Soil particles and ag-
gregates are readily detached and transported, which enlarges
the soil pipe and accelerates the pipeflow and internal erosion
in a self-propagating process (Wilson et al., 2016). This process
will ultimately result in collapse of the pipe roof, thereby
forming sinkholes and gullies (Faulkner et al., 2004;
Wilson et al., 2008; Wilson, 2009, 2011; Faulkner, 2013;
Bernatek-Jakiel et al., 2017). Typical soil pipe collapse features
like sinkholes, flute holes and discontinuous gully windows are
typically precursors to mature gully formation as observed
around the world (Sayer et al., 2006; Faulkner, 2007; Faulkner
et al., 2008; Zhu, 2012; Wilson et al., 2015; Nichols et al., 2016;
Bernatek-Jakiel and Poesen, 2018; Bernatek-Jakiel and
Wrońska-Wałach, 2018). Pipeflow and related seepage pro-
cesses can also be closely associated with gully sidewall failure
and headcut migration processes (Fox and Wilson, 2010;
Tebebu et al., 2010; Frankl et al., 2012, 2016; Nichols
et al., 2016; Bernatek-Jakiel and Poesen, 2018). Since soil pipes
are not visible from the surface until a depression or pipe col-
lapse occurs, pipe erosion has often been overlooked in gully
erosion processes (Faulkner, 2006; Wilson et al., 2018).
Headcut migration is one of the most important processes in
linear erosion features like rills, ephemeral gullies and perma-
nent gullies caused by concentrated overland flow (Meyer
et al., 1975; Römkens et al., 2001). Headcut occurrence and
migration in concentrated flow channels results in significant
increases in sediment yield (Brunton and Bryan, 2000; Xu
et al., 2017). The effects of factors like slope gradient (Ben-
nett, 1999), initial step height (Bennett and Casalí, 2001), soil
stratification (Gordon et al., 2007), soil texture (Wells
et al., 2009a), tail water height and pore water pressures (Wells
et al., 2009b) on headcut migration rates have been examined
in experimental conditions. These experiments determined
headcut migration rates under various initial conditions, from
which analytical models were developed to describe this pro-
cess (Alonso et al., 2002; Wells et al., 2010). Jones (1987) pro-
posed that piping may ‘augment channel extension’ as soil
pipeflow serves as a source, sink and/or link for rills, gullies
and other channel network features across landscapes. How-
ever, interactions of headcuts with pipeflow and perched water
table conditions have yet to be fully considered. Such interac-
tions require further research to simulate gully erosion by pip-
ing, as observed in field conditions around the world
(Crouch, 1983; Crouch et al., 1986; Sayer et al., 2006;
Verachtert et al., 2011; Wilson et al., 2015; Nichols et al., 2016;
Bernatek-Jakiel and Poesen, 2018).
While often overlooked, it is not uncommon for gully
headcuts to have a soil pipe at the head, especially near the
bottom of the scour hole (Wilson et al., 2006; Nichols
et al., 2016). The question can be asked whether the soil pipe
formed the headcut or the deepening of the scour hole revealed
the soil pipe in the profile below the gully. Crouch (1983) mon-
itored the headcut migration processes in an arable catchment
with duplex soils for 3 years, and found that three gullies
retreated due to tunnel erosion with scour hole and back trickle
erosion features. These results also showed that the gully head
retreated rapidly when tunnel development was active, with a
rate of 2.5 m year 1, while the headcut migration rate without
tunnel structure was less than 0.5 m year 1. Swanson
et al. (1989) reported that subsurface erosion (piping and col-
lapse) expanded the gully network by 80 m, which doubled
the gully length over a 6-year period, and delivered an order
of magnitude higher sediment load than that of the surface
drainage net. Nichols et al. (2016) noted that piping was an im-
portant erosion process contributing to gully headcut migration
and sidewall collapse within Walnut Gulch, AZ. Their in-situ
© 2020 John Wiley & Sons, Ltd.
photogrammetric observations showed that soil pipe collapse
advanced the gully headcut 7.4 m in a single event and
accounted for 51% of the total migration length over 10 years
of monitoring. Bernatek-Jakiel and Poesen (2018) noted in their
review that piping can result in gully headcut migrations as
high as 10.5 m in a single event, with average annual migration
rates ranging from 1 to 5 m year 1.
Surface runoff can, and often does, enter directly into soil
pipes through flute holes at the surface (Jones, 1987). However,
a surface source is not necessary for pipeflow to be significant
enough to cause internal erosion. Seepage into the soil pipes
from the surrounding soil can provide sufficient source for
sustained and rapid pipeflow (Vannoppen et al., 2017; Wilson
et al., 2017). Vannoppen et al. (2017) monitored pipeflow dis-
charges from one pipe outlet in the Flemish Ardennes in
Belgium, and found that pipeflow consists of a combination
of rainfall and groundwater but is dominated by old-water
seepage into pipes through hydrochemical analyses of rainfall,
pipeflow and groundwater. Faulkner (2006) and Hardie
et al. (2012) reviewed the correlation of piping and lateral
flows, respectively, to duplex soils (i.e. soils with a hydrauli-
cally contrasting layer). The reason for this correlation is the
propensity for such layers to temporarily perch infiltrating water
and, thereby, foster lateral flow. Seepage could also contribute
to the headcut migration, which had been assumed to be due to
surface flow erosion alone (Huang and Laflen, 1996). Seepage
contributes to headcut migration by providing flow into soil
pipes and fostering self-organization of preferential flow
paths/zones that establish hydrologic connectivity across the
landscape (Sidle et al., 2001; Wilson et al., 2016, 2017).
Even though soil pipeflow and the resulting pipe collapse
due to internal erosion has been widely observed and occa-
sionally associated with headcut migration, understanding of
the mechanisms of these processes and the interactions of soil
pipeflow with convergent flow in the channel on headcut mi-
gration are still lacking. Thus, controlled laboratory experi-
ments that mimic field observations of pipe networks and
perched water conditions are needed. Laboratory procedures
for investigating headcut migration by channelized flow are
well established (Bennett, 1999; Bennett and Casalí, 2001;
Wells et al., 2009b), as are procedures for pipeflow, but no
study to date has combined these processes in laboratory ex-
periments. The objectives of this study were to develop such
laboratory procedures to mimic observations of gully headcuts
with soil pipes observed at Goodwin Creek Experimental Wa-
tershed covered by loess soils (Wilson et al., 2015) and to use
these procedures quantifying the interactions between the con-
vergent surface flow and pipeflow and their controls on
headcut migration dynamics. These interactions were tested
for conditions of upslope inflow into the channel and soil pipe
under seepage and free drainage conditions.
Materials and Methods
Experimental setup and soil bed preparation
Laboratory experiments were conducted in the USDA-ARS Na-
tional Sedimentation Laboratory, Oxford, MS. All experiments
were conducted in a 5.5-m long, 0.165-m wide, non-
recirculating, slope-adjustable flume (Figure 1). The soil used
in this study was Providence silt loam (fine-silty, mixed, active,
thermic Oxyaquic Fragiudalfs), collected from the topsoil of the
Holly Springs Experiment Station in north Mississippi. The
Providence soil is in the same taxonomic series as the Loring,
which are representative of loessic soils with a fragipan subsur-
face that are known to form gullies by pipe collapse (Wilson
Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES

