hillslope hydrologyandrichards

120
An Overview Hillslope Hydrology Mirò - Blue II Riccardo Rigon 2 nd International Summer School on Water Research, Praia a Mare, July 2013 Monday, July 8, 13

Upload: cafe-geoframe

Post on 06-May-2015

440 views

Category:

Education


0 download

DESCRIPTION

The Presentation I gave at the second summer school on Water Resources

TRANSCRIPT

Page 1: Hillslope hydrologyandrichards

An Overview Hillslope HydrologyM

irò -

Blu

e II

Riccardo Rigon

2nd International Summer School on Water Research, Praia a Mare, July 2013

Monday, July 8, 13

Page 2: Hillslope hydrologyandrichards

Goals

• Say what a hillslope is

• Talking about Richards equation

• Say what Hydrology on hillslope is concerned about

• Simplifying Richards’ equation

1

2

• Some reflections

• And Beyond ...

Welcome

R. RigonMonday, July 8, 13

Page 3: Hillslope hydrologyandrichards

Hillslope HydrologyHillslope

Mir

ò -

Blu

e II

Riccardo Rigon

2nd International Summer School on Water Research, Praia a Mare, July 2013

Monday, July 8, 13

Page 4: Hillslope hydrologyandrichards

4

What is a hillslope ?

Mon

tgom

ery

and

Die

tric

h, W

RR

, 19

92

First you have to identify channels

R. Rigon

What we are talking about ?

Monday, July 8, 13

Page 5: Hillslope hydrologyandrichards

5

Orl

an

din

i et

al.,

20

11

R. Rigon

What we are talking about ?

Monday, July 8, 13

Page 6: Hillslope hydrologyandrichards

6

Hillslopes

R. Rigon

What we are talking about ?

Monday, July 8, 13

Page 7: Hillslope hydrologyandrichards

7

Mon

tgom

ery

and

Die

tric

h, 1

98

9

Well ...

R. Rigon

What we are talking about ?

Monday, July 8, 13

Page 8: Hillslope hydrologyandrichards

da

Tar

boto

n: w

ww

.cu

ahsi

.org

8

R. Rigon

Hillslopes

Monday, July 8, 13

Page 9: Hillslope hydrologyandrichards

9

da

Tar

boto

n: w

ww

.cu

ahsi

.org

R. Rigon

Hillslopes

Monday, July 8, 13

Page 10: Hillslope hydrologyandrichards

10

da

Tar

boto

n: w

ww

.cu

ahsi

.org

R. Rigon

Hillslopes

Monday, July 8, 13

Page 11: Hillslope hydrologyandrichards

11

da

Tar

boto

n: w

ww

.cu

ahsi

.org

R. Rigon

Hillslopes

Monday, July 8, 13

Page 12: Hillslope hydrologyandrichards

12

da

Tar

boto

n: w

ww

.cu

ahsi

.org

R. Rigon

Hillslopes

Monday, July 8, 13

Page 13: Hillslope hydrologyandrichards

13

Dolomites- Duron Valley

R. Rigon

Hillslopes

Monday, July 8, 13

Page 14: Hillslope hydrologyandrichards

14

Soil depth

Soil

rocks

Where does water flow ?

R. Rigon

Soil

Monday, July 8, 13

Page 15: Hillslope hydrologyandrichards

15

Keep in mind the complexity

Courtesy of Enzo Farabegoli - Duron catchment

R. Rigon

The complexity of geology (and of gelogists)

Monday, July 8, 13

Page 16: Hillslope hydrologyandrichards

Hillslope HydrologyHydrology

Mir

ò -

Blu

e II

Monday, July 8, 13

Page 17: Hillslope hydrologyandrichards

17

How water moves in hillslopes ?

Turbulent flows - Laminar flows

Both are described by the Navier-Stokes equations

R. Rigon

Fundamentals

Monday, July 8, 13

Page 18: Hillslope hydrologyandrichards

18

2D - de Saint Venant equations with some smart subgrid parameterization (e.g. Casulli, 2009)

1D - Kinematic equationSo many to cite here but ... Liu and Todini, 2002

R. Rigon

Less is more

Navier-Stokes equations are actually never used to do hillslope hydrology

For a synthesis see: abouthydrology.blogspot.com

R. RigonMonday, July 8, 13

Page 19: Hillslope hydrologyandrichards

19

How water moves in hillslopes ?

Turbulent flows - Laminar flows

Darcy flows

R. Rigon

Fundamentals

Monday, July 8, 13

Page 20: Hillslope hydrologyandrichards

20

Darcy equations are OKfor saturated flow

They can be obtained from Navier-Stokes Equation by*:

•introducing a resistance term

•assuming creep flow (neglecting kinetic terms)

•integrating over the Darcy scale

*Whitaker, 1966; Bear, 1988; Narsilio et al., 2009

R. Rigon

Fundamentals

Monday, July 8, 13

Page 21: Hillslope hydrologyandrichards

21

What aboutunsaturated flow

R. Rigon

Fundamentals

Monday, July 8, 13

Page 22: Hillslope hydrologyandrichards

22

One idea isthat we can use Richards’ equation

So, on the earth what is Richards’ equation ?

