Brigham Young University
BYU ScholarsArchive
International Congress on Environmental
Modelling and Software
4th International Congress on Environmental
Modelling and Software - Barcelona, Catalonia,
Spain - July 2008
Jul 1st, 12:00 AM
The role played by mountain tracks on
rainfallinduced shallow landslides: a case study
L. Cascini
S. Cuomo
M. Pastor
Follow this and additional works at: http://scholarsarchive.byu.edu/iemssconference
Cascini, L.; Cuomo, S.; and Pastor, M., "The role played by mountain tracks on rainfallinduced shallow landslides: a case study"
(2008). International Congress on Environmental Modelling and Software. 206.
http://scholarsarchive.byu.edu/iemssconference/2008/all/206
This Event is brought to you for free and open access by the Civil and Environmental Engineering at BYU ScholarsArchive. It has been accepted for
inclusion in International Congress on Environmental Modelling and Software by an authorized administrator of BYU ScholarsArchive. For more
information, please contact [email protected].
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
The role played by mountain tracks on rainfallinduced shallow landslides: a case study
L. Cascinia, S. Cuomoa, M. Pastorb
a
University of Salerno, Department of Civil Engineering, Italy
([email protected], [email protected])
b
Centro de Estudios y Experimentación de Obras Públicas (CEDEX) and ETS de
Ingenieros de Caminos, Madrid, Spain ([email protected])
Abstract: Mountain tracks often play a key role for landslide occurrence inside shallow
soil deposits affecting both superficial and sub-superficial groundwater circulation. This is
essentially related to local changes of topography and soil hydraulic conductivity.
Significant examples of tracks-related landslides were observed during the event that
affected the Campania region (Southern Italy) in May 1998. These phenomena occurred
inside source areas characterised by complex landforms, stratigraphy and groundwater
regime. Referring to this event, the paper addresses the role played by the stratigraphy on
the first-failure stage by using limit equilibrium and stress-strain analyses. The numerical
results outline some important aspects that must be carefully considered in analyses aimed
at an advanced characterisation of landslide source areas.
Keywords: Rainfall, Shallow landslides, Flow-type, Mountain tracks, Cuts.
1.
INTRODUCTION
Rainfall is the most frequent triggering factor for shallow landslides in a variety of
unsaturated soils [Alonso et al., 1995]. For these landslides, the infiltration process and
consequent failure stage are primarily controlled by rainfall patterns (i.e. intensity and
duration) [Rahardjo et al., 2001], soil initial conditions [Tsaparas et al., 2002], and local
natural factors such as slope angle breaks, thinning of deposits and close-ended layers.
Slope instability phenomena can also be induced by anthropogenic factors such as
mountain tracks and cut slopes, especially when acting on singular points of the hillslope.
Mountain tracks are generally associated with changes in topography, discontinuities of
stratigraphy and lower soil hydraulic conductivity values compared to those of the
neighbouring zones [Gasmo et al., 2000; Luce and Black, 2001]. As a consequence, along
the tracks, rainwater can generate surface runoff, ponding conditions and erosion processes
that often produce significant rills and gullies, as reported for well documented cases of
Thailand [Ziegler et al., 2004], Australia [Croke and Mockler, 2001], US Virgin Islands
[McDonald et al., 2001] and Malaysia [Sidle et al., 2004]. Cut slopes, commonly associated
to mountain tracks, can also cause increased erosion processes and volumes of water
runoff. Finally, lateral redistribution of water runoff can occur at the bends of tracks
affecting the adjacent zones [Ziegler et al., 2004].
The effects of anthropogenic factors on superficial and sub-superficial groundwater
circulation are particularly relevant in some areas of the Campania region (Southern Italy)
where pyroclastic deposits overlie carbonate massifs. A paradigm example is provided by
the May 1998 Sarno-Quindici event when some destructive landslides systematically
occurred inside source areas characterised by complex landforms, stratigraphy and
groundwater regime. A comprehensive analysis of the occurred phenomena requires a
careful investigation of the previously mentioned factors. Among these, the paper focuses
on the soil stratigraphy taking into account the typical in-situ conditions and local boundary
conditions related to mountain tracks.
1484
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
2.
