Topographic Effects on

Seismic Ground Motion

GEO Report No. 303

T.K.C. Wong

Geotechnical Engineering Office

Civil Engineering and Development Department
The Government of the Hong Kong
Special Administrative Region
 Table of Contents
Topographic Effects on
Seismic Ground Motion

Table of Contents
GEO Report No. 303

T.K.C. Wong

Table of Contents
Table of Contents

This report was originally produced in September 2011

as GEO Technical Note No. TN 4/2011
 2

© The Government of the Hong Kong Special Administrative Region

Table of Contents
First published, November 2014

Prepared by:

Geotechnical Engineering Office,

Civil Engineering and Development Department,
Civil Engineering and Development Building,
101 Princess Margaret Road,
Homantin, Kowloon,
Hong Kong.

Table of Contents
Table of Contents
Table of Contents
 3

Table of Contents
Preface

In keeping with our policy of releasing information

which may be of general interest to the geotechnical

Table of Contents
profession and the public, we make available selected internal
reports in a series of publications termed the GEO Report
series. The GEO Reports can be downloaded from the
website of the Civil Engineering and Development Department
(http://www.cedd.gov.hk) on the Internet. Printed copies are
also available for some GEO Reports. For printed copies, a
charge is made to cover the cost of printing.

The Geotechnical Engineering Office also produces

documents specifically for publication in print. These include
guidance documents and results of comprehensive reviews.
They can also be downloaded from the above website.

These publications and the printed GEO Reports may be

Table of Contents
obtained from the Government’s Information Services
Department. Information on how to purchase these documents
is given on the second last page of this report.

H.N. Wong
Head, Geotechnical Engineering Office
November 2014
Table of Contents
 4

Table of Contents
Foreword

This Technical Note reviews the key factors that

Table of Contents
influence topographic effects on seismic ground motions. It
presents the results of a parametric study conducted to
investigate the influence of different parameters on the
topographic effects of a soil slope overlying bedrock and the
interaction between topographic effects and soil layer effects.

This Technical Note was prepared by Mr Thomas K.C.

Wong. The work was based largely on Mr Wong’s dissertation
completed during his MSc course at Imperial College London in
the academic year of 2009-2010, together with further work
done during his subsequent debriefing in the Standards &
Testing Division. Mr Lawrence K.W. Shum assisted to review
a draft of this report. Ove Arup & Partners Hong Kong
Limited also gave useful comments on the draft report. All

Table of Contents
contributions are gratefully acknowledged.

Ken K.S. Ho
Chief Geotechnical Engineer/Standards & Testing
Table of Contents
 5

Table of Contents
Abstract

It is well known that surface topography can substantially

affect the amplitude of seismic ground motions during
earthquakes. Observations from destructive earthquakes,
instrumental evidence and numerical studies have shown that

Table of Contents
seismic ground motion can be amplified over convex
topographies such as hills, ridges and crest of slopes. Various
numerical studies have been conducted in the past to analyse
and predict the effects of topographic irregularities on seismic
ground motions, the majority of which consider topographic
irregularity as an isolated feature in a homogeneous half-space.
Despite the qualitative agreement between theory and
observations on topographic effects, these past studies usually
obtained amplification factors that are smaller than those from
instrumental studies.

This Technical Note presents the results of a parametric

study conducted to investigate the effects of different parameters

Table of Contents
on the topographic effects on seismic ground motions for a soil
slope overlying bedrock. This study revealed that the presence
of a soil/rock interface would result in another type of site
amplification effect, namely soil layer effect. The findings of
this study also indicated that as compared to the simple model in
which slopes in a homogeneous half-space are considered, the
presence of a soil/rock interface would result in larger
topographic amplification factors and complicate the
topographic effects. Given the complex interaction between
topographic effects and soil layer effects, these effects should
not be handled separately for the prediction of the overall site
amplification factors. The study findings have implications to
seismic site response analysis.
Table of Contents
 6

Contents

Page
No.

Title Page 1

Preface 3

Foreword 4

Abstract 5

Contents 6

List of Tables 9

List of Figures 10

1 Introduction 13

2 Literature Review 13
2.1 Seismic Waves 13
2.2 Difference between Soil Layer Effects and Topographic 14
Effects
2.2.1 General 14
2.2.2 Soil Layer Effects 17
2.2.3 Topographic Effects 21
2.3 Physical Phenomena Leading to Occurrence of Topographic 21
Effects
2.4 Evidence of Topographic Effects 26
2.4.1 Documented Observations of Earthquake Damage 26
2.4.2 Instrumental Evidence 26
2.5 Previous Numerical Studies on Topographic Effects 38
2.5.1 General 38
2.5.2 Numerical Studies on Ridges and Hills 38
2.5.3 Numerical Studies on Slopes 41
2.6 Parameters Affecting Topographic Effects 45
2.6.1 General 45
2.6.2 Type of Incident Waves 46
 7

Page

Table of Contents
No.

2.6.3 Angle of Incidence 46

2.6.4 Frequency Content of Incident Motion 48
2.6.5 Duration and Type of Time-history Excitations 48
2.6.6 Slope Inclination 50
2.6.7 Three-dimensional Topography Shape 50
2.6.8 Soil Layering 50
2.6.9 Soil Damping 51
2.6.10 Discussion 51

Table of Contents
2.7 Provisions of Topographic Effects in Seismic Codes 54

3 Methodology and Parameters Adopted for This Study 55

3.1 Methodology 55
3.2 Geometry of the Model 55
3.3 Parameters 58
3.4 Time History Excitations 58
3.5 Finite Element Methodology 59
3.5.1 General 59
3.5.2 Domain Reduction Method 63

Table of Contents
3.6 Procedures 63

4 Parametric Study 65
4.1 Topographic Amplification Factors 65
4.2 Effects of Varying Dimensionless Frequency (H/) 69
4.2.1 General 69
4.2.2 Bedrock Depth Z = 250 m 76
4.2.3 Bedrock Depth Z = 125 m 77
4.2.4 Bedrock Depth Z = 500 m 78
4.3 Effects of Varying Slope Inclination (i) 78
Table of Contents

4.4 Effects of Varying Bedrock Depth (Z) 85

4.5 Interaction between Soil Layer Effects and Topographic 86
Effects
 8

Page

Table of Contents
No.

5 Discussion 87
5.1 Comparison with Seismic Code Provisions 87

6 Conclusions 88

7 References 91

Table of Contents
Table of Contents
Table of Contents
 9

List of Tables

Table of Contents
Table Page
No. No.

2.1 Evidence of Topographic Effects: Documented 28

Observations of Earthquake Damage

2.2 Instrumental Evidence of Topographic Effects 31

3.1 Predominant Periods of Input Motion Selected for 60

Analysis and the Corresponding Dimensionless Frequency
Values

3.2 Summary of Analyses Performed in the Present Study and 62

Table of Contents
Selected Analyses from Tripe (2009) for Review in the
Present Study

4.1 Summary of the Results of the Numerical Analysis 70

Performed in This Study

Table of Contents
Table of Contents
 10

List of Figures

Table of Contents
Figure Page
No. No.

2.1 Body Waves: P Wave and S Wave 15

2.2 Surface Waves: Rayleigh Wave and Love Wave 16

2.3 Influence of Frequency on Steady-state Response of 19

Undamped Linear Elastic Layer

2.4 Influence of Frequency on Steady-state Response of 20

Damped, Linear Elastic Layer

Table of Contents
2.5 Sensitivity of Surface Motion to Incidence Angle for 23
Obliquely Incident SV Waves

2.6 Top Amplifications Versus Wedge Internal Angle for 24

Incidence of Plane SH and SV Waves

2.7 Wave Propagation in Cliff Induced by an SV Wave 25

2.8 Schematic Illustration of Incoming SV Waves and 27

Induced Reflected P Wave, Reflected SV Wave and
Rayleigh Wave

2.9 Amplification at the Crest of an Inclined Slope in a 43

Table of Contents
Homogeneous Half-space ( = 1%)

2.10 Amplification at the Crest of an Inclined Slope in a 44

Homogeneous Half-space for a Vertical Harmonic SV
Wave ( = 5%)

2.11 Amplification at the Crest of a Vertical Slope in a 47

Homogeneous Half-space for Inclined Wave

2.12 Effect of the Number of Significant Cycles N on the 49

Horizontal and Vertical Amplification Factors (Ah and Av),
as a Function of Horizontal Distance x from the Slope
Crest (H/ = 2, i = 30,  = 5%)
Table of Contents

2.13 Horizontal Amplifications for Vertically Incident SH 52

Wave in a Slope in a Homogeneous Half-space for
Various Distances Behind the Crest, (a)  = 1% and
(b)  = 5%
 11

Figure Page

Table of Contents
No. No.

2.14 Effect of Soil Damping  on the Horizontal and Vertical 53

Amplification Factors (Ah and Av), as a Function of
Horizontal Distance x from the Slope Crest (H/ = 2,
i = 30, N = 4)

2.15 Provisions for Topographic Effects in the French Seismic 56

Code PS-92

3.1 Geometry of the Two-dimensional Finite Element Model 57

and Domain Reduction Boundaries

3.2 An Example of the Acceleration-time Histories of Chang’s 61

Table of Contents
Motion Used as Input Excitation in This Study

4.1 Topographic Amplifications (Ah and Av) as a Function of 67

Horizontal Distance from Slope Crest (Z = 250 m, i = 45º,
Tp = 0.333 s, H/ = 0.3)

4.2 Topographic Amplifications (Ah and Av) as a Function of 68

Horizontal Distance from Slope Crest (Z = 250 m, i = 45º,
Tp = 10 s, H/ = 0.01)

4.3 Plots of Soil Layer Amplification Factor with 75

Dimensionless Frequency, for Bedrock Depths of 125 m,
250 m and 500 m

Table of Contents
4.4 Horizontal Amplification at Slope Crest for a Vertically 79
Incident SV Wave on a Slope with a Bedrock Depth of
250 m

4.5 Vertical Amplification at Slope Crest for a Vertically 80

Incident SV Wave on a Slope with a Bedrock Depth of
250 m

4.6 Horizontal Amplification at Slope Crest for a Vertically 81

Incident SV Wave on a Slope with a Bedrock Depth of
125 m

4.7 Vertical Amplification at Slope Crest for a Vertically 82

Table of Contents

Incident SV Wave on a Slope with a Bedrock Depth of

125 m

4.8 Horizontal Amplification at Slope Crest for a Vertically 83

Incident SV Wave on a Slope with a Bedrock Depth of
500 m
 12

Figure Page

Table of Contents
No. No.

4.9 Vertical Amplification at Slope Crest for a Vertically 84

Incident SV Wave on a Slope with a Bedrock Depth of
500 m

Table of Contents
Table of Contents
Table of Contents
 13

1 Introduction

Table of Contents
It is well known that surface topography can substantially affect the amplitude of
seismic ground motion. Observations of earthquake damage, instrumental records and
numerical studies have shown that seismic ground motion is amplified over convex
topographies such as hills, ridges and crest of slopes. A considerable amount of research has
been done to analyse and predict the effects of topographic irregularities on seismic ground
motion.

However, the majority of these numerical studies focused on two-dimensional

simulations in which the topographic features are treated as isolated ridges or depressions on
the surface of a homogeneous half-space. Furthermore, seismic input is usually modelled as
single-frequency, vertically propagating incident waves that cannot describe the broad-band
nature of real earthquake motion. Despite the qualitative agreement between theory and
observations on topographic effects, these studies usually obtain amplification factors smaller

Table of Contents
than those from instrumental studies. Some more complicated models which considered
presence of soil layering were able to obtain higher amplification factors, and therefore, some
researchers suggested that soil layering is a possible reason for the discrepancy between
observed and predicted topographic amplification factors. The presence of a soil/rock
interface results in another type of site amplification effects, namely soil layer effects.
Researchers have different views on whether the topographic effects and soil layer effects can
be uncoupled and considered separately for prediction of the overall site amplification factors.

This Technical Note presents the methodology and findings of a parametric study
conducted to study the influence of different parameters on topographic effects on seismic
ground motion for a soil slope overlying rigid bedrock and the interaction between
topographic effects and soil layer effects. The objectives of the study were:

Table of Contents
(a) to investigate the topographic effects on seismic ground
motion for a soil slope overlying rigid bedrock subjected to
seismic input motions of different predominant periods, for
different bedrock depths and slope inclinations;

(b) to study how the topographic effects for a soil slope

overlying rigid bedrock are affected by the presence of the
soil/rock interface underneath the slope toe; and

(c) to investigate whether the topographic effects and soil layer

effects can be uncoupled and handled separately to predict
the overall site amplification factors.
Table of Contents

2 Literature Review
2.1 Seismic Waves

When an earthquake occurs, two types of seismic waves, namely body waves and
surface waves, are produced.
 14

Body waves travel through the interior of the earth and consist of two types: P waves

Table of Contents
and S waves (Figure 2.1).

P waves, also known as primary, compressional, or longitudinal waves, involve

successive compression and rarefaction of the materials through which they pass. They are
analogous to sound waves. The motion of an individual particle that a P wave travels
through is parallel to the direction of travel.

S waves, also known as secondary shear or transverse waves, cause shearing

deformations as they travel through a material. The motion of an individual particle is
perpendicular to the direction of S wave travel. The direction of particle movement can be
used to divide S waves into two components, SV (vertical plane movement) and SH
(horizontal plan movement).

The velocity of body waves depends on the stiffness of the materials they travel

Table of Contents
through. Since geologic materials are stiffer in compression, P waves travel faster than other
seismic waves.

M
Velocity of P wave: νp  ρ
  ..................................... (2.1)

G
Velocity of S wave: νs  ρ
  ..................................... (2.2)

where ρ = Density of the material through which the wave propagates

Constrained modulus =  1     E
 
M =
 1   1  2  
 1 
G = Shear modulus =   E
 2 1    

Table of Contents
E = Young’s modulus
 = Poisson’s ratio

Surface waves are generated by the interaction between body waves and the surface
and surficial layers of the earth. They travel along the earth surface with amplitudes that
decrease roughly exponentially with depth. The most important surface waves, for
engineering purposes, are Rayleigh waves and Love waves (Figure 2.2).

Rayleigh waves, produced by interaction of P and SV waves with the earth’s surface,
involve both vertical and horizontal particle motion. Love waves are generated by the interaction
of SH waves with a soft surficial layer and have no vertical component of particle motion.

It is noted that the interference between S waves and Rayleigh waves is important for
Table of Contents

generation of topographic effects on seismic ground motion (see Section 2.3).

2.2 Difference between Soil Layer Effects and Topographic Effects

2.2.1 General

The destructiveness of seismic ground motion can be significantly affected by local

Table of Contents Table of Contents Table of Contents Table of Contents
15
Figure 2.1 Body Waves: (a) P Wave and (b) S Wave (from Kramer, 1996)
 Table of Contents
Table of Contents
16

Table of Contents
Table of Contents
Figure 2.2 Surface Waves: (a) Rayleigh Wave and (b) Love Wave (from Kramer, 1996)
 17

site conditions, a term which refers to amplification effects that relate to the geometry

Table of Contents
(topography) of the earth surface and the mechanical properties and thickness of surficial
geological formations (soil layering). These conditions may result in large amplification and
significant variation of seismic ground motion, and thus are of a particular significance in the
assessment of seismic risk, microzonation studies, and planning and seismic design of
important facilities.

The most common approach to quantify site effects including topographic effects and
soil layer effects is to compare recordings of ground motion at nearby sites (where only local
site conditions are believed to be different) through either (i) frequency-domain spectral ratios
of Fourier amplitude spectra to give amplification factors at different frequencies, or
(ii) time-domain ratios of peak ground motion parameters such as acceleration or displacement.

It should be emphasised that topographic effects and soil layer effects are two different
phenomena – the former related to geometry of the earth surface and the latter to mechanical

Table of Contents
properties and thickness of surficial geologic materials. In some cases, they can occur alone
without presence of the other. For example, topographic effects alone can occur at a
homogeneous site with surface topographic irregularities while soil layer effects alone can
occur at a flat site with a thick layer of soft soil deposits underneath. In many cases, both
phenomena can occur at the same site simultaneously during an earthquake event. Many
previous studies have investigated whether topographic effects and soil layer effects can
interact with each other or they work independently; however, apparently, there is no
consistent conclusion on this issue.

