1

Assessment of Risks on transportation Networks resulting from

slope Instability and Climate change in the Alps

Main objectives

A. Document current DF/shallow landslide activity in the Alps

B. Consequences on transportation network

C. Definition of RCM future climate scenarios

D. DF response considering future climate conditions and lande use planning

E. Help to practical users users

2

3 key regions

3

1- French Alps

(Haute-Durance, Savoie)

2- Eastern Italian Alps (Tagliamento and Adige rivers)

3- Swiss Alps (Zermatt valley)

Grenoble

Briançon

Geneve

Italy

3- Swiss Alps (Zermatt valley)

Method: point A (Current DF/landslides activity)

Current DF databases (historical archives –tree ring)

Current climate conditions

4

- Observed climate data

- Reanalyses climate data

- 27 RCMs downscaled scenarios

France Italy Switzerland

650 DF from 1970 50 DF from 1990s 250 DF+ from 1850

Methods: point A (DF/landslides activity related to

current climate conditions)

In the 3 regions

– Probabilistic and Deterministic models of occurrence at local scale (Shallow landslide model) considering climatic components

– Probabilistic model of occurrence at meso scale (logit simple or hierarchic model) considering climatic and geomorphologic components

5

6

Main results: landslides local scale

Landslide suseptibily considering different precipitation scenarios

defined from meteo stations

Main results : Landslides local scale

7

Ubaye Valley - Southern French Alps

7 shallow forested landslides sampled in the Riou Bourdoux catchment

759 pine trees - 3096 cores sampled

1298 growth disturbances dated

Main results: landslides local scale

Landslide reactivation related to temperature anomalies

during spring

Evolution suggests a shift from snowmelt-induced landslides

(controlled by winter precipitation) to reactivations controlled

by spring temperatures

8

N = 61

Local scale with current meteorological conditions (I/D of precipitation analysis)

9

Precipitation during each event

Main results point A: occurrence of Debris flows-current meteo

Not a good option

UNIPAD Contribution 10

Main results point A: occurrence of Debris flows-current climate

Meso scale climate variables only

- Pseudo homogeneization of RTM data to reduce mistakes in the database

- Analyze from 1970

No data or year without debris flows ?

Main results point A: occurrence of Debris flows-current climate

Meso scale climate variables only

11

Df occurrence

Precipitation (Number

of rainy days)

Temperature

(maximum summer

temperature)

UNIPAD Contribution 12

Main results point A: occurrence of Debris flows- current climate

Debris flow probability=

1 / (1 + EXP((30,23+1*Tx+0,67*Nrd)))

Meteo factors Value Wald Chi-Square Pr > Chi² % correct 0 % correct 1 % correct

Tx 1,064 5,089 0,024 72,47% 72,31% 72%%

Nrd 0,671 3,066 0,008

Main results point A: DF and landslides activity Local scale

DF and shallow landslides triggered by extreme precipitations

Meso scale

DF and shallow landslides triggered by Temperature + precipitation

13

Main results point A: Meso scale : climate + geomorphic parameters

14

Regional component

(climate variables at annual scale)

Individual component

(catchments characteristics)

Binary probability

for a catchment i for a year t

Statistical modeling: bayesian hierarchical probabilistic model based on logistic regression

logit (pit) = a0 + αi + βt

Main results point A meso scale

15

Individual component

• Elevation

• Area

• Slope

• Lithology

• Land cover

• Permafrost

Regional component

• Number of rainy

days

• Daily max

temperature

• P>10mm/d

• P>20mm/d

• …..

16

Geomorphological

component (R² 0.78) Climatic component

(R² 0.72)

Total explained variance: 0.84

% of total variance 0.29

1- Permafrost 80%

2- Surface area 13%

3- Forest (land cover) 7%

% of total variance 0.55

1- max T in summer 67%

2- Nb of rainy days 33%

Main results point A (meso scale): Role of climatic and geomorphic variables in DF triggerring

Point D: Consequences on transportation network

17

A. Consequences on

transportation network

The 4th of june 2012 a DF event destroyed the road close to Lautaret pass

Method: point D (Impacts of DF/landslides activity on road network)

DF impact databases in the three regions

A comparative analysis in normal and disturbed situation of the network (Accessibility : distance and time)

Susceptibility of the network considering future DF probabilities

Crisis management analysis (In France only)

18

Impacts on transportation: Swiss Alps

19

Loss of accessibility in Zermatt valley

125 DF events since the19th century

20

(>7events/year) = 1/(1+exp (-(-

21,91+0,14*Nrd+0,9Tx)))

Impacts on transportation French Alps

Identification of impacted roads

21

(>7events/year) = 1/(1+exp (-(-

21,91+0,14*Nrd+0,9Tx)))

Main results point D

Loss of accessibility with national and international impacts

The 4th of June 2012 a DF event

destroyed the road close to Lautaret

pass

22

>35 mm

>35 mm during 1h

Rainy event responsible for DF event

23

35 mm <10 y return period

IT1 : Normal situation

IT2 : Via Gap

IT3 : Via St Jean de

Maurienne

Grenoble-Briançon options.