CASE STUDY
2.1 The landslides occurred in May 1998
In May 1998, heavy rainfall [Cascini et al., 2000; Fiorillo and Wilson, 2004] triggered
several destructive landslides of flow-type in unsaturated pyroclastic deposits overlying
the Pizzo d’Alvano limestone
massif (Sarno-Quindici event)
[Cascini, 2004; Cuomo, 2006]
caused by different and
complex
triggering
mechanisms. Cascini et al.
[2008] recognised that the
16% of the 133 analysed
landslide source areas were
related to the presence of
mountain tracks (Fig. 1a, b),
along the hillslopes facing the
town of Quindici (Fig. 1c).
These slope failures mostly
occurred along the flanks of
gullies that were extensively
affected by anthropogenic
modifications especially in the
last decades [Calcaterra et al.,
2003]. Particularly, the failures
occurred at the bends of tracks
(scheme M3a of fig. 1b), later
propagating downwards and
upwards along the hillslopes.
Adjacent zones were also
systematically affected by
landslides (scheme M3b of fig.
1b) due to lateral enlargement
of the original unstable areas.
In some cases, source areas
with compound landforms
Figure 1. Shallow landslides of flow-type connected to
were observed also due to the
mountain tracks during the May 1998 event.
combination of the above
anthropogenic factors and natural discontinuities of the pyroclastic deposits (i.e. bedrock
scarps up to tens of meters tall) (Fig. 2).
The correspondence of some landslide source areas with mountain tracks is widely
emphasised in the current
literature. Di Crescenzo and
Santo [2005] outline that 44%
of these areas were connected
to man-made cuts and tracks,
located at distances lower than
10 m in most cases. Guadagno
et al. [2005] recognise that
49% of the analysed failures
occurred above man-made cuts
and 12% involved fills
downslope the tracks. These
authors interpreted the failures
as
debris
avalanches
phenomena (i.e. characterised
Figure 2. In-situ evidences for some landslides occurred
along the hillslopes facing the town of Quindici.
by triangular shaped source
1485
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
areas) and performed statistical analyses to correlate the height of cuts to other
morphometrical parameters (i.e. apex angle of source areas and slope angle). However,
poor correlations were obtained essentially due to the assumed triggering mechanism and
because some key triggering factors were not included in the analysis (i.e. pore water
pressures, local boundary conditions).
By modelling the first-failure stage of landslides induced by mountain tracks, Guadagno et
al. (2003), referring to steady-state saturated flow conditions, outline that natural or
anthropogenic discontinuities of pyroclastic deposits strongly increase the simulated
displacements for upslope portions of the slope up to failure. The role of the above
discontinuities for landslide occurrence is also emphasised by Crosta and Dal Negro
(2003), on the basis of analyses that model the transient groundwater regime during the
critical rainfall event of 4-5 May 1998.
These contributions, while providing useful insights on the topic, do not exhaustively
analyse the relationships between tracks and landslides occurrence, neglecting some
relevant aspects such as, for instance, the superficial water circulation and the role played
by local stratigraphy. These aspects are addressed in the following sections that provide a
preliminary modelling of the first-failure stage inside the landslide source areas.
2.2 Proposed approach, in-situ conditions and soil properties
Rigorous modelling of superficial and sub-superficial water circulation should take into
account several factors, i.e. rainfall intensity, soil unsaturated conditions, soil hydraulic
conductivity, topography, stratigraphy as well as anthropogenic factors that, in singular
points and/or zones of the hillslope, can produce runoff phenomena and concentrated
superficial water fluxes. To this aim, the physically-based model proposed by Savage et al.
[2003] can be used. However, the lack of an accurate DTM can notably reduce the
potentialities of these models. In such a case, a preliminary insight in the superficial water
circulation can be obtained referring only to topography and computing the contributing
areas along the hillslope (Pack et al., 1998;
Tucker and Bras, 1998). The latter ones
a)
represent, at each point, the upslope catchment
areas and the highest values can be reasonably
assumed as corresponding to the zones where
superficial rainwater concentrates. Once the
hydraulic boundary conditions are outlined for
different zones of the hillslope, pore water
pressures can be computed with an infiltration
model also considering soil unsaturated
conditions through the numerical integration of
the Richards’ equation [Richards, 1931]. The
failure stage can be thus simulated through
either limit equilibrium or stress-strain
analyses, adopting a Mohr-Coulomb criterion
extended to unsaturated conditions [Fredlund
et al., 1978].