Figure 1. Experiment facilities: (a) photograph of the slope-adjustable flume with the soil bed and cameras above and on the flume side for tracking
the headcut migration; (b) photograph of the free drainage (drainage ports open) or seepage (Mariotte device connected to perforated pipes) system
along with surface and subsurface collection system. [Colour figure can be viewed at wileyonlinelibrary.com]

et al., 2015). The soil contains 15% sand, 69% silt and 16%
clay. The soil was first air-dried and mechanically crushed to
pass a 4-mm sieve for removing large aggregates and foreign
materials. Before packing the soil bed, a drainage layer of fine
sand was placed at the bottom of the flume. Perforated pipes
were mounted at the bottom of the flume, covered by highly
permeable fabric, and a 6-cm thick layer of fine sand was
packed above the fabric and then levelled by vibration to cre-
ate a drainage layer. The perforated pipes were either open to
the atmosphere for free drainage conditions or connected to a
Mariotte bottle that imposed a perched water table on the soil
bed to the depth of the preformed step (Figure 1b). Soil was
packed into the soil bed above the drainage layer in 2.5-cm lifts
by measuring the mass of soil needed (which is around 15 kg,
accounting for the moisture content) to achieve the prescribed
bulk density of 1.30–1.35 g cm 3. The soil was evenly spread
along the flume soil bed section and packed into place at the
desired bulk density by the vibration technique employed by
Wells et al. (2010). The vibration was conducted with a 2.75-
m long, 0.165-m wide, 1.6-cm thick aluminium plate. On the
top of the plate, three vibrating motors were mounted, equally
spaced on the surface of the plate. Each layer of the soil bed
was vibrated for 4 min.
When packing the soil bed to a depth of 0.22 m, an alumin-
ium headcut-forming plate was installed 2.3 m downstream of
the upslope inflow, rigid floor on the immediate surface layer
(i.e. final lift packed) (Wells et al., 2010). After the headcut plate
was fixed in position, soil was packed upstream of the plate in
7 kg increments. After packing this final layer, a different 2.3-m
long, 0.165-m wide, 1.6-cm thick aluminium plate with three
vibrating motors was used to vibrate for 2 min. After packing
the soil bed, the soil surface was levelled with the upstream
rigid floor and the headcut-forming plate removed, thereby cre-
ating a 3-cm high preformed step in the channel bed profile
(Figure 2a).
The same procedures were followed for experiments with a
soil pipe, including the creation of a 3-cm step on the surface.
The only exception was packing soil around the soil pipe. Soil
was packed in 15-kg (2.5-cm) increments to a depth of 12.5 cm
(i.e. the depth for creating a soil pipe). A 1.5-cm o.d. (outside
diameter) steel rod was extended through the opening in the
endplate (Figures 2b, c and d) to the upslope inflow point.
The upslope end of the rod was positioned inside a slightly
larger conduit elbow connector with a tube that extended
through the side of the flume and connected to a water reser-
voir (Figure 2e) to provide inflow into the soil pipe. Packing
of the soil bed was then continued as before until the entire
depth of the soil bed was packed. Once the packing was com-
plete, the steel rod was removed by pulling it out of the
endplate, slowly, being careful to maintain alignment with the
soil bed, thereby creating a 1.5-cm i.d. (inside diameter) soil
pipe (linear void) that extended the entire length of the soil
bed. An endoscopic camera was installed in the end of the steel
rod. When sliding out the rod, the camera recorded the process
to verify that a soil pipe was created and did not collapse
(Figure 3). Rod removal was followed by removal of the
endplate, and Figure 4 shows the initial flume outlet after
endplate removal for different treatments.
To prepare the surface soil for erosion by convergent over-
land flow, very fine soil particles (sieved to 1 mm) were spread
evenly over the surface. To get consistent soil moisture, consol-
idate loose soil particles and reduce the spatial variability of the
surface, simulated rainfall was applied to the surface at
20 (±2) mm h 1 for about 6 h until the soil profile was

© 2020 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.

Figure 2. Soil pipe and headcut initiation setups: (a) 3-cm high preformed initial step in channel bed; (b) view from upper end to lower end of soil
bed with removeable rod in place before packing soil above the rod; (c) surface and subsurface runoff collectors, and steel rod sticking out of endplate
(view from outside soil bed); (d) removeable endplate with hole for steel rod (view from inside soil bed); (e) steel rod temporarily inserted into inflow
tube’s elbow connector for supplying inflow into soil pipe. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3. Views inside soil pipe using the endoscope as the steel rod is removed for seepage conditions test: (a) inner end section of soil pipe; (b)
middle section of soil pipe; (c) outlet end section of soil pipe; (d) soil pipe in seepage conditions. [Colour figure can be viewed at wileyonlinelibrary.
com]

completely wetted. The soil bed was tilted to a slope of 5% dur-
ing the pre-wet rainfall, which ensured that no water would
pond on the soil surface. Following pre-wet rainfall application,
the headcut plate and any splashed soil material on the plexi-
glas sidewalls were removed and the slope was then titled to
1% for the formal test. A thin layer of spray sealant (Rust-Oleum
Co., Vernon Hills, IL) was applied to a 30-cm long section of
the interface between the upstream rigid floor and the soil
bed to eliminate unwanted soil erosion near the interface of
the inflow tank and the soil bed, which was similar to the epoxy
resin used by Wells et al. (2009a, b, 2010). The sealant was
allowed to dry before overland flow was applied.
Experimental design
A total of six tests were conducted in this experiment (Table I)
that included three treatments (without a soil pipe, with a soil

