# IJEGE-11_BS-Scheidl-&-Rickenmann

*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-030*

**TopFlowDF - A SIMPLE GIS BASED MODEL TO SIMULATE**

**DEBRIS-FLOW RUNOUT ON THE FAN**

**INTRODUCTION**

fects for the affected communities. It is therefore not

surprising that the risk concept for natural hazards (e.g.

H

*et alii*, 1988; G

decades (f

*et alii*, 2008). Subsequently, this ap-

prediction methods, especially when delineating haz-

ardous areas on the fan. To describe the depositional

characteristics and runout behaviour of debris flows,

several approaches, either based on empirical-statistical

or dynamical methods, have been developed during the

last decades. The complexity of prediction methods for

debris-flow runout on the fan vary from simple one-di-

mensional topographical approaches to two-dimension-

al numerical continuum models. An overview of recent

debris-flow runout models can be found in R

*et alii*2008; s

mainly based on their availability and on the require-

ments of local hazard assessments. In Austria for in-

stance, the criterion to delineate hazardous zones for

potential debris-flow events is based on the maximum

runout distance or maximum inundated area on the fan

(e.g. s

**ABSTRACT**

to delineate possible debris-flow inundation zones on

the fan for hazard mapping purposes. However, sev-

eral case studies indicate that empirical relations may

be valid only for situations which are similar to those

represented by the data used for their development,

and that a more accurate description of the debris-

flow depositional process requires dynamical param-

eters which are often not easily constrained. In this

study we propose that the topography of the potential

deposition area has a major influence in the runout of

a debris flow. We combine a random based flow al-

gorithm, generating a maximum simulation perimeter

on the fan, with the simple dynamical approach of a

constant discharge model to predict the maximum ru-

nout on the fan. The program called TopFlowDF sim-

ulates deposition zones, associated deposition heights

and a spatial distribution of the maximum flow veloc-

ities. The GIS-based simulation model runs with high

resolution (2.5 m x 2.5 m) digital elevation models,

generated for example from LiDAR data, and is tested

with debris-flow events from Switzerland and South

Tyrol. The predicted runout patterns of TopFlowDF

are further compared with the empirical runout pre-

diction method TopRunDF.

**K**

**ey**

**words***: debris flow, predictive modelling, GIS*

*C. SCHEIDL & D. RICkENMANN*

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

*L*

*f*

*v*)

and the flow height (

*h*) above the starting point of the

deposition (with the velocity at deposition

*v*

*d*

*h*

*d*

*θ*

*θ*

*f*

friction slope (

*S*

*fric*

*Resistance to flow approaches*

*L*

*f*

*S*

*fric*

*k*

*fric*

*θ*

*f*

*k*

*fric*

*S*

*fric*

*θ*

*f*

*et alii*(1984) found reasonable agreement between ob-

served runouts on the fan and those predicted by Eq.

(1a) by assuming a constant friction slope (

*S*

*fric*

*v*and

*h*from design

discharge by means of empirical equations.

*L*

*f*

*S*

*fric*

*θ*

*f*

*S*

*fric*

Swiss debris-flow events of 1987 with

*S*

*fric*

*θ*

*f*

*et alii*(2010) found a

*k*

*fric*

events in the Dolomites (Eastern Italian Alps).

*S*

*fric*

*θ*

*f*

Switzerland, they proposed:

*S*

*fric*

*θ*

*f*

the intensity (flow height, flow velocity), of the poten-

tial debris-flow event (BWW/BRP/BUWAL, 1997).

These examples may justify different approaches to

predict the mobility of debris-flow events.

nout on the fan. The delineation of deposition zones,

simulated with TopFlowDF, includes a physical and

empirical component since the model combines the

simple dynamical constant discharge model (H

*et alii*, 1984; t

model against observed debris-flow events in Swit-

zerland and Northern Italy (South-Tyrol). The resist-

ance to flow parameters, obtained by several applica-

tions of the constant discharge model, are discussed

and further compared with back-calculated friction

parameters of TopFlowDF simulation results. Finally

we compare the results of TopFlowDF with results

of the empirical based runout model TopRunDF for

Swiss debris-flow events.