This simplified approach is applied to the May
1998 event, using the available data set that
outlines the presence of three main soil classes
differently arranged along the Pizzo d’Alvano
massif (Fig. 1c). Considering the aim of the
study, parametric analyses are performed,
referring to simplified schemes (Fig. 3) that
reproduce typical stratigraphical settings
observed all over the massif. The schemes
include finer deep ashy soil (Class A) and/or
coarser superficial ashy soil (Class B) also
1486
schemes
1
2
b)
3
ashy B
pumice
ashy A
Figure 3. Typical stratigraphy of pyroclastic
deposits in the study area.
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
eventually including continuous pumice soil layers. The mechanical properties of these
soils in saturated and unsaturated conditions are provided by the scientific literature
(Sorbino and Foresta, 2002; Bilotta et al., 2005; Cascini et al., 2005, among others) and
summarised in table 1.
Table 1. Soil mechanical properties.
Class B ashy soil
Pumice soil
Class A ashy soil
γd
(kN/m3)
γsat
(kN/m3)
n
(-)
ksat
(m/s)
c’
(kPa)
7.30
13.1
0.58
10-5
0
0.69
10
-4
10
-6
6.20
9.10
13.1
15.7
0.66
fb
(°)
ν
(-)
E
(MPa)
÷ 41
20
0.26
5÷7
0
37
20
-
-
5 ÷ 15
32 ÷ 35
20
0.30
1÷3
÷5
f’
(°)
36
y
(°)
10
÷ 20
-
10
÷ 20
γd, (γsat): dry (saturated) unit weight of soil, n: porosity, ksat: saturated soil conductivity, c’: effective cohesion, f’: friction
angle, fb: rate of increase in shear strength due to suction, ν: Poisson’s ratio, E: Young’s modulus, y: dilation angle.
3.
GEOMECHANICAL MODELLING
3.1 Groundwater modelling
The modifications of superficial water circulation induced by mountain tracks were firstly
evaluated, referring to a schematic steep (30°) hillslope, crossed by a steep track (14°) with
one or more bends (Fig. 4a,
a)
4b). The contributing areas
were computed at each point
of the hillslope through a
detailed 0.5m×0.5m digital
elevation model. The highest
contributing areas concentrate
at the bends of tracks (scheme
b)
M3a in Fig. 1 and 4) and
complex scenarios
are
simulated for the scheme
including more than two
M3a M3b
bends. Particularly, the zones
adjacent to the bends are
strongly affected by lateral
redistribution
of
runoff
(scheme M3b in Fig. 1 and 4).
Concisely, both zones near
increasing contributing areas
and adjacent to the bends are
Figure
4.
Contributing
areas along an open slope
outlined as prone to runoff
for different mountain tracks.
phenomena
and
similar
analyses can be extended to real cases if detailed topography (> scale 1:1000) and field
data on the anthropogenic factors are available.
Based on these findings, the groundwater regime was modelled for different zones of the
hillslope, i.e. near and
adjacent to the bends,
considering typical in-situ
conditions
and
the
stratigraphical schemes in
figure 3. Pore water pressures
induced by the 4-5 May 1998
rainfall storm were computed
through the Seep/W Finite
Element code (Geoslope,
2004) using the computational
Figure 5. Computational schemes corresponding to different
portions of the hillslope (schemes M3a and M3b of figure 4c).
schemes of figure 5 that
1487
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
reproduce the schemes M3a and M3b of figures 1 and 4. Concerning scheme M3a, the
rainfall values provided by Cascini et al. [2003] were assumed as boundary condition at the
ground surface while a local ponding was considered downslope of the bend. For scheme
M3b (Fig. 5b), the same ponding condition was assumed taking into account the possible
lateral redistribution of surface runoff highlighted in figure 4b. For scheme M3a, the
obtained results highlight that the ponding condition assumed at the bend of the track and
downslope produce different effects depending on the local stratigraphy (Fig. 6).