© 2020 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES
Figure 4. Initial flume outlet conditions: (a) overland flow for free drainage test (FD&OF); (b) soil pipeflow due to seepage into pipe (SP&OF); (c) soil
pipeflow for upslope inflow combined with seepage (SP&OF&PF). [Colour figure can be viewed at wileyonlinelibrary.com]
Table I. Experimental design
1 1
Pipe existence Water regime Overland inflow rate (l min )
No drained 68.0
No seepage 68.8
Yes drained 68.0
Yes seepage 68.4
Yes seepage 68.9
Yes seepage – 0.98
pipe but no inflow, with a soil pipe having inflow) and two hy-
drologic conditions (free drainage and seepage). All tests were
replicated, thus 12 runs were included. Two tests were con-
ducted with no (N) soil pipe present in the soil bed as a baseline
treatment to illustrate the headcut migration dynamics in upland
surface concentrated flow under free drainage (FD&OF&N) and
seepage (SP&OF&N) conditions without soil pipe interactions.
The seepage conditions involve the establishment of a perched
water table 3 cm below the soil surface which imposes a seep-
age force on any voids below this depth, whether the presence
of a soil pipe or a headcut plunge pool. Note that FD indicates
free drainage and SP indicates seepage, whereas OF and PF in-
dicate overland flow and pipeflow, respectively and all tests
without the ‘N’ designation included a soil pipe present in the
bed. Four tests were conducted with a soil pipe, including two
without inflow into the pipe (FD&OF, SP&OF), another two with
inflow into the pipe, one with (SP&OF&PF) and the other with-
out (SP&PF) overland flow occurring simultaneously.
The test with pipe inflow under seepage conditions
(SP&OF&PF, SP&PF) aimed to represent soil pipe impacts on
headcut migration under the common condition during the ini-
tiation and through the main part of storm hydrographs when
perched water tables are supplying seepage into soil pipes
and potentially to the surface. The conditions of inflow under
free drainage (FD&OF) represent the final stage (recession limb)
of the hydrograph after the water table has dropped below the
pipe depth. Thus, the test of a soil pipe without inflow under
free drainage represents storm events under drier antecedent
conditions that do not produce pipeflow (Wilson et al., 2017).
In contrast, no pipe inflow under seepage (SP&OF) represents
small flow events and/or the initial stages of a significant flow
event in which flow is being generated.
Experimental procedures
Channel flow discharge was controlled by two adjustable in-
take valves and monitored by a flowmeter connected to a data
© 2020 John Wiley & Sons, Ltd.
Pipe inflow rate (l min ) Treatment Replicates
– FD&OF&N 2
– SP&OF&N 2
– FD&OF 2
– SP&OF 2
1.00 SP&OF&PF 2
SP&PF 2
logger (Campbell Scientific Inc., Logan, UT). Upslope channel
inflow was first fed into an inlet tank which was 0.8-m long,
0.4-m wide and 0.3-m deep, which acted as a reservoir to
damp pump-related turbulence. When the tank was filled, wa-
ter spilled onto a 1-m long rigid channel floor located immedi-
ately upstream of a soil bed. The soil bed was 2.75-m long,
0.165-m wide and 0.25-m deep (Figure 1). The sidewalls of
the soil bed were transparent plexiglass to observe the headcut
migration process.
In all tests, the overland flow rate was applied at 68 l min 1,
monitored by a flow meter; for the pipeflow test, pipe inflow
was applied at 1 l min 1, consistent with field observations
(Wilson, 2009), and the pipe inflow rate was controlled by a
small pump at steady rate. The soil bed slope was set to 1%.
Two runoff collectors were set at the end of the soil bed, one
at the soil surface and the other 15 cm below the soil surface
to monitor soil pipeflow. The outflow face of the soil bed also
consisted of plexiglass with a removeable endplate. The
endplate was 18 cm high with a 1.5-cm diameter hole located
15 cm below the top of the plate (i.e. below the soil bed sur-
face), which served as a guide for the placement of the soil pipe
then was removed prior to experiments.
Two acoustic sensors (TS-30S1-IV, Senix Co., Hinesburg, VT)
were installed on the upstream and downstream ends of the soil
bed to monitor the surface flow depth. A drainage system was
installed along the base of the soil bed that consisted of three
interconnected and perforated pipes with 12.7 mm i.d. These
pipes had five free air outlets equally spaced along the flume.
In the experiments under free drainage conditions, these outlets
provided escape routes for both air and water. In the experi-
ments under seepage conditions, the five outlets were con-
nected to a Mariotte bottle to create a constant head on the
soil bed, and a perched water table was simulated to a depth
of 0.22 m right under the preformed step.
During the experiment, lights were installed outside the
flume to view the headcut migration dynamics. A moveable
cart was mounted on rails along the top of the flume and two
Nikon D7100 cameras (Nikon Inc., Melville, NY) were
Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.
attached to the top and side of the cart. As the headcut migrated
upslope, the cart was moved simultaneously for the two cam-
eras to capture video of headcut scour hole morphology
changes. A multiple-intensity rainfall simulator (Meyer and
Harmon, 1979) consisting of two oscillating nozzles was
suspended approximately 4 m above the flume, which was
used for pre-wetting the soil bed to create similar surface con-
ditions before each experiment was conducted.
Overland flow and subsurface flow samples during experi-
ments were captured in 0.5-l glass bottles at 30-s intervals im-
mediately after their initiation to measure the sediment
concentration. The volume of each sample was measured and
then allowed to settle before clear water was decanted. The re-
maining sediment was oven-dried at 105 °C for 24 h and
weighed to determine the total mass of sediment. The duration
of the experiments was determined by the headcut migration
rate, which generally varied from 15 to 30 min. However, when
only pipeflow was conducted, the experiment lasted about
60 min and stopped when an upstream collapse occurred.
Video analysis and scour hole morphology
digitization
During the experiments, the two cameras recorded the top and
side views of the headcut migration dynamics to quantify the
scour hole morphodynamics (Wells et al., 2009a). Scour hole
morphology photos in specified intervals were captured in the
videos and then transferred to Origin 9.0 for digitization. Dur-
ing the digitization process, the same coordinate system was
used to assure that all headcut morphology photos were of
the same scale and relative position. According to the defini-
tion by Bennett et al. (2000) and Wells et al. (2010), the maxi-
mum scour depth (SD), the length from the brinkpoint to the
maximum scour depth (SL), the nappe entry angle (θ) and the
brinkpoint position (M) were measured at different times
(Figure 5). The possible error during digitization and morphol-
ogy measurement induced by lens curvature and operator var-
iance were described by Wells et al. (2009a) and the errors
controlled to be smaller than 5.0%.
Results
Baseline response of headcut erosion without a soil
pipe
Flow rate changes and sediment concentration
When no initial soil pipe existed, the overland flow rate varied
around the designed flow rate of 68 l min 1 (Figures 6a and b),
and the average overland flow rate was 68.8 and 69.1 l min 1
under free drainage and seepage conditions, respectively, cor-
responding to the monitoring results of the flow meter. Figure 7
shows comparisons of the sediment concentration changes for
free drainage and seepage conditions. When no soil pipe exists,
sediment concentrations of overland flow under seepage con-
ditions were 32% larger than that under free drainage
conditions.
Headcut morphodynamics
Compared with the free drainage conditions in Figure 8a, the
scour hole profile under seepage conditions in Figure 8c
showed a larger migration rate. At 15 min, the headcut mi-
grated 20.0 cm under seepage conditions but only 12.0 cm un-
der free drainage conditions. Scour hole depths under seepage
and free drainage conditions were nearly the same. The scour
© 2020 John Wiley & Sons, Ltd.
hole profile under seepage conditions was relatively longer
but shallower. Combined with the result of sediment concentra-
tion, it could be concluded that seepage conditions increased
the lifting force on soil particles and thus soil particle detach-
ment. Thus, headcuts migrated at a higher rate but shallower
depth with higher sediment concentrations under seepage
conditions.
Scour hole morphology parameters
Seepage conditions have larger migration rate compared with
free drainage conditions (Figures 9a and b) when no subsurface
soil pipe exists. The SD results of the FD&OF&N and SP&OF&N
treatments were nearly the same initially, and the values in the
SP&OF&N treatment were a little lower than those in the
FD&OF&N treatment at the end of the experiment, which
reflected the relatively longer but shallower scour hole profile
(Figures 9c and d). For SL, the SP&OF&N treatment kept a
higher rate than the FD&OF&N treatment, which reflected the
higher headcut migration rate and quicker expanding process
(Figures 9e and f). For θ, the SP&OF&N treatment had the
same pattern as the FD&OF&N treatment with time (Figures 9g
and h).
Soil pipe impacts on the headcut migration
Soil pipe altered overland flow into subsurface flow
Note that the endplate used as a guide for the removeable rod
to create the soil pipe was also removed after the rod was re-
moved. Thus, all treatments with a soil pipe involved an open
face at the outlet just as the treatments without the presence
of a soil pipe (Figure 4). For the scenarios of a soil pipe under-
neath the gully bed but without pipe inflow (FD&OF, SP&OF),
upland concentrated surface flow created a headcut that ad-
vanced upslope at 0.41–1.18 cm min 1 (Figures 9a and b) and
the scour hole depth was nearly the same as in the case of a
gully without a soil pipe (Figures 9c and d).
For the case of free drainage with soil pipe existence
(FD&OF), there was no perched water table to provide seepage
into the soil pipe and therefore no pipeflow. However, as
headcut erosion continued, once the headcut scour hole
reached the depth of the soil pipe, overland flow was diverted
from the surface and into the soil pipe downslope of the
headcut as depicted in Figures 5b and c. Thus, even from a
no-upslope inflow condition, pipeflow can become the major
flow path (Figures 6c and d) once a connection is made to
the surface concentrated runoff as proposed by Jones (1987)
and observed by Nichols et al. (2016).
Figure 6c shows the transformation from overland flow to
subsurface flow in the FD&OF treatment. In the FD&OF treat-
ment, the overland flow was abruptly diverted into the soil pipe
at 11.6 min, and overland flow was totally changed to pipeflow
at 16.5 min (Figure 6c, Table II). When the headcut scour hole
initially reached the depth of the soil pipe and began to capture
surface runoff, the 1.5-cm diameter soil pipe on the downslope
side of the scour hole could not handle all the surface flow.
However, flow into the soil pipe created rapid internal erosion
and after 4.9 min overland flow was completely diverted into
subsurface pipeflow.
For the case of seepage with soil pipe existence (SP&OF, Fig-
ure 6d), the perched water table provided seepage into the soil
pipe along its entire length. Thus, as overland flow was occur-
ring there was also minor pipeflow occurring (Figure 4b). Due
to the wetter conditions and upwards seepage gradient, the
headcut migration rate was faster (1.18 cm min 1) than for free
drainage (Table II), the time at which overland flow was
abruptly diverted into the soil pipe was advanced by 2.3 min
Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES
Figure 5. Scour hole morphology digitization and measurement in different treatments: (a) 13 min in seepage conditions without initial soil pipe; (b)
13 min in seepage conditions with initial soil pipe but no pipe inflow; (c) 7 min in seepage conditions with both overland flow and pipeflow applied.
[Colour figure can be viewed at wileyonlinelibrary.com]
Figure 6. Overland and subsurface flow rate in different treatments:
(a) free drainage conditions without soil pipe; (b) seepage conditions
without soil pipe; (c) free drainage conditions with soil pipe but no pipe
inflow; (d) seepage conditions with initial soil pipe but no pipe inflow;
(e) seepage conditions with overland flow and pipe inflow; (f) seepage
conditions with pipe inflow. [Colour figure can be viewed at
wileyonlinelibrary.com]
© 2020 John Wiley & Sons, Ltd.
Figure 7. Overland flow sediment concentration variations with time
for experiments without a soil pipe under free drainage (FD&OF&N)
and seepage (SP&OF&N) conditions. [Colour figure can be viewed at
wileyonlinelibrary.com]
to 9.3 min and the time at which overland flow changed totally
to pipeflow was also advanced by 5.0 min to 11.5 min (Fig-
ure 6c, Table II). Thus, the time required to internally erode
the downslope soil pipe such that it fully captured the overland
flow was greatly reduced to 2.2 min under seepage conditions.
For both conditions, once overland flow was diverted into
pipeflow, the sediment concentrations increased dramatically
(Table II). For free drainage conditions, the sediment concentra-
tion increased from 4.85 to 19.42 g l 1 and for seepage condi-
tions it increased from 6.42 to 33.44 g l 1. This was a
response to the combination of increased scour hole erosion
and internal erosion of the soil pipe. Figure 10 shows the sub-
surface flow sediment concentration variations. The results
showed that the average sediment concentration under seep-
age condition (SP&OF) was 72% larger than that under free
drainage condition (FD&OF), which is the same as the patterns
of overland flow.
Soil pipe impacts on headcut morphodynamics
Compared with the scour hole morphodynamics in the treat-
ment of free drainage without a soil pipe (FD&OF&N), the
scour hole profile of the FD&OF treatment exhibited a higher
SD (Figure 9c) and different morphology styles due to the pres-
ence of a soil pipe (Figures 8a and b). Before 15 min, the scour
hole morphology in these two treatments was evolving in
Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.
Figure 8. Scour hole morphodynamics in different treatments. [Colour figure can be viewed at wileyonlinelibrary.com]
similar patterns and the SD values were nearly 10 cm. After
15 min, the scour hole morphology in these treatments was to-
tally different. In the FD&OF&N treatment without a soil pipe,
the brinkpoint still migrated at the same rate as before (Fig-
ure 9a) and SD kept increasing to a depth of about 12 cm at
19 min (Figure 9c). However, in the FD&OF treatment with soil
pipe, SD experienced a quick increase and finally reached
more than 17 cm at 19 min (Figure 9c). The scour hole profile
morphology also changed from longer and shallower to shorter
and deeper. This morphology change was more obvious in the
seepage condition (Figures 8c and d). Under seepage condi-
tions, the scour hole profiles showed the same pattern as that
under free drainage conditions before 11 min. After 11 min,
two different patterns evolved: in the SP-N treatment without
soil pipe, the SD value increased slowly to a depth of about
11 cm at 15 min; while in the SP-OF treatment with soil pipe,
the SD value increased quickly and finally reached about
20 cm at 15 min (Figure 9d). Obviously, a new deeper scour
hole was formed after overland flow was intercepted by the soil
pipe. Because all overland flow entered the soil pipe, no over-
land flow travelled over the downslope surface of the former
scour hole. For the new scour hole, the sudden increased depth
difference induced by the soil pipe created larger jet flow en-
ergy and further increased SD, which required a new headcut
migration rate to balance the forces.
Scour hole morphology parameter variations with experiment
time
To further quantify the scour hole morphodynamics, the M, SD,
SL and θ variations with experiment time are plotted in Figure 9.
Under free drainage conditions, the brinkpoint position mi-
grated at a relative steady rate with several jumps induced by
the instantaneous detachment and exfoliation of small mass
failure (aggregates) at the brinkpoint position (Figure 9a). For
maximum scour hole depth, the FD&OF treatment experienced
a sudden increase at around 15 min (Figure 9c), which
corresponded well to the phenomenon observed in Figure 8b.
SL represents the distance from the brinkpoint to the maximum
scour depth and showed an increasing trend as the experiment
time went on because of the expanding scour hole (Figure 9e).
When overland flow cut through to the soil pipe, SL experi-
enced a quick decrease as the new scour hole suddenly
© 2020 John Wiley & Sons, Ltd.
shortened this distance, as exhibited in Figure 8b. The nappe
entry angle reflected the incident angle of jet flow induced by
a headcut, which fluctuated at the beginning of the experiment
and then gradually decreased, but after overland flow diverted
to pipeflow in the FD&OF treatment, θ showed a jump to a
higher slope value of around 60° (Figure 9g).
Under seepage conditions, the initial soil pipe had similar
impacts on scour hole morphology parameters as the free
drainage conditions. The brinkpoint position was the same for
SP&OF&N and SP&OF treatments at 15 min of experiment time
(Figure 9b). Compared with SP&OF&N treatment without initial
soil pipe, the SD of treatment SP&OF exhibited an obvious sud-
den increase from 9.4 cm at 11.5 min to 20.6 cm at 15 min (Fig-
ure 9d). In contrasts, the SL of treatment SP&OF experienced a
drop from about 11.5 to only 4.2 cm (Figure 9f), which
corresponded to Figure 8d. Like the free drainage conditions,
θ also showed a jump from 35 to more than 60° when overland
flow was diverted into the soil pipe and a new scour hole pro-
file formed.
Soil pipeflow impacts on headcut migration
Pipeflow impacts on flow rate variations and sediment
concentration
Figure 6e shows the SP&OF&PF treatment with pipe inflow in-
troduced to the initial soil pipe, compared with the FD&OF and
SP&OF treatments (Figures 6c and d). Upslope pipeflow further
accelerated the process of overland flow transfer into subsur-
face flow. For the condition of overland flow with a soil pipe
that was hydraulically active from upslope inflow
(FD&OF&PF), it took 0.5 min for the upslope inflow to exit
the 2.75-m long soil pipe. At 5.3 min in the SP&OF&PF treat-
ment, the headcut scour hole development by overland flow
reached the soil pipe position and all overland flow changed
to subsurface flow with a quick duration of only 0.8 min. Thus,
no overland flow was observed after 6.1 min. This moment was
advanced 5.4 min compared with the SP&OF treatment and
10.4 min compared with the FD&OF treatment (Table II; Fig-
ures 6c, d and e).
Sediment concentration in the SP&OF&PF treatment varied
greatly with time because of rapid changes to the soil pipeflow
Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES

