Geomorphology 238 (2015) 27–36

Contents lists available at ScienceDirect

Geomorphology
journal homepage: www.elsevier.com/locate/geomorph

Parallel retreat of rock slopes underlain by alternation of strata

Fumitoshi Imaizumi a,⁎, Ryoko Nishii b, Wataru Murakami c, Hiromu Daimaru c
a
Graduate School of Agriculture, Shizuoka University, 836 Ohya, Suruga, Shizuoka 422-8529, Japan
b
Public Works Research Institute, 1-6 Minamihara, Tsukuba, Ibaraki 305-8516, Japan
c
Forestry and Forest Products Research Institute, 1 Matsunosato, Tsukuba, Ibaraki 305-8687, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Characteristic landscapes (e.g., cuesta, cliff and overhang of caprock, or stepped terrain) formed by differential
Received 22 August 2014 erosion can be found in areas composed of variable geology exhibiting different resistances to weathering. Paral-
Received in revised form 1 February 2015 lel retreat of slopes, defined as recession of slopes without changes in their topography, is sometimes observed on
Accepted 3 February 2015
slopes composed of multiple strata. However, the conditions needed for such parallel retreat have not yet been
Available online 6 March 2015
sufficiently clarified. In this study, we elucidated the conditions for parallel retreat of rock slopes composed of al-
Keywords:
ternating layers using a geometric method. In addition, to evaluate whether various rock slopes fulfilled the con-
Alternating layers ditions for parallel retreat, we analyzed topographic data obtained from periodic measurement of rock slopes in
Erosion rate the Aka-kuzure landslide, central Japan. Our geometric analysis of the two-dimensional slopes indicates that dip
Geometric model angle, slope gradient, and erosion rate are the factors that determine parallel retreat conditions. However, dip
Japan angle does not significantly affect parallel retreat conditions in the case of steep back slopes (slope
Landslide gradient N 40°). In contrast, dip angle is an important factor when we consider the parallel retreat conditions
Parallel retreat in dip slopes and gentler back slopes (slope gradient b 40°). Geology in the Aka-kuzure landslide is complex be-
cause of faulting, folding, and toppling, but spatial distribution of the erosion rate measured by airborne LiDAR
scanning and terrestrial laser scanning (TLS) roughly fulfills parallel retreat conditions. The Aka-kuzure landslide
is characterized by repetition of steep sandstone cliffs and gentle shale slopes that form a stepped topography.
The inherent resistance of sandstone to weathering is greater than that of shale. However, the vertical erosion
rate within the sandstone was higher than that within the shale, due to direct relationship between slope gradi-
ent and vertical erosion rate in the Aka-kuzure landslide.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
Knowledge of the spatial distribution of weathering and erosion
rates is essential in predicting landscape development and sediment di-
sasters. Therefore, numerous investigations have been conducted on
weathering and erosion processes in various geological and environ-
mental settings (e.g., Coleman, 1981; Maybeck, 1987; Alexandrowicz,
1989; Turkington and Paradise, 2005). Resistance against physical,
chemical, and biological weathering is different not only between geo-
logic units (e.g., Dibb et al., 1983; Pentecost, 1991; Hettema et al,
1998; Mutlutürk et al., 2004) but also within the same geological unit
due to inhomogeneous joint density and the chemical composition of
different rock types (Nicholas and Dixon, 1986; Schmidt, 1989). Charac-
teristic landscapes (e.g., cuesta, cliff and overhang of caprock, or stepped
terrain) formed by differential erosion can be found in areas comprising
multiple strata with different weathering resistances (e.g., Moon and
Selby, 1983; Nicholas and Dixon, 1986; Schmidt, 1989; Matsuoka,
⁎ Corresponding author. Tel./fax: +81 54 238 4845.
E-mail address: imaizumi@cii.shizuoka.ac.jp (F. Imaizumi).
1995; Suzuki, 2002). However, many of these studies have been con-
ducted in regions with a simple coherent formation, and relatively few
have focused on complex geology such as disrupted formations and
deformation.
Parallel retreat of slopes, defined as the recession of slopes without
changes in the slope surface topography, is a typical type of landform
development throughout the world (e.g., Kirk, 1977; Obanawa and
Matsukura, 2008). In cases in which the slopes are not completely flat,
the direction of the parallel retreat can be determined by the travel di-
rection of knick lines. Parallel retreat progresses not only in the vertical
direction but also sometimes horizontally and at right angles to the
slope surface (e.g., Kirk, 1977; Obanawa and Matsukura, 2008). Regard-
less, the erosion rate along the parallel-retreat direction should be spa-
tially constant. However, few studies have considered the dip angle of
the strata, which we believe is an important factor controlling the direc-
tion of the parallel retreat. In regions with high tectonic activity, geolog-
ical strata may be inclined, and the dip angle can affect sediment supply
processes (Chigira and Kiho, 1994; Lin et al., 2004; Nishii and Matsuoka,
2010; Tsou et al., 2011). Thus, an exploration of the relationship be-
tween parallel slope retreat and dip angle is needed to clarify the land-
form development of rock slopes in tectonically active regions.

http://dx.doi.org/10.1016/j.geomorph.2015.02.030
0169-555X/© 2015 Elsevier B.V. All rights reserved.
28 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36

