IJEGE-11_BS-Scheuner-et-alii

(R
evaluation of mitigation measures against debris
flows, there is a strong need for process-based models.
Hazard maps in Switzerland show expected intensities
and return periods of dangerous processes for a certain
area or location by a detailed outlining of areas where
construction of buildings must be prohibited or where
rules and recommendations should be established
(P
method (BWW et alii, 1997). Endangered areas are
classified into one of five classes of equal degree of
hazard which are assigned a color on the final map:
high hazard (red), medium hazard (blue), low hazard
(yellow), residual danger (yellow-white striped) and
no or negligible danger according to current state of
knowledge (white or uncolored). A high precision de-
lineation of endangered areas is thought to be attain-
able and leads finally to a hazard map.
In areas with medium danger, local protection is re-
quired when planning construction projects, whereas
in areas of high danger there is normally a prohibi-
tion on construction (P
method requires delineation of endangered areas by
the expected intensity for a given return period. For
this reason there is a strong need for process-based
models to improve the quality of hazard mapping and
for planning and evaluation of mitigation measures. In
this paper, we present the application of a Swiss mod-
eling system for rapid mass movements. The RAMMS
debris flow model is a two-dimensional model for
simulating the runout distance, velocity, flow depth
and impact pressure of debris flows. It uses the single-
phase Voellmy fluid friction relation which describes
the debris flow resistance as a combination of a dry
Coulomb-type friction and a viscous resistance which
varies with the square of the flow velocity. The model
solves the depth-averaged shallow water equations for
granular flows in 2D using a finite volume scheme. We
present two case studies in which RAMMS was used
by practitioners as a support tool in hazard manage-
ment and planning mitigation measures. The first site
is located in Stechelberg in the western Swiss Alps
and the second one in Walchensee, Bavaria (Germa-
ny). In both examples RAMMS was found to provide
valuable support for practitioners.

following steps:
1. Hazard identification and analysis (including anal-
parameters such as flood discharge or potential
volume of debris. The potential debris volume
is based on geomorphic analysis methods which
have been systematically developed for use in
Switzerland. In this paper, the method proposed
by l
(1997). The three common scenarios in Switzerland
provided a basis for the Mattenbach (see chapter case
studies), described in the following:
• 30 years scenario: return period of 30 years or less.
• 100 years scenario: return period of 30 to 100 years.
• 300 years scenario: return period of 100 to 300 years.
Depending on the case study, the following steps were
added to the described procedure:
• assessment and definition of model parameters to
• Considering planned measurements in hazard as-
tial intensity of debris flow hazard. Examples will be
described later in this paper.
runout problems which could not be solved with exist-
ing 1D runout models. The Voellmy friction relation
tion of land for construction or existent buildings and
therefore significant financial losses for landowners
and communities. Hence it is essential for practition-
ers to accurately establish in particular the boundaries
of red and blue danger levels.
ed methods (e.g. H
of flow paths”. In doing so, hazard areas are deline-
ated directly in field based on flow paths delineated by
topographic features. In the same work step intensities
and debris accumulation are usually assessed, leading
to the definition of an intensity scenario and the as-
sociated hazard level. An inspection of the catchment
as well as a proper definition of breakout and sediment
delivery scenarios are required.
mainly in the case of complex topography together
with vegetation, the consideration of frictional param-
eters as well as the inclusion of mitigation measures
such as sediment retention basin or deflection dams.
In addition, calculations to determine process inten-
sity such as flow velocity or flow heights have to be
carried out, typically using empirical estimation for-
mulas (e.g. R
o’b
models are scarce. Existing models are only partially
suited for detailed hazard assessment because they are
based either on probabilistic or topographic approach-
es (e.g. G
hazard index maps.
substitute for an expert’s opinion, but which surely
may provide considerable support.
eling system RAMMS which was designed partially
with this goal in mind, and subsequently, we present
two case studies in which RAMMS was used to sup-
port practitioners in debris flow hazard assessment.

Lauterbrunnen in Switzerland (Fig. 1). In the upper
part, the catchment mainly consists of limestones
(lower cretaceous of the Doldenhorn nappe). The
middle and the lower part consists of Malm lime-
stones (Mesozoic autochthonous of the Aar massif).
The lower most part (debris fan) consists of alluvi-
ums of the Mattenbach as well as the river Lütschine.
tains permafrost.
the Matta settlement of Stechelberg, most recently in
2004. The channel has to be maintained almost every
year, resulting in costs of several thousand Swiss
francs (oral communication by R. J
2001 (G
eas close to the channel and a community road were
assigned to the red hazard zone. Debris flows initiat-
ing in the upper catchment area are considered to be
the most dangerous hazard. If debris flows reach the
fan apex they may cause lateral overbank flow of wa-
ter and sediment, carry driftwood and cause floods at
both sides of the fan. Moreover sedimentation may
jam the channel cross section and clog bridge cul-
been used to describe the motion of debris flows (e.g.
R
the typical values of the friction coefficients. The ini-
tial application of RAMMS to debris flows was tested
largely in the framework of research theses written by
students (e.g. s
details on the RAMMS model, including further re-
marks on the equations and their numerical solution
are available in C
lution to the depth-averaged equations of motion for
granular flows:
gravitational acceleration; Sg is the driving gravita-
tional; and S
and y directions. RAMMS uses the Voellmy fric-
tion approach, which splits the total friction into a
velocity-independent dry Coulomb term which is
proportional to the normal stress at the base of the
flow (friction coefficient μ) and velocity-dependent
so-called turbulent or viscous friction (coefficient
both the x and y directions):
not of sufficient quality to permit calibration, μ is typi-
cally initially selected to be the same as the local slope on
the area where debris flows have stopped in the past, and
then ξ is selected to provide plausible velocities which
may be based on existing observations, back-calculated
flow velocities estimated by geomorphic methods (such
as super-elevation around channel bends).