Figure 9. Migration of the brinkpoint (M), maximum scour depth (SD), distance from brinkpoint to maximum scour depth (SL) and nappe entry angle
(θ) in different treatments. [Colour figure can be viewed at wileyonlinelibrary.com]

Table II. Parameters before and after overland flow cut through the soil pipe

Treatments

Parameters FD&OF SP&OF SP&OF&PF

Time subsurface pipeflow generated (min) 11.6 9.3 0.5

Time surface flow totally changed to pipeflow (min) 16.5 11.5 6.1
1
Headcut migration rate at overland flow stage (cm min ) 0.55 1.18 0.41
1
Headcut migration rate at subsurface flow stage (cm min ) 0.57 1.83 2.62
1
Average sediment concentration of overland flow (g l ) 4.85 6.42 6.64
1
Average sediment concentration of subsurface flow (g l ) 19.42 33.44 46.68

(Figure 10). In the initial 6 min when subsurface flow was only
induced by upslope pipe inflow, sediment concentration from
the soil pipe varied around 100 g l 1 with a large standard devi-
ation because of internal erosion of aggregate mass failures. By
6.1 min the soil pipe had eroded sufficiently internally that it
was able to intercept all the overland flow. In response, the sed-
iment concentration from the soil pipe quickly dropped to
around 20 g l 1 as a result of the overland flow rate being
considerably larger but with lower sediment concentration than
the upslope pipeflow (Figure 10). But the average sediment
concentration of subsurface flow was still 6.0 times larger than
what had previously been exiting the flume by overland flow,
because the subsurface soil pipe was initially constricted to a
small conduit that experienced greater soil detachment per vol-
ume of flow. Compared with the average sediment concentra-
tion of subsurface flow in the SP&OF treatment, soil pipeflow