Topographic factors such as slope gradient and slope aspect are also
important in terms of controlling weathering and erosion rates
(Bradley, 1963; Alexandrowicz, 1989; Pentecost, 1991; Hall, 1993,
2004). Erosion rates vary over time when the topographic factors asso-
ciated with landform development change, so clarifying the interactions
between weathering and landform development is essential to ade-
quately understand weathering processes (Alexandrowicz, 1989;
Phillips, 2005).
The purpose of this study was to explain the conditions for parallel
retreat of rock slopes comprising alternating layers using a geometric
model. In the model, dip angle of the strata, which is sometimes a key
factor controlling the rate of slope retreat and the slope gradient
(Nicholas and Dixon, 1986; Suzuki, 2002), was considered as one pa-
rameter defining the conditions for parallel retreat. Furthermore, to
evaluate whether our geometric model could be applied to the short-
term erosion rate of rock slopes with complicated geological settings,
we analyzed topographic data obtained from the periodic measurement
of rock slopes in the Aka-kuzure landslide, central Japan, by airborne
light detection and ranging (LiDAR) scanning and ground terrestrial
laser scanning (TLS). The Aka-kuzure landslide, similar to rocky slopes
in many other regions (Selby, 1982; Moon and Selby, 1983; Matsuoka,
1995), consists of alternating strata with stepped terrain. The geology
of this landslide, located in a humid and tectonically active region
where this type of study has rarely been conducted, features alternating
sandstone and shale affected by faulting, folding, and toppling (Kano
and Matsushima, 1988). Because this landslide continuously produces
large sediment volumes after its initial failure (12.2 × 104 m3 year−1 ac-
cording to Chigira and Kiho, 1994), its topographic changes can be de-
tected in a short time period. We also discuss the relationship
between topography and the erosion rate that satisfies parallel retreat
conditions. In this study, we focus on small-scale slope retreat of alter-
nating layers in which hillslope sediment movement such as rockfall
and surface erosion plays an important role in landscape development
(Phillips, 2005). We do not consider fluvial processes, which commonly
play an important role in landscape development on the scale of an en-
tire hillslope or catchment.
2. Conditions for parallel retreat at alternating layers
To improve our understanding of parallel retreat, we discuss here
the conditions for parallel retreat on a two-dimensional (2-D) slope
comprising alternating layers. In the case of inclined strata, lithology
along the horizontal (except in the direction of strike) and vertical di-
rections from a certain point is spatially variable (Fig. 1). Therefore,
the spatial distribution of horizontal and vertical erosion rates is not
simply comparable to that of the lithology. In this paper, we discuss con-
ditions for parallel retreat along dip direction (Fig. 1), because a single
lithology along this direction enables us to derive simple conditions
for parallel retreat. Knick lines of slopes comprising multiple lithologies
usually exist along the boundary where lithology changes (e.g., Selby,
1982; Suzuki, 2002). If knick lines continue to be located on the bound-
ary, the direction of parallel retreat should be consistent with the dip di-
rection. Therefore, we believe that a discussion of parallel retreat in the
dip direction is helpful in terms of understanding the retreat of slopes
comprising multiple lithologies.
One of the key conditions for the parallel retreat of alternating strata
is that the erosion rate in the dip direction is constant among strata:
Ed ðiÞ ¼ const ð1Þ
in which Ed(i) is the erosion rate of the ith stratum along the dip direc-
tion (m yr−1; Fig. 2). Ed(i) can be expressed by
Vi
EdðiÞ ¼ ð2Þ
Di
in which Vi is the cross-sectional area of the eroded rock in the ith stra-
tum per year (m2 yr−1), and Di is the width of the ith stratum (m). In ad-
dition, Vi can be expressed as
V i ¼ Ai EvðiÞ ¼ ai Di EvðiÞ ð3Þ
in which Ev(i) is the erosion rate along the vertical direction for the ith
stratum (m yr− 1), Ai is the vertically projected area of the exposed
part of the ith geology (m), and ai indicates the ratio of Ai to Di. By
using slope gradient θi and dip angle ϕ, ai for back slopes is
cosθi
ai ¼ ð4Þ
sinðθi þ ϕÞ
and ai for dip slopes (θ ≠ ϕ) is
cosθ
ai ¼ : ð5Þ
j sinðθi −ϕÞj

(a) Initial slope surface (b)

Current slope surface
3
2
2
1 3 1

(c) 2 1: Vertical direction

1 2: Horizontal direction
3
3: Dip direction

Fig. 1. Vertical, horizontal, and dip directions of a slope with inclined alternating strata: (a) back slope, (b) dip slope with a dip angle below the slope gradient, and (c) dip slope with a dip
angle that exceeds the slope gradient. In each case, the geology is spatially variable along the vertical and horizontal directions whereas it is constant along the dip direction.
 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36 29

(a) Former Then, the following condition for the parallel retreat is given from
surface Eqs. (1) and (6):
Ai - 1 Ai Ai + 1
Ed(i+1) ai EvðiÞ ¼ const: ð7Þ
θi+1 Di+1
Ed(i) φ Ev(i+1) Note that the width of layer Di, which is an important factor control-
ling the erosion rate of the entire slope (Schmidt, 1989), is not included
(i + 1)th
Ev(i) layer in Eq. (7) as a parameter. Eq. (7) indicates that Ev(i) should be high in the
Ed(i-1) θi strata with a low ai value if the slope is in the parallel retreat state.
φ Di To clarify the influence of dip angle ϕ on the ai value, we investigated
φ θi-1 Ev(i-1) i-th layer the relationship between θi and ai under various ϕ values using sensitiv-
Current ity analysis (Fig. 3). In the case of the back slope, ai has a negative rela-
surface tionship with θi, indicating that Ev(i) should be higher at the strata with a
Di -1
steeper slope surface. In terms of the back slope with steep slope gradi-
(i - 1)th layer ent (i.e., N 40°), dip angle ϕ does not have much influence on the rela-
tionship between ai and θi. In other words, the important parameters
for the parallel retreat condition of the steep back slope are not ϕ, but
rather are ai and θi (Fig. 3a). In the case of a gentle back slope with a
(b) Ai - 1 Ai Ai + 1 Di+1 low dip angle (e.g., θi b 40°, ϕ = 10°), the slope of the relationship be-
tween θi and ai is steep (Fig. 3a). Therefore, even if the difference in
(i + 1)th Ev(i) value between the two rock types is large, the difference in θi be-
Ev(i+1) layer tween those rock types needed for parallel retreat condition is small.
Ed(i+1)
The erosion rate at a right angle to the slope surface can be obtained
i-th layer by Ev(i) cosθi. Changes in cosθi with increases in θi are very close to
θi+1 D changes in ai at the steep back slope (i.e., θi N 40°; Fig. 3a). As such, the
i
Ed(i) erosion rate at a right angle to the slope (=Ev(i) cosθ) is almost constant
among layers when parallel retreat occurs.
Ev(i) θi Di -1 (i - 1)th
layer The influence of ϕ on ai value at the dip slope is markedly different
φ from the case of a back slope (Fig. 3b). Affected by the divergence
Ed(i-1) around θi = ϕ, changes in ai value with increases in θi differ largely
among different ϕ values. Therefore, we need to consider dip angle
φ θi-1 when we discuss the parallel retreat of dip slopes. Slope gradient θi
Ev(i-1) and ai have a negative relationship when θi exceeds ϕ but a positive
Former Current one if θi is less than ϕ. Therefore, to satisfy the condition for parallel re-
surface surface treat expressed as Eq. (7), the erosion rate along the vertical direction
Ev(i) should be higher on steeper slopes for the former case (θi N ϕ),
Fig. 2. Schematic diagram of alternating layers: (a) back slope and (b) dip slope. whereas Ev(i) should be lower on steeper slopes for the latter case
(θi b ϕ). In the case of θi = ϕ, due to preferential erosion of the weaker
lithology, harder rock caprock with wide plateau surfaces and scarps
of weaker rocks appear alternatively, as seen around the world
By substituting Eq. (3) into Eq. (2), the erosion rate along the dip di-
(e.g., Schmidt, 1989).
rection is given as
Note that our model was developed not for the purpose of prediction
or reconstruction of the development of the topography, but rather for
EdðiÞ ¼ ai EvðiÞ : ð6Þ determining whether the terrain is in a state of parallel retreat. The