**FRAMEWORK OF TOPFLOWDF MODEL**

*CONSTANT DISCHARGE MODEL*

*et alii*(1984) and t

the fan (

*L*

*f*

stant discharge from upstream and that deposition starts

at the place where the channel abruptly levels out, it is

therefore also denoted as leading-edge model (v

*et alii*, 2008). The profile of such

*t*and

*t*+

*Δt*is modelled by the trap-

ezoidal shape, shown in Fig. 1 (t

*Fig. 1 - Process of stoppage of forefront of a debris flow*

*(Takahashi, 1991); SP denotes the starting point*

*of the deposition, the flow parameters are de-*

*scribed in the text*

**TopFlowDF - A SIMPLE GIS BASED MODEL TO SIMULATE DEBRIS-FLOW RUNOUT ON THE FAN**

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

*B*

*perimeter*

flow-height and flow-velocity patterns in step two.

each individual flow path(i):

*U*of a debris-flow

Eqs. (1b) and (1c). The resistance to flow is described

by a user defined friction coefficient (

*k*

*fric*

path(i) is then estimated from the predefined starting

point of the deposition (SP) assuming uniform discharge.

For this reason the input peak discharge (

*Q*

*p*

path(i) and probability of each overflowed cell p(cell

*Q*

*path(i)*

*m*the number of cells within each path(i) and

*n*the number of all flow paths, is then estimated by:

*Q*

*path(i)*

by means of Eq. (5), and is reached when the calcu-

lated flow velocity over time equals zero. Instead of a

constant friction slope, the user needs to define a fric-

tion coefficient (

*k*

*fric*

ing on the slope gradient between the actual outflow-

and inflow-cell. It further prevents the flowing mass to

accelerate on the simulated fan. Therefore the highest

velocities exist at the starting point of the deposition

(simulation). The simulations of the deposition heights

are based on the distribution of the total volume in pro-

portion to the overflow probability

*p(cell*

*j*

*)*of each cell.

event at the Glattbach - torrent in Switzerland.

*FLOw ROUTING ALGORITHM*

discharge model with a random based flow algorithm

which is also implemented in the empirical runout predic-

tion model TopRunDF (s

eter

*B*

*perimeter*

an area-volume relation, which was first proposed by

i

*et alii*(1998) to delineate lahar-inundation

*B*

*perimeter*

*k*

*B*

*V*.

*B*

*perimeter*

*k*

*B*

*V*

*B*

*perimeter*

gorithm combined with a Monte Carlo technique as

described in H

*et alii*(2008) and s

is based on multiple (n) individual flow paths(i) with

constant flow width

*b*(= gridsize of the input DTM)

and a probability

*p(cell*

*j*

*)*of each overflowed cell

individual flow pathways are set by the perimeter of

the elevation model or by adverse slopes.

ity coefficient, a starting point of the deposition (fan

apex) and a digital terrain model of the fan area. The

*Fig. 3 - Estimation of the simulation perimeter with multi-*

*ple individual flow pathways. SP denotes the user*

*defined start point*

*C. SCHEIDL & D. RICkENMANN*

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

*v*= 8.84 m/s and

*h*= 1.01 m

based on the observed Volume (

*V*= 8,800 m

*θ*

*f*

*S*

*f*

*θ*

*c*

*S*

*c*

results to 367 m, based on a friction slope of

*S*

*fric*

*Δt*= 0.5 s the differential

solution stops the debris-flow after 80 seconds reaching

a distance of 360 meters. With a smaller time interval,

*Δt*= 0.01, the estimated distance reaches 367 m (in

80 s). The differential approach shows more precise

results if we use smaller timesteps. Therefore, Top-

FlowDF uses a pre-defined time interval of

*Δt*= 0.01.

flow-width as well as average channel slope above the

starting point and a friction coefficient.

*Q*

*p*

*h*were estimated with em-

*b*at the starting point as well as

*S*

*c*

1:25,000 scale topographic maps.

can be downloaded for free, after registration from

*www.debris-flow.at*

**ANALYSIS OF RESULTS**

(Eq. 5) solutions of the constant discharge model for

an observed Swiss debris-flow event at the Glattbach

torrent in 2005 (Figure 5).

*Fig. 4 - Simulation results of TopFlowDF for the Glattbach debris-flow event. a) predicted deposition zones, b) predicted*

*velocity pattern. SP denotes the starting point of the simulation Contour interval is 1 m*

*Fig. 5 - Comparison between analytical (Eq. 1a) and dif-*

*ferential (Eq. 5) solution of the constant discharge*

*model for the observed debris-flow event at the*

*Glattbach (Switzerland)*

**TopFlowDF - A SIMPLE GIS BASED MODEL TO SIMULATE DEBRIS-FLOW RUNOUT ON THE FAN**

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

in South Tyrol were simulated and evaluated with the

observed deposition patterns. For the modelling of the

individual simulation perimeters a back-calculated

mobility coefficient (

*k*

*Bobs*

*B*

*obs*

*V*

*obs*

*k*

*Bobs*

*B*

*obs*

*V*

*obs*

(VAW, 1992). We therefore limited the “input-volume”

to 50,000 m

Saasbach and Seefeldbach, marked with a * in Table 1.

