# IJEGE-11_BS-Bregoli-et-alii

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

*DOI: 10.4408/IJEGE.2011-03.B-091*

**DEVELOPMENT OF PRELIMINARY ASSESSMENT TOOLS TO EVALUATE**

**DEBRIS FLOW HAZARD**

graphic Information System (GIS), in combination

with statistical analysis (e.g., m

*et alii*, 1995)

(e.g., i

*et alii*., 1998; G

*et alii*, 1999).

prehensive field work to determine the hazard in the

debris flow deposition areas.

ent from the mapping that is usually performed by the

basin authority in order to give guidance for the plans

of urban development and, in general, for the manage-

ment of the territories. Actually, in case of early warn-

ing, a comparatively coarser identification of the areas

at risk can be sufficient. In this perspective the hazard,

as a component of risk, may be estimated in a simpli-

fied way and it will be the topic of the present study.

in order to construct a hazard map it is necessary to

estimate, for each elementary portion of the area ex-

amined, the intensity of possible events and the cor-

responding event probability.

through the impact energy of the flow against an ob-

stacle, which depends on the characteristics of flow

depth and velocity (R

**ABSTRACT**

to Debris Flow, one of the main topics is the prelimi-

nary identification of areas at risk. In case of early

warning a coarser identification of areas at risk should

be sufficient. In this perspective, the hazard of phe-

nomena, as component of risk, can be estimated in a

simplified way. In the framework of the IMPRINTS

European Research Project (FP7), a toolbox for fast

assessment of debris flow hazard has been developed.

The aim of this toolbox is to implement different exist-

ing models inside a common package useful for a fast

evaluation of potential hazard. The identification of

hazard is performed by different levels of accuracy, de-

pending on the availability of input data. As an exam-

ple, the result could be achieved by a rough handling

of topographical data but could be improved in quality

by adding geological and hydrological data. Both the

initiation and propagation of the debris flow are mod-

elled. For this study, the methodology has been applied

in a catchment located in the North East of Spain.

**K**

**ey**

**words***: debris flow, hazard assessment, run out, shallow*

*landslide*

**INTRODUCTION**

scale and studies at local scale. Debris flow hazard

*F. BREGOLI, A. BATEMAN, V. MEDINA, F. CIERVO, M. HÜRLIMANN & G. CHEVALIER*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

and save computational time. Past studies have inves-

tigated the occurrence of debris flows and reported

the area of the catchments affected by various de-

bris flows events (s

known to have experienced debris flows. Scheidl

(2009) proposed a threshold value of 25 km

*et alii*(2010) report that many of the Strahler’s

curred, has a similar maximum value.

der to fill the natural depressions that could influence

seriously the execution of the subsequent steps. Be-

cause of that lakes and dams are removed from the

DEM, making those areas as “no data”. It means that

in case of the presence of dams the method is not

valid. A “fillsinks” algorithm is included in the meth-

odology implemented..

*INITIATION MODELS*

*et alii*, 1996; H

*et alii,*2001) but the mobilization from rainfall-trig-

gered landslides (i

*et alii,*1997) seems to be

behaviour of debris flow initiation.

loose sediment layer, thus reducing the factor of safe-

ty. This behaviour is described by the typical Coulomb

failure approach in the infinite slope stability model

(i

*et alii,*1997)

tions are reached after a rainfall having constant inten-

sity and indefinite duration (d

*et alii*, 1995). If

table coincident with the free surface) is also made, a

very simple relation between rainfall and soil trans-

missivity may be derived (d

*et alii*, 1995) for

shallow landslides and his propagation through the paths

pending on the data input availability and on the detail

required by the analysis.

some preliminary results are presented. The validation of

the models is still a work in progress..