(a) (b)
3 15
Dip angle Dip angle
(o) (o)
10 10
30 30
2 50 10 50
70 70
ai,cos θi

cos θi
ai

1 5

0 0
10 30 50 70 90 10 30 50 70 90
θi θi

Fig. 3. Changes in ai with increases in slope gradient θi under given dip angle ϕ: (a) back slopes and (b) dip slopes. Changes in cosθi are also presented in (a).
30 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36

influence of boundary conditions at the upper and lower edges of the
slope on the development of slope topography was not discussed in
this study, because our model does not include terms expressing spatial
continuity of the sediment and water fluxes. Therefore, this study was
conducted far from the upper and lower edges of the slope where the
influence of hydrogeomorphic processes at slope boundaries is consid-
ered limited.
3. Field observation of rock-slope retreat
3.1. Study site
We conducted periodical measurements of the rock slopes in the
Aka-kuzure landslide to verify the findings obtained by our geometric
analysis (Eq. (7)). The Aka-kuzure landslide, with an estimated total
volume of 2.7 × 107 m3 and an area of 4.1 × 105 m2 (Chigira and Kiho,
1994; Chigira, 2000), is a large rockslide–avalanche area in central
Japan (Fig. 4). Historical records are lacking regarding the exact time
of occurrence, but this landslide has existed since at least 100 years
ago (Lin and Oguchi, 2004). Elevation of the landslide ranges from
1200 to 1900 m a.s.l.
The geologic unit belongs to the Cretaceous Shimanto Group and
consists mostly of argillaceous rocks (mainly shale) and sandstone. Nu-
merous linear depressions trend northeast–southwest around the
upper part of the landslide, affected by mass rock creep in this area
(Chigira and Kiho, 1994). Sandstone, alternations of sandstone and
shale, and shale are distributed in and around the Aka-kuzure landslide
(Fig. 5). Basal geology, observed at the northwest part of the landslide
scar, is dip slope trending northeast–southwest. Dip ranges from 40°
to 80° due to faulting and folding. Toppling of the rock slope also affects
dip angle: the surface part of the slope around our measurement area at
the southeastern part of the landslide features back slopes dipping 20°
to 40° southeastward (Chigira and Kiho, 1994; Chigira, 2000). We
could not map detailed geological structure in the landslide area be-
cause of the steep and unstable rock slopes. Therefore, our geological
analyses were mainly focused on surficial lithology, which can be
mapped by eye from outside of the landslide, as well as representative
dip angles throughout the study area.
The Aka-kuzure landslides have produced a large volume of sedi-
ment after the initial failure; annual sediment supply volume in the pe-
riod from 1964 to 1982 was 12.2 × 104 m3 yr− 1, according to aerial
photograph analysis (Chigira and Kiho, 1994). The volume of landslide
sediment evacuated at the time of initial failure would have been
much lower than the current landslide volume (2.7 × 107 m3) because
only a limited amount of sediment can pass through the narrow V-
shaped valley with a small cross-sectional area at the lower end of the
landslide. This indicates that the Aka-kuzure landslide has continuously
been supplying sediment since the initial failure. Active sediment sup-
ply in the entire landslide area prevents the recovery of vegetation,
which results in the rocky and denuded landscape visible today
(Fig. 5b). Affected by weak lithologies, the erosion rate of the Aka-
kuzure landslide (N0.2 m yr−1; Chigira and Kiho, 1994) is much higher
than that in the areas studied previously (generally b0.01 m yr−1;
e.g., Kirk, 1977; Mottershead, 1989; Pentecost, 1991). This is interesting
from a research standpoint, because it means that topographic change
can be observed even on short time scales. The slope gradient of the
landslide (35–80°) exceeds the angle of repose (about 35–38° in the
Aka-kuzure landslide). As such, weathered residues are almost absent
in the sandstone and shale bedrock throughout the year, due presum-
ably to continual rockfall. Therefore, the Aka-kuzure landslide can be
considered a weathering-limited system: direct relationships between
sediment production by weathering and removal as rockfall dominates.

0 200 400 m 138˚ 21’ 20’’ E

Japan Sea
Aka-kuzure
landslide Pacific
35˚ 21’ 30’’ N Ocean
1000
1100
0
100

Oi River
Alluvial fan
0
0
140
130

00
16

0
0

0
0

17
12

00
18

0
130
1500

Site B

Landslide scar Fig. 5

Site A
Geology in the landslide
Sandstone-dominant area
0

Shale-dominant area
190

Alternations of
sandstone and shale

Fig. 4. Topographic and geological map of the Aka-kuzure landslide.