R. Rigon

Fundamentals

Monday, July 8, 13

Page 23: Hillslope hydrologyandrichards

23

Richards’ equation core

is that what it is true is this

Mass conservation (no nuclear reactions) !but actually true if the continuum (a.k.a. Darcy) hypothesis is valid

Process based models

R. RigonR. RigonMonday, July 8, 13

Page 24: Hillslope hydrologyandrichards

Not necessarily this:

24

Se = [1 + (��⇥)m)]�n

Se :=�w � �r

⇥s � �r

C(⇥)⇤⇥

⇤t= ⇥ ·

�K(�w) �⇥ (z + ⇥)

K(�w) = Ks

⇧Se

⇤�1� (1� Se)1/m

⇥m⌅2

SWRC + Darcy-Buckingham

(1907)

ParametricMualem (1976)

Parametricvan Genuchten

(1981)

C(⇥) :=⇤�w()⇤⇥

Process based models

R. RigonMonday, July 8, 13

Page 25: Hillslope hydrologyandrichards

25

To obtain the last slide

One has to assume the validity of the Darcy-Buckingham law:

Darcy-Buckingham Law

Volumetric flow through the surface of the infinitesimal volume

Bu

ckin

gh

am, 1

90

7, R

ich

ard

s, 1

93

1

~Jv = K(✓w)~r h

Fundamentals

Monday, July 8, 13

Page 26: Hillslope hydrologyandrichards

25

To obtain the last slide

One has to assume the validity of the Darcy-Buckingham law:

Darcy-Buckingham Law

Volumetric flow through the surface of the infinitesimal volume

Bu

ckin

gh

am, 1

90

7, R

ich

ard

s, 1

93

1

~Jv = K(✓w)~r h

Fundamentals

Monday, July 8, 13

Page 27: Hillslope hydrologyandrichards

25

To obtain the last slide

One has to assume the validity of the Darcy-Buckingham law:

Darcy-Buckingham Law

Volumetric flow through the surface of the infinitesimal volume

Bu

ckin

gh

am, 1

90

7, R

ich

ard

s, 1

93

1

~Jv = K(✓w)~r h

Fundamentals

Monday, July 8, 13

Page 28: Hillslope hydrologyandrichards

25

To obtain the last slide

One has to assume the validity of the Darcy-Buckingham law:

Darcy-Buckingham Law

Volumetric flow through the surface of the infinitesimal volume

Hydraulic conductivity times gradient of the hydraulic headB

uck

ingh

am, 1

90

7, R

ich

ard

s, 1

93

1

~Jv = K(✓w)~r h

Fundamentals

Monday, July 8, 13

Page 29: Hillslope hydrologyandrichards

26

Ignore soil hysteresisand think of the SWRC as a function that relates water content to matric

pressure

⇤�(⇥)⇤t

=⇤�(⇥)⇤⇥

⇤⇥

⇤t� C(⇥)

⇤⇥

⇤t

Hydraulic capacity of the soil

R. Rigon

Fundamentals

Monday, July 8, 13

Page 30: Hillslope hydrologyandrichards

27

Assume a parametric form of soil water retention curves

Se :=�w � �r

⇥s � �r

Parametricvan Genuchten

(1981)

C(⇥) :=⇤�w()⇤⇥

Se = [1 + (��⇥)m)]�n

But other forms are possible ...

R. Rigon

Fundamentals

Monday, July 8, 13

Page 31: Hillslope hydrologyandrichards

28

A theory for getting hydraulic conductivity from soil water retention curves

K(�w) = Ks

⇧Se

⇤�1� (1� Se)1/m

⇥m⌅2Parametric

Mualem (1976)

But other forms are possible also here...

R. Rigon

Fundamentals

Monday, July 8, 13

Page 32: Hillslope hydrologyandrichards

29

The last representation of mass conservationis just matter of convenience

habits, and ignorance of some phenomena

•variable and changing temperature

•soil freezing

•transition to saturation

•preferential flow

Process based models

R. RigonR. RigonMonday, July 8, 13

Page 33: Hillslope hydrologyandrichards

An example of top down derivationfrom Richards’ equation

Ch

imp

anzee

Con

go p

ain

tin

g

Monday, July 8, 13

Page 34: Hillslope hydrologyandrichards

Iver

son

, 20

00

; Cord

ano e

Rig

on

, 20

08

31

The Richards equation on a plane hillslope

Richardsoniana

R. RigonMonday, July 8, 13

Page 35: Hillslope hydrologyandrichards

Iver

son

, 20

00

; Cord

ano e

Rig

on

, 20

08

32

The Richards equation made dimensionless

Richardsoniana

R. RigonMonday, July 8, 13

Page 36: Hillslope hydrologyandrichards

Iver

son

, 20

00

; Cord

ano e

Rig

on

, 20

08

33

Richards eq. solution expressed in terms of

the asymptotic hydrostatic solution and a transient term:

See also. D’Odorico et al., 2003

Richardsoniana

R. RigonMonday, July 8, 13

Page 37: Hillslope hydrologyandrichards

34

Asymptotic solution

Transient solution

A lot of tricks here !

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 38: Hillslope hydrologyandrichards

35

Depth from surface

Terrain Slope

Water table position

A lot of tricks here !

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 39: Hillslope hydrologyandrichards

and one equation forIver

son

, 20

00

; Cord

ano e

Rig

on

, 20

08

36

So Richards equation is

divided into one equation for

Richardsoniana

R. RigonMonday, July 8, 13

Page 40: Hillslope hydrologyandrichards

37

Interestingly

Water table was not present in the original Richards equation

Hydrostatic hypothesis

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 41: Hillslope hydrologyandrichards

38

In detail:initial condition

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 42: Hillslope hydrologyandrichards

39

In detail:transient situation

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 43: Hillslope hydrologyandrichards

40

In detail:final situation

R. Rigon

Richardsoniana

Monday, July 8, 13

Page 44: Hillslope hydrologyandrichards

41

In turn

“Short term

solution” Taylor’s

expansion

Water table

equation Taylor’s

expansion

Slope normal flow

time scale Lateral flow

time scaleSee also. D’Odorico et al., 2003

Richardsoniana

R. RigonMonday, July 8, 13

Page 45: Hillslope hydrologyandrichards

42

Pay attention to this

Slope normal flow

time scale Lateral flow

time scale

Richardsoniana

R. Rigon

Hydraulic diffusivityD(�) :=K(�)C(�)

Monday, July 8, 13

Page 46: Hillslope hydrologyandrichards

43

Detailsthat can be found in Cordano and Rigon, 2008

in words

•Take the dimensionless Richards equation

•Substitute in it the solution structure (asymptotic plus fast part)