*α, β*and

*γ*) are determined, based on the relations:

*ε*) and negative (

*φ*) volume-

*ε*is defined as

the relation of the total volume within the predicted

deposition zones related to the observed deposition

areas, and

*φ*is defined as the inverse value of the posi-

tive volume-prediction accuracy. The overall evalua-

tion factor (Ω) is then calculated as:

*α - β - γ + ε*

served deposition pattern. On the contrary, a value of

Ω = -2 implies no overlapping between the simulated

and observed deposition area.

*Tab. 1 - Input process parameters for the TopFlowDF simulations of observed debris-flow events in Switzerland (CH) and*

*South-Tyrol (ST)*

*C. SCHEIDL & D. RICkENMANN*

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

*Tab. 2 - Accuracies and adapted friction coefficients for the simulated debris-flow events with TopFlowDF. The possible range*

*of Ω is [-2,2]; SD denotes standard deviation,CH is for Swiss debris-flow events, whereas ST stands for South-*

*Tyrolean (Italy) debris-flow events*

*Fig. 6 - Comparison of ob-*

*served*

*deposition*

*areas and predicted*

*deposition*

*zones*

*with*

*TopFlowDF*

*a) Glattbach, b)*

*Schwendibach,*

*c)*

*Blauseeligraben, d)*

*Piz Caral, e) Heu-*

*gandtal f) Richleren,*

*g) Rotlauibach, h),*

*Gonerli, i) Brich-*

*boden, j) Val Mera*

*1. SP denotes the*

*starting point of the*

*deposition. Contour-*

*interval is 1 m*

**TopFlowDF - A SIMPLE GIS BASED MODEL TO SIMULATE DEBRIS-FLOW RUNOUT ON THE FAN**

maximum discharge as well as flow width - need to

be estimated based on empirical relations or on field

data. Further, a suitable coefficient for

*k*

*fric*

a correlation between the friction slope

*S*

*fric*

*S*

*f*

flow events. Studies of J

*et alii*(1987), m

*et alii*(2000) and R

a higher roughness and friction during depositional

flow, resulting in steeper fan slopes on average and

in a smaller mobility. A more viscous or muddy flow

behaviour, on the other hand, shows higher mobility

and results in smoother and flatter fans. However,

the back-simulated friction coefficients in this study

(k

events. Future back-simulations are necessary to test

6 and 7. If the simulated deposition area differed from

the observed deposition area (agreement of surface

area), the pre-defined friction coefficient was adapted

(

*k*

*fric(sim)*

the best-fit

*k*

*fric(sim)*

area and 76 % of the observed volume, compared to

the observed deposition zones. 37 % of the area and

24 % of the volume were predicted outside the ob-

served deposition zones. The average friction coeffi-

cient amounts to 1.070 +/- 0.044.

**DISCUSSION AND CONCLUDING REMARKS**

on the constant discharge approach (Equations 1a-1c).

*Fig. 7 - Comparison of observed deposition areas and predicted deposition zones with TopFlowDF a) Gerental 3B, b)*

*Saasbach, c) Ri di Gallinoso, d) klammbach, e) Ri di Sozz; f) Draunbergerbach, g) Fanatjoch, h) koglbach, i)*

*Seefeldbach, j) Arundakopfbach. SP denotes the starting point of the deposition. Contour-interval is 1 m*

*C. SCHEIDL & D. RICkENMANN*

plemented in TopFlowDF) of the friction coefficient

and a resistance coefficient, which is constant along

the entire flow path on the fan, as used together with

the analytical solution in Equation 1a.

*B*

*perimeter*

behaves similarly, regarding the selection of a starting

point of the simulation and the use of the Monte-Carlo-

Iteration number (MCI). The starting point corresponds

to a single DTM cell within a cross-section of the chan-

nel close to the fan apex and is sensitive to the level of

detail of the digital terrain model. Using LiDAR data

with an orthogonal gridsize of 2.5 m a maximum differ-

ence of 10 m was determined between the user defined

start point and the observed start point of the deposition

for the simulated debris-flow events in this study.

lations of Swiss debris-flow events. This number was

also used for the simulation in this study. A detailed

discussion of the MCI number and its influence on the

simulation results is given in s

the fan. In contrast to the similar two-dimensional

model TopRunDF (s

proach but include also a simple dynamical runout

model. Hence, higher uncertainties of the simulation

results of TopFlowDF may originate from the selec-

tion of input flow parameters and due to the separation

of the start momentum into multiple flow paths.