© 2020 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.
Figure 10. Subsurface flow sediment concentration variations with
time in different treatments. [Colour figure can be viewed at
wileyonlinelibrary.com]
sediment concentration increased 40% for the SP&OF&PF
treatment (Table II), which illustrates the dramatic role of soil
pipeflow in headcut migration processes.
In the SP&PF treatment with upslope pipe inflow but without
overland flow, the flow rate varied around 1 l min 1 for the
whole experiment duration (Figure 6f) and the sediment con-
centration also varied around an average value of 50 g l 1 (Fig-
ure 10), which displayed the subsurface pipeflow sediment
transport ability without the impacts of overland flow. The ex-
periment with upslope pipeflow under seepage but without
overland flow (SP&PF) also demonstrated the significance of in-
ternal erosion as the pipe enlarged until tunnel collapse created
multiple openings at the surface (Figures 11a, b and c), which
could intercept future overland flow events in the channel (Fig-
ures 11d and e).
Pipeflow erosion impacts on headcut morphodynamics
When upslope inflow into the soil pipe was added, the scour
hole morphology in the SP&OF&PF treatment changed greatly
because of subsurface soil pipe enlargement (Figure 8e), com-
pared with the SP&OF (Figure 8d) and FD&OF (Figure 8b) treat-
ments with a soil pipe but without upslope pipe inflow
(Figure 8e). The scour hole morphology in the SP&OF&PF treat-
ment over the initial 5 min was nearly the same as that in the
SP&OF treatment before overland flow cut to the soil pipe po-
sition. In this treatment, the soil pipe exhibited internal erosion
simultaneous with headcutting by overland flow. Thus, the time
until the scour hole connected with the soil pipe was greatly
advanced to 6.1 min. From 6 to 8 min, the scour hole profile ex-
hibited a boot shape because of a soil block hanging above the
expanded soil pipe created by pipeflow, and the depth of the
scour hole profile increased about 12 cm during the 2 min. At
9 min, the hanging block fell into the pipeflow path and was
transported to the flume outlet. After 9 min, a new scour hole
morphology was formed and a new headcut migration balance
was reached. Due to headcutting being at the initial stages of
scour hole development, the headcut migration rate was much
lower (0.41 cm min 1) at the time of diverting overland flow
into the soil pipe. Once the overland flow was intercepted,
however, the headcut migration rate jumped to 2.62 cm min 1
(Table II).
Pipeflow erosion impacts on scour hole morphology
parameters
Compared with the SP&OF treatment of overland flow with
pipeflow induced by seepage, this invisible subsurface
pipeflow erosion accelerated the growing process of the scour
hole profile compared to the SP&OF&N and FD&OF
© 2020 John Wiley & Sons, Ltd.
treatments. This caused differences in scour hole morphology
parameters (Figures 9b, d, f and h). For the addition of pipe in-
flow, the brinkpoint position in the SP&OF&PF treatment ini-
tially migrated at a similar rate as for the SP&OF treatment
(Figure 9b). Compared with the SP&OF treatment, the SD value
for the SP&OF&PF treatment experienced an earlier sudden in-
crease at about 6 min due to the interception of surface flow
and kept constant over the remaining time (Figure 9d). More-
over, the SL value for the SP&OF&PF treatment also experi-
enced an earlier and similar drop as for the SP&OF treatment
during the process of overland flow being intercepted by the
soil pipe (Figure 9f). The θ value for the SP&OF&PF treatment
was relatively larger compared with other treatments and expe-
rienced a larger jump step from 47 to more than 70° (Figure 9h).
Discussion
Seepage role in headcut migration processes
Seepage is a common phenomenon where the rate of infiltra-
tion exceeds the rate of percolation through soil layers. Duplex
soils (Faulkner, 2006) can restrict deep percolation, thereby
forming temporary perched water tables. Such conditions foster
seepage into preferential flow paths (Tomlinson and Vaid, 2000;
Fox and Wilson, 2010), such as macropores, and establish con-
nectivity among macropores and with soil pipes in a process
described as self-organization (Sidle et al., 2001). Under seep-
age conditions, the exfiltration gradient works against gravita-
tional forces and reduces contact stress between soil particles,
thus increasing soil erodibility and exponentially decreasing
the soil resistance to fluvial erosion (Owoputi and Stolte, 2001).
Al-Madhhachi et al. (2014) demonstrated with a mini-jet device
that seepage forces significantly impact scour hole depth.
Wilson et al. (2016) described how soil pipes are
self-propagating in that internal erosion of the preferential flow
path in response to self-organizing processes develops into pro-
gressively larger and more connected soil pipe networks with
time. As demonstrated in the SP&OF experiments, pipeflow
can be generated by seepage from a perched water table that
can greatly influence the propensity of the headcut to scour
into the soil pipe and subsequently the headcut migration rate.
Under seepage conditions, the soil pipe expanded along its en-
tire length by internal erosion. Thus, when the scour hole cut
down to the soil pipes, the pipe had enlarged such that the
scour hole was almost instantly much deeper than the condi-
tion in which a pipe existed without seepage conditions (Fig-
ures 8b and d). The increased scour hole depth increased the
scour hole energy and thereby accelerated the headcut migra-
tion and the sediment transport.
Such seepage into soil pipes creates an upslope contributing
area for the pipe, in addition to surface runoff interception into
the soil pipe network upslope. The combination of upslope in-
flow plus seepage even more dramatically impacts the connec-
tivity of headcut scour holes with soil pipes and the subsequent
headcut migration rate (Uchida et al., 1999, 2002; Wil-
son, 2011; Figure 8e) as the SP&OF&PF treatment
demonstrated.
Seepage is known to accelerate headcut migration and in-
crease sediment concentrations in runoff on hillslopes (Huang
and Laflen, 1996; Gabbard et al., 1998; Zheng et al., 2000).
In treatments without a soil pipe below the channel bed, seep-
age still increased the headcut migration rate (Figures 8a and c,
9a and b) and sediment concentration (Figure 7). However, the
scour depth was decreased, which corresponds well with labo-
ratory experiments by Wells et al. (2009b) on a fine sandy clay
loam (Atwood series) concentrated flow channel under four
Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES
Figure 11. Top view of pipe collapse features induced by pipeflow and internal erosion: (a–c) pipe collapse process; (d, e) pipe collapse converting
overland flow into subsurface pipeflow. [Colour figure can be viewed at wileyonlinelibrary.com]
hydrologic conditions (drained, desiccated, water table at 30
and 80 mm). They showed that the presence of even a small
perched water table resulted in increased headcut migration
rates and sediment discharges but smaller scour hole depths,
which they related to changes in the soil erodibility coefficient
(Wells et al., 2009b). This study on a Providence silt loam chan-
nel with a water table that extended to right above the initial
soil surface found similar results on the role of seepage in ac-
celerating headcut migration and increasing sediment concen-
tration when a soil pipe was not present and much more
dramatic impacts when a soil pipe was present.
The role of soil pipe and pipe erosion in headcut
migrating dynamics
The headcut migration processes in the treatments with and
without the presence of a soil pipe were nearly identical until
the scour hole reached the depth of the soil pipe position. As
previously discussed, there were differences between the hy-
drologic condition imposed on the bed and soil pipe, whether
free drainage or seepage. However, the biggest difference ob-
served in headcut migration among treatments was the pres-
ence of a soil pipe below the channel bed. This dramatic
difference was in response to the headcut scour hole
connecting to the soil pipe below the channel bed, which
© 2020 John Wiley & Sons, Ltd.
diverted overland flow into the pipe. Once this occurred, the
flow and headcut processes changed substantially. The over-
land flow entering the scour hole experienced a sudden drop
from the original scour hole depth and then entered into the
soil pipe on the downslope side. The increased drop height in-
creased the jet flow energy (Liu et al., 2017), which enhanced
the scouring process, as illustrated in Figures 8b and d.
Wilson et al. (2017) noted that individual soil pipes behave
significantly differently in timing of pipeflow, rates (stage) and
durations of pipeflow even within the same catchment and
same storm events. This, they reasoned, was due to differences
in their ability to intercept surface flow and differences in their
hydrologic connectivity with the perched water table through
their upslope network of interconnected soil pipes. Overland
flow interception by soil pipe collapses is a common feature
in catchments with pipe collapses (Wilson et al., 2015, 2017;
Nichols et al., 2016). However, no one to date had addressed
the impacts that this interception has on erosion of the scour
hole or downslope internal erosion of the pipe.
Alonso et al. (2002) developed equations to predict the mag-
nitude of headcut scour depth and the headcut migration rate
using jet impingement theory. Their predictions were success-
fully compared to their experimental database. Their analytic
model of headcut erosion has been tested and modified ac-
cordingly for various conditions like soil stratification (Gordon
et al., 2007), soil texture (Wells et al., 2009a) and upstream sed-
iment loads (Wells et al., 2010). The influence of a soil pipe
Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.