•Here you obtain two coupled equations

•Further expand the solution structure in Taylor series

•Consider the terms which have the same expansion exponent in

•Solve each equation

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 47: Hillslope hydrologyandrichards

44

Neglecting those detailsthat can be found in Cordano and Rigon, 2008

Zeroth perturbation order

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 48: Hillslope hydrologyandrichards

45

Neglecting those detailsthat can be found in Cordano and Rigon, 2008

Zeroth perturbation order

R. Rigon

1D-Richards equation A source term

(exchange with water table)

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 49: Hillslope hydrologyandrichards

46

Neglecting those detailsthat can be found in Cordano and Rigon, 2008

Water Table equation

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 50: Hillslope hydrologyandrichards

47

Neglecting those detailsthat can be found in Cordano and Rigon, 2008

Zeroth perturbation order

First perturbation order

+ analogous for d*

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 51: Hillslope hydrologyandrichards

48

Integrating zeroth order solution in the column

Making a long story short

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 52: Hillslope hydrologyandrichards

49

Integrating zeroth order solution in the column

R. Rigon

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 53: Hillslope hydrologyandrichards

50

Integrating zeroth order solution in the column

Making a long story short

Topkapi* model Liu and Todini, 2002

R. Rigon

*With some interpretation

Richardsoniana - Iversoniana

Monday, July 8, 13

Page 54: Hillslope hydrologyandrichards

51

Integrating first order solution slope-parallel

Making a long story short - II

Boussinesq equation(e.g. Cordano and Rigon, 2013)

R. Rigon

: dimensionless transmissivities

: drainable porosity

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 55: Hillslope hydrologyandrichards

52

Making a long story short - III

R. Rigon

Figure represents a map of a small catchment, river network and a hillslope (hollow type, in gray). The distance of any point (P in the figure) in the hillslope to the channel head (C in the figure) is evaluated along the path drawn following the steepest descent (the dashed line). The characteristic length of the hillslope L (the length of x axis in Figure) is the mean of hillslope to channel distance for any point in the hillslope. The x axis used in the paper is downward parallel to mean topographic gradient, the axis y is normal to x (parallel to contour lines in a planar hillslope) and the z axis orthogonal to the x and y axes downward.

Integrating again over the lateral dimensionfrom Boussinesq

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 56: Hillslope hydrologyandrichards

53

Integrating Boussinesq

Making a long story short - III

HsBTroch et al. 2003

R. Rigon

: is the so called width function

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 57: Hillslope hydrologyandrichards

54

Hillslopes Width function

R. Rigon

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 58: Hillslope hydrologyandrichards

55

Simplifying HsB assuming stationarity of fluxesand neglecting diffusive terms

Making a long story short - IV and V

TopogO’Loughlin, 1986

R. Rigon

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 59: Hillslope hydrologyandrichards

56

Simplifying HsB assuming stationarity of fluxesand neglecting diffusive terms

Making a long story short - IV and V

assuming an exponential decay of vertical hydraulic conductivity

TopmodelBeven and Kirkby, 1979

R. Rigon

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 60: Hillslope hydrologyandrichards

57

Take home message

We can use Richards equation at various degree of simplification:

•1D (if we think that just slope-normal infiltration counts

•1D + 2D Boussinesq (Beq) if we want to account for lateral flow

* On this I will come back later

R. Rigon

Richardsoniana - Iversoniana - and beyond

Monday, July 8, 13

Page 61: Hillslope hydrologyandrichards

58

Take home message

We can use various simplification of either 1D and 2D Beq together:

• 1D Complete + 2D asymptotic- stationary• 1D linearized + 2D asymptotic- stationary• 1D bulk* + 2D asymptotic- stationary

• 1D Complete + 2D full Beq• 1D linearized + 2D full Beq• 1D bulk* + 2D full Beq

* On this I will come back later

R. Rigon

Richardsoniana - Iversoniana - and beyond beyond

Monday, July 8, 13

Page 62: Hillslope hydrologyandrichards

59

Take home message

We can also try a kinematic approximation of the Boussinesq equation, and

therefore:

• 1D Complete + 2D Kinematic• 1D linearized + 2D Kinematic• 1D bulk* + 2D Kinematic

* On this I will come back later

R. Rigon

Richardsoniana - Iversoniana - and beyond beyond

Monday, July 8, 13

Page 63: Hillslope hydrologyandrichards

1D linear + 2D asymptotic a.k.a D’Odorico et al., 2005

Mir

ò. T

he-

nig

hti

ngal

e-s-

son

g-a

t-m

idn

igh

t-an

d-t

he-

morn

ing-r

ain

Monday, July 8, 13

Page 64: Hillslope hydrologyandrichards

C(⇥)⇤⇥

⇤t=

⇤z

⇤Kz

�⇤⇥

⇤z� cos �

⇥⌅+ Sr

In literature related to the determination of slope stability this equation

assumes a very important role because fieldwork, as well as theory, teaches

that the most intense variations in pressure are caused by vertical infiltrations.

This subject has been studied by, among others, Iverson, 2000, and D’Odorico

et al., 2003, who linearised the equations.

61

The Richards Equation!

R. Rigon

Linearize it !

Monday, July 8, 13

Page 65: Hillslope hydrologyandrichards

The analytical solution methods for the advection-dispersion equation

(even non-linear), that results from the Richards equation, can be found

in literature relating to heat diffusion (the linearised equation is the

same), for example Carslaw and Jager, 1959, pg 357.

Usually, the solution strategies are 4 and they are based on:

- variable separation methods

- use of the Fourier transform

- use of the Laplace transform

- geometric methods based on the symmetry of the equation (e.g.

Kevorkian, 1993)

All methods aim to reduce the partial differential equation to a system

of ordinary differential equations

62

Th

e R

ich

ard

s Eq

uat

ion

1

-D

R. Rigon

Linearize it !