a lower average overall evaluation factor (Ω) for the

results with TopFlowDF are noticed. Table 3 lists all

area- and volume-accuracies as well as the total evalu-

ation factors for the simulation results of TopFlowDF

and TopRunDF respectively, for the 14 Swiss debris-

flow events used in s

lated with TopRunDF, more accurate results can be

achieved, compared to the simulation results of Top-

FlowDF. However, the difference between the mean

values of Ω and Ω* remains small, compared to its

possible range between -2 to 2. Moreover, the qual-

ity of the visual evaluation of the predicted deposition

zones of TopFlowDF compared to the observed depo-

sition areas (Figures 6 and 7) appears to be similar.

*Tab. 3 - Comparison of TopFlowDF and TopRunDF based on the accuracies of predicted deposition zones.The results based*

*on TopRunDF (S*

*cheiDl*

*& r*

*icKeNmANN*

*, 2010) are denoted with **

**TopFlowDF - A SIMPLE GIS BASED MODEL TO SIMULATE DEBRIS-FLOW RUNOUT ON THE FAN**

Tyrol. Markus Zimmermann provided the original

maps which document the debris-flow events of 1987

in Switzerland. The Swiss Federal Office of Topog-

raphy (swisstopo) provided the LiDAR DTM for the

simulations of this study, which are based on DTM-

AV and DOM-AV ©2008 swisstopo (DV033492.2).

**ACKNOWLEDGEMENTS**

diction of debris flows”. The Swiss Federal Office for

the Environment (FOEN) supported the documenta-

tion and analysis of the debris-flow events 2005 in

Switzerland. The Department 30 - Hydraulic Engi-

neering of the Autonomous Province of Bozen - Bol-

**REFERENCES**

Landschaft (BUWAL), EDMZ Nr. 804.201d, Bern.

*Predicting lahar-inundation zones: Case study in west 32 Mount Pinatubo, Philippines.*

**37**: 331-372.

*Field and laboratory investigations of runout distances of debris flows in the*

*Dolomites (Eastern Italian Alps)*- Geomorphology,

**115**: 294 - 304.

*The application of the risk concept to debris flow hazards*. Geomechanik und

**2**: 120-129, Ernst & Sohn, Berlin.

*A review of scale dependency in landslide hazard and risk analysis*. In: G

*Landslide hazard and risk.*75-138. John Wiley & Sons.

*Methoden zur Analyse und Bewertung von*

*Naturgefahren*. Umwelt-Materialien Nr.

**85**, Naturgefahren. Bundesamt für Umwelt, Wald und Landschaft (BUWAL), Bern.

*Quantitative analysis of debris torrent hazards for design of remedial measures.*

**21**: 663-677.

*Evaluation of approaches to calculate debris-flow parameters*

*for hazard assessment.*Engineering Geology,

**102**: 152-163.

**110**: 972-984.

*Some relationships between debris flow motion and microtopography for the kamikamihori fan,*

*north japan alps*. In: b

*Catchment Experiments in Fluvial Geomorphology*. 447-464,

*Debris-flow runout predictions based on the average channel*

*slope (acs)*. Engineering Geology,

**98**: 29-40.

*Empirical Relationships for Debris Flows*. Natural Hazards,

**19**: 47-77.

*Runout prediction methods.*In: J

*Debris-Flow Hazards and Related*

*Phenomena*. 305-324, Praxis Springer, Berlin - Heidelberg.

*Modelle zur Abschätzung des Ablagerungsverhaltens von Murgängen.*Wasser Energie

**102**: 17-26, Baden.

*Empirical prediction of debris-flow mobility and deposition on fans.*Earth Surface

**35**: 157-173.

*Debris-flow runout and deposition on the fan*. In: b

*Tracking past events on fans and cones*.

*Gefahrenzonenplan - Fluch oder Segen?*Festschrift und Tagungsband, 30 Jahre Gefahrenzonenplan“,

**152**: S. 93-104.

*Study on the deposition of debris flows, part 1- deposition due to abrupt change of bed slope.*

*Debris flow*. IAHR Monograph Series. A.A.Balkema / Rotterdam / Brookfield.

*C. SCHEIDL & D. RICkENMANN*

*Debris flow control structures for forest engineering*. Working paper, Ministry of Forest Research

*Murgänge 1987, Dokumentation und Analyse*. Bericht Nr.:

**97**.6 der Versuchsanstalt für Wasserbau, Hydrologie