 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36
(a)
Fig. 5. View of the Aka-kuzure landslide. (a) Full view of the Aka-kuzure landslide. Upper left of the photograph shows the landslide scar and lower right shows the alluvial cone (June 13,
2008). (b) View around measurement site at the Aka-kuzure landslide (June 3, 2011). Dashed line indicates boundary of lithologies in the area with alternating sandstone and shale — ss
and sh indicate sandstone and shale zones, respectively.
Rocks detached from rock slopes are directly transported into gullies
deeply dissecting the landslide scar. Sediment accumulating in the
gullies is periodically evacuated by debris flows and forms a large allu-
vial cone (volume: 5 × 105 m3) at the downslope end of the landslide
(Ikeda et al., 1993; Chigira and Kiho, 1994; Sidle and Ochiai, 2006;
Fig. 5a).
The area around the Aka-kuzure landslide receives abundant
rainfall with a mean annual precipitation of 2800 mm for the period
1993–2002 (Imaizumi et al., 2010). Heavy rainfall events (i.e., total
rainfall N 100 mm) occur during the Baiu rainy season (June and July)
and in the autumn typhoon season (late August to early October). Win-
ter snowfall occurs from December to March, and precipitation in this
period accounts for only about 15% of the total annual precipitation. Be-
cause of the steep topography, the annual maximum depth of snow
cover on slopes in the landslide is less than 20 cm and most of it melts
within 1 week of snow fall.
We conducted periodic field measurements of the topography of the
rock slopes at the southern part of the Aka-kuzure landslide. The rock
slopes in the measurement area are back slopes composed of alternating
sandstone and shale. The thickness of each stratum in and around this
site ranges from 3 to 15 m. The slope gradient of the sandstone
(60–80°) is generally steeper than that of the shale (35–45°). Therefore,
the rock slope in the measurement area presents a stepped terrain as a
whole.
3.2. Survey method
The stepped terrain in the landslide was measured four times from
2003 to 2011. Geodetic surveys in 2003 and 2007 were performed by
airborne LiDAR scanning (Table 1) conducted after typhoon season
(usually from August to October) and before the snow season (from
the middle of December). The accuracy of the measurements by laser
devices is affected by both device performance (accuracy of irradiation
angle, degree of laser diffusion, and accuracy of timing measurement
when laser pulse returns from ground objects) and measurement
Table 1
General information related to DEMs.
Measurement day Type of measurement
October 18, 2003 Airborne
December 1, 2007 Airborne
October 22, 2010 Ground
June 3, 2011 Ground
a
Ministry of Agriculture, Forestry and Fisheries, Japan.
31
(b)
ss
sh
ss
sh
ss
sh
conditions (measurement distance, slope angle, and type of ground ob-
jects). The maximum error in our airborne LiDAR scanning along the
vertical direction was about 30–40 cm. The positioning of aircraft was
other sources of potential measurement errors.
The other two measurements (2010 and 2011) were performed
using ground TLS. The TLS measurements were done with a portable
laser scanner (Optech Inc., ILRIS-3D) featuring a measuring range of
1200 m and an accuracy of 7 mm (for a measurement distance of
100 m) from both sides of the landslide (sites A and B, Fig. 4) in October
2010 and from the west side only (site A) in June 2011. Slopes in the
Aka-kuzure landslide are composed of repetitions of concave–convex
topography affected by deep gully incision after the initial formation
of the landslide. It was not possible to scan the topography of shaded
slopes by TLS. We also determined the laser reflection ratio to obtain
basic data for an estimation of the spatial distribution of the surface li-
thology. Because point cloud data obtained by TLS measurement are
presented as a relative coordinate system from the laser scanner, we
conducted static GPS positioning at two ground control points (GCPs)
and converted the coordinate system of the TLS data into that of the air-
borne LiDAR data (JGD 2000 orthogonal coordinate system). The GCPs
were located at the head of the landslide. It was not possible to set up
any GCPs in the landslide itself due to impossibility of access. We then
interpolated the point cloud data by kriging and created two digital el-
evation models (DEMs) from the data in 2010: one created from the
point cloud data measured from two sites (sites A and B) and one
from only a single site (site A). The former DEM covers west- and
east-facing slopes and has a larger error ratio due to the combination
of point clouds obtained at two different sites. This DEM was compared
with that derived from airborne LiDAR data, whereas the latter DEM
was created only for calculating the topographical changes in the period
from 2010 to 2011. In this period, changes in the elevation were much
smaller than in the other periods because of the short length of the pe-
riod (about 7 months). Errors in the measurements and analysis need to
be less than the erosion rate to detect such small changes in the topog-
raphy. By creating a DEM from the data measured at a single site, we
Grid size (m) Distance between laser spots (m) Performer of the measurement
0.5 b1.0 MAFFa
1.0 b1.5 University of Tsukuba
0.5 b0.3 Authors of this paper
0.1 b0.2 Authors of this paper
32 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36
avoided potential errors stemming from arranging point cloud data ob-
tained at multiple sites. This DEM covers west-facing but not east-facing
slopes. To ensure precise comparison with the 2010 DEM (from the sin-
gle measurement site), close attention was paid to coordinate system
conversion of the 2011 point cloud data via matching of stable points
free of erosion (a total of six points in the landslide) at the same loca-
tions as in the 2010 DEM, rather than a precise arrangement of GCPs
for static GPS positioning. Thus, the 2011 DEM is suitable for comparison
with the 2010 DEM and has larger errors when compared with DEMs
from 2003 to 2007. A comparison of photographs taken in 2010 and
2011 showed that no erosion occurred at the points used for matching
the 2000 and 2011 data.
4. Results
4.1. Survey results
The laser reflection ratio of the sandstone colored white or light gray
was generally higher compared to the dark gray shale. Therefore, spatial
distribution of the sandstone strata can be detected as high-reflection
areas on the laser reflection map obtained by TLS measurement. Loca-
tions of the high-reflection areas correspond to steep sandstone cliffs
in the topographic map (Fig. 6). In contrast, low-reflection areas are
generally located on the gentler shale slopes.
The erosion rate in the vertical direction was calculated from the dif-
ference in elevation of two DEMs measured in different periods divided
by the length of time between the two periods (Fig. 7a–c). Spatial pat-
tern of the erosion rate agrees with that of the surficial lithology; active
erosion zones located at steep sandstone cliffs (generally N0.2 m yr−1)
and inactive erosion zones located at gentle shale slopes (generally
b0.2 m yr−1) alternated in the measurement area in all periods. In the
period from 2003 to 2007, a wide and deep erosional feature (width
of about 30 m, depth of N1 m) was identified in the southwest part of
the measurement area. Because similar erosion was not identified at
the same location in other periods, this erosion is considered an episodic
Reflectivity 0 5 10 15 20m
Max: 255
Min: 0
60
17
80
17
00
18
00 20
18 18
40
18
18
Fig. 6. Topography and distribution of the laser reflectivity in the area with alternating
sandstone and shale obtained by TLS measurement in October 2010. High reflection
areas (red) correspond to sandstone steep cliffs and low reflection areas (blue) are gener-
ally located on the shale gentle slopes.
slope failure and is not related to the parallel retreat in the time–space
scale of our observation. Average erosion rate in the measurement
area in each period ranges from 0.22 to 0.51 m yr−1 (Table 2). Erosion
rate in the period from 2003 to 2007 is the highest because of the
deep slope failure in the southwest part of the measurement area. Aver-
age erosion rate, excepting this deep slope failure, in the period from
2003 to 2007 (0.28 m yr−1) was similar to that in other periods. There-
fore, a retreat of the measurement area can be considered a chronic pro-
cess on a time scale of several years.
Errors in our measurement and analysis of DEMs were evaluated
using standard errors calculated from the difference in elevation at sta-
ble areas between two different periods (Table 2). Stable areas used for
the evaluation of errors were found based on periodic photography at
the measurement area in different periods. The standard errors were
calculated for steeper terrains (N 50°) and gentler terrains (≤50°) sepa-
rately to check the effect of the slope gradient on the errors. Both stan-
dard errors in gentler and steeper terrains (ranging 0.02–0.12 and
0.04–0.18 m, respectively) were lower than the changes in the topogra-
phy (erosion rate multiplied by the duration of period) in each period,
indicating that our analysis has sufficient accuracy to detect changes
in the topography in the Aka-kuzure landslide. Standard errors in the
steeper terrains were larger than those in the gentler terrains
(Table 2). Similar trends were also found in the evaluation of errors in
DEMs measured by airborne LiDAR scanning (Su and Bork, 2006;
Ashraf et al., 2012).
Differences in the erosion rate between the two lithologies are ap-
parent in the changes in the longitudinal profile of topography
(Fig. 8a). The erosion rate is spatially variable in each stratum, but that
in the sandstone strata is generally higher than that in the shale strata
(Fig. 8b). As shown in Fig. 8, the average erosion rates in the period
from 2003 to 2010 in sandstone strata range from 0.27 to 0.43 m yr−1
(average: 0.33 m yr−1), much higher than that in shale strata, which
range from 0.15 to 0.22 m yr−1 (average: 0.20 m yr−1, Fig. 8c). A signif-
icant difference in the mean erosion rate between sandstone and shale
was identified by t-test (p-value b 0.01). Standard deviations of the ero-
sion rates in each stratum were also different between sandstone and
shale; the standard deviation in the sandstone strata (0.09–0.10) is
markedly higher than that in the shale strata (0.02–0.07). The higher
standard deviations in the sandstone strata are likely affected by the
larger measurement and analysis errors in steeper topography
(Table 2).
Looking at the entire longitudinal topography in the measurement
area, the topography maintains a stepped terrain in all periods despite
the different erosion rates between sandstone and shale (Fig. 8a). In
previous studies, the retreat of slopes was classified into parallel retreat,
declining retreat, and a combination of the two (e.g., Obanawa and
Matsukura, 2008). Because slope gradient, height, and horizontal
width of each sandstone and shale section do not clearly change, paral-
lel retreat seems to dominate at the alternating strata in the Aka-kuzure
landslide. The retreat direction of the stepped terrain roughly corre-
sponds with the dip angle of the strata (about 30°; Fig. 8a); steep sand-
stone cliffs went back along the dip direction of the strata.
4.2. Application of parallel retreat condition to the survey results
The erosion rate along the dip direction, Ed(i), in the measurement
area was calculated from the erosion rate along the vertical direction
(Fig. 7a–c) using Eqs. (4) and (6) (Fig. 7d–f). The dip angle around the
measurement area is spatially variable (20° to 40°), and for this calcula-
tion we set it to 30°, the mean of the dip-angle range. The contrast of
6 0
Ed(i) between sandstone and shale was much less clear than that along
the vertical direction (Fig. 7d–f).
Ed(i) in sandstone and shale in the period 2003 to 2010 is similar on
the longitudinal map (Fig. 8b). Difference in the mean erosion
rate between sandstone and shale was not clear in the t-test result (p-
value = 0.11). The average Ed(i) in each sandstone stratum ranges
 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36 33