2.2.2 Soil Layer Effects

Soil layer effects relate to the thickness and stiffness of soil layers from ground surface

Table of Contents
down to bedrock at a particular site.

The influence of soft soil layers on the intensity of ground shaking and earthquake
damage has been known for many years. Numerous observations after destructive
earthquakes around the world have illustrated the effects of local soil conditions resulting in
non-uniform distribution of damage. In recent years, the availability of strong-motion
records has also shown the soil layer effects on amplitude, frequency composition, and
duration of ground motions and has allowed soil layer effects to be measured quantitatively.
Examples of earthquakes showing significant soil layer effects include the 1985 Mexico City
earthquake and the 1989 Loma Prieta (California) earthquake (Kramer, 1996).

The fundamental phenomenon responsible for amplification of seismic ground motion

due to soil layer effects is the trapping of seismic energy due to the impedance contrast
between bedrock and overlying soil. The interference between these trapped waves leads to
Table of Contents

resonant patterns, the shape and frequency of which are related to the thickness and the
geometrical and mechanical characteristics of the soil. The amplitude of amplification is
related to the impedance contrast between the surface layers and the underlying bedrock, as
well as to the soil material damping.

One-dimensional ground response analysis shows that for a uniform layer of isotropic,
linear elastic soil overlying rigid bedrock, the magnitude of the transfer function, F1(),
 18

which describes the ratio of displacement amplitudes at the top and bottom of the soil layer, is

Table of Contents
given by the following equation.

1
F1      .............................................. (2.3)
cos  H /  s 

where  = angular frequency of the seismic excitation

H = thickness of soil layer
vs = shear-wave velocity of soil

When the angular frequency of the seismic excitation, , is equal to multiples of

 , the magnitude of the transfer function, F1(), will become infinity (Figure 2.3).
     s

   
 2   H 

In other words, infinite amplification will occur.

Table of Contents
For a more realistic case of uniform, damped soil on rigid bedrock, one-dimensional
ground response analysis shows that the magnitude of the transfer function will reach a local
maximum whenever the angular frequency of the seismic excitation, , is very close to
multiples of  2    Hs  , but will never reach a value of infinity (Figure 2.4). The
   

frequencies that correspond to the local maxima are the natural frequency of the soil deposit.

The nth natural frequency of the soil deposit is given by:

  2n  1    s
 n    H , n = 0, 1, 2, …∞ ............................ (2.4)
 2

Since the peak amplification factor decreases with increasing natural frequency, the

Table of Contents
greatest amplification factor will occur approximately at the lowest natural frequency, also
 s
known as the fundamental frequency,  0  .
2H

The period of vibration corresponding to the fundamental frequency is called the

fundamental site period:

4H
Ts  or T0   ................................................... (2.5)
νs

The fundamental site period, which depends only on the thickness and shear wave
velocity of the soil, provides a very useful indication of the period of vibration at which the
most significant amplification can be expected.
Table of Contents

Similarly, the nth natural site period of the soil deposit is given by:

4H
Tn  , n = 0, 1, 2, …∞ ................................ (2.6)
2n  1 s

When the bedrock is rigid, any downward-travelling waves in the soil will be
 Table of Contents
I

Table of Contents
19

Table of Contents
Table of Contents
Figure 2.3 Influence of Frequency on Steady-state Response of Undamped Linear Elastic Layer (from Kramer, 1996)
 Table of Contents
Table of Contents
20

Table of Contents
Table of Contents
Figure 2.4 Influence of Frequency on Steady-state Response of Damped, Linear Elastic Layer (from Kramer 1996)
 21

completely reflected back toward the ground surface by the rigid layer, thereby trapping all of

Table of Contents
the elastic energy within the soil layer. However, if the bedrock is elastic,
downward-travelling waves that reach the soil-rock boundary will be only partially reflected;
part of their energy will be transmitted through the boundary to continue travelling downward
through the rock, and the elastic energy of these waves will effectively be removed from the soil
layer. The free surface motion amplitudes will become smaller than those for the case of rigid
bedrock. Therefore, the amplification ratio depends on the bedrock stiffness (or the impedance
ratio). A stiffer bedrock (a higher impedance ratio) will result in a greater amplification.

2.2.3 Topographic Effects

Topographic effects are associated with the presence of surface topographic

irregularities, such as hills, ridges, canyons, cliffs and slopes, which can have effects on the
intensity, frequency composition and duration of ground motions during earthquakes.

Table of Contents
It is obvious that topographic effects are more complicated than soil layer effects, due
to their truly two- or three-dimensional nature. Although the effects of topographic
irregularities have been studied by many researches, only few seismic codes (e.g. Eurocode 8)
contain provisions for surface topography effects, despite their significance in engineering
practice. This is primarily due to the complexity of the problem that involves a large number
of governing parameters, the quantitative discrepancies between theoretical and instrumental
studies, and the lack of adequate instrumental data to justify rigorous regression analysis.
The physical phenomena explaining occurrence of topographic effects on seismic ground
motion are described in Section 2.3.

2.3 Physical Phenomena Leading to Occurrence of Topographic Effects

Table of Contents
The topographic effects due to surface topography such as hills, ridges and slopes can
be recognised by observations of structural damage after destructive earthquakes.
Instrumental data as well as analytical and numerical studies also allow identification and
understanding of the topographic effects from a quantitative perspective.

From field observations, instrumental data and theoretical studies, the following
general pattern of topographic effects on seismic ground motion can be observed:

(a) amplification of seismic motion on convex topographies, e.g.

ridge crests and cliffs;
(b) de-amplification over concave topographic features, e.g.
canyons and hill toes; and
Table of Contents

(c) complex amplification and de-amplification patterns on hill

slopes that result in significant differential motions.

In theory, these topographic effects are primarily a result of scattering or diffraction of

the incident seismic waves by topographic irregularities. According to Bard &
Riepl-Thomas (2000), these effects are related mainly to the following three physical
phenomena:
 22

(a) Body and surface waves are diffracted at and propagated

Table of Contents
outward from the topographic feature and this leads to
interference patterns between the direct and diffracted
waves. However, these diffracted waves generally have
smaller amplitudes on the surface than the direct body
waves.

(b) Surface motion is sensitive to the incidence angle and this

sensitivity is especially large for SV waves near the critical
angle of incidence (defined as that which produces a
refracted wave that travels parallel to the ground surface).
Therefore, the slope angle can produce significant variations
in surface motions (Figure 2.5). Kawase & Aki (1990)
suggested that this effect was a contributing cause to the
peculiar damage distribution observed on a mild slope

Table of Contents
during the 1987 Whittier Narrows (California) earthquake.
 
(c) Reflections of seismic waves can result in focusing or
defocusing of energy along the topographic surface.
Sanchez-Sesma (1990) provided insights into this effect by
means of a wedge-shaped medium. If this wedge has an
angle of 360/n and is subjected to plane SH waves
(Figure 2.6), then the response can be computed by
considering the multiple wave reflections within the wedge.
All these waves interfere positively at the vertex and the
resulting amplitude of motion at the vertex is n times greater
than the incident wave. However, instrumental proof of
such focusing/defocusing effects in strong seismic shaking

Table of Contents
has not been adequate because the lack of dense
strong-motion arrays on or near topographic features.

The phenomenon of wave diffraction had been observed by Ohtsuki & Harumi (1983)
in their numerical study on the effect of cliff topography on seismic SV waves. The results
of their numerical study (Figure 2.7) showed that when the incident SV wave travelling
vertically in the elastic half medium encounters the lower ground surface, it produces a
reflected SV wave (SV1), which returns into the half-space. Later when the incident SV
wave encounters the upper ground surface, it produces another reflected SV wave (SV2),
which also returns into the half-space. While the incident SV wave reaches the lower corner
of the slope, a Rayleigh wave (R1) is produced from the lower corner of the slope, and it
propagates along the slope and upper surface to the right. Later when the upcoming incident
SV wave reaches the upper corner of the slope, another Rayleigh wave (R2) is also produced
at the upper corner of the slope, and it propagates along the upper surface ahead of the R1
Table of Contents

wave. Figure 2.7 clearly shows that Rayleigh waves, which consist of both horizontal and
vertical displacements, are produced in the vicinity of the slope of the cliff. The incident SV
waves and Rayleigh waves combine near the slope crest resulting in a zone of amplification.

Similarly, Bouckovalas & Papadimitriou (2005), in their numerical analyses for the
seismic response of steep slopes under vertically propagating SV waves, mentioned that the
topographical effects could be attributed to the reflection of the incoming SV waves on the
 Table of Contents
Table of Contents
57
23

Table of Contents
Table of Contents
Figure 2.5 Sensitivity of Surface Motion to Incidence Angle for Obliquely Incident SV Waves (from Bard & Riepl-Thomas, 2000)
 Table of Contents
Table of Contents
58
24

Table of Contents
Table of Contents
Figure 2.6 Top Amplifications Versus Wedge Internal Angle for Incidence of Plane SH and SV Waves (from Sanchez-Sesma, 1990)
 25

Table of Contents
Table of Contents
Table of Contents
Table of Contents

Figure 2.7 Wave Propagation in Cliff Induced by an SV Wave (from Ohtsuki &
Harumi, 1983)
 26

inclined free surface of the slope (Figure 2.8), which leads to reflected P and SV waves

Table of Contents
impinging obliquely at the free ground surface behind the crest, as well as Rayleigh waves.
All these induced waves arrive with a time lag and a phase difference at the different points of
the ground surface so that their superposition to the incoming SV waves may lead either to
amplification or to de-amplification of the horizontal seismic motion.

From the above, it can be seen that topographic effects on seismic ground motion are
complicated. They are not caused by one single physical phenomenon and they are also
two- or three dimensional in nature.

2.4 Evidence of Topographic Effects

2.4.1 Documented Observations of Earthquake Damage

Table of Contents
Many documented observations from destructive earthquakes have repeatedly show
concentration of heavy damage near the crest of cliffs and slopes, or near the tops of hills and
ridges. Examples of such observations can be found in the 1985 Chile earthquake (Celebi,
1987 & 1991), 1987 Whittier Narrows (California) earthquake (Kawase & Aki, 1990), 1994
Northridge (California) earthquake (Sitar et al, 1997), 1995 Aegion (Greece) earthquake
(Bouckovalas et al, 1999), 1999 Parnitha (Athens) earthquake (Gazetas et al, 2002), and 1999
'Eje Cafetero' (Colombia) earthquake (Restrepo & Cowan, 2000).

A summary of the documented observations that have been identified and reviewed as
part of this study is presented in Table 2.1.

These documented observations clearly show that buildings located at the top of hills
and at the crest of slopes suffer much more intensive damage than those located at the base.
However, it should be noted that these observations cannot provide any quantitative

Table of Contents
information concerning the amplitude of the topographic effects on seismic ground motion.

2.4.2 Instrumental Evidence

Apart from observations of earthquake damage, there is very strong instrumental

evidence showing that surface topography affects considerably the amplitude and frequency
contents of seismic ground motion. These instrumental evidence usually come from strong
motion records of major earthquakes and also data obtained from instrumented arrays during
aftershocks, small-magnitude earthquakes and mining blasting.

A summary of the instrumental studies that have been identified and reviewed as part
of this study is presented in Table 2.2.
Table of Contents

Many of the instrumented arrays were installed after major destructive earthquakes and
therefore, these instrumental studies were generally limited to low amplitude recordings of
aftershock events. In some cases, instruments were installed in seismically active regions or
areas with mining or quarry blasting operations for a sufficiently long period of time so that
abundant instrumental data can been obtained for small-magnitude earthquakes and blastings.
As a result, instrumental evidence of topographic amplification of seismic ground motion is
 27

Table of Contents
Table of Contents
Table of Contents
Table of Contents

Figure 2.8 Schematic Illustration of Incoming SV Waves and Induced Reflected P Wave,
Reflected SV Wave and Rayleigh Wave (from Bouckovalas & Papadimitriou,
2005)
 Table of Contents
Table 2.1 Evidence of Topographic Effects: Documented Observations of Earthquake Damage (Sheet 1 of 3)

Cases Reference Types of topography Seismic source Observations

Lambesc, Provence, Levret et al (1986) Hillocks, steep 1909 Lambesc  More serious damage was observed in
France, 1909 ridges and slopes (Provence) earthquake places situated either at the top of a hill, at
(Ms=6.0) the end of a steep ridge, or on the slopes of

Table of Contents
escarpments.

Pacoima Dam, Boore (1972) Ridge 1971 San Fernando  Churned ground and overturned boulders
California, 1971 (California) were observed following earthquakes.
earthquake (M=6.6) These observations, indicating accelerations
greater than 1.0 g, were noted only on ridge
crests or other topographically high

28
features.

Table of Contents
Canal Beagle, Celebi (1987 & Ridges, canyons and 1985 Chile  The 4-storey buildings in the canyon did not
Chile, 1985 1991) hilltops earthquake (Ms=7.8) suffer any damage.
 All of the 4- and 5-storey buildings on the
ridges were extensively damaged, some
beyond repair.

Table of Contents
 Table of Contents
Table 2.1 Evidence of Topographic Effects: Documented Observations of Earthquake Damage (Sheet 2 of 3)

Cases Reference Types of topography Seismic source Observations

Puente Hills, Kawase & Aki Hill (300m high, 1987 Whittier  The most heavily damaged area was the
Whitter Narrows, (1990) 2400m wide) Narrows (California) northern part of the city of Whittier.
California, 1987 earthquake (ML=5.9)  Damage was concentrated along the slope

Table of Contents
of the hill rather than at the hilltop, which
was not expected.
 The incident angle for the heavily damaged
area roughly corresponds to the critical
angle for the realistic Poisson's ratio and the
reported focal depth.
 A combination of the topographic

29
irregularity with a critically incident
SV-wave source may cause a sharply
localized amplification.

Table of Contents
Pacific Palisades, Sitar et al (1997) Costal bluffs 1994 Northridge  The most severely damaged zone occurred
California, 1994 (40 m to 60 m high; (California) within about 50 m of the slope crest, a
45° to 60° steep) earthquake (Mw=6.7) dimension approximately equal to the
height of the bluffs.
 Most of the remaining damage occurred
within about 100 m of the slope crest.
 Most severe damage occurred nearest the
steepest slope.

Table of Contents
 Table of Contents
Table 2.1 Evidence of Topographic Effects: Documented Observations of Earthquake Damage (Sheet 3 of 3)

Cases Reference Types of topography Seismic source Observations

Aegion, Greece, Bouckovalas A vertical drop of 1995 Aegion  Structural damage was concentrated in the
1995 et al (1999) about 90 m formed by earthquake, Greece central part of the city situated behind the
a normal fault (Ms=6.2) fault escarpment.
 Little damage was observed in the harbour

Table of Contents
located below the fault escarpment.

Adames, Kifisos Gazetas et al Slope (30° steep, 1999 Parnitha  Structural damage was concentrated in two
Canyon, Greece, (2002) 40 m high) (Athens) earthquake regions: one located 10 to 50 m from the
1999 (Ms=5.9) slope crest, and the other at a distance of
about 200 to 300 m from it.
 The observed concentration of damage

30
could be attributed to combined effects of
topography and soil heterogeneity.

Table of Contents
Armenia, Colombia, Restrepo &  Southern part - 1999 'Eje cafetero'  A development of mainly 2-storey houses
1999 Cowan (2000) Deep gullies with earthquake, Colombia and 4-storey buildings in the southeast part
urban development (Mw=6.2) of the city suffered partial or total collapse
on ridge of 90% of the dwelling.
 Northern part -  The northern part of the city had relatively
Broad ridges and light damage.
fewer stream  Despite being in the northern part of the
gullies city, some buildings built on the crest of
steep roadside cut slopes still suffered
appreciable damage.

Table of Contents
 Lateral spreading was observed around the
crest of some ridges in rural areas.
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 1 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

Pacoima Boore (1972) Ridge Strong motion 1971 San N/A  High acceleration up to 1.25 g was
Dam, record Fernando recorded.
California, (California)  The wavelengths of the recorded seismic
1971 earthquake

Table of Contents
energy were comparable to the
(M=6.6) dimensions of the mountain (Kagel
Mountain) directly above the station.