nel is steep, except of the section from the fan apex
downwards. It is evident that the torrent is able to
form debris flows (G
from other debris flow events at similar catchment
areas. In this manner the slope’s friction parameter
and the density of potential debris flows have been
roughly assessed (Tab. 1). After completion of the
first debris flow modeling, the estimated parameters
were refined by on site mapping of debris accumula-
tions at the slope.
2).
• 30-years scenario: It is expected that the total
retention basin (Tab. 2). Neither hydraulic nor
also cause overbank flooding and debris flow deposi-
tion (G
with a retention capacity of 1,900 m
tion construction was optimized with regard to the
retention volume. Compared to the maximum poten-
tial debris volume (Tab. 2) the construction is not ca-
pable to stop all of the total debris load. It is possible
that, due to topography, part of the debris load may
leave the channel above the retention construction
(G
outbursts above the planned debris retention con-
struction was considered. Due to expected costs, this
measurement was not planned any further.
tion of debris on the hazard areas after completion
of the debris retention construction. The aim was
to analyze in which manner the missing protection
dike influences the endangered area (e.g. changes in
the hazard map). The hazard map shown in Figure 2
was already finished and published before modelling
with RAMMS. The RAMMS model was solely used
to support the expert’s opinion with the objective to
obtain repeatable model results and to confirm there-
with the assumed hazard areas after the protection
measures were completed. The evaluation was man-
dated by the local water authorities.
RAMMS
Lütschine (900 m a.s.l.). It mainly consists of rock
faces and debris slopes. Between the rock faces and
the alluvial fan we find a section covered with grass
and bush vegetation. Here part of the channel runs on
loose material. On the fan apex, there is significant

at 895 m a.s.l. or is deposited at the left side of
the cone. Sedimentation of approx. 270 m
tude of 895 m a.s.l.. Due to the higher bank at the
right waterside debris and water can not leave the
channel to the orographic right side and do not af-
fect the Matta settlement. Debris flows can leave
the channel at the left side of the cone. It is very
probable that debris and flood material will block
the bridge. Large amounts of debris and water
discharge near the bridge at the altitude of 895 m
a.s.l. are expected.
scribed above, RAMMS was used to support the as-
sessment of possible outbursts at the fan apex. As a
basis, a high precision digital terrain model with a
2x2 m grid was used. The altitude-accuracy is 0.5 m
(swisstopo, 2005).
are still possible, even when the protection measures
are built. Additionally, the modelling shows that out-
bursts to the orographic left side of the fan apex are
possible as well. The hazard map shown in Figure 2,
published before the RAMMS modelling was carried
out, shows a correct, but more spacious classifica-
tion of the red and partially of the blue hazard zone.
Most part of the blue and yellow hazard zones are not
shown in the results of RAMMS modelling due to a
process change from debris flow to overbank sedi-
mentation and inundation. This process change had
to be delineated in the field by the expert’s opinion
and cannot be modelled within the RAMMS model.
parent decision making. The modelling results with
RAMMS show that an outburst is likely for both sides
of the fan apex. Hence, the modelling results support
the experts opinion derived from the “method of flow
paths” and the conclusion of the hazard map shown in
Figure 2. The model served as a support tool for the
near the channel and the bridge at the altitude of
895 m a.s.l. This scenario was not calculated with
RAMMS.
tion basin construction at the orographic right side
of the cone are possible. It is expected that most of
the debris load can be held back by the sediment
retention basin (Tab. 2). There are no failure points
located below the sediment retention basin. To cal-
culate the model with RAMMS we took a single
debris flow surge of a size of 300 m
discharge. It is able to hold a maximum load of
1,900 m
flow outbursts above the planned sediment re-
tention basin at the orographic right side of the
cone are expected to deposit their debris load on
parts of the farmland. Debris flows do not affect
residential buildings (Figure 2). To calculate the
model with RAMMS we took a single debris flow