(a) 2003–2007 0 5 10 15 20 m
(d) 2003–2007 0 5 10 15 20 m

2 0
17
Vertical 1740 Erosion rate 40
17
erosion rate along dip
(m yr-1) direction
< -1 (m yr-1)

0
0

8
8
-1
60 6<0 -1

17
17
170 17 -1

1 0
>1 1
>1

0
0

184
184
0

0
0

0
18

18
0 0
182 182

6 0 6 0
18 18

(b) 2007–2010 0 5 10 15 20 m (e) 2007–2010 0 5 10 15 20 m

20
17
Vertical 740 0
60 Erosion rate
74
1
erosion rate 17 along dip1
(m yr-1) direction
< -1 (m yr-1)
-10 <
6 60-1
170 17-1
80

80
17

17
1 0
>1 1
>1

00 20 00 0
18 18 18 182

1840 184
0
1

(c) 2010–2011 0 5 10 15 20 m
(f) 2010–2011 0 5 10 15 20 m
20
17
60

Vertical 740 0
17

1 Erosion rate74
erosion rate along dip1
(m yr-1) direction
< -1 Fig. 8 (m yr-1)
0

-10
8

< -1
17

6 60
170 17 -1

1 0
80

>1
17

1
>1
0
184

0
184
00

0
0
18

18

0
182 18
2 0

6 0
18

Fig. 7. Spatial distribution of erosion rate along vertical direction and dip direction on rock slopes in the Aka-kuzure landslide: (a) erosion rate along vertical direction from 2003 to 2007,
(b) erosion rate along vertical direction from 2007 to 2010, (c) erosion rate along vertical direction from 2010 to 2011, (d) erosion rate along dip direction from 2003 to 2007, (e) erosion
rate along dip direction from 2007 to 2010, and (f) erosion rate along dip direction from 2010 to 2011. A positive value indicates the erosion of sediments while a negative value indicates
the deposition of sediments. The dashed line in (c) indicates the longitudinal line in Fig. 8. Counter lines in the figure were drawn on the basis of measurement data at the end of each
period.

from 0.12 to 0.23 m yr−1 (average 0.16 m yr−1), similar to that in each
shale stratum ranging from 0.11 to 0.15 m yr−1 (average: 0.13 m yr−1,
Fig. 8d). Standard deviation of the erosion rate in sandstone (0.01–0.04)
did not clearly differ from that in shale (0.01–0.05). Consequently, the
conditions for parallel retreat, described in Eq. (7), are almost fulfilled
in the Aka-kuzure landslide. Although the geology in the Aka-kuzure
is complex because of folding, faulting, and toppling, the short-term
erosion rate in the Aka-kuzure landslide agreed with our parallel retreat
model using the surficial lithologic and topographic data. Spatial vari-
ability of the Ed(i) between strata was less clear than that of Ev(i) (see
34 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36