and/or pipeflow below the channel bed on headcut migration
described in this paper is a possible option for further modifying
the analytic model of headcut erosion for factors like pipe posi-
tion, pipe diameter, seepage and pipeflow rate. More studies
need to be conducted to gain sufficient knowledge of these
complex phenomena in order to develop new models for pre-
diction of piping impacts on headcut migration of channels.
Pipe collapse induced by pipeflow
The combination of pipe inflow and seepage resulted in tunnel
collapse upslope of the headcut scour hole that diverted the
surface runoff into the soil pipe. The pipe collapse process
was also observed during the experiment without overland
flow (Figures 11a, b and c), which corresponded well to the
pipe collapse features observed in the field (Wilson
et al., 2015, 2018). In field conditions, subsurface pipeflow
may still be active, but not observed from the surface except
in these collapsed pipe openings, even when overland flow is
not active (Wilson et al., 2017). Subsurface flow can also occur
through soil pipes when perched water tables either are not
measurable or not hydrologically connected across the land-
scape (Wilson et al., 2017). Such pipeflows continuously erode
these underground conduits until the surface soil cannot sup-
port the soil and tunnel collapse occurs. The collapsed soil
blocks can provide abundant soil material for sediment trans-
port through the pipe (Wilson et al., 2018), similar to what is
observed in channels with mass failure of sidewalls (Qin
et al., 2018). However, unlike mass failures in open channels,
soil pipes can become clogged which temporarily blocks
pipeflow and causes pressure increases within the pipe and ad-
jacent soil until the plug is removed (Kosugi et al., 2004;
Midgley et al., 2013; Wilson and Fox, 2013). Wilson (2011) ob-
served, in laboratory flumes with soil pipes, that pipe clogging
can foster pipe collapse and surges in pipeflow and sediment
concentration. These surges in flow and sediment concentra-
tions observed previously (Wilson, 2009, 2011) for pipeflow
were also observed in these experiments when upslope
pipeflow was combined with seepage and resulted in high
(>100 g l 1) sediment concentrations. However, when over-
land flow was captured by the scour hole and diverted into
the soil pipe, the sediment concentration was dramatically de-
creased (<20 g L 1) due to dilution by the low-concentration
overland flow.
There were many interesting phenomena revealed in the ex-
periment that were similar to those observed in the field.
Figure 12 shows comparisons of these laboratory simulations
with field observations. For example, when pipeflow was initi-
ated under a perched water table condition, bubbles were ob-
served exiting at the pipe outlet as air was displaced from the
soil pipe prior to pipe inflow reaching the outlet (Figure 12c).
Yamasaki et al. (2017) demonstrated the importance of
entrapped air in water flow through closed soil pipes in labora-
tory experiments. The phenomenon of entrapped air escaping
has been observed in the field (Figure 12c), similar to this labo-
ratory observation (Figure 12d). It has been speculated that this
process of entrapped air movement can cause linear voids in
the soil that could develop into soil pipes. Another process ob-
served was mass failure of soil aggregates that can block or plug
the soil pipe. This phenomenon is often observed in the field
(Faulkner, 2007; Wilson et al., 2015, 2017, 2018;
Bernatek-Jakiel and Poesen, 2018) in which flow is forced to
the surface (Figure 12e), thereby creating sediment mounds at
the surface around the pipe collapse (Figure 12f). The fluctua-
tions in sediment concentration and increased headcut migra-
tion observed in this study were consistent with field
observations (Swanson et al., 1989; Sayer et al., 2006; Nichols
et al., 2016), in part due to these mass failures followed by
flushing of the plug.

Figure 12. Comparison of laboratory observations with field investigations at Walnut Gulch (a) and Goodwin Creek Experimental Watershed (b–f).
Observations of runoff interception (a) by Nichols et al. (2016) and (b) by Wilson; (c) field observations of air bubbles escaping to surface as seepage
enters soil pipe; (d) laboratory observation in this study of entrapped air escaping the soil pipe; (e) field observation of artesian flow from pipe to sur-
face; (f) field observation of pipeflow exiting the surface and creating a surface sediment mound. [Colour figure can be viewed at wileyonlinelibrary.
com]