Kagel Davis & West Mountain Instrumental Aftershocks Spectral ratios  Frequency-dependent amplification of
Mountain (1973) (Kagel data (A total of (M=2.6-3.2) (crest to base) the motion at the crest relative to the base
and Mountain - 4 stations on the from the 1971 of pseudo- was observed at both mountains.
Josephine 1400 ft high) top and the base San Fernando relative  At Kagel Mountain, amplification factors

31
Peak, (Josephine of the two (California) velocity in frequency domain were typically more
California, Peak - mountains) earthquake response than 10 and up to 30.
1971 3500 ft high) spectrum
 At Josephine Peak, amplification factors

Table of Contents
in frequency domain were about 3 to 5.
 Amplifications in time domain were
smaller than those in frequency domain.
The peak amplitude ratios (crest-to-base)
were smaller than the peak spectral
ratios.
 The wavelengths corresponding to largest
amplification were comparable to
mountain half-width.
 The amplifications were larger than the

Table of Contents
values predicted by theoretical models.
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 2 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

NASA Rogers et al Mountain Instrumental Underground Time-domain  Observed amplifications were about 25%
Mountain, (1974) (6 km long, data nuclear crest/base which was in good agreement with
Nevada 2 km wide) explosion ratios of peak theoretical predictions.
ground

Table of Contents
velocities

Tien Shan Tucker et al Ridges and Instrumental 25 earthquakes Spectral ratios  Rock outcrop sites on ridges had spectral
mountains, (1984) tunnels data from (ML=1.6-4.2) (crest to amplification ratios with respect to a
Garm seismometers tunnel) nearby tunnel (not with respect to the
Region, ridge base) as high as 8.
USSR,  Amplification was higher for horizontal
1976-1978 than for vertical motions, and was

32
frequency-dependent.

Coalinga Celebi (1991) Ridges and Instrumental Aftershocks Spectral ratios  Amplification of motions was up to 10.

Table of Contents
anticline, gulleys data from a from the 1983 (ridge to  The effects were frequency-dependent.
California, temporary Coalinga gulley)
1983 closely spaced (California)
accelerographs earthquake

Canal Celebi (1987 & Ridges, Instrumental Several Spectral ratios  Horizontal amplification was
Beagle, 1991) canyons & data from a aftershocks (ridge to frequency-dependant and occurred at
Chile, 1985 hilltops temporary dense from the 1985 valley) frequency range of 2 to 4 Hz.
array Chile  Amplification typically was up to 10,
earthquake few even reached 20.
 The frequency range of 2 to 4 Hz

Table of Contents
correlated well with the fundamental
frequencies of the heavily damaged 4-
and 5-storey buildings.
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 3 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

Superstition Celebi (1991) Mountain Instrumental 11 aftershocks Spectral ratios  Horizontal motions were amplified by as
Mountain, (top and data from three from the 1987 (top to flank) much as a factor of 20.
California, flanks) temporary Superstition  The effects were frequency-dependent.
1987 digital recording Hills

Table of Contents
stations (California)
earthquake

Robinwood Hartzell et al Ridge Instrumental 1989 Loma Spectral ratios  Topographic effect with amplification
Ridge, (1994) (110 m high) data from seven Prieta (crest to base) from 1.5 to 4.5 in the frequency range
California, seismograph earthquake from 1.0 to 3.0 Hz was seen, part of
1989 stations (Ms=7.1) which may be caused by local site effects.
These wavelengths were comparable to

33
the base width of the ridge.
 Amplifications of up to a factor of 5

Table of Contents
were seen at higher frequencies and are
attributed to local site effects.

Epire, Chavez-Garcia An elongated Instrumental 68 small Horizontal-to-  Maximum amplification at mountain top
Northern et al (1996) hill data from 10 seismic events vertical computed from HVSR was moderate.
Greece, (200 m high, seismometers (M=1.7-4.5) spectral ratios  Observed and theoretical amplifications
1989 1000 m (HVSR) were similar and were below a factor of
wide) 5.

Tarzana, Celebi (1995) Small hill Strong motion 1987 Whittier N/A  High horizontal peak acceleration (0.61 g)
California, (20 m high, record Narrows was recorded, much larger than the

Table of Contents
1987 500 m long, (California) expected values for an earthquake of that
200 m wide) earthquake magnitude.
(Ms=6.1)
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 4 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

Tarzana, Celebi (1995) Small hill Strong motion 1994 N/A  Unusually high horizontal peak
California, (20 m high, record Northridge acceleration (1.82 g) was recorded, much
1994 500 m long, (California) larger than those estimated by
200 m wide) earthquake attenuation relationships.

Table of Contents
Instrumental Aftershocks Spectral ratios  Amplitude of motions and spectral ratios
data from a from the 1994 (crest to base) decreased from the hilltop towards the
temporary dense Northridge more level part of the hill.
array (California)
earthquake

Tarzana, Spudich et al Small hill Strong motion 1994 N/A Unusually high horizontal peak

34

California, (1996) (15 m high, record Northridge acceleration (1.78 g) was recorded, while
1994 500 m long, (California) two other stations located within 2 km of
130 m wide) earthquake the station at Tarzana recorded

Table of Contents
(Mw=6.7) considerably smaller accelerations.

Instrumental Aftershocks Spectral ratios  Amplifications of about 4.5 and 2 were

data from a from the 1994 (crest to base) observed for horizontal components
dense array Northridge oriented perpendicular and parallel to the
(California) long axis of the hill respectively.
earthquake  The numerical study by Bouchon &
(M=1.7-2.9) Barker (1996) obtained an amplification
of about 1.6.

Table of Contents
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 5 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

Sourpi, Pederson et al Ridge Instrumental 14 local and Spectral ratios  The average spectral ratios showed
Greece, (1994b) (300 m high, data from seven regional (crest to base) amplifications of 1.5 to 3.
1992 5 km long, seismometers shallow  The horizontal components were more
2.5 km wide) earthquakes

Table of Contents
amplified than the vertical component.
 The observed spectral ratios were
modest and within the range predicted
by numerical simulations.

Mont Saint Pederson et al Ridge (25° Instrumental 7 teleseismic Spectral ratios  The average spectral ratio was up to 4.
Eynard, (1994b) steep, 400 m data from two events and 5 (crest to flank)  The observed amplifications were
France, high, 7-8 km seismometers local/regional modest and within the range predicted

35
1993 long, 2-3 km (Mb=1.4-2.0) by numerical simulations.
wide) events

Table of Contents
Kitheron LeBrun et al Hill (700 m Instrumental 51 earthquakes Spectral ratios  The E-W elongation of the hill produced
Mountain, (1999) high, 6 km data from seven (ML=0.5-3.6) a larger amplification in the N-S
Corinth, long, 3 km seismometers component than in the E-W one.
Greece, wide) across the ridge  Spectral ratios were smaller than 5 for
1993 all stations, except one which showed a
larger (almost 10) amplification on both
horizontal components.
 The vertical component showed much
smaller effect than the horizontal one.

Table of Contents
 Table of Contents
Table 2.2 Instrumental Evidence of Topographic Effects (Sheet 6 of 6)

Types of Measure of
Cases Reference Source of data Seismic source Observations
topography amplification

Nice, Nechtschein et Two ridges Instrumental Micro- Spectral ratios  Amplification and attenuations were
France,1994 al (1995) (about 200 m data from 11 earthquakes mainly observed on the horizontal
(Locations: and 400 m seismic stations and mining components. The vertical component did
Castillon & high) with along the slopes explosions not exhibit any strong effect.

Table of Contents
Piene) steep slopes (total 46  Significant amplifications for ground
on both sides events) motions was observed at the ridge crest,
with a peak value around 10.
 Attenuation was observed near the base.
 Ground motion showed a strong
variability along the slopes.
Amplification at ridge top was largest

36

along the horizontal direction
perpendicular to the ridge axis.

Table of Contents
Table of Contents
 37

not much for strong and destructive seismic shaking, and therefore it seems not possible to

Table of Contents
derive reliable conclusions of a general nature.

Nevertheless, unlike documented observations, these instrumental studies can quantify

the amplitudes of topographical amplification of seismic ground motion. Typically, the
amplitudes of topographical amplification are calculated directly by either the time-domain
ratios of peak ground motion parameters (usually acceleration or displacement), or the
frequency-domain spectral ratios of Fourier amplitude spectra, of a particular location on
topography to a selected reference site which is assumed as the free-field condition. It is
noted that in these studies, the base of ridges or hills, or the slope toe, is usually selected as
the reference site.

The following general observations are obtained from the instrumental studies
reviewed in this study:

Table of Contents
(a) Some strong ground motion records show unusually high
horizontal peak accelerations which are much higher than
those estimated by attenuation relationships for an
earthquake of that magnitude. Many researchers believe
that these unusually high horizontal accelerations are related
to the surface topography of the sites. Examples are the
recorded PGA values of 1.25 g at Pacoima Dam in the 1971
San Fernando (California) earthquake (Boore, 1972) and of
1.82 g on a small and relatively flat hill in Tanzana during
the 1994 Northridge (California) earthquake (Celebi, 1995;
Spudich et al, 1996).

(b) The crest-to-base spectral ratios (see Section 2.2.1)

Table of Contents
consistently show a frequency-dependent amplification of
the horizontal components of ground motion. In general,
the wavelengths corresponding to amplification frequencies
are comparable to the dimensions of the topography (e.g.
ridge width). There are cases where modest amplifications
are observed (e.g. Pederson et al, 1994b; LeBrun et al,
1999). There also exist many observations of
amplification factors up to 10 (e.g. Celebi, 1991;
Nechtschein et al, 1995). In some cases, the observed
amplification factors even reach the order of 20 to 30 (e.g.
Davis & West, 1973; Celebi, 1987 & 1991).
 
(c) The time-domain crest-to-base ratios show relatively
smaller amplification factors than frequency-domain
Table of Contents

spectral ratio.
 
(d) The amplification of the vertical motion is much smaller
than the horizontal motion.
 
As instrumental studies have provided valuable quantitative information on the
amplitudes of the topographic amplification on seismic ground motion, a comparison between
 38

these instrumental (experimental) results and theoretical (numerical) results has been made.

Table of Contents
The findings of such comparison will be presented in Section 2.5.

2.5 Previous Numerical Studies on Topographic Effects

2.5.1 General

Prompted by observational and instrumental evidence, a large number of numerical

studies have been conducted using different numerical methods in an attempt to model,
quantify and predict the topographic effects on seismic ground motions, and to compare the
theoretical amplifications with observations. In addition, some numerical parametric studies
have been performed to identify the parameters that affect topographic effects on seismic
ground motion and to investigate their influence. Typically these numerical studies show:

Table of Contents
(a) systematic amplification of seismic motion over convex
topographies such as hills and ridges,

(b) de-amplification over concave topographic features such as

canyons and hill toes,

(c) complex amplification and de-amplification patterns on hill

slopes that result in significant differential motions,

(d) amplification being frequency dependent, and

(e) higher amplification on the horizontal components than

vertical.

Table of Contents
In general, there is a qualitative agreement between theoretical and observed
topographic amplifications. However, from a quantitative viewpoint, there exist clear
discrepancies between theory and observations.

2.5.2 Numerical Studies on Ridges and Hills

The majority of numerical studies on ridges and hills focus on two-dimensional

simulations in which the topographical features are treated as isolated ridges on the surface of
homogeneous half-spaces, usually under further simplified assumptions of vertically
propagating incident waves and harmonic steady-state excitation.

Geli et al (1988) presented a comprehensive review of previous numerical studies on

Table of Contents

ridges and hills, and they found that most of them gave consistent results:

(a) The time-domain crest-to-base acceleration amplification

ratios remain below two.

(b) The frequency-domain crest-to-base amplification ratios are

up to three and the peaks occur when wavelength is about
 39

equal to the ridge width.

Table of Contents
 
(c) Varying amounts of amplification and attenuation occur
along the surface of the slope from the crest to the base.
 
(d) The amplification is lower for incident P waves than for
incident S waves.
 
(e) The amplification is slightly larger for in-plane horizontal
motion (SV waves) than for anti-plane motion (SH waves).

Geli et al (1988) also noted that most of the numerical studies reviewed considerably
underestimate amplifications observed in the field, which mostly range from 2 to 10, and up
to as much as 30. In the same paper, they carried out a detailed examination of several cases
and they considered that the discrepancy between observations and theory may be due to

Table of Contents
subsurface soil layering and/or neighbouring ridges, and suggested that a more complex
numerical model incorporating subsurface soil layering and neighbouring ridges could reduce
the discrepancy.

In this connection, Geli et al (1988) analysed a more complex configuration to evaluate

the effects of subsurface layering and/or nearby ridges to incident SH waves and they arrived
at conclusions similar to those of the previous researchers. In particular, their results showed
that:

(a) The topographic effect itself is difficult to isolate from other

effects, like subsurface soil layering, and therefore the
topographical amplification cannot be predicted by a priori
estimations based solely on topography.

Table of Contents
(b) The high crest-to-base amplification ratios observed in the
field usually cannot be matched even with complex
two-dimensional structures with incident plane SH waves,
which suggests that more complex models are needed to
incorporate more complex wave fields (e.g. SV, surface) and
three-dimensional geologic configurations.

There are also other numerical studies which involve relatively more complex models.
Bard & Tucker (1985) investigated the SH response of a set of ridges with irregular
subsurface layering. Bouchon & Barker (1996) and Bouchon et al (1996) investigated the
response of three-dimensional homogeneous hills. In general, these three-dimensional
models lead to slightly higher crest-to-base amplification ratios than two-dimensional models
and they show a higher amplification on the horizontal component along the direction of
Table of Contents

motion perpendicular to the ridge axis.

In summary, based on the comparison of instrumental and theoretical results, the

following observations can be made:

(a) A qualitative agreement exists between theory and

observations about the existence of seismic motion
 40

amplification at ridges and mountain tops, and

Table of Contents
de-amplification at the base of hills. The amplification is
generally larger for horizontal components than for the
vertical component.

(b) This amplification (or de-amplification) phenomenon is

frequency-dependent. There is a rather good qualitative
agreement between instrumental observations and
theoretical results for the relationship between mountain
width and the approximate frequency range where
amplification is significant.
 
(c) However, from a quantitative viewpoint, clear discrepancies
exist between theory and observations. Cases have been
reported in which field measurements exhibit only weak to

Table of Contents
modest amplifications at ridge crests or hilltops and fit fairly
well with the numerical results (e.g. Rogers et al, 1974;
Pedersen et al, 1994b; LeBrun et al 1999). On the other
hand, numerous cases also exist in which the observed
amplifications are significantly larger than the theoretical
predictions obtained from sophisticated, two- or
three-dimensional models (e.g. Bouchon & Barker, 1996;
Bouchon et al, 1996). Numerous observations have been
made of spectral amplifications around 10, and up to the
range of 20 to 30 for some cases (e.g. Davis & West, 1973;
Celebi, 1987 & 1991; Nechtschein et al, 1995), but such
amplitudes are rarely predicted by numerical models.

Table of Contents
The confusing results from quantitative comparison between observed and theoretical
amplifications could be attributed to the following:

(a) Internal structure (e.g. subsurface layering, lateral variations

of geology) of the topography features is not adequately
accounted for in numerical models.

(b) Three-dimensional shape of the topography features and

neighbouring ridges are not adequately taken into account in
numerical models.
 
(c) It is important to note that the crest-to-base spectral ratios
are not measurements of absolute amplifications with
respect to a half-space. Motions at stations at the base of a
Table of Contents

hill or ridge can be de-amplified while those at the slope or

flank can be amplified or de-amplified as a function of
frequency. This means in practice that amplification on
mountain tops estimated by comparison with a reference
station at the base or flank can be severely biased and very
large observed amplifications can be obtained.
 41

2.5.3 Numerical Studies on Slopes

Table of Contents
While some of the procedures and concepts developed for the analysis of ridges and
hills may be extended to steep slopes, there are significant differences between the response
of steep slopes and the response of rock ridges simulated as homogeneous half-spaces. The
most important differences are the semi-infinite nature of material in the horizontal direction
behind the slope crest and the potential for soil amplification of the motions in the
one-dimensional (i.e. vertical) sense.