Monday, July 8, 13

Page 66: Hillslope hydrologyandrichards

⇥ ⇥ (z � d cos �)(q/Kz) + ⇥s

Iver

son

, 20

00

; D’O

dori

co e

t al

., 2

00

3,

Cord

ano a

nd

Rig

on

, 20

08

63

s

The Richards equation on a plane hillslope

R. Rigon

Linearize it !

Monday, July 8, 13

Page 67: Hillslope hydrologyandrichards

Assuming K ~ constant and neglecting the source terms

⇤⇥

⇤t= D0 cos2 �

⇤2⇥

⇤t2

64

The Richards Equation 1-D

C( )@

@t= Kz 0

@2

@z2

D0 :=Kz 0

C( )

D’O

dori

co e

t al

., 2

00

3

R. Rigon

Linearize it !

Monday, July 8, 13

Page 68: Hillslope hydrologyandrichards

The equation becomes LINEAR and, having found a solution

with an instantaneous unit impulse at the boundary, the

solution for a variable precipitation depends on the

convolution of this solution and the precipitation.

65

The Richards Equation 1-D

D’O

dori

co e

t al

., 2

00

3

R. Rigon

Linearize it !

Monday, July 8, 13

Page 69: Hillslope hydrologyandrichards

66

The Richards Equation 1-D

D’O

dori

co e

t al

., 2

00

3

R. Rigon

Linearize it !

Monday, July 8, 13

Page 70: Hillslope hydrologyandrichards

For a precipitation impulse of constant intensity, the solution can be

written:

⇥0 = (z � d) cos2 �

D’O

dori

co e

t al

., 2

00

3

67

= 0 + s

s =

8<

:

qKz

[R(t/TD)] 0 t T

qKz

[R(t/TD)�R(t/TD � T/TD)] t > T

The Richards Equation 1-D

R. Rigon

Linearize it !

Monday, July 8, 13

Page 71: Hillslope hydrologyandrichards

In this case the equation admits an analytical solution

D’O

dori

co e

t al

., 2

00

3

68

R(t/TD) :=⇤

t/(� TD)e�TD/t � erfc�⇤

TD/t⇥

s =

8<

:

qKz

[R(t/TD)] 0 t T

qKz

[R(t/TD)�R(t/TD � T/TD)] t > T

TD :=z2

D0

The Richards Equation 1-D

R. Rigon

Linearize it !

Monday, July 8, 13

Page 72: Hillslope hydrologyandrichards

D’O

dori

co e

t al

., 2

00

3

69

TD

TD

TD

TD

Th

e R

ich

ard

s Eq

uat

ion

1

-D

R. Rigon

Linearize it !

Monday, July 8, 13

Page 73: Hillslope hydrologyandrichards

70

Second message

Why using other simplifying assumptions (like Horton’s or Green-Ampt), if we have this ?

R. Rigon

Forget them!

Monday, July 8, 13

Page 74: Hillslope hydrologyandrichards

Just kidding!

Monday, July 8, 13

Page 75: Hillslope hydrologyandrichards

72

Did you care about hypotheses ?

Is it for any occasion realistic ? Look at the following sandy-loam:

Hypotheses counts

R. RigonMonday, July 8, 13

Page 76: Hillslope hydrologyandrichards

72

Did you care about hypotheses ?

Is it for any occasion realistic ? Look at the following sandy-loam:

Hypotheses counts

R. RigonMonday, July 8, 13

Page 77: Hillslope hydrologyandrichards

constant diffusivity

73

The Decomposition of the Richards equation

is possible under the assumption that:

Time scale of infiltration

soil depth

time scale of lateral flow

hillslope length

reference conductivity

reference hydraulic capacity

Iver

son

, 20

00

; C

ord

ano a

nd

Rig

on

, 2008

Hypotheses counts

R. RigonMonday, July 8, 13

Page 78: Hillslope hydrologyandrichards

Assuming hydrostatic conditions

74

Initial condition is then:

Consequently, at surface

Hypotheses counts

R. RigonMonday, July 8, 13

Page 79: Hillslope hydrologyandrichards

75

For the sandy-loam soilassuming the water table at one meter depth

we have a vertical variation of hydraulic conductivity of one order of magnitude !

Hypotheses counts

R. RigonMonday, July 8, 13

Page 80: Hillslope hydrologyandrichards

76

D which characterizes the time scales of flow is varying with depth

Hypotheses counts

R. Rigon

So a D0 reference cannot be significant

Monday, July 8, 13

Page 81: Hillslope hydrologyandrichards

77

Therefore

at surface

so, lateral flow at the water table level has the same time scale vertical flow at the surface (at least if we believe to Richards’ equation)

Hypotheses counts

R. RigonMonday, July 8, 13

Page 82: Hillslope hydrologyandrichards

78

X - 52 LANNI ET AL.: HYDROLOGICAL ASPECTS IN THE TRIGGERING OF SHALLOW LANDSLIDES

Figure 2: Experimental set-up. (a) The infinite hillslope schematization. (b) The initial suction head profile.