Table 2 relationship on the back slope (Fig. 3). Therefore, erosion rate Ev(i) at
Average erosion rate along vertical direction in the measurement area. Standard errors on the steeper layer should be higher than that at the gentle layer to
DEM analyses at the stable areas in the Aka-kuzure landslide, calculated from the differ-
ence in elevation of two DEMs measured in different periods, were also listed in the table.
satisfy the condition for parallel retreat expressed as Eq. (7). In the
Aka-kuzure landslide, the erosion rate of the steeper sandstone exceeds
Period Erosion rate (m yr−1) Standard error that of the gentler shale (Figs. 7a–c and 8), corresponding to the rela-
at stable areas
tionship given by Eqs. (4) and (7). However, the inherent weathering
b50° ≥50° resistance of the sandstone is generally higher than that of the shale
October 18, 2003–December 1, 2007 0.51 0.07 0.07 around the Aka-kuzure landslide (Nagai and Nakamura, 2000). Joint
December 1, 2007–October 22, 2010 0.31 0.12 0.18 spacing in the sandstone (generally N 10 cm) is much wider
October 22, 2010–June 3, 2011 0.22 0.02 0.04 than that in the shale (generally form several millimeters to 10 cm)
that has high-density joints along slaty cleavage (Kano et al., 1986,
1991; Matsuoka and Sakai, 1999). Sandstone sediment which accumu-
mean erosion rates in Fig. 8c, d). In addition, spatial variability of Ed(i) in lated at the talus cone and in gullies is much smaller than shale sedi-
a stratum is also smaller than that of Ev(i) (see standard deviations of the ment accumulation, indicating that the shale is susceptible to surficial
erosion rate in Fig. 8c, d). Thus, spatial representation of Ed(i) is better rock weathering activities such as diurnal freeze–thaw that actively
than that of the Ev(i) in the Aka-kuzure landslide. occurs in the region (Imaizumi et al., 2006). As with the Aka-kuzure
landslide, a difference in slope gradient among lithologies is found
5. Discussion in other areas: lithologies with higher weathering resistance (or
higher strength) usually present as steep slopes, whereas lithologies
Slope gradient is an important factor when we discuss parallel re- susceptible to weathering (or lower strength) present as gentler slopes
treat because it determines the ai value in Eq. (7) expressing parallel re- (Selby, 1982; Moon and Selby, 1983; Matsuoka, 1995). Therefore,
treat conditions. Eq. (4) indicates that θi and ai have a negative a positive relationship between weathering resistance and slope

1830
(a)
Elevation of ground surface (m)

Shale
1820
Sand-
stone
Shale
Sand-
stone
1810

Shale
1800
2003
2007
Sandstone 2010
1790
0
Erosion rate

(b)
(m yr-1)

0.2

0.4 Along vertical direction

Along dip direction
0.6
0
(c) Along vertical direction
0.2
in each stratum (m yr-1)
Average erosion rate

0.4

0.2

(d) Along dip direction

0.4

0 5 10 15 20 25
Horizontal distance (m)

