Napoli 8-11-06-09

Analysis of Saturated and Unsaturated Landslide Triggering through the Distributed Hydrological

Model GEOtop

First Italian Workshop on Landslides - Napoli 8-11 June, 2008

Riccardo Rigon, Silvia Simoni, Cristiano Lanni, Giuseppe Formetta

Monday, June 25, 12

Landslide an approach by GEOtop

R. Rigon, S. Simoni, C. Lanni, G. Formetta and A. Tarantino

2

... you know what I'm craving? A little perspective. That's it. ...Anton Egò

What’s for

•We tried to to have a model that could approach consistently the modeling of

water fluxes and head in a hillslope.

•To get landslide triggering ... well, and other processes.

•To be able to interpret field measures.

•And to cope the model with remote sensed data.

Monday, June 25, 12



3

snow, ice, permafrost

water cycle

shallow landslides

This is it !

Monday, June 25, 12



4Rigon et al., JHM, 2006, Bertoldi et al., JHM, 2006, Simoni, 2007, Endrizzi, 2007

http://www.geotop.org

The GEOtop project1. GEOtop is a distributed hydrological model, which integrates water and energy budget in complex terrain [Rigon et al. 2006].

2. It performs energy balance and water balance, computing energy fluxes between soil and atmosphere, subsurface and surface flows [ Bertoldi et al., 2006].

Monday, June 25, 12





5

Slope stability is assessed through the probabilistic and dynamic module GEOtop-FS [Simoni et al, 2007].

The GEOtop project

Data Assimilation

RadiationEnergyBudget

Vegetation

SubsurfaceFlows

SurfaceFlows

Snow&

Cryosphere

I/O

GEOtop 2010

Monday, June 25, 12



Liquid water

Soil particles

Frozen water

Air gas

Mass Volume

Ms Vs

Vag

Vi

Vlw

VspMsp

Mlw

Mi

Mag

The Basic, however, is the dynamics of

the soil water column

6

Monday, June 25, 12



Mass Budget of Condensed Water

The equation does not come alone. It fulfills some constraints:

⇤t (⇥lw�lw + ⇥i�i) +⇤ · (⇥lw⌃Jlw + ⇥i

⌃Ji)� ⇥lwslw � ⇥isi = 0

or

�r � �lw � ⇥se

�r � �lw + �i � ⇥s

7

Monday, June 25, 12




Some simplifications are readily made:

- The flux of ice is negligible, i.e.

- The source/sink of ice is also neglibile, i.e.

Thus the equation reduces to

⇥Ji � 0

si � 0

⇧t (⇥lw min(⇤se,max(�r, �lw)) + ⇥i min(max(0, �i), ⇤s) +⇤ · (⇥lw�Jlw)� ⇥lwslw = 0

8

Monday, June 25, 12



9

True: we also have to parametrize soil water retention curves and hydraulic conductivity [L T-1] (please do not call permeability [L2]).

For the first we use the van Genucthen (1980) scheme for the second Mualem (1976)


If the ice is neglected, for the present case, the conventional Richard’s equation is obtained

⇧t (⇥lw min(⇤se,max(�r, �lw))) +⇤ · (⇥lw�Jlw)� ⇥lwslw = 0

Monday, June 25, 12



A Flash back ...

Two main contenders debated in the last decade about modeling shallow landslides. Just to personalize this, as it is common nowadays

10

into the geological/gemorphological community

The West Coast guys The USGSes

Monday, June 25, 12



11

A Flash back ... into the geological/gemorphological community

Bill and Dave assert that shallow landslides can be approximately explained by saturation excess hydrology, lateral flow, and infinity slope stability. They produced the widely used and cited SHALSTAB model.

Montgomery and Dietrich, 1994

Monday, June 25, 12



12

A Flash back ... into the geological/gemorphological community

Dick Iverson (S. Baum, J. Godt) insists that the transient vertical effects counts in building a sufficient pore pressure to destabilize hillslope. This broughto TRIGRS

Iverson, 2000; Baum et al., 2002

Monday, June 25, 12



13

A Flash back ... with intrusion of hydrologists

One Italian gang tried to reconcile the fighters observing that all

of that above derives from the Richards equation with various

degree of simplifications, and proposed a linear simplified

theory with analytical solutions available that superimpose the

SHALSTAB theory with Iverson’s one, to be convoluted to

precipitations. D’Odorico et al., WRR, 2005; Cordano and Rigon, WRR, 2008

Monday, June 25, 12



14

A Flash back ... with intrusion of hydrologists

I.M.H.O., the linear theory works close to saturation but has some

problems of parameter characterizations in more unsaturated

cases, which caused some headaches to us.