© 2020 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES
Conclusions
Physical experiments were conducted to examine the impacts
of a soil pipe below a channel bed and seepage, as well as flow
into and through a soil pipe, on headcut migration. The follow-
ing results were observed. (1) For the baseline treatments with-
out a soil pipe, seepage conditions increased the sediment
concentration by 42% and the headcut migration rate by
47% compared with free drainage conditions. (2) When a soil
pipe was present below the soil bed, once the headcut scour
hole contacted the soil pipe, the scour hole depth increased
and combined with the overland flow being diverted into the
pipeflow downslope of the headcut, these changes accelerated
the headcut migration rate. (3) The process of overland flow be-
ing diverted into the soil pipe occurred sooner and resulted in
higher headcut migration rates in the presence of a perched
water table (seepage). (4) When soil pipes are fed by seepage
and an upslope contributing area (i.e. pipe inflow with seep-
age), whether there is overland flow through the channel or
not, tunnel collapse will create surface openings for
intercepting surface runoff above or in the absence of headcut.
This study quantitatively demonstrated the importance of
pipeflow in accelerating headcut migration processes and the
role of seepage and the presence of a soil pipe below a channel
even without pipeflow from upslope. The laboratory observa-
tions were consistent with field observations and led to better
understanding of these processes. There is still a need for fur-
ther studies under various soil and boundary settings.
Acknowledgements—This study was supported by the National Key
R&D Programme of China (Grant No. 2016YFE0202900), the National
Natural Science Foundation of China (Grants No. 41571263,
41790424, 41730645 and 41907060) and the Strategic Priority Re-
search Programme of the Chinese Academy of Sciences (Grants No.
XDA23060502 and XDA23100401).
Data availability statement
The data sets used and analysed during the current study are
available from the corresponding author on reasonable request.
Conflict of interest
The authors declare no conflict of interest.
References
Al-Madhhachi AT, Fox GA, Hanson GF. 2014. Quantifying the erodibil-
ity of streambanks and hillslopes due to surface and subsurface
forces. Transactions of the ASABE 57(4): 1057–1069. https://doi.org/
10.13031/trans.57.10416.
Alonso CV, Bennett SJ, Stein OR. 2002. Predicting head cut erosion and
migration in concentrated flows typical of upland areas. Water
Resources Research 38: 1303–1317. https://doi.org/10.1029/
2001WR001173.
Bennett SJ. 1999. Effect of slope on the growth and migration of
headcuts in rills. Geomorphology 30(3): 273–290. https://doi.org/
10.1016/S0169-555X(99)00035-5.
Bennett SJ, Casalí J. 2001. Effect of initial step height on headcut devel-
opment in upland concentrated flows. Water Resources Research 37
(5): 1475–1484. https://doi.org/10.1029/2000WR900373.
Bennett SJ, Alonso CV, Prasad SN, Römkens MJM. 2000. Experiments
on headcut growth and migration in concentrated flows typical of
upland areas. Water Resources Research 36: 1911–1922. https://
doi.org/10.1029/2000WR900067.
© 2020 John Wiley & Sons, Ltd.
Bernatek-Jakiel A, Poesen J. 2018. Subsurface erosion by soil piping:
significance and research needs. Earth-Science Reviews 185:
1107–1128. https://doi.org/10.1016/j.earscirev.2018.08.006.
Bernatek-Jakiel A, Wrońska-Wałach D. 2018. Impact of piping on gully
development in mid-altitude mountains under a temperate climate: a
dendrogeomorphological approach. Catena 165: 320–332. https://
doi.org/10.1016/j.catena.2018.02.012.
Bernatek-Jakiel A, Jakiel M, Krzemień K. 2017. Piping dynamics in
mid-altitude mountains under a temperate climate: Bieszczady
Mts., Eastern Carpathians. Earth Surface Processes and Landforms
42: 1419–1433. https://doi.org/10.1002/esp.4160.
Brunton DA, Bryan RB. 2000. Rill network development and sediment
budgets. Earth Surface Processes and Landforms 25: 783–800. https://
doi.org/10.1002/1096-9837(200007)25:7<783::AID-ESP106>3.0.
CO;2-W.
Bryan RB, Jones JAA. 1997. The significance of soil piping processes:
inventory and prospect. Geomorphology 20(3–4): 209–218. https://
doi.org/10.1016/S0169-555X(97)00024-X.
Castillo C, Gómez JA. 2016. A century of gully erosion research: ur-
gency, complexity and study approaches. Earth-Science Reviews
160: 300–319. https://doi.org/10.1016/j.earscirev.2016.07.009.
Chappell NA. 2010. Soil pipe distribution and hydrological functioning
within the humid tropics: a synthesis. Hydrological Processes 24(12):
1567–1581. https://doi.org/10.1002/hyp.7579.
Crouch RJ. 1983. The role of tunnel erosion in gully head progression.
Journal of the Soil Conservation Service of New South Wales 39:
149–155.
Crouch RJ, McGarity JW, Storrier RR. 1986. Tunnel formation processes
in the Riverina area of N.S.W. Australia. Earth Surface Processes and
Landforms 11: 157–168. https://doi.org/10.1002/esp.3290110206.
Dunne T, Malmon DV, Mudd SM. 2010. A rain splash transport equa-
tion assimilating field and laboratory measurements. Journal of Geo-
physical Research – Earth Surface 115: F01001. https://doi.org/
10.1029/2009JF001302.
Faulkner H. 2006. Piping hazard on collapsible and dispersive soils in
Europe. Soil Erosion in Europe: 537–562. https://doi.org/10.1002/
0470859202.ch40.
Faulkner H. 2007. Improvements to the dispersion status of piped gully
soils following reworking and stabilisation by vegetation. Catena 70:
410–415. https://doi.org/10.1016/j.catena.2006.11.007.
Faulkner H. 2013. Badlands in marl lithologies: a field guide to soil dis-
persion, subsurface erosion and piping-origin gullies. Catena 106:
42–53. https://doi.org/10.1016/j.catena.2012.04.005.
Faulkner H, Alexander R, Teeuw R, Zukowskyj P. 2004. Variations in
soil dispersivity across a gully head displaying shallow sub-surface
pipes, and the role of shallow pipes in rill initiation. Earth Surface
Processes and Landforms 29: 1143–1160. https://doi.org/10.1002/
esp.1109.
Faulkner H, Alexander R, Zukowskyj P. 2008. Slope–channel coupling
between pipes, gullies and tributary channels in the Mocatán catch-
ment badlands, Southeast Spain. Earth Surface Processes and Land-
forms 33: 1242–1260. https://doi.org/10.1002/esp.1610.
Fox GA, Felice RG. 2014. Bank undercutting and tension failure by
groundwater seepage: predicting failure mechanisms. Earth Surface
Processes and Landforms 39: 758–765. https://doi.org/10.1002/
esp.3481.
Fox GA, Wilson GV. 2010. The role of subsurface flow in hillslope and
streambank erosion: a review of status and research needs. Soil Sci-
ence Society of America Journal 74(3): 717–733. https://doi.org/
10.2136/sssaj2009.0319.
Fox GA, Felice RG, Midgley TL, Wilson GV, Al-Madhhachi AT. 2014.
Laboratory soil piping and internal erosion experiments: evaluation
of a soil piping model for low compacted soils. Earth Surface Pro-
cesses and Landforms 39: 1137–1145. https://doi.org/10.1002/
esp.3508.
Frankl A, Poesen J, Deckers J, Haile M, Nyssen J. 2012. Gully head re-
treat rates in the semi-arid highlands of Northern Ethiopia. Geomor-
phology 173: 185–195. https://doi.org/10.1016/j.geomorph.2012.
06.011.
Frankl A, Deckers J, Moulaert L, Van Damme A, Haile M, Poesen J,
Nyssen J. 2016. Integrated solutions for combating gully erosion in
areas prone to soil piping: innovations from the drylands of Northern
Ethiopia. Land Degradation & Development 27: 1797–1804. https://
doi.org/10.1002/ldr.2301.
Earth Surf. Process. Landforms, (2020)
 X. XU ET AL.
Gabbard DS, Huang C, Norton LD, Teinhrdt GCS. 1998. Landscape po-
sition, surface hydraulic gradients and erosion processes. Earth Sur-
face Processes and Landforms 23: 83–93. https://doi.org/10.1002/
(SICI)1096-9837(199801)23:1<83::AID-ESP825>3.0.CO;2-Q.
Gordon LM, Bennett SJ, Wells RR, Alonso CV. 2007. Effect of soil strat-
ification on the development and migration of headcuts in upland
concentrated flows. Water Resources Research, 43(7): W07412.
https://doi.org/10.1029/2006WR005659
Govers G, Giménez R, Van Oost K. 2007. Rill erosion: exploring the re-
lationship between experiments, modelling and field observations.
Earth-Science Reviews 84(3–4): 87–102. https://doi.org/10.1016/j.
earscirev.2007.06.001.
Hardie MA, Doyle RB, Cotching WE, Lisson S. 2012. Subsurface lateral
flow in texture contrasts (duplex) soils and catchments with shallow
bedrock. Applied and Environmental Soil Science 2012: 861358.
https://doi.org/10.1155/2012/861358.
Huang CH, Laflen JM. 1996. Seepage and soil erosion for a clay loam
soil. Soil Science Society of America Journal 60(2): 408–416.