Ashford et al (1997) presented a comprehensive review of previous numerical studies

that specifically considered seismic response of soil slopes. Many researchers (Idriss &
Seed, 1967; Idriss, 1968; Kovacs et al, 1971; Sitar & Clough, 1983) suggested that if the
motions were amplified in the vicinity of the slope crest, the natural period of the
one-dimensional soil column behind the crest is responsible for larger amplification of the
input motion than the amplification due to slope geometry. Therefore, when a soil layer over

Table of Contents
bedrock is considered, the amplitudes of the amplification of seismic ground motions near the
crest of slopes will be different from the case of slopes with a homogenous half-space.

Among the published studies specific to topographical effects on seismic response of

steep slopes, the recent ones performed by Ashford et al (1997), Ashford & Sitar (1997), and
Bouckovalas & Papadimitriou (2005) are the most extensive and most relevant to the scope of
the present study.

Ashford et al (1997) conducted a frequency-domain parametric study to evaluate the

significance of topographic effects on the seismic response of steep slopes in a homogeneous
half-space (i.e. with no impedance contrasts). Both vertically propagating SH and SV waves
were considered. The key findings of their study are summarised below:

Table of Contents
(a) The topographical effects of a steep slope on the seismic
response of that slope can be normalised as a function of the
ratio of the slope height (H) and wavelength of the motion
().

(b) For both SH and SV waves, the magnitude of the response

at slope crest is significantly reduced by increased damping,
particularly at higher frequencies. However, the
amplification of the motion at the crest over that in the free
field behind the crest is relatively unaffected by damping.
 
(c) The peak topographical effect occurs at H/ = 0.2 (defined
as ‘topographic frequency’ by the authors) for both
vertically propagating SH and SV waves. The
Table of Contents

amplification at slope crest (relative to the free-field behind

the crest) at the topographic frequency is in the order of
25% for SH waves, and 55% for SV waves (Figures 2.9a
and 2.9b). The topographical amplification on the vertical
component has a monotonic increase with normalised
frequency and peaks at H/ = 1.0 for SV waves
(Figure 2.9c).
 42

(d) The topographical effects, as reflected by the initial peak

Table of Contents
amplifications of horizontal response at the topographic
frequency, tend to increase with slope angle (Figures 2.9a
and 2.9b).

In an attempt to investigate the influence of soil amplification on topographic effects,

Ashford et al (1997) also carried out a parametric study on the response of a slope in a soil
layer overlying a homogeneous half-space, with an impedance ratio of three. They
concluded that the natural frequency of the site behind the crest dominates the response,
which agrees with the findings of Sitar & Clough (1983). They also established that if the
natural frequency of the site is approximately equal to the topographic frequency, then the
response at slope crest would be amplified by over 50% relative to the free-field motion, and
this amount of amplification is similar to the amount they observed at the topography
frequency of the stepped half-space. Based on these results, Ashford et al (1997) suggested
that the effects of natural frequency and those of topography may work independently and

Table of Contents
hence the effects of topography can be handled separately from amplification due to the
natural frequency of the soil layer behind the slope crest.

Bouckovalas & Papadimitriou (2005) also conducted a similar parametric study to

evaluate the topographical effects of step-like slopes in homogeneous material on seismic
response. The numerical analyses were performed with the Finite Difference method and
only vertically propagating SV waves and homogeneous half-space were considered.
Despite the use of distinctly different methodologies of analyses (i.e. Finite Difference
method versus generalised consistent boundary method), Bouckovalas & Papadimitriou (2005)
obtained very similar results to that from Ashford et al (1997) (see Figure 2.10), and their key
findings are as follows:

(a) Even a purely horizontal excitation, such as vertically

Table of Contents
propagating SV waves, can result in considerable parasitic
vertical motion at the ground surface near the slope. The
results of the parametric analyses show that the vertical
parasitic component of seismic motion may become
comparable to the horizontal free-field motion.

(b) The topography aggravation of the horizontal ground

motion fluctuates intensely with distance away from the
crest of the slope, alternating between amplification and
de-amplification within very short horizontal lengths.
Similarly, the topographic apparition of parasitic vertical
motion is also intensely variable with distance. The
authors highlighted that actual ground motion recordings
near slopes must therefore be obtained via very dense
Table of Contents

seismic arrays.
 
(c) The horizontal ground motion is de-amplified at the toe of
the slope and amplified near the crest. As a result,
topography aggravation may be seriously overestimated,
when calculated as the peak seismic ground motion at the
crest over that at the toe of the slope. The authors believed
 43

Table of Contents
(a) Horizontal amplification for a vertically incident SH wave

Table of Contents
Table of Contents
(b) Horizontal amplification for a vertically incident SV wave

Table of Contents

(c) Vertical amplification for a vertically incident SV wave

Figure 2.9 Amplification at the Crest of an Inclined Slope in a Homogeneous

Half-space ( = 1%) (from Ashford et al, 1997)
 44

Table of Contents
Table of Contents
(a) Horizontal amplification

Table of Contents

(b) Vertical amplification

Table of Contents

Figure 2.10 Amplification at the Crest of an Inclined Slope in a Homogeneous

Half-space for a Vertical Harmonic SV Wave ( = 5%) (from Bouckovalas
& Papadimitriou, 2005)
 45

that this overestimation may explain, at least in part, why

Table of Contents
field measurements (without appropriate free field selection)
of topography aggravation are usually significantly higher
than analytical predictions.

Bouckovalas & Papadimitriou (2005) also studied the effects of different parameters
(including slope angle i, dimensionless frequency of excitation H/, hysteretic damping ratio
of soil , and number of significant excitation cycles, N) on the topographical effects on
seismic ground motion. More details will be given in Section 2.6.

It should be emphasised that the amplitudes of topographical amplification of steep

slopes given in both on Ashford et al (1997) and Bouckovalas & Papadimitriou (2005) were
obtained from the analyses of step-like slopes in homogenous half-space, without an
impedance contrast (i.e. bedrock).

Table of Contents
Tripe (2009) carried out a parametric study to investigate topographic amplification in
the presence of soil layer effects using the Imperial College Finite Element Program (ICFEP)
(Potts & Zdravkovic, 1999). His analysis considered a soil slope in a homogeneous layer
overlying bedrock. Tripe (2009) made the following conclusions in his study:

(a) With presence of bedrock, the amplification of ground

motion relative to the input motion at bedrock has two
components, one directly related to the amplification due to
soil layer effects and the other to the interaction of the
topographic effects with the soil layer effects. The second
component increases with reducing bedrock depth.

(b) Similar trends in the behaviour of topographic effects were

Table of Contents
observed compared with Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005), but the magnitude of
amplification was considerably larger due to the interaction
with soil layer effects.

As only one bedrock depth was analysed in his parametric study, Tripe (2009) could
not make any general quantitative assessment of the effects of bedrock depth on the
magnitude of amplification.

From literature review, it remains uncertain whether for steep slopes, the geometry of
topographic irregularity and the subsurface soil conditions can be handled separately, as
suggested by Ashford et al (1997), or they have to be modelled simultaneously to obtain
accurate estimates of the amplification levels.
Table of Contents

2.6 Parameters Affecting Topographic Effects

2.6.1 General

Based on the findings of the numerical studies reviewed, the following factors which
can affect the amplitude and extent of topographic effects on seismic ground motion are
 46

identified:

Table of Contents
(a) characteristics of incident wave (e.g. wave type, angle of
incidence, frequency content, duration);

(b) topographic geometry (e.g. slope height, slope inclination,

three-dimensional shape);

(c) subsurface soil layering; and

(d) soil parameters.

2.6.2 Type of Incident Waves

Table of Contents
The incidence of SH waves has been studied more frequently, as reflection and
diffraction of SH waves does not generate other wave types, in contrast to incident P or SV
waves. For incident S waves, it has been shown that the amplification of SV waves is
usually higher than that of the SH waves. According to Ashford et al (1997), a vertical cliff
subjected to vertically propagating SH waves resulted in a maximum amplification at the
surface in the order of 1.25 (with respect to the free-field), whereas for the same configuration,
a maximum amplification in the order of 1.5 was observed for vertically propagating SV
waves (Figure 2.9). Qualitatively, the pattern of distribution of soil amplification and
de-amplification of seismic motion as a function of the distance from the cliff is similar for
SH and SV waves.

2.6.3 Angle of Incidence

Table of Contents
The majority of the numerical studies reviewed considered vertically propagating
waves. This is a reasonable assumption in earthquake engineering because seismic waves
are refracted towards the vertical as they travel upwards from materials of higher density to
materials of lower density. However, this assumption may not be fully valid for ridges or
mountains in homogenous rock extending to great depth and subject to relatively shallow
earthquakes.

Ashford & Sitar (1997) performed a parametric study to study the effect of the incident
angle on topographic effects for a vertical slope (Figure 2.11). Their study was limited to
angles smaller than the critical angle. When the angle of incidence of the propagating
S waves coincides with the critical angle, a transformation takes place, and a single P wave
arises and propagates along the slope. They observed that the horizontal motion at the slope
crest is amplified for waves travelling into the slope and attenuated for waves travelling away
Table of Contents

from the slope, as compared to the motion in the free field behind the slope crest. This
amplification can be as much as twice that due to vertically propagating waves. In contrast,
the vertical response due to SV waves appeared to be independent of the direction of wave
propagation. The authors suggested that the amplification of inclined SV waves travelling
into the slope may partially explain field observations of failures on slopes facing in a
particular direction, while slopes in the same material, but of different orientation, showed no
distress.
 47

Table of Contents
(a) Horizontal amplification for an inclined SH wave

Table of Contents
Table of Contents
(b) Horizontal amplification for an inclined SV wave

Table of Contents

(c) Vertical amplification for an inclined SV wave

Figure 2.11 Amplification at the Crest of a Vertical Slope in a Homogeneous Half-space

for Inclined Wave (from Ashford & Sitar, 1997)
 48

2.6.4 Frequency Content of Incident Motion

Table of Contents
Topography effects are very significant for wavelengths comparable to the geometric
characteristics of the irregularity, whereas they are negligible for very low frequencies, i.e.
very long wavelengths (Ohtsuki & Harumi, 1983; Geli et al, 1988; Ashford et al, 1997;
Bouckovalas & Papadimitriou, 2005).

For slope type topographies, it has been observed that the surface response is primarily
controlled by diffraction of incident waves at the slope surface, the effect of which depends on
the ratio of the dimensionless frequency (H/). According to Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005) (Figures 2.9 and 2.10), who studied topographical
effects on the seismic response of steep slopes in a homogeneous half-space, horizontal
amplification at slope crest peaks at H/ = 0.2 and 0.7 for vertically incident SH waves and
peaks at H/ = 0.2 and 1.0 for vertically incident SV waves. For the vertical component,
there is a monotonic increase of the vertical amplification with the dimensionless frequency

Table of Contents
of the vertical incident SV waves, and the vertical component is almost zero for very low
frequencies (H/ < 0.05).

2.6.5 Duration and Type of Time-history Excitations

The majority of the parametric studies reviewed used a single type of time-history
excitations, most commonly harmonic motions. Bouckovalas & Papadimitriou (2005) is the
only study that considered different types of time-history excitations and evaluated the effects
of their duration on seismic response.

Bouckovalas & Papadimitriou (2005) studied the seismic response of steep slope in a
homogenous half-space under vertically propagating SV waves. In their study, most of the

Table of Contents
parametric analyses were performed either with a harmonic excitation of 20 to 40 uniform
cycles, or with a Chang’s signal excitation aimed to simulate the limited duration as well as
the gradual rise and decrease of shaking amplitude. Their results show that the number of
significant excitation cycles N has a relatively minor overall effect on the amplifications of
peak accelerations (Figure 2.12).

Bouckovalas & Papadimitriou (2005) also performed analyses on the same geometry
of a step-like slope using two actual earthquake records and a Chang’s signal excitation with
similar characteristics of predominant period and number of significant cycles and scaled to
the same peak ground acceleration. They observed that all three analyses provide
compatible results, with the Chang’s signal excitation leading to rather conservative estimates
of the topographic amplification of the peak ground accelerations. A possible explanation
for this observation is that the much wider frequency content of actual earthquake excitations
may prevent the topography from generating large amplification at a particular frequency as
Table of Contents

the different frequency components of real earthquake excitations can develop different
patterns of amplifications and de-amplifications and have different magnitudes of
amplification/de-amplification during earthquakes.
 Table of Contents
Table of Contents
49

Table of Contents
Table of Contents
Figure 2.12 Effect of the Number of Significant Cycles N on the Horizontal and Vertical Amplification Factors (Ah and Av), as a
Function of Horizontal Distance x from the Slope Crest (H/ = 2, i = 30,  = 5%) (from Bouckovalas & Papadimitriou,
2005)
 50

2.6.6 Slope Inclination

Table of Contents
Ashford & Sitar (1997) and Bouckovalas & Papadimitriou (2005) studied the effect of
inclination of cliff slope on the response at the crest to vertically propagating SV and SH waves
(Figures 2.9 and 2.10). They observed that as the slope becomes less steep from vertical to
30, the magnitude of horizontal amplification (relative to free-field) at the first peak occurring
at H/ = 0.2 decreases from 25% to about 15% for SH waves and from 55% to 15% for
SV waves.

2.6.7 Three-dimensional Topography Shape

Bouchon et al (1996) investigated the effects of three-dimensional shape of a hill on

amplification of seismic ground motion and its frequency dependency by considering a hill in
an elliptical shape on plan, subject to vertical shear waves polarised along the minor axis and

Table of Contents
along the direction of elongation of the hill respectively. They found that for incident shear
waves polarized along the short minor axis, the amplification at the top of the hill was stronger
and the maximum amplification was about two when the wavelength of input excitation is
equal to the height of the hill, while for incident shear waves polarized along the direction of
elongation of the hill, the maximum amplification was about 1.75 and occurred when the
wavelength is equal to four times the height of the hill (the longest of the four wavelengths
considered in the study). Bouchon & Barker (1996) also obtained the same finding in their
three-dimensional modelling specifically for the case of Tarzana Hill using the same numerical
methods.

Apart from numerical modelling, the effects of three-dimensional topography shape on

topographic amplification can also be observed in instrumental studies. For example,
Nechtschein et al (1995) observed that for ridges having a marked elongation, the amplification

Table of Contents
effects at ridge top were the largest along the horizontal direction perpendicular to the ridge
axis.

2.6.8 Soil Layering

The majority of the literature reviewed considered topographic features in a

homogenous half-space, and therefore they did not consider additional amplification due to soil
layer effects. As mentioned in Section 2.5.1, some researchers (e.g. Geli et al, 1988)
suggested that the discrepancy between observed and theoretical amplifications could be
attributed to the presence of subsurface layering that had not been considered in the model, and
some work have been done to study how the presence of soil layer overlying homogeneous
half-space can influence the topographic effects on seismic ground motions (e.g. Sitar &
Clough, 1983; Geli et al, 1988; Ashford et al, 1997; Graizer 2009). In general, these studies
Table of Contents

were able to predict higher amplification ratios than those considering homogeneous half-space
only.

Strong soil amplification is expected when the incoming wavefield is rich in frequencies
close to the natural frequency of the soil column behind the slope crest. Sitar & Clough (1983)
and Ashford et al (1997) obtained large amplification of horizontal motion at slope crest
 51

(relative to input motion) when the frequency of the input motion was equal to the natural

Table of Contents
frequency of the soil layer behind the crest. Their results show that the natural frequency of
the site behind the crest dominates the surface response and the topographic amplification at
the slope crest was small compared to the soil layer amplification in the free field. Therefore,
the role of soil layering on the surface response of topographic features can be very important.

However, as mentioned before, the interaction between topographic effects and soil
layer effects on amplification of seismic ground motion behind slope crest is still an unsettled
issue. Ashford et al (1997) suggested that the topographic effects can be handled separately
from amplification due to the natural frequency of the layer behind the crest of the slope
whereas some (e.g. Geli et al, 1988) considered that the topographic effects are difficult to
isolate from subsurface layering and the amplification of ground motion cannot be predicted by
a priori estimations based solely on topography.

Table of Contents
2.6.9 Soil Damping

In the parametric study by Ashford et al (1997), the values of critical damping ratios ()
used in the analyses varied from 1% to 20%. They found that increased damping
significantly reduces the response of both the free field and of the slope, particularly at higher
frequencies, but damping has very little effect on the amplification of the motion at the crest as
compared to the free field behind the crest (Figure 2.13).

