An prediction model for post-

earthquake debris flow runout zones in

the Wenchuan earthquake area

State Key Laboratory of Geohazard Prevention, Chengdu

University of Technology

ZHU JingZHU Jing

1. Introduction1. Introduction

◆ Debris flows - one of the most dangerous geomorphologic processes that occur in mountainous areas.

◆ The May 12, 2008 Wenchuan earthquake happened three years ago. The heavy rain events have induced massive debris flows.

◆ With the ability to rapidly erode and transport large amounts of material, debris flows have the potential for massive destruction and may be the most hazardous consequence of earthquake-related erosion.

◆ The main objective of the present study was to develop a feasible and verifiable approach to hazard assessment of potential debris flow runout zones on the fans in the Wenchuan earthquake area.

地震前影像

地震后影像

Debris flow event on Sep. 24, 2008 in Beichuan city

2008年 6 月 12日 2008年 9 月 24日

Photograph showing the source areas, channels, and depositional

fans of Shenjia gully debris flow.

a b c

DF16

DF15

RS Imagery from three different data shows the landslide and debris flow development near Beichuan city

◆ These debris flows in the Wenchuan earthquake

area showed an increase in flow volume and

discharge, causing debris flow runout zones to be

much larger than usual.

For this reason, existing methods for the prediction of

the characteristics of runout zones of debris flows

were not applicable for the debris flows in the

earthquake-affected region.

General considerations

◆ In addition, existing empirical models for prediction

of the runout distance on the fans rely on input

parameters that are often difficult to estimate,

including volume, velocity, and frictional factors..

General considerations

Recent catastrophic debris flows indicate a need to improve our understanding of erosion processes following the violent earthquake, and a demand for predictive models that provide critical information on the location and magnitude of these potential disasters.

Therefore, it was necessary to develop a new model for future risk management in the reconstruction areas. We hope to develop a model to estimate debris flow runout zones from easily measurable topographic parameters and sediment supply in a drainage basin.

Our study focused on an area situated in the northeast of the earthquake's epicentre.

46 debris flow gullies with well-defined debris flow fans were chosen and investigated in the study area.

Variables that could potentially influence debris flow runout

zones were measured and used in multiple regression

analysis. Variables were divided into two main categories:

catchment characteristics and the volume of sediment

supply.

Catchment morphological characteristics was

measured in ArcGIS using 25-meter digital elevation

models (DEMs). The runout zones on the fans were

extracted from the SPOT5 images (October 14, 2008)

Data sources

The catchment area, A, is the area enclosed by the

ridge defined by the highest elevations

surrounding the stream.

The catchment relief, H, reflects the gradient of the

channel in a basin and is determined by

dividing the change in elevation between the

top of the debris flow scar and the beginning of

the debris flow deposit.

Data sources

The maximum runout distance, Lf ， is defined as the

length between the onset of deposition point and the lowest point of the deposit on the alluvial fan. This variable is assumed for deposition generally occurring outside the channel and downstream of the fan apex.

The maximum runout lateral width, Bf, is defined as the width of the debris flow lateral spreading in the depositional zone.

Data sources

H

Lf

Bf

Landslide scar

Depositional zone

Basin divide

Parameters measured in the debris flow catchments

DF1DF2

DF18

DF6

DF11

Volume of sediment supply (VL)

The volume of a landslide, VL, is generally estimated

by multiplying the area covered by the deposit by an

estimated average thickness.

To determine the volume of sediment supply in

debris flow source areas, the post-earthquake aerial

photographs taken on May 18, 2008, with a

resolution of 0.3 m, could be used to identify the

landslide deposit regions at least 10 m2 in size.

Data sources

Forty-nine landslides of different scales in the Weijia

gully and Sujia Gully catchments were selected as

the samples for the determination of landslide

thickness.

t =1.432 Ln( SL )-4.985

Where ：

t average landslide thickness (m)

SL landslide area (m2)

Data sources

Empirical relationship between the landslide surface area and

the landslide thickness in the debris flow catchments

Model generation

A multiple regression technique was used to develop an

empirical model for the determination of the characteristics of

debris flow runout zones.

Lf = 0.36A0.06+0.03(VL∙H)0.54-0.18

Bf = 0.40A0.08+0.04(VL∙H)0.35-0.23 Where:

Lf is the predicted maximum runout distance (km)

Bf is the maximum width (km)

VL is the volume of removable sediment in the catchment (106m3)

A is the surface of the debris flow catchment area (km2)

H is the catchment internal relief (km)

Hongchun gully

Shaofang gully

Xiaojia gully

Wangyimiao gully

Mozi gully

Low- altitude aerial photo taken on Augest 15, 2010 showing some debris flow fans produced by the catastrophic event near Yingxiu town on August 14, 2010.

0

0. 1

0. 2

0. 3

0. 4

0. 5

0 0. 1 0. 2 0. 3 0. 4 0. 5Observed (km)

Pred

icte

d (

km)

Maximum runout distanceMaximum lateral width

Validation

To validate the reliability of the statistical model, an independent test of the regression models with 17 catchments yielded generally good results and met the requirements for determination of debris flows runout zones in the Wenchuan earthquake areas.

The major advantage of the empirical relationship was its

simplicity. The only necessary input data were the

topographic parameters and the loose sediment supply. In

contrast, several potential limitations should be considered as

following:

--The main limitations are related to the DEM resolution and

to the estimate of the landslide debris volume.

--The multiple regression was established based on the

limited datasets of surveyed debris flows without taking into

account the specific catchment characteristics that may

influence runout behaviours.

Discussion

--The multiple regression does not account for event volume,

velocity, material properties, or fan morphology, so other,

more rigorous, analyses may be required to provide more

accurate estimates of runout zones.

In spite of the limitations mentioned above, test results

showed that the proposed method could be of potential utility

for practical applications in the Wenchuan earthquake area

and other similar seismic areas where its suitability can be

demonstrated through validation and experience.

Discussion

To enhance the accuracy of prediction, a high priority is the

better understanding and description of depositional

characteristics and runout behaviour of debris flows.

More observations about debris flow events will allow a

refinement of the empirical methods.

In addition, site-specific investigations and knowledge of the

history and magnitude of debris flow events still remains the

most important basis for prediction of any debris flow runout

zone, regardless of the approach used.

Discussion