To make a long story short, we did our best with simplified

theories*, but finally we had to observe that going directly to a

numerical non linear model was the right and conceptually

simpler choice.

*I am pretty sure that Frattini and Crosta, 2009 does it better

Monday, June 25, 12



15

•Area: 13.4 km2

•Min elevation: 924 m

•Max elevation: 2890 m

•Two main ephemeral stream

A case studyRio Corda, Italy, Trentino

•No steep, but hilly Oregon: an Alpine small catchment

Monday, June 25, 12



16


A landslide happened in June 2008 and we tried to reproduce that event

Monday, June 25, 12



17


We investigated the water pressures in the yellow point.

Monday, June 25, 12



18



Psi @ point 1

time (days/month)

wa

ter

su

ctio

n [m

m]

30/04 03/05 06/05 09/05 12/05 15/05 18/05 21/05 24/05 27/05 30/05 02/06 05/06 08/06 11/06

-2500

-1000

01000

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]Monday, June 25, 12



18



Psi @ point 1

time (days/month)

wa

ter

su

ctio

n [m

m]

30/04 03/05 06/05 09/05 12/05 15/05 18/05 21/05 24/05 27/05 30/05 02/06 05/06 08/06 11/06

-2500

-1000

01000

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]Monday, June 25, 12



18



Psi @ point 1

time (days/month)

wa

ter

su

ctio

n [m

m]

30/04 03/05 06/05 09/05 12/05 15/05 18/05 21/05 24/05 27/05 30/05 02/06 05/06 08/06 11/06

-2500

-1000

01000

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]

15.05.08

Monday, June 25, 12



18



Psi @ point 1

time (days/month)

wa

ter

su

ctio

n [m

m]

30/04 03/05 06/05 09/05 12/05 15/05 18/05 21/05 24/05 27/05 30/05 02/06 05/06 08/06 11/06

-2500

-1000

01000

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]

15.05.08

Monday, June 25, 12



18



Psi @ point 1

time (days/month)

wa

ter

su

ctio

n [m

m]

30/04 03/05 06/05 09/05 12/05 15/05 18/05 21/05 24/05 27/05 30/05 02/06 05/06 08/06 11/06

-2500

-1000

01000

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]

22.05.0815.05.08

Monday, June 25, 12



19


We investigated the water pressures in the yellow point.

Psi point 1

time (days/month)

wate

r suction [m

m]

30/04 05/05 10/05 15/05 20/05 25/05 30/05

−2500

−1500

−500

500

1500

0.025 m0.1 m0.25 m0.5 m0.825 m1.2 m

05

10

15

rain

[m

m/h

]

Monday, June 25, 12



20


Method in Simoni et al., HP, 2008

Monday, June 25, 12



20


Method in Simoni et al., HP, 2008

Monday, June 25, 12



21

A simple comparison between GEOtop and SHALSTABRio Corda, Italy, Trentino

Monday, June 25, 12



22


Monday, June 25, 12



23


Top Index - Saturated Top Index - Not saturatedGEOtop -Saturated 26.9 15.04

GEOtop Not Saturated 15.04 43.02

Monday, June 25, 12



24


Monday, June 25, 12



25

Conjectures on the differencesRio Corda, Italy, Trentino

Differences are due to:

•Non stationarity of the response

•Downslope reinfiltration of surface water

•The treatment of surface water in GEOtop that set a boundary conditions for

the water table elevation at the channel network

This is apparent, for instance, after Cordano and Rigon, 2008.

Monday, June 25, 12



26

A theoretical investigation to assess these behavior

Geometry

Soil Type

Antecedent Soil Moisture Conditions

Rainfall intensities and volumes

The aim is to investigate the role of some factors that control the processes

of pore-water pressure redistribution and, hence, the factor of safety (FS) of

the slope.