Bouckovalas & Papadimitriou (2005), in their parametric study, also obtained similar
results. The damping ratio of soil () has insignificant effect on the amplification of the
motion at slope crest (Figure 2.14). They also observed that material damping can affect the
extent of the area where the response is governed by the topographic irregularity, which
becomes more confined in the vicinity of the slope as material damping increases.

Table of Contents
2.6.10 Discussion

From the literature review, a range of parameters affecting the topographic effects on
seismic ground motions have been identified. However, it is noted that the majority of the
numerical and parametric studies examined topographic features in a homogenous half-space,
and therefore they were unable to investigate the interaction between topographic effects and
soil layer effects.

Although a few numerical studies have considered the presence of subsurface soil layer,
they were not performed based on sufficient parameters so that a general overall idea of the
interaction between topographic effects and soil layer effects can be obtained. Therefore,
from a quantitative point of view, it remains uncertain about how the presence of a soil layer
Table of Contents

overlying bedrock would influence the amplitude of topographic amplification at slope crest,
especially when the topography is subjected to incident seismic waves having the same period
as the site periods of the one-dimensional soil column at free-field behind the slope crest which
results in soil layer amplification of the incident seismic waves.
 Table of Contents
Table of Contents
52

Table of Contents
(a) (b)

Table of Contents
Figure 2.13 Horizontal Amplifications for Vertically Incident SH Wave in a Slope in a Homogeneous Half-space for Various
Distances Behind the Crest, (a)  = 1% and (b)  = 5% (from Ashford et al, 1997)
 Table of Contents
Table of Contents
53

Table of Contents
Table of Contents
Figure 2.14 Effect of Soil Damping  on the Horizontal and Vertical Amplification Factors (Ah and Av), as a Function of Horizontal
Distance x from the Slope Crest (H/ = 2, i = 30, N = 4) (from Bouckovalas & Papadimitriou, 2005)
 54

2.7 Provisions of Topographic Effects in Seismic Codes

Table of Contents
Based on the review of seismic codes carried out by Bouckovalas & Papadimitriou
(2005 & 2006) and Tripe (2009), it was found that only the Eurocode 8 (EC-8) (British
Standards Institution, 2004) and the French seismic code PS-92 (AFNOR, 1995) have
provisions for considering topographic effects of seismic ground motion, particularly for
two-dimensional topographic irregularities.

The EC-8 specifies that for important structures on or near slopes, topographic
amplification effects should be taken into account. According to the (Informative) Annex A
of EC-8 Part 5, the topographic amplification factor, ST, considered to be independent of the
fundamental period of vibration, is multiplied as a constant scaling factor to the ordinates of
the elastic response spectrum. The code gives the following recommendations for
topographic amplification factor ST for two-dimensional topographic irregularities:

Table of Contents
 ST is applied as a constant scalar factor at all frequencies.

 ST is applied for slopes with height greater than 30 m and

average slope angle greater than 15.

 For isolated cliffs and slopes, a value ST is ≥ 1.2 should be

used for sites near the top edge.

 For ridges with crest width significantly less than the base
width, a value ST ≥ 1.4 should be used near the top of the
slopes for average slope angles greater than 30, and a value
ST ≥ 1.2 should be used for smaller slope angles.

Table of Contents
 In the presence of a loose surface layer, the value of ST
should be increased by at least 20%.

 The value of ST may be assumed to decrease as a linear

function of the height above the base of the cliff or ridge,
and to be unity at the base.

It is noteworthy that the EC-8 only specifies the minimum values of topographic
amplification factor ST for different cases, however it does not provide any clear guideline on
how to determine the value of ST.

As the topographic amplification factor is multiplied as a constant scaling factor with

the ordinates of the elastic design response spectrum, topographic amplification is considered
frequency-independent and applied separately to soil layer effects in the EC-8. It is also
Table of Contents

noted that the EC-8 prescribes that topographic amplification is applied “near the top edge” or
“near the top of the slopes” and hence, the distance to the free field is relatively small but not
clearly defined.

The PS-92 code specifies that “except if the effect of the topography on the seismic
motion is directly taken into account using a dynamic calculation based on a proper
idealization of the relief, a multiplying coefficient  called site response factor or topography
 55

factor will be used”. The PS-92 code provides an empirical method to determine the

Table of Contents
magnitude of the topographic factor, based on the gradients of the ground at upslope and
downslope sides of the topographic convex point, with a maximum value of 1.4. The PS-92
code also prescribes an empirical method to estimate the distances to the free field where
topographic effect can be neglected, according to which these distances rarely exceed the
height of the slope. Like the EC-8, the PS-92 code prescribes conditions for the geometric
dimensions of the slope, below which topographic effect of seismic ground motion should be
ignored. The provisions contained in the PS-92 code for topographic effects are summarized
in Figure 2.15.

3 Methodology and Parameters Adopted for This Study

3.1 Methodology

Table of Contents
A time-domain two-dimensional finite element parametric study using the Imperial
College Finite Element Program (ICFEP) (Potts & Zdravkovic, 1999) has been carried out to
evaluate the topographic effects at the crest of a soil slope overlying rigid bedrock, for
different bedrock depths (and hence site frequencies) and different slope inclinations, under
vertically propagating SV waves.

The parametric study has included incident waves with frequencies equal to the
fundamental and first harmonic frequencies of the one-dimensional soil column at free-field
behind the slope crest so that the topographic and soil layer effects can occur behind the slope
crest simultaneously and their interaction can be investigated.

In this study, the values of dimensionless frequency (H/) were varied by using input
motions of different predominant periods and the variation of topographic amplifications with
H/ for different slope inclinations and bedrock depths were obtained. This allows a direct

Table of Contents
comparison of the results obtained from this study with those from Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005).

3.2 Geometry of the Model

The geometry of the two-dimensional finite element model is presented in Figure 3.1.
A soil slope of 50 m in height (H = 50) was considered, and the width of the mesh was set at
1000 m (i.e. 20H) whereas the height of the mesh was varied and equal to the thickness of the
soil layer above the rigid bedrock (Z). In this analysis, three different thicknesses of the soil
layer were modelled, i.e. Z = 125 m, 250 m and 500 m.

The domain reduction method (DRM) developed by Bielak et al (2003) in conjunction

Table of Contents

with the standard viscous boundaries of Lysmer & Kuhlemeyer (1969) were used as the
boundary conditions in this study. This approach can approximate free-field conditions at
the lateral boundaries of the mesh.

The slope height, soil parameters and frequencies of the input motions were selected to
cover a sufficiently wide range of the normalised height of the slope (H/). Following
Ashford et al (1997) and Bouckovalas & Papadimitriou (2005), a range of H/ = 0.01 to 1.0
 Table of Contents
Topography factor  is required when H  10 m and i  I/3:

Table of Contents
=1 for I – i  0.40
 = 1 + 0.8 (I – i - 0.4) for 0.40  I - i  0.90
 = 1.40 for I – i  0.90

b = 20I or (H + 10)/4, whichever is the smaller

a = H/3

56
c = H/4

Table of Contents
Note: i = tangent of upslope angle
I = tangent of downslope angle

Table of Contents
Figure 2.15 Provisions for Topographic Effects in the French Seismic Code PS-92
 Table of Contents
Table of Contents
H
i

57
e

Table of Contents
Soil-rock interface

Table of Contents
Figure 3.1 Geometry of the Two-dimensional Finite Element Model and Domain Reduction Boundaries
 58

was studied. The values of H/ were varied by using input motions of different predominant

Table of Contents
periods, rather than changing the slope height or shear wave velocity, so that the same
geometry could be used and the same site periods of the soil layer could be maintained for
different analyses. In addition, the predominant period (Tp) of the input motions were
specified within the range from 0.1 to 10 s, which practically covers the large majority of
possible earthquake events.

Apart from vertical slope, analyses were also performed for slopes with inclinations
i = 45, 30 and 10, to study the effects of the slope inclination on the topographic
amplifications.

3.3 Parameters

In this parametric study, the soil was modelled as a linear elastic material, and thus

Table of Contents
shear modulus remained constant. The material damping is of the Rayleigh type and
therefore it is frequency-dependent. In order to have the same level of soil damping for all
frequencies of the input motions, the Rayleigh damping coefficients had to be specified
according to the input frequencies of the motions. A damping ratio () of 5% was adopted in
this analysis.

The soil properties adopted in this parametric study are the same as those used in
Bouckovalas & Papadimitriou (2005): shear-wave velocity, vs = 500 m/s, Poisson’s ratios
 = 1/3 and mass density  = 2000 kg/m3.

3.4 Time History Excitations

Table of Contents
As the ICFEP analyses use the time-domain method, the input excitations have to be
discretised to give a time history of acceleration to be introduced at the base of the mesh.
Obviously a too large time step will result in an inaccurate solution. In this study, a
time-step equal to 1/40 of the predominant period of the input motions was adopted in
discretisation of the input acceleration-time history to ensure that an accurate solution can be
obtained.

In order to investigate the effects of the frequency of input motions on topographic

amplifications, artificial input motions with single predominant frequency have to be used,
rather than actual seismic excitations which have very wide frequency spectra. In this
analysis, wavelets of Chang’s motion, presented in Equation 3.1 below, were used with the
aim of simulating the limited duration and the gradual rise and decrease of shaking amplitude
of actual earthquakes.
Table of Contents

 t   2t 
a t    e t sin   ............................................ (3.1)
 Tp 
 

where α,  and  are constants controlling the shape of the envelope of the acceleration-time
history and Tp is the predominant period of the motions.
 59

In this parametric study, the adopted predominant period of the Chang’s motion (Tp)

Table of Contents
ranges from 0.1 to 10 s, a range practically covering the large majority of possible earthquake
events. This range of Tp produced the range of H/ = 0.01 to 1.0 (Table 3.1), as all analyses
were performed for a slope of height H = 50 m and uniform shear wave velocity vs = 500 m/s.

There are two possible approaches of specifying the duration of the input motions.
The first one is to maintain the same number of cycles and hence to have different durations
for motion records of different predominant period. The other is to maintain the same
duration but vary the number of cycles for different motion records. In this investigation, the
first approach was adopted because it reflects better the nature of real earthquake records,
which show shorter durations when the motions have high frequency contents, i.e. shorter
periods.

By varying the values of α,  and  in Equation 3.1, all the input motions in this
investigation were adjusted to have a maximum amplitude of unity and 12 number of

Table of Contents
excitation cycles. An example of the acceleration-time histories of Chang’s motion used in
this study is shown in Figure 3.2.

The raw acceleration-time histories, given by Equation 3.1, had to be first corrected
using the program SeismoSignal v4.0 (SeismoSoft, 2010) developed by Seismosoft, and these
baselined acceleration-time histories were then introduced as the base excitations in the
ICFEP analyses.

Table 3.2 presents a summary of the different slope geometries and acceleration-time
history cases that have been analysed in this study.

3.5 Finite Element Methodology

Table of Contents
3.5.1 General

The parametric study was carried out with the Imperial College Finite Element
Program (ICFEP) using two-dimensional dynamic time domain analysis in plane strain. The
time-integration was performed with the generalised- method (Chung & Hulbert, 1993).

Mesh elements have to be small enough to give a sufficiently fine mesh so that the
motion of seismic waves, especially those with short wavelength, can be modelled with
sufficient accuracy. In this analysis, a maximum element dimension equal to λ/10 was
adopted, in line with the values discussed by Kontoe (2006). Therefore, input motion with
the shortest wavelength (i.e. largest frequency or shortest period) is the most critical to
determination of element dimension. As the shortest wavelength considered in this
investigation is 50 m, the maximum dimension of mesh elements in this analysis should be
Table of Contents

limited to 5 m. However, outside the middle half of the mesh, i.e. beyond 250 m (5H) from
the mid-point of the slope, the maximum horizontal dimension of mesh element was increased
to λ/5 (10 m) to reduce the total number of elements and hence computational time. This is
considered acceptable because very accurate modelling is relatively not important in this
region where the topographic irregularity is far away and in which topographic effects are
expected to be very small.
 Table of Contents
Table 3.1 Predominant Periods of Input Motion Selected for Analysis and the Corresponding Dimensionless Frequency Values

Predominant period, Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1

Table of Contents
Frequency, f (Hz) 0.1 0.25 0.5 0.75 1 1.5 2 3 5 10

Wavelength,  (m) 5000 2000 1000 666.7 500 333.3 250 166.7 100 50

60
Dimensionless frequency, H/

Table of Contents
0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1

Note: vs = Shear-wave velocity = 500 m/s

H = Height of slope = 50 m

Table of Contents
 Table of Contents
 2t 
a t    e t t  sin  
T 

Table of Contents
 p 

61

Table of Contents
Predominant Period (Tp) = 1 s
Parameters Used:  = 2,  = 1.5,  = 5

Table of Contents
Figure 3.2 An Example of the Acceleration-time Histories of Chang’s Motion Used as Input Excitation in This Study
 Table of Contents
Table 3.2 Summary of Analyses Performed in the Present Study and Selected Analyses from Tripe (2009) for Review in the Present
Study

Predominant period of input motion, Tp (s)

Bedrock Slope 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1

depth, Z inclination, i
(m) (degrees) Dimensionless frequency, H/

Table of Contents
0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1

90          

45 o  o  o  o o o o
125
30 o  o  o  o o o o

62
10 o  o  o  o o o o

90          

Table of Contents
45          
250
30          

10          

500 90          

Legends :
 Analysis performed in the present study  No analysis performed in the present study

Table of Contents
o Analysis performed in Tripe (2009) and the results reviewed in the present study
Analysis corresponding to the fundamental and second site periods of the soil column behind the slope crest
 63

3.5.2 Domain Reduction Method

Table of Contents
The domain reduction method is a two-step procedure. A detailed description of the
implementation and benefits of the domain reduction method is given in Kontoe et al (2009).

In this investigation, the first step of the DRM considered a one-dimensional finite
element model corresponding to the soil column extending to the bedrock behind the slope
crest (slope topography was not considered). A time history of horizontal acceleration was
applied along the bottom mesh boundary as the seismic excitation. Due to the finite element
node constraints imposed, no absorbing boundary condition can be specified at the bottom
boundary together with the excitation. Vertical displacements were restricted along the
bottom and lateral boundaries whereas horizontal displacements were restricted along the
bottom boundary after the end of the base excitation. During the step I analyses, the
incremental displacements were calculated at various depths of the one-dimensional model.
These were then used in the step II analyses to calculate the corresponding equivalent forces.

Table of Contents
The second step was performed on a two-dimensional model incorporating the slope
topography. The seismic excitation was directly introduced into the computational domain
in the form of equivalent forces calculated in the first step. These forces were applied to the
corresponding nodes of the step II model located within the boundaries e and , which are
schematically indicated on Figure 3.1. As the excitation is in the form of forces in the step II
analyses, absorbing boundary conditions can be applied at the bottom of the mesh together
with the excitation. In the two dimensional model, normal and tangential dashpots were
applied along the lateral boundaries, whereas vertical and horizontal displacements were
restricted along the bottom boundary of the mesh which represents the surface of rigid
bedrock. The dashpots and the domain reduction are combined to prevent reflections from
the lateral boundaries, by approximating free-field conditions at the lateral sides of the mesh.

Table of Contents
However, the above approach of using the DRM in the dynamic finite element analysis
results in a major limitation to the investigation of this study. As the one-dimensional finite
element analysis of the soil column extending to the bedrock behind the slope crest was used
in the DRM to calculate the equivalent forces which were introduced in the subsequent
two-dimensional analysis, the ground motions in front of the slope crest could not be
modelled accurately in the two-dimensional analysis and therefore they are not considered in
this study.

3.6 Procedures

In summary, the procedures of the parametric analysis in this investigation consist of

two stages, and are summarised as follow:
Table of Contents

Stage 1 – One-dimensional Modelling

The first stage of the analysis consists of the following steps:

(a) Generation, discretisation and baseline processing of the

acceleration-time history.
 64

(b) One-dimensional ICFEP dynamic analysis of the soil

Table of Contents
column behind the slope crest (step I of the DRM).

 The analysis results were used in the step II of the

DRM to calculate the equivalent forces to be applied in
the step II analyses.

 The one-dimensional analyses also gave the free-field

response of the ground, which is free from any
topographic effect.