Figure 3: The soil-pixel hillslope numeration system (the case of parallel shape is shown here). Moving from 0 to 900 (the total number of

soil-pixels), corresponds to moving from the crest to the toe of the hillslope

Table 1: Physical, hydrological and geotechnical parameters used to characterize the silty-sand soil

Parameter group Parameter name Symbol Unit ValuePhysical Bulk density ⇥b (g/cm3) 2.0

% sand - - 60% silt - - 40

Hydrological Saturated hydraulic conductivity Ksat (m/s) 10�4

Saturated water content �sat (cm3/cm�3) 0.39Residual water content �r (cm3/cm�3) 0.155

water retention curve parameter n [�] 1.881water retention curve parameter � (cm�1) 0.0688

Geotechnical Effective angle of shearing resistance ⇤0 � 38Effective cohesion c0 kN/m2 0

D R A F T September 24, 2010, 9:13am D R A F T

The OpenBook hillslope in a 3D simulation

Comparing with 3D

R. Rigon

Lan

ni

and

Rig

on

, un

pu

bli

shed

Monday, July 8, 13

Page 83: Hillslope hydrologyandrichards

79

X - 54 LANNI ET AL.: HYDROLOGICAL ASPECTS IN THE TRIGGERING OF SHALLOW LANDSLIDES

(a) DRY-Low (b) DRY-Med

(c) DRY-High (d) WET-Low

(e) WET-Med (f) WET-High

Figure 5: Values of pressure head developed at the soil-bedrock interface at each point of the subcritical parallel hillslope. The slope of

the pressure head lines represents the mean lateral gradient of pressure

D R A F T September 24, 2010, 9:13am D R A F T

Simulations result

Comparing with 3D

R. Rigon

Lan

ni

and

Rig

on

, un

pu

bli

shed

Monday, July 8, 13

Page 84: Hillslope hydrologyandrichards

80

At the beginning the pressure is constant

along the whole transect (except for

phenomena at the divide’s edge

Comparing with 3D

R. RigonMonday, July 8, 13

Page 85: Hillslope hydrologyandrichards

81

After a certain amount of time (25h in this

simulation) pressures along the slope

differentiate. With a little of analysis we

c a n d i s t i n g u i s h t w o r e g i o n s o f

differentiation. One controlled by the

boundary conditions at the bottom.

The second generated by lateral water

flow accumulation.

Comparing with 3D

R. RigonMonday, July 8, 13

Page 86: Hillslope hydrologyandrichards

82

LANNI ET AL.: HYDROLOGICAL ASPECTS IN THE TRIGGERING OF SHALLOW LANDSLIDES X - 55

(a) (b)

Figure 6: Temporal evolution of the vertical profile of hydraulic conductivity (a) and hydraulic conductivity at the soil-bedrock interface

(b) of a soil-pixel located in the mid-slope zone. Results are shown for the case representing DRY antecedent soil moisture conditions, Low

rainfall intensity and parallel hillslope shape of the subcritical (gentle) case

D R A F T September 24, 2010, 9:13am D R A F T

Hidraulic conductivity is varying by three order of magnitude

at the bedrock interface.

The key to understand this phenomenology

Lan

ni

et a

l., 2012

Comparing with 3D

R. RigonMonday, July 8, 13

Page 87: Hillslope hydrologyandrichards

83

X - 56 LANNI ET AL.: HYDROLOGICAL ASPECTS IN THE TRIGGERING OF SHALLOW LANDSLIDES

(a) (b)

(c) (d)

Figure 7: Transient pore pressure profiles related to points No. 300, 450 and 600 for the DRY -Low (a), DRY -High (b) and WET �High

(c) cases, and soil hydraulic conductivity function inferred using the Mualem model (d). The vertical lines at 0 pressure head indicate the

position of the water table at different timing

D R A F T September 24, 2010, 9:13am D R A F T

X - 56 LANNI ET AL.: HYDROLOGICAL ASPECTS IN THE TRIGGERING OF SHALLOW LANDSLIDES

(a) (b)

(c) (d)

Figure 7: Transient pore pressure profiles related to points No. 300, 450 and 600 for the DRY -Low (a), DRY -High (b) and WET �High

(c) cases, and soil hydraulic conductivity function inferred using the Mualem model (d). The vertical lines at 0 pressure head indicate the

position of the water table at different timing

D R A F T September 24, 2010, 9:13am D R A F T

Another view

R. Rigon

Comparing with 3D

Monday, July 8, 13

Page 88: Hillslope hydrologyandrichards

84

When simulating is understanding

cou

rtes

y of

E. C

ord

ano

T’L can be very small indeed .....

Interpretations

R. RigonMonday, July 8, 13

Page 89: Hillslope hydrologyandrichards

85

Understanding from simulations

At the beginning of the infiltration process the situation in surface is

marked by the blue line, the situation at the bedrock is marked by the

red line

cou

rtes

y of

E. C

ord

ano

R. Rigon

Interpretations

Monday, July 8, 13

Page 90: Hillslope hydrologyandrichards

86

When lateral flow start we are in the following situation

cou

rtes

y of

E. C

ord

ano

Understanding from simulations

R. Rigon

Interpretations

Monday, July 8, 13

Page 91: Hillslope hydrologyandrichards

87

At the beginning

The condition of the perturbative derivation are verified

cou

rtes

y of

E. C

ord

ano

R. Rigon

Interpretations

Monday, July 8, 13

Page 92: Hillslope hydrologyandrichards

88

At the end

cou

rtes

y of

E. C

ord

ano

Conditions for lateral flow are dominating. Actually the same

phenomenology deducted by the perturbation theory! But obtained for a

different reason.

R. Rigon

Interpretations

Monday, July 8, 13

Page 93: Hillslope hydrologyandrichards

89

Take home message:

Never fully believe on the magic of simplifications

Detailed physics in models can help

R. Rigon

Magic ad Mermeids do not exist (Sponge Bob)

Monday, July 8, 13

Page 94: Hillslope hydrologyandrichards

90

Lateral Flow

•Can be fast, ... very fast, much faster than what happens in vadose

conditions

•In fact, to have the effects just described, we have to believe to the form

that Soil Water retention Curves have.

•Other soils behave differently

•If macropores or cracks are present, vertical infiltration can still remain

faster

R. Rigon

Interpretations

Monday, July 8, 13

Page 95: Hillslope hydrologyandrichards

91

Inappropriate numerics (or gridding)

Can hide it!