Fig. 8. Temporal changes in longitudinal profile in the area with alternating sandstone and shale: (a) longitudinal profiles of ground surface from 2003 to 2010, (b) erosion rate along ver-
tical direction and dip direction between 2003 and 2010, and (c) average erosion rate in each stratum along the vertical direction and (d) that along the dip direction between 2003 and
2010. Geology is identified on the basis of the spatial distribution of laser reflectivity obtained by TLS measurement. The inclination of the geology in Fig. 8a indicates the dip angle in this
area. Location of the longitudinal line is drawn in Fig. 7c.
 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36
gradient, found in the Aka-kuzure landslide, is considered a non-
exceptional case.
Steep topography seems to be a factor facilitating the weathering
and erosion processes of the strong sandstone for the following reasons.
First, steep cliffs receive higher radiation in winter due to the lower sun
angle (Hall, 2004), resulting in a higher rock temperature in the day-
time. In contrast, decreases in rock temperature on winter mornings
are more significant on steeper slopes having a larger surface area per
unit area on the horizontal plane, affected by the higher thermal radia-
tion. These meteorological effects may accelerate diurnal freeze–
thaw processes, which are considered to dominate the weathering
and erosion processes around the Aka-kuzure landslide (Nagai and
Nakamura, 2000; Imaizumi et al., 2006). Second, the expansion of
rock, caused by denudation of surficial rock (Bradley, 1963), easily oc-
curs when the surface area increases, thus accelerating the erosion
rate. In addition, weathered rock easily detaches on steeper slopes be-
cause of the larger gravitational force (Antoniou and Lekkas, 2010), pos-
sibly leading to the high erosion rate. Consequently, a positive
relationship between slope gradient and erosion rate is needed to satis-
fy the condition for parallel retreat in areas in which stronger lithology
presents steeper terrain.
Weathered residues, which cover bedrock and affect the weathering
rate by changing the thermal and moisture conditions in other areas
(Coleman, 1981; Trustrum and DeRose, 1988; Phillips, 2005), are absent
on the bedrock in the measurement area. Such simple retreat conditions
may also relate to fulfillment of our parallel retreat conditions. In the
gentler areas covered by the weathered residuals, depth of residuals at
a specific location is not just controlled by the weathering rate of the
rock, but by influx and efflux of the sediment (Heimsath et al., 1997,
2001). In such areas, weathering rate is affected by the sediment flux
between surrounding strata together with topographic and lithological
properties of the strata.
Thickness of strata is another factor affecting fulfillment of parallel
retreat. If the thickness of sandstone or shale strata is similar to
the joint spacing, sediment removal can occur in adjacent strata.
In such cases, recession rate is not simply controlled by the rock proper-
ties and topography in the specific stratum, but also by erosion in sur-
rounding strata. Because thickness of strata in the Aka-kuzure
landslide (generally 2–10 m) is much larger than the joint spacing
(b1 m), erosion associated with that in adjacent strata is considered
limited.
Multiple types of sediment transport processes usually contribute to
the development of topography in other areas (Roberts and Church,
1986; Imaizumi et al., 2014). Relationships between sediment transport
rate and topographic factors (e.g., slope gradient, contributing area) dif-
fer among transport process types (Matsuoka, 2001; Imaizumi et al.,
2014). Spatial nonuniformity in the contribution of each sediment
transport process could increase spatial variability in the erosion rate
within each layer, and likely interrupt fulfillment of the simple parallel
retreat model. In the steep Aka-kuzure landslide, rockfall solely trans-
ports sediment in both the sandstone and shale strata in the measure-
ment area. This simple sediment transport system fulfilled our
geometric model.
Long-time landform development of the Aka-kuzure landslide may
not satisfy parallel retreat conditions, because spatial distributions of
sandstone and shale at surficial and basal layers are different due to
folding. Changes in climatic and topographic factors such as thermal
and rainfall conditions and topography of the entire slope (Phillips,
2005), as well as episodic erosional processes identified only in the pe-
riod from 2003 to 2007 in our short-period monitoring, possibly affect
landscape development in the longer timescale. In addition, we need
to consider other factors, such as downslope gully erosion, for the dis-
cussion of the development of the topography at larger spatial scales in-
cluding an entire slope or a catchment (Phillips, 2005). Therefore,
fulfillment of parallel retreat conditions depends on both time and spa-
tial scales of landscape development.
35
6. Conclusions
In this study, we explained the conditions for parallel retreat of rock
slopes composed of alternating layers by comparing the erosion rate
along the dip direction among layers. From the development of the geo-
metric model for parallel retreat conditions and application of the
model to the topographic data obtained from the periodic measure-
ments in the Aka-kuzure landslide, central Japan, we conclude:
(1) In the case of steep terrain (slope gradient N 40°) with back-slope
geology, dip angle does not significantly affect the relationship
between slope gradient and erosion rate, which fulfills the paral-
lel retreat condition. In this case, the erosion rate at a right angle
to the slope surface is also almost constant among strata. In con-
trast, dip angle is an important factor when we consider the con-
dition of parallel retreat in dip slopes and gentler back slopes
(slope gradient b 40°).
(2) Spatial distribution of the erosion rate at alternating layers in the
Aka-kuzure landslide, which is composed of an accretionary
complex affected by folding and toppling, roughly fulfilled our
condition for parallel retreat. This implies that our model can ex-
plain the short-term parallel retreat of rock slopes with compli-
cated geology.
(3) In the Aka-kuzure landslide, alternation of steep sandstone cliffs
and gentle shale slopes forms a stepped terrain as a whole. The
inherent resistance of sandstone for weathering is higher than
that of shale; however, the erosion rate in the vertical direction
at the sandstone was higher than that at the shale, because of
the positive relationship between slope gradient and erosion
rate in the vertical direction in the Aka-kuzure landslide. This in-
dicates that the relationship between slope gradient and erosion
rate also plays an important role in the development of the
stepped terrain.
In this study, we focused on hillslope retreat and did not investigate
fluvial processes in gullies. Further study on the fluvial processes as well
as the interaction between hillslope and fluvial processes is needed to
fully elucidate landscape development in steep rocky terrain.
Acknowledgment
This work was supported by JSPS KAKENHI Grant Number 24780141
and Research and Education Funding for the Japanese Alps Inter-
Universities Cooperative Project, MEXT, Japan. The Ohigawa Chisan
Center, Ministry of Agriculture, Forestry and Fisheries, kindly provided
us with topographic data. We appreciate the Tokushu Tokai Paper Co.,
Ltd., land owner around the Aka-kuzure landslide, for allowing us to ac-
cess the site. Gratitude is also expressed to Yasuhiro Ogawa, Takashi
Miyamae (Forestry and Forest Products Research Institute), Toru Endo,
Akira Takinami, Yoshikazu Endo and Yusuke Ueji (Ikawa University For-
est, University of Tsukuba) for their help.
References
Alexandrowicz, Z., 1989. Evolution of weathering pits on sandstone tors in the Polish
Carpathians. Z. Geomorphol. NF 33, 275–289.
Antoniou, A.A., Lekkas, E., 2010. Rockfall susceptibility map for Athinios port, Santorini Is-
land, Greece. Geomorphology 118, 152–166. http://dx.doi.org/10.1016/j.geomorph.
2009.12.015.
Ashraf, M.I., Zhao, Z., Bourque, C.P.A., Meng, F.R., 2012. GIS-evaluation of two slope-
calculation methods regarding their suitability in slope analysis using high-