 The results of the two-dimensional ICFEP dynamic

analyses would be normalised against this free-field
response for determination of the topographic
amplification factors.

Table of Contents
 
(c) Equivalent-linear site response analysis of the
one-dimensional soil column behind the slope crest, using
the EERA program and a soil damping ratio () of 5%.

 The EERA program version 2000 was developed by the

University of Southern California (USC, 2000).

 The results of EERA analysis were compared with the

results of the ICFEP one-dimensional dynamic analysis
to confirm that the Rayleigh coefficients used in the
ICFEP analysis gave the correct level of damping (5%).

Table of Contents
The above procedures were repeated for different selected parameters of slope
inclination (i), predominant period of input motion (Tp) and bedrock depth (Z).

Stage 2 – Two-dimensional Modelling

The second stage of the analysis consists of the following steps:

(a) Two-dimensional ICFEP static analysis to model removal of

soil material in front of the slope in a single excavation
stage.

(b) Two-dimensional ICFEP dynamic analysis (step II of the

DRM).
Table of Contents

 The equivalent forces calculated in the one-dimensional

ICFEP dynamic analysis (step I of the DRM) were
introduced in the two-dimensional dynamic analysis as
the seismic excitation.

 The two-dimensional analyses gave the response

(acceleration-time history) at different locations (nodes)
 65

on ground surface behind the slope crest. The peak

Table of Contents
acceleration values were normalised against the
corresponding free-field response obtained from the
one-dimensional analyses for determination of the
topographic amplification factors.

4 Parametric Study
4.1 Topographic Amplification Factors

In this parametric study, the topographic amplification factors at a particular point of

ground surface behind the slope crest were determined by the acceleration values at that point
normalised against the free-field horizontal acceleration of the ground behind the slope crest,
which is free from any topographic effects. The free-field response was obtained from the

Table of Contents
one-dimensional ICFEP dynamic analysis of the soil column behind the crest.

It is important to note that in this study, comparison is made between the acceleration
value at the point of interest and the free-field horizontal acceleration of the ground behind the
slope crest. This approach is the same as that adopted in Bouckovalas & Papadimitriou
(2005). If comparison is made directly with the magnitude of acceleration of the input
excitation, it would not be able to determine the topographic amplification factors because the
ground motion has been modified by both topographic and soil layer effects (i.e. topographic
and soil layer effects are coupled).

The topographic amplification factors, Ah and Av for horizontal and vertical motions
respectively, were calculated using the following simple equations:

ah a
and A  .......................................... (4.1)

Table of Contents
Ah 
a h , ff a h , ff

where ah and av are the peak horizontal and peak vertical accelerations at the point of interest;
ah,ff is the peak free-field horizontal acceleration.

It should be noted that ah,ff is used for normalisation of both ah and av, because the
free-field vertical acceleration, av,ff, is zero for a vertically propagating SV wave.

From Equation 4.1, it is important to note that topographic de-amplification (Ah < 1)
means that the horizontal acceleration value at the point of interest is smaller than that of the
free-field motion behind the slope crest. It does not necessarily mean that the horizontal
acceleration value is smaller than that of the input excitation at rock level.
Table of Contents

Similarly, the one-dimensional soil layer amplification factor, A1D, was determined by
the following equation:

a h , ff
A1D  ...................................................... (4.2)
a h,in

where ah,in is the peak acceleration of the input motion.

 66

When the topographic amplification factors, Ah and Av, are plotted against the distance

Table of Contents
along the ground surface behind the slope crest, the following trends of topographic effects on
seismic ground motion can be indentified:

(a) The horizontal ground motion behind the slope crest is

modified by topographic effects and becomes different from
the free-field response.

(b) Parasitic vertical motion is generated at the ground surface

behind the slope crest, even though the input motion, a
vertically propagating SV wave, has no vertical component.
This vertical component of seismic motion may become
comparable to the horizontal free-field motion in some cases.

(c) In some cases, the magnitudes of topographic effects on

Table of Contents
horizontal and vertical motions fluctuate with distance away
from the slope crest, resulting in zones of alternating
amplification and de-amplification of the horizontal motion
along the ground surface behind the slope crest (Figure 4.1).

(d) The magnitudes of topographic effects generally decrease

with distance away from the slope crest (Figure 4.1).

(e) When the wavelength of the input motion is very large

compared with the height of the slope (very small H/), the
topographic effects on the ground motion response are
insignificant (Figure 4.2).

Table of Contents
(f) By comparing the results obtained from the same
dimensionless frequency but different slope inclination and
bedrock depth, it can be seen that the pattern of topographic
effects depends primarily on the dimensionless frequency of
the input motion (H/) whereas the problem geometry (slope
inclination and bedrock depth) can only affect the magnitudes
of the topographic effects. One exception is that the bedrock
depth can also control the site periods of the soil column
behind the slope crest, leading to soil layer amplification which
can significantly influences the pattern of topographic effects.
 
(g) When the predominant frequency of input motion is equal to
the site periods of the soil column behind the slope crest, soil
amplification of the input motion occurs, and this results in a
Table of Contents

large horizontal free-field motion. In this case, the

magnitude of horizontal ground motion at the slope crest
becomes smaller than the horizontal free-field motion. In
other words, topographic de-amplification occurs. However,
the magnitude of horizontal ground motion at the slope crest
could still be larger than that of the input excitation. More
details of this finding will be given in the following sections.
 67

Table of Contents
Table of Contents
(a) Horizontal Amplification

Table of Contents

(b) Vertical Amplification

Table of Contents

Figure 4.1 Topographic Amplifications (Ah and Av) as a Function of Horizontal Distance
from Slope Crest (Z = 250 m, i = 45, Tp = 0.333 s, H/ = 0.3)
 68

Table of Contents
Table of Contents
(a) Horizontal Amplification

Table of Contents

(b) Vertical Amplification

Table of Contents

Figure 4.2 Topographic Amplifications (Ah and Av) as a Function of Horizontal Distance
from Slope Crest (Z = 250 m, i = 45, Tp = 10 s, H/ = 0.01)
 69

The values of the peak horizontal acceleration (ah) and peak vertical acceleration (av)

Table of Contents
at the slope crest obtained from the numerical analysis in this parametric study are presented
in Table 4.1. The values of the peak accelerations of the input motion (ah,in) and the peak
free-field horizontal accelerations (ah,ff) are also presented in the same table. From these
values, the one-dimensional soil layer amplification factor (A1D) and the topographic
amplification factors for horizontal and vertical motions (Ah and Av) can be determined using
Equations 4.1 and 4.2. Table 4.1 can illustrate the effects of dimensionless frequency (H/),
slope inclination (i) and bedrock depth (Z) on the topographic effects of seismic ground
motion and also the interaction between soil layer effects and topographic effects. These
will be discussed in detail in the coming sections.

4.2 Effects of Varying Dimensionless Frequency (H/)

4.2.1 General

Table of Contents
In order to allow a direct comparison of the results obtained from this investigation
with those from previous studies carried out by Ashford et al (1997) and Bouckovalas &
Papadimitriou (2005), the results of the analyses of this study are presented as a function of
the dimensionless frequency (H/).

The dimensionless frequency (H/) can be expressed in terms of the period of input
motion (T), slope height (H) and shear-wave velocity of soil (vs) as follow:

H fH 1 H
   .................................................. (4.3)
 s T s

According to Equation 2.6, the natural site periods of the soil layer with thickness Z is
given by:

Table of Contents
4Z
Tn  , n = 0, 1, 2, …∞
2n  1 s

Equation 4.3 becomes:

H

 2n  1 s 
H

2n  1 H , n = 0, 1, 2, …∞ .................... (4.4)
 4Z s 4Z

As a result, the values of H/λ corresponding to the fundamental (n = 0) and second

(n = 1) site periods of the soil layer can be determined.
Table of Contents

The one-dimensional soil layer amplification factor, A1D, was determined for different
bedrock depths, and its variation with the dimensionless frequency is presented in Figure 4.3.
It can be seen that in all cases, a sharp amplification peak occurs at the dimensionless
frequency corresponding to the fundamental site period of the soil column behind the slope
crest and also another peak of smaller magnitude occurs at the dimensionless frequency
corresponding to the second site period.
 Table of Contents
Table 4.1 Summary of the Results of the Numerical Analysis Performed in This Study (Sheet 1 of 5)

H/ 0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1
Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1
ah,in 1.005 - 0.999 - 0.982 - 1.002 1.003 1.005 1.005
ah,ff 1.017 - 1.435 - 7.105 - 1.146 2.426 1.527 0.922

Table of Contents
ah,crest 1.018 - 1.428 - 6.443* - 3.019 1.580 1.847 1.347

Z = 125 av,crest 0.004 - 0.065 - 1.199 - 3.328 2.285 2.129 1.490

i = 90º ah,crest/ah,in 1.013 - 1.429 - 6.560 - 3.012 1.576 1.839 1.341
A1D 1.013 - 1.436 - 7.235 - 1.143 2.419 1.520 0.918
Ah 1.000 - 0.995 - 0.907 - 2.634 0.651 1.210 1.461

70
Av 0.004 - 0.045 - 0.169 - 2.904 0.942 1.395 1.616
ah,in 1.005 - 0.999 - 0.982 - 1.002 1.003 1.005 1.005

Table of Contents
ah,ff 1.017 - 1.435 - 7.105 - 1.146 2.426 1.527 0.922
ah,crest - - 1.366 - 5.082* - 2.183 1.663 1.682 0.670

Z = 125 av,crest - - 0.004 - 0.175 - 1.865 1.407 0.583 0.395

i = 45º ah,crest/ah,in - - 1.368 - 5.175 - 2.178 1.658 1.675 0.667
A1D 1.013 - 1.436 - 7.235 - 1.143 2.419 1.520 0.918
Ah - - 0.952 - 0.715 - 1.905 0.685 1.102 0.727
Av - - 0.003 - 0.025 - 1.627 0.580 0.382 0.429

Table of Contents
 Table of Contents
Table 4.1 Summary of the Results of the Numerical Analysis Performed in This Study (Sheet 2 of 5)

H/ 0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1
Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1
ah,in 1.005 - 0.999 - 0.982 - 1.002 1.003 1.005 1.005
ah,ff 1.017 - 1.435 - 7.105 - 1.146 2.426 1.527 0.922

Table of Contents
ah,crest - - 1.355 - 4.824* - 2.127 1.545 1.892 1.273

Z = 125 av,crest - - 0.012 - 0.332 - 1.849 1.053 0.763 0.295

i = 30º ah,crest/ah,in - - 1.356 - 4.912 - 2.122 1.541 1.883 1.267
A1D 1.013 - 1.436 - 7.235 - 1.143 2.419 1.520 0.918
Ah - - 0.944 - 0.679 - 1.856 0.637 1.239 1.381

71
Av - - 0.008 - 0.047 - 1.613 0.434 0.500 0.320
ah,in 1.005 - 0.999 - 0.982 - 1.002 1.003 1.005 1.005

Table of Contents
ah,ff 1.017 - 1.435 - 7.105 - 1.146 2.426 1.527 0.922
ah,crest - - 1.368 - 5.028* - 1.732 1.582 1.728 1.020

Z = 125 av,crest - - 0.010 - 0.381 - 1.581 0.605 0.632 0.123

i = 10º ah,crest/ah,in - - 1.369 - 5.120 - 1.728 1.578 1.720 1.015
A1D 1.013 - 1.436 - 7.235 - 1.143 2.419 1.520 0.918
Ah - - 0.953 - 0.708 - 1.511 0.652 1.132 1.106
Av - - 0.007 - 0.054 - 1.380 0.249 0.414 0.134

Table of Contents
 Table of Contents
Table 4.1 Summary of the Results of the Numerical Analysis Performed in This Study (Sheet 3 of 5)

H/ 0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1
Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1
ah,in 1.005 1.005 0.999 1.004 0.982 1.001 1.002 1.003 1.005 1.005
ah,ff 1.057 1.424 7.227 1.845 1.123 2.494 1.433 1.256 0.921 0.433

Table of Contents
ah,crest 1.054 1.395 6.552* 2.655 1.849 2.069 2.374 1.592 1.036 0.642

Z = 250 av,crest 0.003 0.023 0.483 0.474 1.274 0.923 0.836 1.755 1.233 0.665
i = 90º ah,crest/ah,in 1.049 1.389 6.559 2.643 1.882 2.068 2.368 1.588 1.032 0.639
A1D 1.052 1.417 7.235 1.837 1.144 2.492 1.430 1.252 0.917 0.431
Ah 0.997 0.980 0.907 1.439 1.646 0.830 1.656 1.268 1.125 1.481

72
Av 0.002 0.016 0.067 0.257 1.135 0.370 0.583 1.398 1.339 1.535
ah,in 1.005 1.005 0.999 1.004 0.982 1.001 1.002 1.003 1.005 1.005

Table of Contents
ah,ff 1.057 1.424 7.227 1.845 1.123 2.494 1.433 1.256 0.921 0.433
ah,crest 1.052 1.379 6.192* 2.410 1.584 1.783 1.701 1.396 0.947 0.320

Z = 250 av,crest 0.001 0.001 0.080 0.215 0.775 0.708 0.697 0.517 0.250 0.199
i = 45º ah,crest/ah,in 1.047 1.372 6.199 2.399 1.613 1.782 1.697 1.392 0.943 0.318
A1D 1.052 1.417 7.235 1.837 1.144 2.492 1.430 1.252 0.917 0.431
Ah 0.995 0.968 0.857 1.306 1.411 0.715 1.187 1.112 1.029 0.738
Av 0.001 0.001 0.011 0.116 0.691 0.284 0.486 0.412 0.271 0.458

Table of Contents
 Table of Contents
Table 4.1 Summary of the Results of the Numerical Analysis Performed in This Study (Sheet 4 of 5)

H/ 0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1
Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1
ah,in 1.005 1.005 0.999 1.004 0.982 1.001 1.002 1.003 1.005 1.005
ah,ff 1.057 1.424 7.227 1.845 1.123 2.494 1.433 1.256 0.921 0.433

Table of Contents
ah,crest 1.051 1.375 6.123* 2.365 1.536 1.743 1.651 1.415 1.044 0.605

Z = 250 av,crest 0.001 0.004 0.144 0.241 0.705 0.688 0.800 0.311 0.276 0.154
i = 30º ah,crest/ah,in 1.046 1.368 6.130 2.355 1.564 1.741 1.647 1.412 1.039 0.603
A1D 1.052 1.417 7.235 1.837 1.144 2.492 1.430 1.252 0.917 0.431
Ah 0.995 0.966 0.847 1.282 1.367 0.699 1.152 1.127 1.133 1.397

73
Av 0.001 0.003 0.020 0.130 0.628 0.276 0.558 0.248 0.300 0.355
ah,in 1.005 1.005 0.999 1.004 0.982 1.001 1.002 1.003 1.005 1.005

Table of Contents
ah,ff 1.057 1.424 7.227 1.845 1.123 2.494 1.433 1.256 0.921 0.433
ah,crest 1.052 1.380 6.307* 2.397 1.502 1.754 1.718 1.396 1.007 0.477

Z = 250 av,crest 0.001 0.006 0.236 0.309 0.580 0.568 0.615 0.296 0.112 0.019
i = 10º ah,crest/ah,in 1.047 1.374 6.314 2.386 1.529 1.753 1.714 1.392 1.003 0.475
A1D 1.052 1.417 7.235 1.837 1.144 2.492 1.430 1.252 0.917 0.431
Ah 0.995 0.970 0.873 1.299 1.337 0.703 1.199 1.112 1.094 1.101
Av 0.001 0.004 0.033 0.168 0.517 0.228 0.429 0.236 0.121 0.044

Table of Contents
 Table of Contents
Table 4.1 Summary of the Results of the Numerical Analysis Performed in This Study (Sheet 5 of 5)

H/ 0.01 0.025 0.05 0.075 0.1 0.15 0.2 0.3 0.5 1
Tp (s) 10 4 2 1.333 1 0.667 0.5 0.333 0.2 0.1
ah,in 1.005 1.005 0.999 1.004 0.982 1.001 1.002 1.003 1.005 1.005
ah,ff 1.246 7.492 1.142 2.465 1.404 1.253 1.071 0.788 0.433 0.104

Table of Contents
ah,crest 1.235 7.455* 1.278 2.920 1.893 1.833 1.520 1.023 0.481 0.156

Z = 500 av,crest 0.005 0.224 0.256 0.298 0.293 0.865 0.867 0.912 0.584 0.157
i = 90º ah,crest/ah,in 1.229 7.419 1.280 2.907 1.928 1.831 1.516 1.020 0.479 0.155
A1D 1.240 7.456 1.144 2.454 1.430 1.252 1.068 0.786 0.431 0.103
Ah 0.991 0.995 1.119 1.185 1.348 1.463 1.420 1.298 1.111 1.496

74
Av 0.004 0.030 0.224 0.121 0.208 0.691 0.810 1.157 1.349 1.514

H = Height of slope = 50 m ah,in = Peak acceleration of input motion

Table of Contents
 = Wavelength of input motion ah,ff = Peak free-field horizontal acceleration
Tp = Predominant period of input motion ah,crest = Peak horizontal acceleration at slope crest
Z = Bedrock depth av,crest = Peak vertical acceleration at slope crest
i = Slope inclination A1D = One-dimensional soil layer amplification factor
Ah & Av = Topographic amplification factors for horizontal and vertical motions

Legend:
Analysis corresponding to the fundamental and second site periods of the soil column behind the slope crest
Notes: (1) Results of the analysis performed for (Z = 125 m, i = 45), (Z = 125 m, i = 30) and (Z = 125 m, i = 10) are extracted from

Table of Contents
Tripe (2009).
(2) Results shown with * refer to the largest peak horizontal acceleration at slope crest.
 Table of Contents
Table of Contents
75

Table of Contents
Table of Contents
Figure 4.3 Plots of Soil Layer Amplification Factor with Dimensionless Frequency, for Bedrock Depths of 125 m, 250 m and 500 m
 76

It has been shown that topographic amplification factors for both horizontal and

Table of Contents
vertical motions vary with the dimensionless frequency (H/λ). However, such relationship is
not the same for the three different bedrock depths analysed in this study.