**DEBRIS FLOW HAZARD ASSESSMENT**

**METHODOLOGIES**

common methodology able to describe the phenomenon

in a wide range of cases. A flow chart of methodology is

described in Fig. 1.

the debris flows triggered by rainfall of particular in-

tensity and duration, the probability of occurrence of

an event may be related to the return period of the trig-

gering rainfall. The connection between the rainfall

and its effects can be reconstructed by the simulation

of the different processes, which take place during the

spatial and temporal evolution of the flow from its

mobilization to its stop. In this sense we can distin-

guish two distinct phases: the first aims to estimate the

volume potentially mobilized by a given precipitation,

with an assigned return period, (initiation models), the

second has the objective to estimate the area of inva-

sion and the resulting intensity (propagation models)

*Fig. 1 - debris flow hazard assessment methodology’s flow*

*chart*

**DEVELOPMENT OF PRELIMINARY ASSESSMENT TOOLS TO EVALUATE DEBRIS FLOW HAZARD**

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

(IDF). An example of those curves is given in Figure 2.

*PROPAGATION MODELS*

trajectories of debris flow. G

*et alii*(2008) combined a D8 flow routing algo-

ly applied in the European Alps and the Spanish Pyr-

enees catchments. The model used here is a modifica-

tion of the previous one with the incorporation of local

flow velocity computation and a stopping mechanism

to obtain a flow path of propagation for each point,

and subsequently

*n*

*P*

*n*

containing information on the probability of each cell

of the DEM to be affected by a future debris flow. The

result depends strongly on the DEM resolution and on

the number of iterations, which is recommended to be

set to 10

*et alii*, 2008)

Debris Flow (1955):

*q*is the rainfall intensity, T is the soil transmissiv-

ity, )α is the slope, a/b is the cumulated area per width of

flow,

*p*

*z*is the thickness of soil,

*φ*is the soil internal friction angle

and

*p*

ruptures and it is not possible to compute the total un-

stable debris volume. For these reasons this approach

is here called “

*qualitative-steady state”*.

*h*and the

thickness of the soil layer

*z,*may be derived by equa-

tion [2] (d

*et alii*, 1995).

*F*

*γ*

quently the corresponding return period is not defined.

To overcome this difficulty, the duration of the rainfall

event is fixed equal to the time necessary for the soil

to reach to steady state condition. A simple relation to

evaluate such interval time is proposed in equation [4]

(P

*et alii*, 2010):

*quantita-*

tive-steady state”.

tive-steady state”

*Fig 2 - IDF curves in the Upper Llobregat Basin in Cata-*

*lonia, Spain*

*F. BREGOLI, A. BATEMAN, V. MEDINA, F. CIERVO, M. HÜRLIMANN & G. CHEVALIER*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

based on the shallow flow hypothesis and is depth in-

tegrated. A bi-dimensional approach is used for mo-

mentum conservation..

• constant density flow is considered

• no pore pressure effect is considered

• Terrain curvatures are neglected

• Steep slopes are considered

• Multiple rheologies are implemented

sets including a DEM and a raster defining the initial

extension and volume of the debris flow. The accuracy

of calibration of the rheological parameters and the

computational time requirement represent the major

drawback of this technique, but the outputs can be di-

rectly used to generate intensity maps, since velocity

and flow depth are simulated within the entire study

area. The computational cost also increases consider-

ably with the number of initiation points..

**MULTILEVEL APPROACH**

Tab. 1). Obviously the quality of output depends on the

system applied: detailed input data allow for the use of

more complex models and give better results. The three

systems proposed (S_mall, M_edium and L_arge) are

composed by two models, an initiation model and a

propagation model.

*v*is velocity of the mixture,

*s*is the flow path

line,

*μ*

*k*is the “tur-

*μ*

*k*should be defined by backanalysis, but typi-

and the total debris flow volume (C

*β*is the reach angle,

*H*is the gradient between

centre of mass of landslide and fan,

*Lmax*is the travel

distance and

*V*is the volume in m

involved in equation 7 are described

and does not includes the depth of deposit. However

the velocity, useful for hazard assessment, is estimated.