Different Values of these features are chosen as described below

Monday, June 25, 12



27


Lanni et al., 2009, in preparation

Monday, June 25, 12



27


Lanni et al., 2009, in preparation

Monday, June 25, 12



28

Boundary Conditions

On the sides AB, BC, CD, DEOmogeneous Neumann Boundary

Conditions

On the sides AF, EF (switch of the b.c.)Neumann or Dirichlet Boundary Conditions,

depending from moisture conditions at the

first soil layer

Monday, June 25, 12



29

Computer Experiment Set-up

Two cases analyzed: steep and gentle slope

i. STEEP SLOPE, when the angle of the slope is bigger than the frictional angle of the soil

ii. GENTLE SLOPE, when the angle of the slope is smaller (or is the same) than the frictional angle of the soil

€

tanφ'tanα1

= 0.7

€

tanφ'tanα2

=1.0

Monday, June 25, 12



30

Two cases analyzed: sandy soil and sandy-silt soil

Through physical properties and soil texture it is possible toget the shape parameters of the van Genuchten model usingVereecken PTF (1989)

SANDY SOIL

S1

€

φ'= 35o

€

% sand = 80

€

% silt = 20

€

Ksat =10−4m /s

SANDY-SILT SOIL

S2

€

φ'= 30o

€

% sand = 40

€

% silt = 60

€

Ksat =10−6m /s

The Computer Experiment Set-up

Monday, June 25, 12



31

€

ψ(z) =ψbottom + γw ⋅ H − z( )

Chosen so as to obtain the following values of initial safety factor of the slope:

FS=1.05 for CI1 initial condition



The Computer Experiment Set-upInitial Conditions

Monday, June 25, 12



32

0.7 if steep slope case1.0 if gentle slope case

0.35 (steep) 0.05 (gentle) for CI10.40 (steep) 0.10 (gentle) for CI20.50 (steep) 0.20 (gentle) for CI3

€

FS =tanφ 'tanα

+γwψ( )0.45 γwψb( )0.55

γ ⋅ h ⋅ sinα ⋅ cosαtanφ'


The above prescriptions turn into:

Monday, June 25, 12



33

In unsaturated medium the total stress is defined as:

€

σ = σ − ua( ) + χ ψ( ) ua − uw( ) Bishop (1959)


χ valued according to Khalili & Kabbaz empirical relation (1998)

€

χ =ua − uwua − uw( )b

⎡

⎣ ⎢

⎤

⎦ ⎥

−0.55

if ua − uw( )b < ua − uw( )b

€

χ =1

€

if ua − uw( )b ≥ ua − uw( )b

€

ua − uw( )b FL−2[ ]where:

is the air-entry value matrix suction

Monday, June 25, 12



34

Some Results

1D means that soil failure happens above the bedrock

2D means that soil failure happens at the bedrock

work in progress

Soil Type Slope Initial Conditions Rainfall Duration BehaviorCoarse Steep Wet Short 1DCoarse Steep Wet Long 1DCoarse Steep Dry Long 2D· · · · ·Fine Gentle Wet Any 1D

Probably this also affect the position in the hillside where the failure can happen.

Monday, June 25, 12



35

Some Further Results Resultswork in progress by I. Pretto and C. Lanni

Monday, June 25, 12



36

Some Further Results Resultswork in progress by I. Pretto and C. Lanni

In real life the water table has a curvature due to the channal drainage

Monday, June 25, 12



37

PostGISPostgres

Webservices

WMSWFS-TWPS

Beyond GEOtop: The complete GEOFRAME framework

OpenMI

J-Console Engine

JGrass

uDigEclipse RCP

H2 spatial

UIBuilder

GRASS

GIS engine

The Horton Machine

Models

BeeGIS

www.slideshare.net/GEOFRAMEcafe/geoframe-a-system-for-doing-hydrology-by-computer

Monday, June 25, 12

http://www.slideshare.net/GEOFRAMEcafe/geoframe-a-system-for-doing-hydrology-by-computer

http://www.slideshare.net/GEOFRAMEcafe/geoframe-a-system-for-doing-hydrology-by-computer



Next year

38

First Steps into GEOFRAME: First Componentization

WaterBudget

EnergyBudget

MeteoForcings

The first version

+ + I/O

Monday, June 25, 12



Next year

39

I/OWaterBudget

EnergyBudget

MeteoForcings

The second version

First Steps into GEOFRAME: Second Componentization

Monday, June 25, 12



40

Further splitting will follows

I/O Data Assimilation

RadiationEnergyBudget

Vegetation

SubsurfaceFlows

SurfaceFlows

Snow&

Cryosphere

Monday, June 25, 12

Hoping that Bill, Dick Dave, forgive me!Thank you for your attention

Free GIS: www.jgrass.orgFree Hydrological Model: www.geotop.org

Search also GEOFRAME Cafe on Slideshare

Monday, June 25, 12

http://www.jgrass.org

http://www.jgrass.org



Napoli 8-11-06-09

Education

shalstab rio

computer experiment

geotop mass

lw jlw

surface water

condensed

water pressures

water suction