R. Rigon

Interpretations

Monday, July 8, 13

Page 96: Hillslope hydrologyandrichards

Further investigationsM

ach

ael

Leon

g -

Cu

ttin

g t

he

tim

e w

ith

a k

nif

e, 2

01

2

Monday, July 8, 13

Page 97: Hillslope hydrologyandrichards

93

CAPITOLO 5. IL BACINO DI PANOLA

Figura 5.2: Rappresentazione della profondita del suolo del pendio di Panola.

costante su un campione prelevato a 10 cm di profondita, risulta pari a 64 [cm/h]; per cio che concerneil valore della conducibilita idraulica a saturazione del bedrock, non esistono misure dirette e↵ettuatesu campioni prelevati in sito; tuttavia si stima che il suo valore sia 2-3 ordini di grandezza inferiorerispetto a quella del terreno soprastante. Entrambi i valori di conducibilita idraulica satura (del bedrocke del terreno) saranno comunque oggetto di calibrazione numerica all’atto delle simulazioni svolte conGEOtop, utilizzando come valori di partenza quelli qui citati.Va infine evidenziata la presenza di cinque macropori (si veda la fig. 5.2 per la loro posizione), didiametro compreso tra i 10 ed i 60 mm, che a�orano sulla trincea e che contribuiscono attivamentealla generazione del deflusso.

5.2 Analisi delle precipitazioni e del deflusso subsuperficiale

Il versante di Panola e tipicamente ben drenato e rimane insaturo per la maggior parte dell’anno.Durante il pediodo di studio (gennaio - fine maggio 2002), il contenuto d’acqua medio del terrenomostra un andamento fortemente stagionale, passando da valori relativi prossimi all’80% del periodoinvernale, a valori di circa il 40% ad inizio estate, calando poi ulteriormente nei mesi piu caldi: in fig.(5.3) si riporta l’andamento appena descritto.Il deflusso sub-superficiale misurato presso la trincea e intermittente e si verifica solo in risposta aprecipitazioni intense; uno studio approfondito della relazione tra entita delle precipitazioni e deflussosub-superficiale e stato condotto da Tromp-van Meerveld e McDonnell, (2006a) [24], e nel seguito se neriportano i risultati, necessari a comprendere la dinamica idrologica del pendio.Nello studio condotto da Tromp-van Meerveld e McDonnell sono stati esaminati 2 anni di dati dideflusso sub-superficiale, dal 19 febbraio 1996 al 10 maggio 1998, periodo nel quale si sono verificati147 temporali. L’analisi correlata delle precipitazioni e del deflusso indica che il 22% degli eventi

79

Panola’s hillslope

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 98: Hillslope hydrologyandrichards

94

Terrain surface Bedrock surface Soil depth varies

Dep

ress

ion

Soil (sandy loam) Bedrock

Ksat = 10-4 m/s Ksat = 10-7 m/s

Panola’s hillslope

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 99: Hillslope hydrologyandrichards

95

Q (m

3 /h)

t=9h

t=18h

t=22h

With a rainfall of 6.5 mm/h and a duration of 9 hoursLa

nn

i et

al., 2

01

1

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 100: Hillslope hydrologyandrichards

96

t=6h t=9ht=7h t=14h

Lan

ni

et a

l., 2

01

1

With a rainfall of 6.5 mm/h and a duration of 9 hours

Tromp Van Meerveld et al., 2006 call it filling and spilling

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 101: Hillslope hydrologyandrichards

97

Q (m

3 /h)

t=9h

t=18h

t=22h

Lan

ni

et a

l., 2

01

1

With a rainfall of 6.5 mm/h and a duration of 9 hours

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 102: Hillslope hydrologyandrichards

98

1D

3D

No role played by hillslope gradient

First Slope Normal infiltration works

Then Lateral flow start

Infiltration front propagate

Drainage is controlled by the bedrock form

As in the open book caseLa

nn

i et

al., 2

01

1

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 103: Hillslope hydrologyandrichards

99

Now we want a model that can run 100 times faster

In which we obviously use all the machinery of the Richards’ equation, i.e. hydraulic conductivity and soil

water retention curves

R. Rigon

Richards equation is still valid here ?

Monday, July 8, 13

Page 104: Hillslope hydrologyandrichards

100

Introducing the concept of concentration timein subsurface flow

we have the distances from the channels

R. Rigon

Variations

Monday, July 8, 13

Page 105: Hillslope hydrologyandrichards

101

If we assume that water just move laterally in saturated conditions, we can use Darcy law for getting the

velocities

possibly in its more traditional form:

R. Rigon

Variations

Monday, July 8, 13

Page 106: Hillslope hydrologyandrichards

102

If we assume that water just move laterally in saturated conditions, we can use Darcy law for getting the

velocities

And assuming Dupuit approximation, i.e. hydrostatic distribution of pressures

R. Rigon

Variations

Monday, July 8, 13

Page 107: Hillslope hydrologyandrichards

103

Then:

Time = Lengths/velocity

And, for any point:

is the max residence time*

R. Rigon

Variations

*The operator means that we are looking for the maximum of T choosing it from all the possible path that we can define upstream of the point i

Monday, July 8, 13

Page 108: Hillslope hydrologyandrichards

104

The largest time

is the concentration time

Up to concentration time

The area contributing to the discharge is not the TOTAL upslope area

R. Rigon

Variations

Monday, July 8, 13

Page 109: Hillslope hydrologyandrichards

105

The area contributing to the discharge is not the TOTAL upslope area

Lan

ni

et a

l., 2

01

2a

R. Rigon

!(Steady state)

Monday, July 8, 13

Page 110: Hillslope hydrologyandrichards

106

Actually there is a second issue

Water table cannot “exist” everywhere

C. Lanni et al.: Modelling shallow landslide susceptibility 3961

Fig. 1. A flow chart depicting the coupled saturated/unsaturated hydrological model developed in this study.