precision LiDAR digital elevation models. Hydrol. Process. 26, 1119–1133.
Bradley, W.C., 1963. Large-scale exfoliation in massive sandstones of the Colorado
Plateau. Geol. Soc. Am. Bull. 74 (5), 519–527. http://dx.doi.org/10.1130/0016-
7606(1963)74[519:LEIMSO]2.0.CO;2.
Chigira, M., 2000. Geological structure of large landslides in Japan. J. Nepal Geol. Soc. 22,
497–504.
36 F. Imaizumi et al. / Geomorphology 238 (2015) 27–36
Chigira, M., Kiho, K., 1994. Deep-seated rockslide–avalanches proceeded by mass rock
creep of sedimentary rocks in the Akaishi Mountains, central Japan. Eng. Geol. 38,
221–230. http://dx.doi.org/10.1016/0013-7952(94)90039-6.
Coleman, S.M., 1981. Rock-weathering rates as functions of time. Quat. Res. 15, 250–264.
http://dx.doi.org/10.1016/0033-5894(81)90029-6.
Dibb, T.E., Hughes, D.W., Poole, A.B., 1983. The identification of critical factors affecting
rock durability in marine environments. Q. J. Eng. Geol. Hydrogeol. 16, 149–161.
http://dx.doi.org/10.1144/GSL.QJEG.1983.016.02.08.
Hall, K., 1993. Rock moisture data from Livingston Island (maritime antarctic) and impli-
cations for weathering processes. Permafr. Periglac. Process. 4, 245–253. http://dx.
doi.org/10.1002/ppp.3430040306.
Hall, K., 2004. Evidence for freeze–thaw events and their implications for rock weathering
in northern Canada. Earth Surf. Process. Landf. 29, 43–57. http://dx.doi.org/10.1002/
esp.1012.
Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., Finkel, R.C., 1997. The soil production func-
tion and landscape equilibrium. Nature 388, 358–361. http://dx.doi.org/10.1038/
41056.
Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., Finkel, R.C., 2001. Stochastic processes
of soil production and transport: erosion rates, topographic variation and
cosmogenic nuclides in the Oregon Coast Range. Earth Surf. Process. Landf. 26,
531–552. http://dx.doi.org/10.1002/esp.209.
Hettema, M.H.H., Wolf, K.-H.A.A., De Pater, C.J., 1998. The influence of steam pressure on
thermal spalling of sedimentary rock: theory and experiments. Int. J. Rock Mech. Min.
Sci. 35, 3–15. http://dx.doi.org/10.1016/S0148-9062(97)00318-5.
Ikeda, H., Iseya, F., Kodama, Y., 1993. Sedimentation on an alluvial cone in the Upper Oi
River, Central Japan. Bull. Tsukuba Univ. For. 9, 149–173 (in Japanese with English
abstract).
Imaizumi, F., Hattanji, T., Hayakawa, Y.S., 2010. Channel initiation by surface and subsur-
face flows in a steep catchment of the Akaishi Mountains, Japan. Geomorphology 115,
32–42. http://dx.doi.org/10.1016/j.geomorph.2009.09.026.
Imaizumi, F., Sidle, R.C., Togari-Ohta, A., Shimamura, M., 2014. Temporal and spatial var-
iation of infilling processes in a landslide scar in a steep mountainous region, Japan.
Earth Surf. Process. Landf. http://dx.doi.org/10.1002/esp.3659 (printing).
Imaizumi, F., Sidle, R.C., Tsuchiya, S., Ohsaka, O., 2006. Hydrogeomorphic processes in a
steep debris flow initiation zone. Geophys. Res. Lett. 33, L10404. http://dx.doi.org/
10.1029/2006GL026250.
Kano, K., Matsushima, N., 1988. The Shimanto belt in the Akaishi Mountains eastern part
of southwest Japan. Mod. Geol. 12, 97–126.
Kano, K., Muramatsu, T., Hirota, Y., 1986. Deformation style of the upper Cretaceous
Shimanto supergroup in the southern Akaishi Mountains, central Japan. Geosci.
Rep. Shizuoka Univ. 12, 89–114 (in Japanese with English abstract).
Kano, K., Nakaji, M., Takeuchi, S., 1991. Asymmetrical melange fabrics as possible indica-
tors of the convergent direction of plates: a case study from the Shimanto Belt of the
Akaishi Mountains, central Japan. Tectonophysics 185, 375–388. http://dx.doi.org/10.
1016/0040-1951(91)90455-2.
Kirk, R.M., 1977. Rates and forms of erosion on intertidal platforms at Kaikoura Peninsula,
South Island, New Zealand. N. Z. J. Geol. Geophys. 20, 571–613.
Lin, Z., Oguchi, T., 2004. Drainage density, slope angle, and relative basin position in
Japanese bare lands from high-resolution DEMs. Geomorphology 63, 159–173.
http://dx.doi.org/10.1016/j.geomorph.2004.03.012.
Lin, C.W., Shieh, C.L., Yuan, B.D., Shieh, Y.C., Liu, S.H., Lee, S.Y., 2004. Impact of Chi–Chi
earthquake on the occurrence of landslides and debris flows: example from the
Chenyulan River watershed, Nantou, Taiwan. Eng. Geol. 71, 49–61. http://dx.doi.
org/10.1016/S0013-7952(03)00125-X.
Matsuoka, N., 1995. Rock weathering processes and landform development in the Sør
Rondane Mountains, Antarctica. Geomorphology 12, 323–339. http://dx.doi.org/10.
1016/0169-555X(95)00013-U.
Matsuoka, N., 2001. Solifluction rates, processes and landforms: a global review. Earth Sci.
Rev. 55, 107–134. http://dx.doi.org/10.1016/S0012-8252(01)00057-5.
Matsuoka, N., Sakai, H., 1999. Rockfall activity from an alpine cliff during thawing periods.
Geomorphology 28, 309–328. http://dx.doi.org/10.1016/S0169-555X(98)00116-0.
Maybeck, M., 1987. Global chemical weathering of surficial rocks estimated from river
dissolved loads. Am. J. Sci. 287, 401–428. http://dx.doi.org/10.2475/ajs.287.5.401.
Moon, B.P., Selby, M.J., 1983. Rock mass strength and scarp forms in Southern Africa.
Geogr. Ann. 65A, 135–145.
Mottershead, D.N., 1989. Rates and patterns of bedrock denudation by coastal salt
spray weathering: a seven-year record. Earth Surf. Process. Landf. 14, 383–398.
http://dx.doi.org/10.1002/esp.3290140504.
Mutlutürk, M., Altindag, R., Türk, G., 2004. A decay function model for the integrity loss of
rock when subjected to recurrent cycles of freezing–thawing and heating–cooling.
Int. J. Rock Mech. Min. Sci. 41, 237–244. http://dx.doi.org/10.1016/S1365-1609(03)
00095-9.
Nagai, O., Nakamura, H., 2000. The large-scale landslide of Mt. Shichimen — the study on
the history of the landslide and the enlargement process. J. Jpn. Landslide Soc. 37 (2),
20–29 (in Japanese with English abstract).
Nishii, R., Matsuoka, N., 2010. Monitoring rapid head scarp movement in an alpine
rockslide. Eng. Geol. 115, 49–57. http://dx.doi.org/10.1016/j.enggeo.2010.06.014.
Nicholas, R.M., Dixon, J.C., 1986. Sandstone scarp form and retreat in the Land of Standing
Rocks, Canyonlands-National-Park, Utah. Z. Geomorphol. 30, 167–187.
Obanawa, H., Matsukura, Y., 2008. Cliff retreat and talus development at the caldera wall
of Mount, St. Helens: computer simulation using a mathematical model. Geomor-
phology 97, 697–711. http://dx.doi.org/10.1016/j.geomorph.2007.09.013.
Pentecost, A., 1991. The weathering rates of some sandstone cliffs central Weald, England.
Earth Surf. Process. Landf. 16, 83–91. http://dx.doi.org/10.1002/esp.3290160109.
Phillips, J.D., 2005. Weathering instability and landscape evolution. Geomorphology 67,
255–272. http://dx.doi.org/10.1016/j.geomorph.2004.06.012.
Roberts, R.G., Church, M., 1986. The sediment budget in severely disturbed
watersheds, Queen Charlotte Ranges, British Columbia. Can. J. For. Res. 16,
1092–1106. http://dx.doi.org/10.1139/x86-189.
Sidle, R.C., Ochiai, H., 2006. Landslides: Processes, Prediction, and Land Use: American
Geophysical Union Water Resources Monograph 18. American Geophysical Union,
Washington, DC.
Schmidt, K.H., 1989. The significance of scarp retreat for Cenozoic landform evolution on
the Colorado Plateau, USA. Earth Surf. Process. Landf. 14, 93–105. http://dx.doi.org/
10.1002/esp.3290140202.
Selby, M.J., 1982. Rock mass strength and the form of some inselbergs in the central
Namib Desert. Earth Surf. Process. Landf. 7, 489–497. http://dx.doi.org/10.1002/esp.
3290070509.
Su, J.G., Bork, E.W., 2006. Influence of vegetation, slope and LIDAR sampling angle on DEM
accuracy. Photogramm. Eng. Remote. Sens. 72, 1265–1274.
Suzuki, T., 2002. Rock control in geomorphological processes: research history in Japan
and perspective. Trans. Jpn Geomorphol. Union 23 (2), 159–160.
Trustrum, N., DeRose, R., 1988. Soil depth–age relationship of landslides on deforested
hillslopes, Taranaki, New Zealand. Geomorphology 1, 143–160. http://dx.doi.org/10.
1016/0169-555X(88)90012-8.
Tsou, C.Y., Feng, Z.Y., Chigira, M., 2011. Catastrophic landslide induced by Typhoon
Morakot, Shiaolin, Taiwan. Geomorphology 127, 166–178. http://dx.doi.org/10.
1016/j.geomorph.2010.12.013.
Turkington, A.V., Paradise, T.R., 2005. Sandstone weathering: a century of research and in-
novation. Geomorphology 67, 229–253. http://dx.doi.org/10.1016/j.geomorph.2004.
09.028.