pogenic modification of land is considered low

*et alii*, 2008). FLATModel is a two-dimen-

with analytical, experimental and real test cases. It is

a complete model that include basal entrainment of

sediments, stop and go phenomenon, dynamical cor-

rection of the evolution of fan slope and different fluid

models including laminar rheologies (Bingham, Her-

schel-Bulkley), granular flows (Coulomb, Voellmy)

*Fig. 3 - scheme of a debris flow reach angle and variables*

*involved*

*T*

*ab. 1 - Multilevel approach at debris flow hazard assess-*

*ment.*

**DEVELOPMENT OF PRELIMINARY ASSESSMENT TOOLS TO EVALUATE DEBRIS FLOW HAZARD**

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

Model). In order to run such model it is necessary

to estimate the rheological properties of solid-liquid

mixture in motion. The results of the computation are

the flow depth and velocity for each numerical cell of

the affected area. The result of the L_arge system is a

map with quantitative assessment of classes of debris

flow intensity corresponding to fixed return periods.

**HAZARD ASSESSMENT**

ability is given by the combination of the initiation

and the propagation models. Depending on the model

adopted, the hazard is defined as below.

*HAZARD IN THE S_MALL MODEL*

Taking in account the works of d

It has to be remarked that values of Log(q/T)= 1 are

possible due to particular local values and truncation

errors. Such value has to be neglected.

zone and the invaded areas are assigned the same level

of hazard defined in Tab. 3.

*HAZARD IN THE M_EDIUM MODEL*

plied, for the width of study area and for the lack of

input data. The compilation of input data files and the

running of the system are very fast. The assessment

of possible unstable areas is performed by the infinite

slope stability model in which the water pore pressure

is estimated, in a simplified way, by assuming steady-

state groundwater flow (d

*et alii*, 1995) and

plified approach only the morphological description

of the basin is needed and the result does not depend

on rainfall data. As a consequence it is not possible to

define a return period for the event. The given result

is simply a qualitative map of the area most prone to

debris flow initiation

a map with qualitative assessment of classes of debris

flow intensity corresponding to hypothetical scenarios.

on a given rainfall in the hypothesis of steady-state

groundwater flow (d

*et alii*, 1995). With the

tion it is possible to obtain a map of instable area with

given return period..

edium system is a map with qualitative assessment

of classes of debris flow intensity corresponding to

fixed return periods.

the M_edium system while the propagation model is

*Tab. 2 - Data input required by the different proposed*

*methods. DEM, Digital Elevation Model; ,, inter-*

*nal friction angle; c’, cohesion; k, hydraulic con-*

*ductivity; z, soil thickness; q, rainfall intensity; D,*

*rainfall duration*

*Tab. 3 - Definition of hazard in the S_mal model*

*T*

*ab. 4 - Definition of hazard in the M_edium model*

*F. BREGOLI, A. BATEMAN, V. MEDINA, F. CIERVO, M. HÜRLIMANN & G. CHEVALIER*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

*HAZARD IN THE L_ARGE MODEL*

definition of the intensity is done in Tab. 6 following

G

*et alii*(2005).

**IMPLEMENTATION OF ALGORITMS**

GNU/GPL open source code. Most of the algorithms

require basic common Geographic Information Sys-

tem (GIS) tools (i.e. slope, curvature, aspect). These

tools are provided through the SEXTANTE GIS library

(G

job. Algorithms themselves are included inside SEX-

TANTE to be available to extern applications. Notice

that SEXTANTE is not a GIS but a library that could

be accessed from different open source as well as com-

mercial GIS. A command line application to use SEX-

TANTE library without GIS is developed in that study.

The data exchange formats of information are ESRI

ASCII for raster and ESRI shapefile for vectorial.

**APPLICATION OF METHODOLOGIES**

(Spain). The outlet is located immediately upstream of

the Baells Reservoir in the municipality of Berga. The

catchment considered has an area of about 350 km

the world is a 30x30 meters, coming from the database

of ASTER_GDEM (2008).