35

1

2

3 4

Figure 2. The concept of hydrological connectivity. Lateral subsurface flow occurs at point 5

(x,y) when this becomes hydrologically connected with its own upslope contributing area 6

A(x,y). 7

8

Fig. 2. The concept of hydrological connectivity. Lateral subsurfaceflow occurs at point (x,y) when this becomes hydrologically con-nected with its own upslope contributing area A(x,y).

storage of soil moisture needed to produce a perched watertable (i.e. zero-pressure head) at the soil–bedrock interface(Fig. 3); and I [LT�1] is the rainfall intensity assumed to beuniform in space and time. Computation of V0 and Vwt re-quire the use of a relationship between soil moisture content✓ [�] and suction head [L], and a relationship between and the vertical coordinate (positive upward) z [L] (Fig. 3).By using the assumption that the suction head profile (z)

changes from one steady-state situation to another over time,

36

1

2

3 4

Figure 3. i(z) and i(z) are, respectively, the initial water content and the initial suction 5

head vertical profiles. wt(z) and wt(z) represents the linear water content and suction 6

head vertical profiles associated with zero-suction head at the soil-bedrock interface. 7

8

Fig. 3. ✓i(z) and i(z) are, respectively, the initial water contentand the initial suction head vertical profiles. ✓wt(z) and wt(z) rep-resents the linear water content and suction head vertical profilesassociated with zero-suction head at the soil–bedrock interface.

the relation between [L] and z [L] is that of hydraulicequilibrium:

= (z = 0) + z = b + z, (2)

where b = (z = 0) is the suction head at the soil–bedrockinterface. Bierkens (1998) argued that this assumption isvalid for a shallow system where redistribution of soil-water

www.hydrol-earth-syst-sci.net/16/3959/2012/ Hydrol. Earth Syst. Sci., 16, 3959–3971, 2012

Lan

ni

et a

l., 2

01

2b

R. Rigon

!(Steady state)

Monday, July 8, 13

Page 111: Hillslope hydrologyandrichards

107

Ii.e. time to water table development

Twt(x,y):= [Vwt(x,y)-V0(x,y)]/I

Initial conditions(hydrostatic slope normal)

boundary conditions(including rainfall, I)

t> Twt(x,y)

YES

NO

Lan

ni

et a

l., 2

01

2

Slope Normalunsaturated flow

A heuristic model

for each timestep

Faster is better

R. RigonMonday, July 8, 13

Page 112: Hillslope hydrologyandrichards

108

YES

t> Tmaxwt(x,y)

hydrologicallyconnectedA(x,y) >0

YES

NO hydrologicallydisconnected

A(x,y) =0

A heuristic modelLa

nn

i et

al., 2

01

2

R. Rigon

Faster is better

Monday, July 8, 13

Page 113: Hillslope hydrologyandrichards

109

YES

update soil pressure

start lateral flow update soil pressure

next timestep

A heuristic modelLa

nn

i et

al., 2

01

2

R. Rigon

Faster is better

Monday, July 8, 13

Page 114: Hillslope hydrologyandrichards

110

* Is not completely true.

I question also of personal attitude:

I understand (fluid) mechanics through

equations and I try to interpret observations

through equations.

Someone else (i.e. many of my students)

simply did not have the training for that and

prefer to rebuilt the physics of the problem by

small pieces.

This has a certain appealing to many (especially

to natural scientists and geologists), and can

indeed be useful to see thing from different

perspectives.

Dood

ley,

Mu

ttle

y, a

nd

th

eir

flyi

ng m

ach

ines

R. Rigon

Attitudes

Monday, July 8, 13

Page 115: Hillslope hydrologyandrichards

111

3968 C. Lanni et al.: Modelling shallow landslide susceptibility

40

1

2

3

Figure 7. Patterns of Return period TR (years) of the critical rainfalls for shallow landslide 4

triggering  (i.e.,  FS≤1)  and  associated  levels  of  landslide  susceptibility  obtained  by  means  5

of QDSLaM. 6

7

Fig. 7. Patterns of return period TR (years) of the critical rainfalls for shallow landslide triggering (i.e. FS 1) and associated levels oflandslide susceptibility obtained by means of QDSLaM.

Table 3. Percentages of catchment area (C) and observed landslide area (L) in each range of critical rainfall frequency (i.e. return period TR)for QDSLaM.

Susceptibility Pizzano Fraviano Cortina

TR level C

aL

bC

aL

bC

aL

b

Years Category % % % % % %

Uncond Unstable 9.9 60.2 7.7 77.7 8.5 56.80–10 Very high 20.3 26.9 16.1 18.5 13.5 39.210–30 High 7.8 0.0 5.6 1.5 5.8 4.030–100 Medium 6.0 9.7 5.9 2.3 6.7 0.0> 100 Low 42.9 3.2 53.5 0.0 54.7 0.0Uncond. stable Very low 13.1 0.0 11.2 0.0 10.8 0.0

aC = catchment area; b L= landslide area

high slope failure hazard) of CI-SLAMwith that of the quasi-dynamic model QDSLaM (Borga et al., 2002b).QDSLaM is based on coupling a hydrological model to

a limit-equilibrium slope stability model to calculate thecritical rainfall necessary to trigger slope instability at anypoint in the landscape. The hydrological model assumesthat flow infiltrates to a lower conductivity layer and fol-lows topographically-determined flow paths to map the spa-tial pattern of soil saturation based on analysis of a “quasi-dynamic” wetness index. With respect to the model pro-posed in this paper, QDSLaM does not consider the follow-ing aspects: (i) vertical rainwater infiltration into unsaturatedsoil; (ii) analysis of the connectivity to compute the quasi-dynamic wetness index; and (iii) soil depth variability. All re-maining aspects of the modelling framework are described ina consistent way by the two models. Both models have beenapplied to the three catchments by using the same parameters

set. A map of shallow landsliding susceptibility obtained byusing QDSLaM is reported in Fig. 7, whereas Table 3 reportsthe corresponding percentages of slope-stability categories interms of catchment area and observed landslide area in eachrange of critical rainfall frequency (i.e. return period TR). Ex-amination of Fig. 7 and of Table 3 shows the considerableimpact of the areas considered unconditionally unstable inQDSLaM. The percentage of topographic elements consid-ered unconditionally unstable ranges from 7.7% (Fraviano)to 9.9% (Pizzano). This overrepresentation is reported bothin the lower and in the upper portions of the catchments,where high local slope values are present. CI-SLAM doesnot predict unconditionally unstable locations. In fact, thecontribution of negative pressure head (Eq. 15a) ensures thestability of steeper topographic elements (i.e. locations withtan � � tan '

