Download - A distributed physically based model to predict timing and spatial distribution of rainfall-induced shallow landslides

Transcript
Page 1: A distributed physically based model to predict timing and spatial distribution of rainfall-induced shallow landslides

A distributed physically based model to predict timing and spatial distribution ofrainfall-induced shallow landslides.

Grigorios G. Anagnostopoulos and Paolo BurlandoInstitute of Environmental Engineering, ETH Zurich, Switzerland

correspondence: [email protected]

1. IntroductionA distributed physically based model, raster-basedand continuous in space and time, was devel-oped in order to investigate the interactions be-tween surface and subsurface hydrology and shal-low landslides initiation. In this effort emphasis isgiven to the modelling of the temporal evolution ofhydrological processes and their triggering effectsto soil slip occurrences.

2. Hydrological ComponentParticular weight is given to the modeling of hy-drological processes in order to investigate the hy-drological triggering mechanisms and the impor-tance of continuous modeling of water balance todetect timing and location of soil slip occurrences.

• Evapotranspiration is computed using thePriestley-Taylor method.

• Root water uptake is taken into account.

• Surface run-off is routed using the kinematicwave approach.

• The 3-D flow of water through soil and theresulting water balance is considered, by tak-ing into account both saturated and unsatu-rated conditions

• Soil hydraulic hysteresis is also included be-cause it can be crucial for the continuous sim-ulation of soil water content during stormand inter-storm periods.

(i,j-1,0)

(i,j-1,1)

(i,j-1,2)

(i,j-1,k)

(i,j-1,n)

(i,j-1)

(i,j)

(i-1,j)

qinflow

Hpondedqsurface

(i,j+1,0)

(i,j+1,1)

(i,j+1,2)

(i,j+1,k)

(i,j+1,n)

(i,j+1)

qoutflow

qinf

3. Geotechnical ComponentA multidimensional limit equilibrium analysis isutilized for the computation of the stability of ev-ery cell by taking into account the basic principlesof unsaturated soil mechanics.

• The failure surface is assumed to be planarand plastic limit equilibrium conditions areconsidered.

• At the upslope and downslope faces we as-sume that active and passive stresses respec-tively are developed.

• At the lateral faces earth pressures at rest de-velop and the shear resistance is also takeninto account.

• Root cohesion is considered at both lateralfaces and at the base of the column.

θ

Τ Ν

W

φ

Fa

A

σv

Ka σvP

σv

Kp σv

Fp

φ

zx

y

F0

F'0

x

z

φ φ

0

σv

K0 σv

Active stresses

Passive stresses

Τlat

Τlat

4. Soil depth modelingSoil depth is one of the most significant parameterscontrolling the factor of safety (FS) , especially fordepths of less than 1.5 m, within which small vari-ations produce very rapid changes in the FS. Theapproach of Pelletier et al. is to solve numericallythe steady-state form of the landscape evolution.

@h@t = ⇢b

⇢sP0

p1 + |rz|2e

� h

(h0

p1+|rz|2 +

D3r⇣

hrz1�(|rz|/Sc)2

The parameters were determined by searchingthrough the parameter space the parameter setthat minimizes the root-mean-square difference be-tween predicted and measured soil depth data.

5. Test case: Napf catchmentNapf catchment is located in Kanton Bern, Switzerland. Itspans over an area of 2, 5 km

2, 48 % of which is forested. A3-hour precipitation event on 15-16 July 2002 caused manysoil slips.

A 3x3 m DEM is available and the soil map of Switzer-land is used for the identification of the soil classes present inthe catchment. As meteorological input the historical recordof the Napf station was used, which is located 5 km north ofthe cathcment.

The output of our model is tested against a state-of-the-artmodel (TRIGRS, Baum and Godt 2010) and against the inven-tory of observed landslides.

• The model correctly captures most of the observed land-slides (True Positive Rate: 43,2% against 23,5% of TRI-GRS).

• It reduces the overestimation of landsliding cells, whichis a main artifact of most of the existing models (slidingarea: 6.5% against 13% of TRIGRS).

• The continuous modeling of soil moisture and the inclu-sion of many feedback mechanisms improved the predic-

tive ability of the model. The timing is also affected com-pared to TRIGRS mainly due to the more detailed hydro-logical component.

References[1] G.G. Anagnostopoulos, P. Burlando, (2011). Object-oriented computational framework for the simulation of variably saturated flow, using a reduced complexity model, Submitted in Environmental Modelling

& Software

[2] R.L. Baum, J.W. Godt, (2010). Estimating the timing and location of shallow rainfall-induced landslides using a model for transient, unsaturated infiltration. Journal of Geophysical Research, Vol 115

[3] J. Pelletier, C. Rasmussen (2009). Geomorphically based predictive mapping of soil thickness in upland watersheds. Water Resources Research, Vol 45

Grigoris Anagnostopoulos
Abstract No: NH 31B-1548