*S_MALL MODEL APPLICATION*

• z = 1 m

• Z = φ.48 rad

but only with the internal friction angle. In that case

the internal friction angle should be increased till

450 to counteract the absence of cohesion (d

*at alii*, 1994).

slope stability. It is a common result that that tech-

nique is overestimating zones of failure (as also re-

ported in previous studies as C

*et alii*, 2008).

*Tab. 5. - Definition of hazard in the L_arge model*

*Tab. 6 - Definition of inensity in the L_arge model (G*

*ArciA*

*et alii, 2005)*

*Fig. 4 - Upper Llobregat basin location*

**DEVELOPMENT OF PRELIMINARY ASSESSMENT TOOLS TO EVALUATE DEBRIS FLOW HAZARD**

*et alii*, 2010). In the framework of the same

been done, including the reconnaissance of initiation

points, total volume involved and depositional area

through standard studies as well as dendrochronology

studies on affected trees. The event of 2006 is taken in

account for the present simulation (Fig. 7)

depth of the flow..

The volume of initiation is estimated to be 2000 m3.

The Voellmy rheology for granular flow is used and

the rheological parameters are estimated by back anal-

ysis and settled as below:

• turbulent friction term: C = 10 m

tion of flow at the end of the fan is registered. That

trend seems to be possible in the future, due to the

reconnaissance of a new flooded path emphasized by

recent events.

qualitatively and not in term of return period..

In that case spatial distributed values of cohesion,

estimated from geological maps. In that region no soil

map is available and a reclassification of a geological

map has been done. The use of geotechnical parameters

coming from geology maps is not appropriate for the

methodologies presented (v

*r*= 2200 kg/m

In Figure 6.b it is shown the result after running the

stochastic model of propagation. In that case the result

is given in term of return period.

cal parameters, it would be more interesting to move

in a basin where those parameters are well known

*L_ARGE MODEL APPLICATION*

Llobregat that is suffering debris flow activity. In that

spot the Technical University of Catalonia, thanks to

the National Research Project DEBRISCATCH, has

*Fig. 5 - S_mall model application. a) Result of the qualita-*

*tive steady state initiation model; b) hazard after*

*the propagation with the stochastic model*

*Fig. 6 - M_edium model application. a) Result of the*

*quantitative steady state initiation model; b) haz-*

*ard after propagation with the stochastic model*

*F. BREGOLI, A. BATEMAN, V. MEDINA, F. CIERVO, M. HÜRLIMANN & G. CHEVALIER*

to choose the most appropriate method according to

their data and needs. A toolbox has been developed to

facilitate users in the application of the methodologies

for their test beds.

The first step of the process is the assessment of event

intensity and consequent hazard; it requires math-

ematical and numerical modelling of the debris flow.

**DISCUSSION AND CONCLUDING RE-**

**MARKS**

bris flow and flash floods. Methodologies of differ-

ent level of accuracy have been developed, requiring

different level of elaboration, manual work and data

accuracy. The project has shown that hazard assess-

ment delineation is a large computational process sea-

soned with an important manual effort for the user.

*F*

*ig. 7 - Initiation and depositional areas of debris flow for*

*the event of 2006 in the Ensija’s Catchment*

*Fig. 8 - FLATModel simulation result in the Ensija’s*

*Catchment. Maximum velocity recorded du-*

*ring the simulation*

*Fig. 9 - FLATModel simulation result in the Ensija’s*

*Catchment. Maximum depth recorded during*

*the simulation*

*F*

*ig. 10 - Result of the intensity analysis using the output of*

*FLATModel*

**DEVELOPMENT OF PRELIMINARY ASSESSMENT TOOLS TO EVALUATE DEBRIS FLOW HAZARD**

too much depending on the geotechnical parameters

and it would be more interesting to move in a basin

where those parameters are known.

*et alii*, 2008,

*et alii*, 2008) and the case of studies pre-

tween the model and the field studies.