https://doi.org/10.2136/sssaj1996.03615995006000020011x.
Jones JAA. 1987. The effects of soil piping on contributing area and ero-
sion patterns. Earth Surface Processes and Landforms 12(3): 229–248.
https://doi.org/10.1002/esp.3290120303.
Kosugi K, Uchida T, Mizuyama T. 2004. Numerical calculation of soil
pipeflow and its effect on water dynamics in a slope. Hydrological
Processes 18: 777–789. https://doi.org/10.1002/hyp.1367.
Leopold LB, Miller JP. 1956. Ephemeral Streams: Hydraulic Factors and
their Relation to the Drainage Net, Vol. 282. US Government Printing
Office: Washington, D.C.
Liu QJ, Wells RR, Dabney SM, He JJ. 2017. Effect of water potential and
void ratio on erodibility for agricultural soils. Soil Science Society of
America Journal 81(3): 622–632. https://doi.org/10.2136/
sssaj2016.11.0369.
Meyer LD, Harmon WC. 1979. Multiple-intensity rainfall simulator for
erosion research on row sideslopes. Transactions of ASAE 22(1):
100–103. https://doi.org/10.13031/2013.34973.
Meyer LD, Foster GR, Nikolov S. 1975. Effect of flow rate and canopy
on rill erosion. Transactions of ASAE 18: 905–911.
Midgley TL, Fox GA, Wilson GV, Felice RC, Heeren DM. 2013. In situ
pipeflow experiments on contrasting streambank soils. Transactions
of the ASABE 56: 479–488. https://doi.org/10.13031/2013.42685.
Nichols MH, Nearing M, Hernandez M, Polyakov VO. 2016. Monitor-
ing channel head erosion processes in response to an artificially in-
duced abrupt base level change using time-lapse photography.
Geomorphology 265: 107–116. https://doi.org/10.1016/j.
geomorph.2016.05.001.
Owoputi LO, Stolte WJ. 2001. The role of seepage in erodibility. Hydro-
logical Processes 15(1): 13–22. https://doi.org/10.1002/hyp.153.
Poesen J. 2018. Soil erosion in the anthropocene: research needs. Earth
Surface Processes and Landforms 43: 64–84. https://doi.org/10.1002/
esp.4250.
Poesen J, Nachtergaele J, Verstraeten G, Valentin C. 2003. Gully ero-
sion and environmental change: importance and research needs. Ca-
tena 50: 91–133. https://doi.org/10.1016/S0341-8162(02)00143-1.
Qin C, Zheng F, Wells RR, Xu X, Wang B, Zhong K. 2018. A laboratory
study of channel sidewall expansion in upland concentrated flows.
Soil & Tillage Research 178: 22–31. https://doi.org/10.1016/j.
still.2017.12.008.
Rodzik J, Furtak T, Zgłobicki W. 2009. The impact of snowmelt and
heavy rainfall runoff on erosion rates in a gully system, Lublin Up-
land, Poland. Earth Surface Processes and Landforms 34:
1938–1950. https://doi.org/10.1002/esp.1882.
Römkens MJM, Helming K, Prasad SN. 2001. Soil erosion under differ-
ent rainfall intensities, surface roughness, and soil water regimes. Ca-
tena 46: 103–123. https://doi.org/10.1016/S0341-8162(01)00161-8.
Sayer AM, Walsh RPD, Bidin K. 2006. Pipeflow suspended sediment
dynamics and their contribution to stream sediment budgets in small
rainforest catchments, Sabah, Malaysia. Forest Ecology and Manage-
ment 224: 119–130. https://doi.org/10.1016/j.foreco.2005.12.012.
Sidle RC, Noguchi S, Tsuboyama Y, Laursen K. 2001. A conceptual
model of preferential flow systems in forested hillslopes: evidence
of self-organization. Hydrological Processes 15: 1675–1692. https://
doi.org/10.1002/hyp.233.
© 2020 John Wiley & Sons, Ltd.
Sun L, Fang H, Qi D, Li J, Cai Q. 2013. A review on rill erosion process
and its influencing factors. Chinese Geographical Science 23(4):
389–302. https://doi.org/10.1007/s11769-013-0612-y.
Swanson ML, Kondolf GM, Boison PJ. 1989. An example of rapid gully
initiation and extension by subsurface erosion: coastal San Mateo
County, California. Geomorphology 2(4): 393–403. https://doi.org/
10.1016/0169-555X(89)90023-8.
Tebebu TY, Abiy AZ, Zegeye AD, Dahlke HE, Easton ZM, Tilahun SA,
Collick AS, Kidnau S, Moges S, Dadgari F, Steenhuis TS. 2010. Sur-
face and subsurface flow effect on permanent gully formation and
upland erosion near Lake Tana in the northern highlands of
Ethiopia. Hydrology and Earth System Sciences 14: 2207–2217.
https://doi.org/10.5194/hess-14-2207-2010.
Tomlinson SS, Vaid YP. 2000. Seepage forces and confining pressure ef-
fects on piping erosion. Canadian Geotechnical Journal 37(1): 1–13.
https://doi.org/10.1139/t99-116.
Uchida T, Kosugi K, Mizuyama T. 1999. Runoff characteristics of
pipeflow and effects of pipeflow on rainfall–runoff phenomena in a
mountainous watershed. Journal of Hydrology 222(1–4): 18–36.
https://doi.org/10.1016/S0022-1694(99)00090-6.
Uchida T, Kosugi KI, Mizuyama T. 2002. Effects of pipeflow and bed-
rock groundwater on runoff generation in a steep headwater catch-
ment in Ashiu, central Japan. Water Resources Research 38(7): 24-
1–24-14. https://doi.org/10.1029/2001wr000261.
Vannoppen W, Verachtert E, Poesen J. 2017. Pipeflow response in
loess-derived soils to precipitation and groundwater table fluctua-
tions in a temperate humid climate. Hydrological Processes 38:
586–596. https://doi.org/10.1002/hyp.11049.
Verachtert E, Maetens W, Van Den Eeckhaut M, Poesen J, Deckers J.
2011. Soil loss rates due to piping erosion. Earth Surface Processes
and Landforms 36(13): 1715–1725. https://doi.org/10.1002/
esp.2186.
Wells RR, Alonso CV, Bennett SJ. 2009a. Morphodynamics of headcut
development and soil erosion in upland concentrated flows. Soil Sci-
ence Society of America Journal 73(2): 521–530. https://doi.org/
10.2136/sssaj2008.0007.
Wells RR, Bennett SJ, Alonso CV. 2009b. Effect of soil texture,
tailwater height, and pore water pressure on the morphodynamics
of migrating headcuts in upland concentrated flows. Earth Surface
Processes and Landforms 34: 1867–1877. https://doi.org/10.1002/
esp.1871.
Wells RR, Bennett SJ, Alonso CV. 2010. Modulation of headcut soil ero-
sion in rills due to upstream sediment loads. Water Resources Re-
search 46: W12531. https://doi.org/10.1029/2010WR009433.
Wilson GV. 2009. Mechanisms of ephemeral gully erosion caused
by constant flow through a continuous soil-pipe. Earth Surface Pro-
cesses and Landforms 34: 1858–1866. https://doi.org/10.1002/
esp.1869.
Wilson GV. 2011. Understanding soil-pipeflow and its role in ephem-
eral gully erosion. Hydrological Processes 25: 2354–2364. https://
doi.org/10.1002/hyp.7998.
Wilson GV, Fox GA. 2013. Internal erosion during soil pipeflow: state
of the science for experimental and numerical analysis. Transactions
of the ASABE 56(2): 465–478. https://doi.org/10.13031/2013.42667.
Wilson GV, Cullum RF, Römkens MJM. 2006. Pipe flow impacts on
ephemeral gully erosion. In Proceedings of 8th Federal Interagency
Sedimentation Conference, Reno, Nevada.
Wilson GV, Cullum RF, Römkens MJM. 2008. Ephemeral gully erosion
by preferential flow through a discontinuous soil-pipe. Catena 73(1):
98–106. https://doi.org/10.1016/j.catena.2007.09.008.
Wilson GV, Rigby JR, Dabney SM. 2015. Soil pipe collapses in a loess
pasture of Goodwin Creek Watershed, Mississippi: role of soil prop-
erties and past land use. Earth Surface Processes and Landforms 40:
1448–1463. https://doi.org/10.1002/esp.3727.
Wilson GV, Rigby JR, Ursic M, Dabney SM. 2016. Soil pipeflow
tracer experiments: 1. Connectivity and transport characteristics. Hy-
drological Processes 30: 1265–1279. https://doi.org/10.1002/
hyp.10713.
Wilson GV, Nieber JL, Fox GA, Dabney SM, Ursic M, Rigby JR. 2017.
Hydrologic connectivity and threshold behavior of hillslopes with
fragipans and soil pipe networks. Hydrological Processes 31(13):
2477–2496. https://doi.org/10.1002/hyp.11212.
Earth Surf. Process. Landforms, (2020)
 ROLE OF SOIL PIPE AND PIPEFLOW IN HEADCUT MIGRATION PROCESSES
Wilson GV, Wells RR, Kuhnle R, Fox GA, Nieber J. 2018. Sediment de-
tachment and transport processes associated with internal erosion of
soil pipes. Earth Surface Processes and Landforms 43: 45–63. https://
doi.org/10.1002/esp.4147.
Xu X, Zheng F, Wilson GV, Wu M. 2017. Upslope inflow, hillslope gra-
dient and rainfall intensity impacts on ephemeral gully erosion. Land
Degradation & Development 28(8): 2623–2635. https://doi.org/
10.1002/ldr.2825.
Yamasaki T, Imoto H, Hamamoto S, Nishimura T. 2017. Determination
of the role of entrapped air in water flow in a sloped soil pipe using a
© 2020 John Wiley & Sons, Ltd.
laboratory experiment. Hydrological Processes 31: 3740–3749.
https://doi.org/10.1002/hyp.11302.
Zheng FL, Huang C, Norton LD. 2000. Vertical hydraulic gradient and
run-on water and sediment effects on erosion processes and sedi-
ment regimes. Soil Science Society of America Journal 64: 4–11.
https://doi.org/10.2136/sssaj2000.6414.
Zhu TX. 2012. Gully and tunnel erosion in the hilly Loess Plateau re-
gion, China. Geomorphology 153: 144–155. https://doi.org/
10.1016/j.geomorph.2012.02.019.
Earth Surf. Process. Landforms, (2020)