0 for cohesionless soils) that would be otherwiseclassified as unconditionally unstable by QDSLaM (as well

Hydrol. Earth Syst. Sci., 16, 3959–3971, 2012 www.hydrol-earth-syst-sci.net/16/3959/2012/

Lan

ni

et a

l., 2

01

2

However, it works

R. Rigon

Faster is better if it works (Klemes fogive me!)

Monday, July 8, 13

Page 116: Hillslope hydrologyandrichards

Further investigations -IIM

ach

ael

Leon

g -

Cu

ttin

g t

he

tim

e w

ith

a k

nif

e, 2

01

2

Monday, July 8, 13

Page 117: Hillslope hydrologyandrichards

113

CAPITOLO 5. IL BACINO DI PANOLA

Figura 5.4: Immagine tratta da Tromp-van Meerveld e McDonnell, (2006a) [24]; (a) deflusso sub-superficiale totale per i segmenti in cui e stata suddivisa la trincea e (b) numero di eventi meteorici cheproducono deflussi misurabili.

5.2.1 Il ruolo dei macropori

Secondo l’interpretazione di Tromp-van Meerveld e McDonnell, il flusso attraverso i cinque macroporicontribuisce in maniera significativa alla generazione del deflusso totale: rispetto al flusso totalemisurato durante i 147 eventi meteorici, il 42% deriva proprio da essi; in particolare il macroporodenominato M14 (si veda la fig. 5.2 per la sua localizzazione) e responsabile del 25% del deflussototale durante il periodo di analisi. Si e dimostrato come sussista una relazione lineare molto robusta(r2=0.96) tra il deflusso sub-superficiale totale ed il deflusso totale a carico dei macropori (fig. 5.5),il che suggerisce che vi sia un meccanismo comune che innesca entrambi i processi. Va evidenziatocomunque come vi sia una forte stagionalita nella caratterizzazione del deflusso ad opera dei macropori,che sarebbero responsabili del 50% del deflusso totale durante l’autunno e del 41% durante l’inverno,mentre in primavera ed estate contribuiscono solo per lo 0-2 %.

5.2.2 Soglia di risposta del pendio

Le analisi condotte da Tromp-van Meerveld e McDonnell suggeriscono come vi sia una soglia abbastanzanetta per innescare significativi (> 1mm) deflussi sub-superficiali nel versante di Panola: deflussisignificativi avvengono solo a seguito di eventi meteorici il cui apporto sia maggiore di 55 mm di

81

Tro

mp

Van

Mee

rvel

d e

t al

., 2

00

6

And finally macropores

R. Rigon

Macropores

Monday, July 8, 13

Page 118: Hillslope hydrologyandrichards

114

Macropore Flow InitiationWater supply to the macropores

InteractionWater transfer between macropores and the surrounding soil matrix

M. W

eile

r, f

rom

Moch

a p

roje

ct

Macropores!

R. Rigon

Macropores

Monday, July 8, 13

Page 119: Hillslope hydrologyandrichards

115

CAPITOLO 5. IL BACINO DI PANOLA

0.00

0.02

0.04

0.06

0.08

0.10

Flussi alla base del pendio - Simulazione 0

date (dd/mm) 2002po

rtate

[l/s

]

01/01 11/01 21/01 31/01 10/02 20/02 02/03 12/03 22/03 01/04 11/04 21/04 01/05 11/05 21/05

Flussi misuratiSimulazione 0

Figura 5.16: Confronto tra flussi misurati e computati attraverso la Simulazione 0 presso la trinceaalla base del pendio.

0.00

0.02

0.04

0.06

0.08

0.10

Simulazione 0 - evento 6 febbraio

date (dd/mm) 2002

porta

te [l

/s]

05/02 06/02 07/02 08/02 09/02 10/02 11/02 12/02

Flussi misuratiSimulazione 0

0.00

0.02

0.04

0.06

0.08

0.10

Simulazione 0 - evento 30 marzo

date (dd/mm) 2002

porta

te [l

/s]

29/03 30/03 31/03 01/04 02/04 03/04 04/04 05/04 06/04 07/04

Flussi misuratiSimulazione 0

Figura 5.17: Confronto tra flussi misurati e computati attraverso la Simulazione 0 presso la trinceaalla base del pendio: a sinistra si riporta l’evento del 6 febbraio 2002, a destra quello del 31 marzo.

puo essere causata da diversi fattori, quali un’errata assegnazione delle caratteristiche del suolo o delbedrock, oppure un errore nello stabilire la condizione iniziale circa la quota della falda.Un aspetto decisamente importante da considerare, tanto in questi risultati quanto in quelli presentatisuccessivamente, e che nella creazione della geometria di calcolo 3D utilizzata da GEOtop non estato possibile inserire la presenza dei macropori che si a↵acciano sulla trincea alla base del pendio.Tale mancanza puo avere un peso notevole all’atto di calcolare i volumi defluiti, in quanto, comeevidenziato alla sezione (5.2.1), nel periodo invernale essi contribuiscono a generare circa il 40% del

95

Da

Prà

, 2

01

3

Certainly the volumes of water cannot be simulated with the only Richards equation

No way!

R. Rigon

Macropores

Monday, July 8, 13

Page 120: Hillslope hydrologyandrichards

Thank you for your attention

116

G. U

lric

i -

R. Rigon

Slides on http://abouthydrology.blogspot.com

Monday, July 8, 13