**ACKNOWLEDGEMENTS**

(IMproving Preparedness and RIsk maNagemenT

for flash floods and debriS flow events), SEVENTH

FRAMEWORK PROGRAMME THEME 6.1.3.3,

ENVIRONNEMENT: Preparedness and risk manage-

ment for flash floods including generation of sediment

and associated debris flow, Grant agreement n°: FP7-

ENV-2008-1-226555.

the Spanish Ministry of Education.

• The M_edium system

• The L_arge system

The choice of the proper system may be done de-

study area and the computational effort; the simple

models (S_mall and M_edium) well meet requirements

of early warning system, while the complex models,

like the L_arge, are more useful for the compilation of

detailed hazard maps to be used for territorial planning

The S_mall model is known from previous similar

studies, which is overestimating the zone of failure

(C

*et alii*., 2008) due to the low accuracy of the

curate study has to be done to validate the methodol-

ogy. The main issue, in that case, is the lack of data in

**REFERENCES**

*Comparing models of debris-flow susceptibility in the alpine environment*. Geomor-

**94:**353– 378.

*Geomorphology-based debris flow hazard recogni-*

*tion in the Easter Pyrenees*

*– a fluvial approach.*International workshop on EU Flood directive implementation in Mediter-

ranean zone: Tools and challenges for efficient risk management, Barcelona, 36.

*The angle of reach as a mobility index for small and large landslides*. Canadian Geotechnical Journal,

**33**(2)

**:**260–271.

*Recognition, classification and mechanical description of debris flows*. Earth-Science Re-

**40:**209-227.

*A physically based model for the topographic control on shallow landsliding*. Water

**30**(4)

**:**1153–1171.

*dfwalk—Ein Murgang-Simulationsprogramm zur Gefahrenzonierung.*PhD-Thesis, University of Berne, 144

*Debris Flows of December 1999 in Venezuel*a. In: J

*Debris-flow*

*Hazards and Related Phenomena.*Chapter 20

*Sextante Java library for geospatial analysis.*Junta de Extremadura, European Commission,

*Landslide hazard evaluation: a review of current techniques*

*and their application in a multi-scale study, central Italy*. Geomorphology,

**31:**181-216.

*A Review of the Classification of Landslides of the Flow Type.*

**VII**(3)

**:**221-238.

*F. BREGOLI, A. BATEMAN, V. MEDINA, F. CIERVO, M. HÜRLIMANN & G. CHEVALIER*

*Set-up of debris-flow monitoring*

*stations in the Eastern Pyrenees*. Preliminary results and first experiences. General Assembly of the European Geosciences

Union, Vienna. Geophysical Research Abstract,

**12**: 5351.

*A. Evaluation of approaches to calculate debris-flow parameters*

*for hazard assessment.*Engineering Geology,

**102:**152-163.

*Debris-flow mobilization from landslides*. Annu. Rev. Earth Planet. Sci.

**25:**

*Objective delineation of lahar-inundation hazard zones.*Geological

**110:**972–984.

*Estimation of debris flow magnitude in the eastern Italian Alps.*Earth Surface Process and

**29:**207-220.

*Statistical and simulation models for mapping debris-flow hazard*. In: C

*Geographical Information Systems in Assessing Natural Hazards*. Kluwer Academic Publishers, Dordrecht, The

*FLATModel: a 2D finite volume code for debris-flow modelling. Application*

*to events occurred in the Eastern Pyrenees*. Int. J. Sediment Research,

**23**(4): 348–360. (http://dx.doi.org/10.1016/S1001-

6279(09)60006-8)

*Valutazione della pericolosità connessa a fenomeni di correnti detritiche*. Proceedings of the

*Runout prediction methods*. In:

*Debris-flow Hazards and Related Phenomena.*Springer, Berlin: 305-

*Valley incision by debris flows: Evidence of a topographic signature.*Water Resources Re-

**39:**1089.

*Prediction of debris-flow mobility and deposition on torrential fans*. PhD thesis, Institute of Mountain Risk

*Spatial data for landslide susceptibility, hazard, and vulnerability*

*assessment: An overview.*Engineering Geology,

**102:**112–131.