4.2.2 Bedrock Depth Z = 250 m

The results of the horizontal response for the bedrock depth Z = 250 m are presented in
Figure 4.4. It is clear that the trend of topographic effects for the bedrock depth Z = 250 m
are quite different from those given in Ashford et al (1997) and Bouckovalas & Papadimitriou
(2005). The following main differences amongst the various studies are observed:

(a) For all slope inclinations, topographic de-amplification

(Ah < 1) occurs at the slope crest at H/ = 0.05 and 0.15. It
should be noted that this de-amplification is relative to the

Table of Contents
free-field response behind the slope crest and the horizontal
acceleration value at the slope crest is still larger than that of
the input excitation (i.e. the ratio of ah,crest/ah,in is still larger
than unity, see Table 4.1). On the other hand, both Ashford
et al (1997) and Bouckovalas & Papadimitriou (2005)
obtained topographic amplification at these two H/ values.
It is interesting to note that these H/ ratios correspond to
the fundamental and second site periods of the 250 m thick
soil column behind the slope crest.

(b) For all slope inclinations, the peak horizontal amplifications

occurs at H/ = 0.1. This is different from Ashford et al
(1997) and Bouckovalas & Papadimitriou (2005) who

Table of Contents
reported that peak amplification occurs at H/ = 0.2 in all
analysed cases. It is interesting to note that the peak
horizontal amplifications occur at an H/ ratio in-between
those corresponding to the fundamental and second site
periods of the 250 m thick soil layer.
 
(c) The magnitudes of the amplification peaks at H/ = 0.1
obtained in this study are much higher than those at H/ =
0.2 obtained from Ashford et al (1997) and Bouckovalas &
Papadimitriou (2005). For example, an amplification
factor of around 40% was obtained for 45 slope in this
study but the corresponding value at H/ = 0.2 given in
Ashford et al (1997) was only about 15%.
Table of Contents

The above discrepancies occur within the range of H/ ratios between 0.05 and 0.15,
which are the fundamental and second site periods of the 250 m thick soil layer. Outside this
range, the results of this study have similar trend and magnitudes of horizontal topographic
amplification compared with Ashford at al (1997) and Bouckovalas & Papadimitriou (2005).

The results of the vertical response for the bedrock depth Z = 250 m are presented in
Figure 4.5. Again, the results from Ashford et al (1997) for slopes in a homogeneous
 77

half-space are superimposed as background for comparison. Figure 4.5 shows that, for all

Table of Contents
slope inclinations, the vertical topographic amplifications at the slope crest at the fundamental
and second site periods of the 250 m thick soil column (H/ = 0.05 and 0.15) are much
smaller, compared with Ashford et al (1997) and Bouckovalas & Papadimitriou (2005).
De-amplification for vertical motion (Av < 0) is not possible according to its definition (see
Equation 4.1). In addition, there exist a sharp peak of the vertical topographic amplification,
for all slope inclinations, at H/ = 0.1, which is in-between those corresponding to the
fundamental and second site periods of the 250 m thick soil layer.

In summary, the trend and magnitudes of both horizontal and vertical topographic
effects at the crest of a soil slope overlying rigid bedrock are different from those for a slope
in a homogeneous half-space. It is reasonable to postulate that the differences are due to the
soil layer effects that have significantly affected the magnitude of the horizontal ground
response of the soil column, which is taken as the free-field response for evaluation of the
two-dimensional topographic effects.

Table of Contents
4.2.3 Bedrock Depth Z = 125 m

For bedrock depth Z = 125 m, only one analysis (vertical slope) was performed. The
analysis results are presented in Figures 4.6 and 4.7 for the horizontal and vertical response at
the slope crest respectively.

The horizontal response shows de-amplification (Ah < 1) when the predominant period
of the input motions is equal to the fundamental and second site periods of the soil column
behind the slope crest (H/λ = 0.1 and 0.3 in this case). Again, this de-amplification is
relative to the free-field response behind the slope crest and the horizontal acceleration value
at the slope crest is still larger than that of the input excitation (i.e. the ratio of ah,crest/ah,in is

Table of Contents
still larger than unity, see Table 4.1). Although the peak amplification for the bedrock depth
Z = 125 m occurs at H/λ = 0.2 which is in agreement with Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005), this result should be treated with caution. In fact, this
agreement is just a coincidence because H/λ = 0.2 is in-between the values corresponding to
the fundamental and second site periods of the 125 m thick soil layer.

In addition, the vertical topographic amplifications at the slope crest at the

fundamental and second site periods of the 125 m thick soil column are much smaller than the
values reported in Ashford et al (1997) and Bouckovalas & Papadimitriou (2005). A sharp
peak of the vertical topographic amplification also exists at H/ = 0.2, which is again
in-between those corresponding to the fundamental and second site periods of the 125 m thick
soil layer.

A comparison between the results for the bedrock depths Z = 250 m and Z = 125 m
Table of Contents

shows that they have the same pattern in the variations between the horizontal and vertical
topographic amplifications and the dimensionless frequency (H/), taking into account the
natural and second site periods of the soil column behind the slope crest. However, the
bedrock depth Z = 125 m has much larger magnitudes of both the horizontal and vertical
topographic amplification factors than the bedrock depth Z = 250 m.
 78

4.2.4 Bedrock Depth Z = 500 m

Table of Contents
Analysis for a vertical slope was also carried out for a much deeper bedrock depth Z =
500 m. Figures 4.8 and 4.9 show the analysis results for the horizontal and vertical response
at the slope crest respectively.

It is interesting to note that for bedrock depth Z = 500 m, the analysis results of both
the horizontal and vertical topographic amplifications obtained from this study are very
similar to those given in Ashford et al (1997) and Bouckovalas & Papadimitriou (2005). The
aforementioned differences due to the presence of soil/rock interface observed for bedrock
depths Z = 125 m and Z = 250 m are insignificant in this case. Therefore, it is reasonable to
conclude that with a bedrock depth Z = 500 m (or a ratio of the soil layer thickness to the
slope height (Z/H) equal to 10), the soil/bedrock interface has little influence on the
topographic effects on ground motion at the slope crest. In other words, the model behaves
like a homogeneous half-space even with an impedance contrast.

Table of Contents
There is one point concerning Figure 4.8 worth mentioning. Originally, the
horizontal amplification factor at H/ = 0.2 obtained in this study by using Chang’s signal
excitation is lower than the value expected from Ashford et al (1997). Considering that the
computational model adopted by Ashford et al in their study was based on steady-state
condition, another analysis for H/ = 0.2 was conducted using harmonic motion with the same
peak amplitude, predominant period and duration as the previously adopted Chang’s signal
excitation. The purpose of this additional analysis is to model the steady-state behaviour.
Interestingly, the results of this additional analysis using harmonic input motion fit very well
with the value given in Ashford et al (1997). With this additional analysis included, the
horizontal topographic amplification factor for bedrock depth Z = 500 m reaches maximum at
H/ = 0.2, and the peak value is around 1.5.

Table of Contents
The good match between the analysis results obtained in this study for bedrock Z =
500 m and those obtained from Ashford et al (1997) and Bouckovalas & Papadimitriou (2005)
for a slope in homogeneous half-space shows that although three different methodologies of
analyses are used (finite element, finite difference and generalised consistent transmitting
boundary methods respectively), they produce practically identical results. This helps to
verify the computational model adopted in this study.

4.3 Effects of Varying Slope Inclination (i)

It can be seen that with decreasing slope inclination, the magnitudes of both the
horizontal topographic amplification at the first peak and of the vertical topographic
amplification at the slope crest decrease.
Table of Contents

This is in agreement with the findings from Ashford et al (1997) and Bouckovalas &
Papadimitriou (2005).
 79

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The horizontal topographic amplification factor is determined by

normalising the horizontal acceleration value at that point against the
free-field horizontal acceleration (i.e. not against the acceleration of the
Table of Contents

input excitation).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space.

Figure 4.4 Horizontal Amplification at Slope Crest for a Vertically Incident SV Wave
on a Slope with a Bedrock Depth of 250 m
 80

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The vertical topographic amplification factor is determined by

normalising the vertical acceleration value at that point against the
free-field horizontal acceleration (i.e. not against the acceleration of the
Table of Contents

input excitation).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space.

Figure 4.5 Vertical Amplification at Slope Crest for a Vertically Incident SV Wave on a
Slope with a Bedrock Depth of 250 m
 81

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The horizontal topographic amplification factor is determined by

normalising the horizontal acceleration value at that point against the
free-field horizontal acceleration ( i.e. not against the acceleration of the
Table of Contents

input excitation ).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space .

Figure 4.6 Horizontal Amplification at Slope Crest for a Vertically Incident SV Wave
on a Slope with a Bedrock Depth of 125 m
 82

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The vertical topographic amplification factor is determined by

normalising the vertical acceleration value at that point against the
free-field horizontal acceleration (i.e. not against the acceleration of the
Table of Contents

input excitation).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space.

Figure 4.7 Vertical Amplification at Slope Crest for a Vertically Incident SV Wave on a
Slope with a Bedrock Depth of 125 m
 83

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The horizontal topographic amplification factor is determined by

normalising the horizontal acceleration value at that point against the
free-field horizontal acceleration (i.e. not against the acceleration of the
Table of Contents

input excitation).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space.

Figure 4.8 Horizontal Amplification at Slope Crest for a Vertically Incident SV Wave
on a Slope with a Bedrock Depth of 500 m
 84

Table of Contents
Table of Contents
Table of Contents

Notes: (1) The vertical topographic amplification factor is determined by

normalising the vertical acceleration value at that point against the
free-field horizontal acceleration (i.e. not against the acceleration of the
Table of Contents

input excitation).
(2) Background curves coloured black are extracted from Ashford et al
(1997) which are applicable to slopes in a homogeneous half-space.

Figure 4.9 Vertical Amplification at Slope Crest for a Vertically Incident SV Wave on a
Slope with a Bedrock Depth of 500 m
 85

4.4 Effects of Varying Bedrock Depth (Z)

Table of Contents
Previous parametric studies, including Ashford et al (1997) and Bouckovalas &
Papadimitriou (2005), considered slopes in a homogeneous half-space and thus they did not
consider the influence of bedrock depth on topographic effects.

From Table 4.1 and by comparing among Figures 4.4 to 4.9, it can be seen that the
bedrock depth has the following effects on the topographic effects on ground motion at slope
crest:

(a) The presence of a soil/rock interface results in soil layer

effects, and the bedrock depth controls the site periods of
the soil column behind the slope crest. Therefore, the
bedrock depth determines the dimensionless frequencies at
which amplification of the input motion due to soil layer

Table of Contents
effects occurs. This in turn affects the magnitudes of the
topographic amplification factors and also their variation
patterns with dimensionless frequency, because the
horizontal motion at the ground surface of the soil column,
which is subject to soil layer effects, is taken as the
free-field response for evaluation of the topographic
amplification.

(b) From (a) above, it is expected that when the predominant

period of the input motion is equal or close to the site
periods of the soil column behind the slope crest, very large
horizontal response at the ground surface of the soil column
is obtained. This large free-field response results in a

Table of Contents
horizontal topographic de-amplification (Ah < 1) and a small
vertical topographic amplification (Av) of the ground motion
at the slope crest. It should be noted that this horizontal
topographic de-amplification is relative to the free-field
response behind the slope crest and the horizontal
acceleration value at the slope crest is still larger than that of
the input excitation (i.e. the ratio of ah,crest/ah,in is still larger
than unity, see Table 4.1).

(c) Similarly, the bedrock depth also controls the dimensionless

frequency at which the peak horizontal and vertical
topographic amplifications occur. It is observed that these
peak topographic amplifications occur at an H/ ratio
in-between those corresponding to the fundamental and
Table of Contents

second site periods of the soil column behind the slope crest.
 
(d) With a very deep bedrock (Z = 500 m), the influence of the
soil/rock interface on the topographic effects becomes
insignificant.
 
(e) The magnitudes of horizontal and vertical topographic
 86

amplifications increase with decreasing bedrock depths, and

Table of Contents
much larger amplification factors are obtained in this study
compared with the ones given in Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005) who studied slopes in
a homogeneous half-space.

4.5 Interaction between Soil Layer Effects and Topographic Effects

It has been demonstrated that with the presence of a soil/rock interface, soil layer
effects and topographic effects can occur simultaneously, unless the bedrock is very deep (for
example, 10 times the slope height).

From Table 4.1 and Figures 4.3, 4.4 and 4.6, the following observations concerning the
interaction between topographic effects and soil layer effects can be made:

Table of Contents
(a) Even though the topographic amplification factors are
determined by normalising the two-dimensional analysis
results against the free field response, the soil layer effects
still influence the magnitudes of topographic amplifications
as well as the variation patterns of the topographic
amplifications with dimensionless frequency.

(b) The magnitude of peak amplification due to soil layer

effects is larger than that of peak horizontal topographic
amplification. For example, for bedrock depth Z = 250 m,
the maximum soil layer amplification factor is 7.2 (at H/ =
0.05) while the maximum horizontal topographic

Table of Contents
amplification factor is 1.65 (at H/ = 0.1 for vertical slope).
 
(c) The peak soil layer amplification and the peak horizontal
topographic amplification do not occur at the same
dimensionless frequency. In fact, when large amplification
(relative to the input excitation) due to soil layer effects
occurs, the predominant period of the input motion is equal
to the site periods of the soil column behind the slope crest
and topographic de-amplification of horizontal motion
(relative to the free-field response) (i.e. Ah < 1) occurs.

(d) Where there is little or no amplification due to soil layer

effects, the horizontal topographic amplification is larger for
the case of a soil slope overlying bedrock, than the case of a
Table of Contents

slope in a homogeneous half-space. The maximum

horizontal and vertical topographic amplification factors
increase with decreasing bedrock depth.
 
(e) Because the peak soil layer amplification and the peak
horizontal topographic amplification do not occur
simultaneously and the magnitude of peak amplification due
 87

to soil layer effects is larger than that of peak horizontal

Table of Contents
topographic amplification, it follows that the actual
magnitude of maximum horizontal ground acceleration at
the slope crest is dominated by soil layer effects (i.e. when
the predominant period of input motion is equal to the
fundamental site period of the soil column behind the slope
crest), rather than topographic effects. This refers to the
analysis results shown with * in Table 4.1.
 