near the fan apex. The model indicates that the flows
are expected to stop above the settlements and that
substantial quantities of the debris load will not reach
the residential area. Therefore, the previously described
additional reflection dam was deemed unnecessary and
planning was stopped. This led to lower costs for the
construction of countermeasures and therefore to finan-
cial savings for the community.
flows from the so-called Reissenwand (Fig. 3). On
behalf of the local authorities a detailed hazard as-
sessment was carried out followed by the design
of protective countermeasures. Model calculations
were made using RAMMS as a support tool.
RAMMS
the altitude of 1,380 m a.s.l. down to the Walchsee
at the altitude of 800 m a.s.l. Below the steep scarp
at the Reissenwand the channel runs on bare rock,
then the channel runs through massive slope debris
accumulations resulting from intensive rock fall out
of the Reissenwand. The sediment accumulations
constitute a large debris reservoir
channel the debris cone is covered with vegetation.
With a short distance above the road there are large,
the channel is active in terms of debris flow processes.
based on experienced data from other debris flow
events (s
were estimated (Table 3). We subsequently refined
the estimated parameters based on information on
mapped traces from debris flow deposits at the slope
based on observed patterns and tendencies and was
calibrated with the experienced and limiting values
of 58 debris flow events. This method is an empiri-
cal, system based and strongly process oriented es-
timation procedure. The debris load calculation is
made for single homogenous channel sections where
slope and channel processes are distinguished.
(> 60%) running on bare rock, there is a distinct shift
in direction of the channel axis at the fan ape, where
bank erosion and the potential for overbank flow en-
danger the orographic right side of the cone. Similar-
ly, shortly before the forest, the channel may change
its position (channel slope < 20%), leading to a new
flow path towards the road. Such channel avulsions
and consequent directional changes are expected es-
pecially when the channels slope is low and sediment
is deposited (e.g. z

wedge shaped retention dam may be capable of hold-
ing back debris above the road. Water will be drained
off controlled by a culvert (concrete pipe). To pre-
vent jams during frequent events an appropriate de-
bris screen has to be installed above the culvert. In
this case plain water can be drained of over the peak
of the dam in a controlled manner.
Therefore the dynamic impact is expected to be
small, however debris accumulation will certainly
elevate the static stress. Due to back-filled material
behind the retention dam debris deposits are expect-
ed to accumulate up to 2 m thickness. Besides flow
velocity and height of deposition, the weight of the
back-filled material and the energy-impact of a de-
bris flow impacting on the dam have to be taken into
consideration.
the bank from erosion. By this means the probability
of channel outbursts can be minimized.
where the flows run out at the altitude of 860 m a.s.l.
Only small amounts of debris are expected to reach
the forest. Small grain diameter sediment loads and
water may reach the road and by flowing through the
forest mobilize leaves and branches as well, which
may cause jams at the culvert and therefore initiate
flooding of the road (depth of inundation < 0.5 m).
up to 1m high. It is expected, that large areas above
the road will be covered by debris, because the flow
velocity is strongly reduced by the presence of forest
and long flat reaches, and a forest road. Moreover it
is possible that debris flow leave the channel and fan
out above the forest. Trace of previous events on the
slope support this scenario.
with estimation formulas in common use, higher
flow velocities have to be expected at the altitude of
950 m a.s.l. The reasons why we expect this phenom-
enon are the decisive declination of the channel (>
60%) as well as the long channel section running on
bare rock. We expect flow velocity of up to 8 m/s.
These circumstances facilitate bank erosion. Togeth-
er with morphological and topographical characteris-
tics of the debris fan as well as blockages caused by
bank failures, an outburst at the fan apex and a debris
flow further south of the main direction is possible.
Changes in topography during a debris flow (bank
failures, erosion processes) are not yet incorporated
in RAMMS. Therefore, notwithstanding the model-
ling results (Fig. 4) the southern part of the fan is po-
tentially endangered, as is the northern part. Such a
debris flow on the southern part of the fan apex may
follow the still visible former debris channel down
to the road
described. Events with a return period of 30 years do
not endanger the road and therefore only protection
measures preventing road flooding have to be taken
into account (especially maintenance of the culvert).

are necessary to establish the plausibility of model
results. In the discussed case studies we found that
a process change from debris flow to inundation can
hardly be represented by the modelling system. The
flood-prone areas had to be delineated in the field by
an expert’s assessment. Furthermore channel-bed or
bank erosion is not yet incorporated. Due to this con-
straint it is not possible to assess break-out points
solely by the modelling system.
field. Additional developments, currently in the test-
ing phase, include the possibility to use an input hy-
drograph instead of a block release, and to include the
influence of channel-bed erosion on the flow properties.
tool for experts evaluating natural hazards. The model
results provide estimates of flow paths, maximum
flow depths and velocities. Using these results, it is
possible to estimate the dynamic forces for use in
evaluating the dimensions of protection measures.
pirical formulas often used in hazard assessment and
may lead to an additional degree of certainty in the
results, especially in cases where model verification
with good-quality field data is possible.
Landscape (BUWAL), Biel.