Particularly, if only ashy B soils are present (scheme 1 of Fig. 6) a local and slight increase
of pore water pressures can be simulated; strong variations are conversely computed due to
presence of both ashy A and B soils (scheme 2 of Fig. 6). Finally, for slopes with
continuous pumice soil
layers (scheme 3 of Fig.
scheme 1
scheme 2
6), pore water pressures
track
track
increase involving larger
portions of the slopes
-4
-4
uw < 0
ponding
than in the previous
ponding
cases. Similar results are
-1
uw > 0
uw < 0
12
obtained for scheme
numbers refer to u (kPa)
numbers refer to u (kPa)
M3b (i.e. corresponding
to zones adjacent to the
M3a
schemes
scheme 3
bends), thus highlighting
1
2
3
that the groundwater
uw < 0
regime is primarily
-8
ponding
controlled by both the
ashy B
assumed
boundary
pumice
6
conditions at the ground
uw > 0
ashy A
numbers
refer
to
u
(kPa)
surface
and
the
stratigraphy rather than
Figure 6. Pore water pressure induced by rainfall and a local ponding
by local geometrical
condition for the scheme M3a of figure 5a with different
discontinuities.
stratigraphies.
w
w
w
3.2 Slope stability analyses
For the scheme M3a, limit equilibrium analyses (Fig. 7) were performed through the
Slope/W code (Geoslope, 2004) assuming as input the computed pore water pressures. The
obtained results show that for scheme 1 of figure 6 (only ashy B soils) a local ponding
condition does not cause any failure, as the slope safety factor (FS) is greater than one (Fig.
7a). Conversely, for the scheme 2 of figure 6 (ashy A and B soils) failure is simulated
along slip surfaces involving zones both downslope and upslope the cut slope (Fig. 7b).
Similar results are obtained for the scheme 3 of figure 6 with a slip surface involving both
the upslope and downslope zones of the track.
Similar limit equilibrium analyses were performed for scheme M3b. Figure 8 provides the
results obtained, at the
a) scheme 1
b) scheme 2
M3a
same time instant of
FS=1.10
FS=0.89
figure 7, for different
stratigraphical settings
track
assuming an equal local
track
ponding
condition.
Particularly, for scheme
ponding
ponding
1 (only ashy B soils)
failure conditions are
not simulated (Fig. 8a);
conversely
these
ashy B
conditions
involve
ashy A
different zones of the
slope in the case of
Figure 7. Limit equilibrium analyses for the scheme M3a of figure 5a.
scheme 2 (only ashy A
1488
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
soils) and scheme 3
M3b
a) scheme 1 b) scheme 2 c) scheme 3
(ashy A and B and
FS=0.97
FS=1.25
FS=0.88
pumice soils) (Fig. 8c).
The obtained results
clearly outline the role
played
by
the
po
stratigraphy on the
po
nd
nd
po
ing
ing
nd
failure onset. Moreover,
ing
comparing scheme 2 of
figure 7 with scheme 2
ashy B
of figure 8 it derives
pumice
that the slope safety
ashy A
factors are about the
same
(respectively
Figure 8. Limit equilibrium analyses for zones of hillslope far away
from bends of tracks.
FS=0.89 and FS=0.88)
independent of the
presence of a geometrical discontinuity inside the pyroclastic deposit.
1
2
displacements
(m)
displacements (m)
To validate the previous results, pore water pressures were used as input for uncoupled and
coupled
stress-strain
analyses. For scheme
after 10 hours
after 20 hours
after 24 hours
M3b, Figure 9 shows
the
displacements
P
computed
for
an
homogeneous
soil
displacement (m)
deposit (scheme 1 of
0.0005 – 0.002
Fig. 6). It is worth
0.002 – 0.0035
noting that in the case
0.0035 – 0.005
of rainfall only as
M3b
boundary
condition,
0.005
Point Ponly rainfall
M3
ponding
the failure onset is not
0.004
simulated. On the
contrary, including a
0.003
ponding condition, the
0.002
simulated
rainfall
scheme 1
0.001
displacements
are
characterised by quite
0.000
ashy B
0
3
6
9
12 15 18 21 24 27
high gradients leading
time (hours)
(hours)
time
to failure conditions
Figure 9. Stress-strain analyses for zones far away from the bends
after about 24 hours
(scheme 1 of Figure 8).