24

Normal way

Road destroyed by the DF event

2 solutions

+ 1h30; 36€

+ 1h10; 66€

NCAR/ASP Thompson Lecture Series

Institutionnal vulnerability : crisis management.

Reconstitution of decision-making and organizational process :

25

26

Impacts on transportation: Italy

Identification of impacted roads with local stakeholders (Tyrol region)

Probability of dysfonction in the future

Point C: Definition of RCMs future scenarios

All simulations are based on the A1B emission scenario

24 RCMs until 2050 and 17 RCMs until 2100 from the EU-FP6 project ENSEMBLES.

Error correction.

27

Main results point C: error corrections

28

Raw (orange) and corrected (blue) precipitation distributions

at station St. Valent (Tirol). Left: Light and moderate

precipitation; Right: Heavy precipitation.

29

after Météo-France, 2011

Near future (2020-2050)) Far future (2070-2100) Annual precipitation sum changes

Annual maximum temperature changes

+2°C +4°C

slightly more slightly less

Main results point C

Future climate change

Main results point C: Triggering climate parameters in the future

30

Climate change signal of precipitation types . The number of stations with increasing (arrow up),

decreasing (arrow down), and no change (horizontal arrow (-1 % to +1 %)) for precipitation

frequency of different thresholds is shown. Different colors represent the numbers of stations.

Main objectives

A. Document current DF/shallow landslide activity in the Alps

B. Consequences on transportation network

C. Definition of RCM future climate scenarios

D. DF response considering future climate conditions and lande use planning

E. Help to practical users

31

Point D local scale: Results

32

Current period Future period

Model of DF triggering based on Rcms data

24 RCMs until 2050

17 RCMs until 2100

A1B scenario Model of DF triggering forced by future climate scenarios

2100

2050

Point D meso scale: Results

Model of DF triggering calibrated on Safran data

33

Current period

Future period

Interchanging Safran with RCMs downscaled data

Model of DF triggering based on Rcms data

24 RCMs until 2050

17 RCMs until 2100

A1B scenario

Model of DF triggering forced by future climate scenarios

Step 1 Step 2

Step 3

Point E: Help to practical users

34

1. Volume and run out estimates from MassMov2D

(physical model) for case studies

2. Crisis amanagment analysis

3. Fonctional disturbance analysis regional and

international scale

In France

In Italy

In Switzerland

1. Volume and run out estimates from MassMov2D (physical

model) for case studies

2. Probability of future dysfonction based on climate scenarios

for cases studies

1. Fonctional disturbance analysis regional and international scale

2. Technical investigations

From a case study example: The rif Blanc df event on 4th of June 2012 Volume and run out estimates from a deterministic model

35

36

Volume estimated from peak flow calculation :

(from the video of the 2nd DF at 10am)

1. Number of pictures per second on

the video (30p/sec).

2. Section of the flow

(L x h x WP).

3. Choose a representative block

(B). transported distance/ time

between :

Position 1 at the time T

Position 2 at the time T+1

4. Qp = WS (m²) x velocity (m/s)

5. Rickenmann volume estimation :

Qp = 0.1V0.83

V = 10 760

Debris flow volume (2nd)

= 11 000m3

Help to local stakeholders

37

Sensitivity tests of local protections to DF on the Highway France-Italy

<15 year return period <30 year return period

Sensitivity to climate change of the debris flows which are able to reach the road system along the considered road

38

Help to local stakeholders in italy

Summer

Fall

Conclusions

Current relationships between DF/landslides and climate depends on the considered region

Slope processes strongly impact transportation network

Perception of DF Risk depends on the region

Mitigation is not perfect ! (No by pass, underestimation of protected constructions)

Good relationhsip with public stakeholders more difficult with private component

Climate would have a stronger influence than geomorphic component on DF activity

Future climate change will impact slope processes but how? More data are needed to clarify change in Frequency/ magnitude

39