(Fig. 9).
4.
CONCLUSION
Anthropogenic factors often influence the superficial groundwater circulation during
rainfall. Particularly, mountain tracks and cut slopes cause erosion processes and runoff
that can locally trigger slope instability phenomena. Sometimes, multiple failures develop
inside large areas according to complex mechanisms not easy to recognise and/or to model.
Examples of significant landslide source areas related to anthropogenic factors were
recorded in the Campania region (Southern Italy) in May 1998, when several and multiple
landslides affected shallow unsaturated pyroclastic deposits covering a carbonate massif.
Considering the complexity of the occurred phenomena, few contributions in the scientific
literature address the topic. In order to obtain a preliminary insight in the role played by
different factors on the triggering mechanisms, groundwater and slope instability analyses
were performed, in a parametric form, referring to simplified and bi-dimensional schemes
of the landslide source areas.
Groundwater modelling highlights that ponding conditions strongly affect the transient
1489
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
pore water pressures, mostly depending on the local stratigraphical setting. Limit
equilibrium and stress-strain analyses outline that geometrical discontinuities act as an
important aggravating factor for slope stability conditions. However, stratigraphy is
confirmed as the key factors for the triggering of large instability phenomena for all the
landslide source areas. Consequently, more advanced numerical models than those used in
the present paper should take stratigraphy into account together with more elaborate
methodologies such as, for instance, the three-dimensional analysis of slope stability
conditions.
REFERENCES
Alonso, E., Gens, A., Lloret, A., Delage, C., Effect of rain infiltration on the stability of
slopes. Proc. 1th Int. Conference on Unsaturated Soils, Alonso&Delage eds, Paris, 241249, 1995.
Bilotta, E., Cascini, L., Foresta, V., Sorbino, G., Geotechnical characterization of
pyroclastic soils involved in huge flowslides. Geotechnical and Geological Engineering,
23, 365-402, 2005.
Calcaterra, D., Parise, M., Palma, B., Combining historical and geological data for the
assessment of the landslide hazard: a case study from Campania, Italy. Natural Hazard
and Earth System Sciences, 3, 13-16, 2003.
Cascini, L., The flowslides of May 1998 in the Campania region, Italy: the scientific
emergency management. Italian Geotechnical Journal, 2, 11-44, 2004.
Cascini, L., Cuomo, S., Sorbino, G., Flow-like mass movements in pyroclastic soils:
remarks on the modelling of triggering mechanisms. Italian Geotechnical Journal, 4,
11-31, 2005.
Cascini, L., Cuomo, S., Guida, D., Typical source areas of May 1998 flow-like mass
movements in the Campania region, Southern Italy, Engineering Geology, 96, 107-125,
2008.
Cascini, L., Guida, D., Nocera, N., Romanzi, G., Sorbino, G., A preliminary model for the
landslides of May 1998 in Campania Region. Proc 2nd Int. Symposium on Geotechnics
of Hard Soil-Soft Rock – Napoli, Balkema, 3, 1623-1649, 2000.
Cascini, L., Sorbino, G., Soil suction measurement over large areas: a case study. Proc. 3rd
Int. Conference on Unsaturated Soils, Recife (Brasil), Balkema 2, 829-834, 2002.
Cascini, L., Sorbino, G., Cuomo, S., Modelling of flowslides triggering in pyroclastic soils.
Proc. Int. Conference on “Fast Slope Movements - Prediction and Prevention for Risk
Mitigation”, Napoli, Patron Ed. 1, 93-100, 2003.
Croke,J., Mockler, S., Gully initiation and road-to-stream linkage in a forested catchment,
Southeastern Australia. Earth Surface Processes and Landforms, 26(2), 205-217, 2001.
Crosta, G.B., Dal Negro, P., Observations and modelling of soil slip-debris flow initiation
processes in pyroclastic deposits: the Sarno 1998 event. Natural Hazards and Earth
System Science, 3, 53-69, 2003.
Cuomo, S., Geomechanical modelling of triggering mechanisms for flow-like mass
movements in pyroclastic soils. PhD dissertation at the University of Salerno, Italy, pp.
274, 2006.
Di Crescenzo, G., Santo, A., Debris slides–rapid earth flows in the carbonate massifs of the
Campania region (Southern Italy): morphological and morphometric data for evaluating
triggering susceptibility. Geomorphology, 66, 255-276, 2005.
Fiorillo, F., Wilson, R.C., Rainfall induced debris flows in pyroclastic deposits, Campania
(Southern Italy). Engineering Geology, 75, 263- 289, 2004.
Fredlund, D.G., Morgenstern, N.R., Widger, R.A., The shear strength of unsaturated soils.
Canadian Geotechnical Journal, 15, 313-321, 1978.
Gasmo, M., Rahardjo, H., Leong, E.C., Infiltration effects on stability of a residual soil
slope. Computers and Geotechnics 26 (2), 145-165, 2000.
Geoslope, User’s guide. GeoStudio 2004, Version 6.13. Geo-Slope Int. Ltd. Calgary,
Canada, 2004.
Guadagno, F.M., Martino, S., Scarascia Mugnozza, G., Influence of man-made cuts on the
stability of pyroclastic covers (Campania-Southern Italy): a numerical modelling
approach. Environmental Geology, 43, 371-384, 2003.
1490
L. Cascini et al. / The role played by mountain tracks on rainfall-induced shallow landslides: a case study
Guadagno, F.M., Forte, R., Revellino, P., Fiorillo, F., Focareta, M., Some aspects of the
initiation of debris avalanches in the Campania Region: the role of morphological slope
discontinuities and the development of failure. Geomorphology, 66, 237-254, 2005.
Luce, C.H., Black, T.A., Spatial and temporal patterns in erosion from forest roads.
Influence of Urban and Forest Land Uses on the Hydrologic-Geomorphic Responses of
Watersheds. M.S. Wigmosta and S.J. Burges, Editors. American Geophysical Union,
Washington, D.C., 165-178, 2001.
MacDonald, L.H., Sampson R.W., Anderson, D.M., Runoff and road erosion at the plot
and road segment scales, St John, US Virgin Islands. Earth Surface Processes and
Landforms, 26, 251-272, 2001.
Montgomery, DR, Dietrich, WE., A physically based model for the topographic control on
shallow landsliding. Water Resources Research, 30( 4), 1153-1171, 1994.
Pack, R.T., Tarboton, D.G., Goodwin, C.N., The SINMAP Approach to Terrain Stability
Mapping. 8th Congress of the International Association of Engineering Geology,
Vancouver, British Columbia, Canada, 1998.
Rahardjo, H., Li, X.W., Toll, D.G., Leong, E.C., The effect of antecedent rainfall on slope
stability. Geotechnical and geological Engineering [19], 371-399, 2001.
Richards, L.A., Capillary conductions of liquids through porous mediums. Journal of
Applied Physics 1, 318-333, 1931.
Savage W.Z., Godt J.W., Baum R.L., A model for spatially and temporally distributed
shallow landslide initiation by rainfall infiltration. in press. Proc. 3rd Int. Conf. on
Debris Flow Hazards Mitigation: Mechanics, Prediction, and Assessment, Davos,
Switzerland, 179-187, 2003.
Sidle, R., Sasaki, S., Otsuki, M., Noguchi, S., Rahim, A., Sediment pathways in a tropical
forest: effects of logging roads and skid trails. Hydrological Processes, 18, 703-720,
2004.
Sorbino,G., Foresta,V., Unsaturated hydraulic characteristics of pyroclastic soils. Proc. 3rd
International Conference on Unsaturated Soils, Recife (Brasil), Balkema, 1, 405-410,
2002.
Tsaparas, I., Rahardjo, H., Toll, D.G., Leong, E.C., Controlling parameters for rainfallinduced landslides, Computers and Geotechnics, 1, 1-27, 2002.
Tucker, G.E., Bras, R.L., Hillslope processes, drainage density, and landscape
morphology. Water Resources Research, 34(10), 2751-2764, 1998.
Ziegler, A.D., Giambelluca, T.W., Sutherland, R., Nullet, M., Yarnasarn, S., Pinthong, J.,
Preechapanya, P., Jaiaree, S., Toward understanding the cumulative impacts of roads in
upland agricultural watersheds of northern Thailand. Agriculture, Ecosystems and
Environment [104], 145-158, 2004.
1491