(f) On the other hand, as vertical ground motion is generated by
topographic effects only, the actual magnitude of the
maximum vertical ground acceleration at the slope crest is
independent of soil layer effects and the site periods of the
soil column behind the slope crest.

Table of Contents
Based on the above, it can be concluded that the presence of a soil/rock interface
underneath slope toe has complicated the topographic effects compared to the simple case of a
slope in a homogeneous half space that has been studied by many researchers such as Ashford
et al (1997) and Bouckovalas & Papadimitriou (2005). This complex interaction between
topographic effects and soil layer effects suggests that these effects should be considered
simultaneously. It is noteworthy that this is in contrast to the conclusion made by Ashford
et al (1997).

5 Discussion
5.1 Comparison with Seismic Code Provisions

EC-8 recommends that the topographic amplification factor ST should be applied as a

Table of Contents
constant scalar factor to the elastic response spectrum, which implies that topographic effects
are considered separately to soil layer effects. This study has demonstrated, in principle, that
topographic effects and soil layer effects are not independent, and hence they should be
considered simultaneously due to their complex interaction.

This study has also shown that at an H/λ value in between those corresponding to the
fundamental and second site periods of the soil column behind the slope crest, very large
horizontal topographic amplification would occur, particularly when the bedrock is shallow.
In these circumstances, the topographic amplification factors recommended by the EC-8 may
be too low. However, it could be argued that as the peak soil layer amplification and the
peak horizontal topographic amplification do not occur simultaneously (in other words, these
two effects amplify different frequency ranges of the seismic motion), it may not be
reasonable to take into account the soil layer effect by adopting an elastic response spectrum
Table of Contents

that corresponds to a particular ground type as recommended by EC-8 and at the same time,
to using a large amplification factor to account for topographic effect.

It is of interest to note that if only the simple cases of slopes in a homogeneous

half-space or slopes with a very deep bedrock are considered, then the magnitudes of peak
horizontal topographic amplification obtained from previous numerical analyses (see
Section 2.5) and the current study are quite similar in order with the factors recommended by
 88

the seismic codes reviewed in this study (i.e. EC-8 and PS-92).

Table of Contents
It should be noted that this study has investigated the topographic effect on ground
surface accelerations and did not consider the acceleration response spectra. Therefore, the
direct comparison of the results from this study with the EC-8’s recommendations which
apply to the ordinates of elastic response spectra should be taken as approximate only.

As mentioned in Section 2.7, the EC-8 does not give clear guidance on the extent of
the area over which topographic effect should be considered, while the empirical relation
given in the PS-92 would generally result in a distance to the free field being less than the
height of the slope. However, according to the results of this study, both of these seismic
codes have apparently underestimated the distance from the slope crest to the free field where
topographic effect could be ignored.

Table of Contents
6 Conclusions

An extensive parametric study has been conducted to investigate the topographic

effects on seismic ground motion for a soil slope overlying rigid bedrock, by carrying out
time-domain finite element analyses using the Imperial College Finite Element Program
(ICFEP).

The parametric study has shown that the topographic effects for the simulated soil
slope overlying rigid bedrock are very similar in certain aspects with those reported in
previous studies for slopes in a homogeneous half-space (e.g. Ashford et al, 1997;
Bouckovalas & Papadimitriou, 2005). These similarities include:

(a) The horizontal ground motion behind the slope crest is

Table of Contents
modified by topographic effects and becomes different from
the free-field response.

(b) Parasitic vertical motion is generated at the ground surface

behind the slope crest, even with an input motion with no
vertical excitation. This vertical component of seismic
motion may become comparable to the horizontal free-field
motion in some cases.
 
(c) In some cases, zones of alternating amplification and
de-amplification of the horizontal motion along the ground
surface behind the slope crest are observed.
 
(d) The magnitudes of topographic effects generally decrease
Table of Contents

with distance away from the slope crest.

 
(e) In cases where the wavelength of the input motion is very
large compared with the height of the slope (the
dimensionless frequency H/ is very small), the topographic
effects on ground motion is insignificant.
 
 89

(f) With decreasing slope inclination, the magnitudes of both

Table of Contents
the horizontal topographic amplification at the first peak and
of the vertical topographic amplification decrease.

On the other hand, the soil/rock interface included in the model of this study has
complicated the topographic effects by introducing additional soil layer amplification. This
study has shown that compared with the topographic effects for slopes in a homogeneous
half-space obtained in Ashford et al (1997) and Bouckovalas & Papadimitriou (2005), the
inclusion of soil/rock interface has resulted in the following differences:

(a) Horizontal topographic de-amplification occurs at the slope

crest at dimensionless frequencies (H/λ) corresponding to
the fundamental and second site periods of the soil column
behind the slope crest, and therefore this pattern is
dependent on the bedrock depth. It must be re-iterated that

Table of Contents
this de-amplification is relative to the free-field response
behind the slope crest and the horizontal acceleration value
at the slope crest is still larger than that of the input
excitation. It is noted that at these H/λ values,
amplification of the input motion occurs due to soil layer
effects, and this results in a large horizontal free-field
motion and the horizontal ground motion is de-amplified at
the slope crest. The physical phenomenon explaining
occurrence of this horizontal topographic de-amplification
at the slope crest is not examined in this study. However, it
is logical to consider that the one-dimensional soil layer
amplification is a pure resonance at the free field, which is
interfered with at the crest because of the more complex

Table of Contents
geometry there. On the other hand, both Ashford et al
(1997) and Bouckovalas & Papadimitriou (2005) did not
obtain horizontal topographic de-amplification at any H/λ
values.

(b) Peak horizontal amplifications occurs at an H/λ value in

between those corresponding to the fundamental and second
site periods of the soil column behind the slope crest. On
the other hand, both Ashford et al (1997) and Bouckovalas
& Papadimitriou (2005) reported that the peak horizontal
amplification occurs at H/λ = 0.2 in all analysed cases.
 
(c) Similar to points (a) and (b) above, the vertical topographic
amplifications at the slope crest at dimensionless
Table of Contents

frequencies (H/λ) corresponding to the fundamental and

second site periods of the soil column behind the slope crest
are much smaller than the values reported in Ashford et al
(1997) and Bouckovalas & Papadimitriou (2005). In
addition, a sharp peak of the vertical topographic
amplification also exists at an H/λ value in between those
corresponding to the fundamental and second site periods of
 90

the soil column behind the slope crest.

Table of Contents
 
(d) For all slope inclinations, the maximum horizontal and
vertical amplification factors obtained in this study increase
with decreasing bedrock depths, and they are much larger
than those obtained from Ashford et al (1997) and
Bouckovalas & Papadimitriou (2005), who studied slopes in
a homogeneous half-space.

In addition, this study has the following findings concerning the interaction between
topographic effects and soil layer effects:

(a) The occurrence of soil layer effects can significantly

influence the magnitudes of the topographic amplifications
and the variation patterns of the topographic amplifications

Table of Contents
with the dimensionless frequency.

(b) The factor of peak amplification due to soil layer effects is

larger than that of peak horizontal topographic
amplification.
 
(c) Peak soil layer amplification and peak horizontal
topographic amplification do not occur at the same
dimensionless frequency.

(d) Maximum horizontal ground acceleration at the slope crest

appears when the peak soil layer amplification occurs, rather
than the peak horizontal topographic amplification. This

Table of Contents
suggests that the actual magnitude of the maximum
horizontal ground acceleration at the slope crest is
dominated by soil layer effects.
 
(e) As vertical ground motion can be generated from vertically
propagating SV wave by topographic effects only, the actual
magnitude of the maximum vertical ground acceleration at
the slope crest is controlled by topographic effects and not
affected by soil layer effects.

This study has shown that the presence of a soil/rock interface underneath the slope toe
has complicated the topographic amplification response compared to the simple case of slopes
in a homogeneous half space that have been studied by many researchers such as Ashford et al
(1997) and Bouckovalas & Papadimitriou (2005). The presence of a soil/rock interface has
Table of Contents

resulted in larger topographic amplification factors. In addition, topographic irregularities

and soil layering amplify different frequency components of seismic excitation. The
complex interaction between topographic effects and soil layer effects suggests that they
should not be handled separately for prediction of the overall site amplification factors. This
is in contrast to the conclusion made by Ashford et al (1997).

It should be noted that this parametric study considered a simplified model only.
 91

Some other parameters (e.g. type and direction of incident wave and three-dimensional

Table of Contents
topography shape) that can influence topographic effects on seismic ground motion were not
considered in this study. These factors can make actual topographic effects more
complicated than the ones predicted in this study.

7 References

AFNOR (1995). Règles de Construction Parasismique - Règles PS Applicables Aux

Bâtiments, Dites Règles PS 92. Normes NF P 016-013, Norme Francaise.

Ashford, S.A. & Sitar, N. (1997). Analysis of topographic amplification of inclined shear
waves in a steep coastal bluff. Bulletin of the Seismological Society of America, 87
(3), pp 692-700.

Table of Contents
Ashford, S.A., Sitar, N., Lysmer, J. & Deng, N. (1997). Topographic effects on the seismic
response of steep slopes. Bulletin of the Seismological Society of America, 87 (3),
pp 701-709.

Bard, P.Y. & Riepl-Thomas, J. (2000). Wave propagation in complex geological structures
and their effect on strong ground motion. In: Kausel E. and Manolis G. D. (eds.)
Wave Motion in Earthquake Engineering. Southampton, WIT Press, pp 37-95.

Bard, P.Y. & Tucker, B.E. (1985). Underground and ridge site effects: a comparison of
observation and theory. Bulletin of the Seismological Society of America, 75 (4),
pp 905-922.

Bielak, J., Loukakis, K., Hisada, Y. & Yoshimura, C. (2003). Domain reduction method for

Table of Contents
three-dimensional earthquake modelling in localized regions, part I: theory. Bulletin
of the Seismological Society of America, 93 (2), pp 817-824.

Boore, D.M. (1972). A note on the effect of simple topography on seismic SH waves.
Bulletin of the Seismological Society of America, 62 (1), pp 275-284.

Boore, D.M., Harmsen, S.C. & Harding, S.T. (1981). Wave scattering from a step change in
surface topography. Bulletin of the Seismological Society of America, 71 (1),
pp 117-125.

Bouchon, M. & Barker, J.S. (1996). Seismic response of a hill: the example of Tarzana,
California. Bulletin of the Seismological Society of America, 86 (1A), pp 66-72.

Bouchon, M., Schultz, C.A. & Toksoz, M.N. (1996). Effect of three-dimensional
Table of Contents

topography on seismic motion. Journal of Geophysical Research, 101 (B3),

pp 5835-5846.

Bouckovalas, G.D., Gazetas, G. & Papadimitriou, A.G. (1999). Geotechnical aspects of the
1995 Aegion, Greece, earthquake. In: Proceedings of the Second International
Conference on Geotechnical Earthquake Engineering, June 1999, Lisbon. Rotterdam,
Balkema, volume 2, pp 739-748.
 92

Bouckovalas, G.D. & Papadimitriou, A.G. (2005). Numerical evaluation of slope topography

Table of Contents
effects on seismic ground motion. Soil Dynamics and Earthquake Engineering, 25,
pp 547-558.

Bouckovalas, G.D. & Papadimitriou, A.G. (2006). Aggravation of Seismic Ground Motion
due to Slope Topograhy. In: Proceedings of the First European Conference on
Earthquake Engineering and Seismology, 3-8 September 2006, Geneva Switzerland.

British Standards Institution (2004). Eurocode 8: Design of Structures for Earthquake

Resistance. Part 5: Foundations, Retaining Structures and Geotechnical Aspects.
BS EN 1998-5:2004. London, British Standards Institution.

Celebi, M. (1987). Topographical and geological amplifications determined from

strong-motion and aftershock records of the 3 March 1985 Chile earthquake. Bulletin
of the Seismological Society of America, 77 (4), pp 1147-1167.

Table of Contents
Celebi, M. (1991). Topographical and geological amplification: case studies and
engineering implications. Structural Safety, 10, pp 199-217.

Celebi, M. (1995). Northridge (California) earthquake: Unique ground motions and

resulting spectral and site effects. In: Proceedings of the Fifth International
Conference on Seismic Zonation, 17-19 October, 1995, Nice.

Chavez-Garcia, F.J., Sanchez, L.R. & Hatzfeld, D. (1996). Topographic site effects and
HVSR. A comparison between observations and theory. Bulletin of the Seismological
Society of America, 86 (5), pp 1559-1573.

Chung, J. & Hulbert, G.M. (1993). A time integration algorithm for structural dynamics with

Table of Contents
improved numerical dissipation: the generalized- method. Journal of Applied
Mechanics, 60, pp 371-375.

Davis, L.L. & West, L.R. (1973). Observed effects of topography on ground motion.
Bulletin of the Seismological Society of America, 63 (1), pp 283-298.

Gazetas, G., Kallou, P.V. & Psarropoulos P.N. (2002). Topography and soil effects in the Ms
5.9 Parnitha (Athens) earthquake: the case of Adames. Natural Hazards, 27,
pp 133-169.

Geli, L., Bard, P.Y. & Jullien, B. (1988). The effect of topography on earthquake ground
motion: a review and new results. Bulletin of the Seismological Society of America,
78 (1), pp 42-63.
Table of Contents

Graizer, V. (2009). Low-velocity zone and topography as a source of site amplification

effect on Tarzana hill, California. Soil Dynamics and Earthquake Engineering, 29,
pp 324-332.

Griffiths, D.W. & Bollinger, G.A. (1979). The effect of Appalachian Mountain topography
on seismic waves. Bulletin of the Seismological Society of America, 69 (4),
pp 1081-1105.
 93

Hartzell, S.H., Carver, D.L. & King, K.W. (1994). Initial investigation of site and

Table of Contents
topographic effects at Robinwood Ridge, California. Bulletin of the Seismological
Society of America, 84 (5), pp 1336-1349.

Idriss, I.M. (1968). Finite element analysis for the seismic response of earth banks.
Journal of the Soil Mechanics and Foundations Division, ASCE, 94 (SM3), pp 617-636.

Idriss, I.M. & Seed, H.B. (1967). Response of earth banks during earthquakes. Journal of
the Soil Mechanics and Foundations Division, ASCE, 93 (SM3), pp 61-82.

Kawase, H. & Aki, K. (1990). Topography effect at the critical SV-wave incidence: possible
explanation of damage pattern by the Whittier Narrows, California, earthquake of
1 October 1987. Bulletin of the Seismological Society of America, 80 (1), pp 1-22.

Kontoe, S. (2006). Development of Time Integration Schemes and Advanced Boundary

Table of Contents
Conditions for Dynamic Geotechnical Analysis. Ph.D. Thesis, Imperial College,
London.

Kontoe, S., Zdravkovic, L. & Potts, D.M. (2009). An assessment of the domain reduction
method as an advanced boundary condition and some pitfalls in the use of
conventional adsorbing boundaries. International Journal for Numerical and
Analytical Methods in Geomechanics, 33, pp 309-330.

Kovacs, W.D., Seed, H.B. & Idriss, I.M. (1971). Studies of seismic response of clay banks.
Journal of the Soil Mechanics and Foundations Division, ASCE. 97 (SM2),
pp 441-455.

Kramer, S.L. (1996). Geotechnical Earthquake Engineering. New Jersey, Prentice-Hall.

Table of Contents
LeBrun, B., Hatzfeld, D. Bard, P.Y. & Bouchon M. (1999). Experimental study of the
ground motion on a large scale topography hill at Kitherion (Greece). Journal of
Seismology, 3, pp 1-15.

Levret, A., Loup, C. & Goula, X. (1986). The Provence earthquake of June 11th, 1909
(France): A new assessment of near-field effects. In: Proceedings of the Eighth
European Conference on Earthquake Engineering, Lisbon, September 1986, volume 2,
pp 4.2/79-4.2/86.

Lysmer, J. & Kuhlemeyer, R.L. (1969). Finite dynamic model for infinite media. Journal
of the Engineering Mechanics, ASCE, 95 (EM4), pp 859-877.

Nechtschein, S., Bard, P.Y., Gariel, J.C., Meneroud, J.P., Dervin, P., Cushing, M., Gaubert, C.,
Table of Contents

Vidal, S. & Duval, A.M. (1995). A Topographic Effect Study in the Nice Region. In:
Proceedings of the Fifth International Conference on Seismic Zonation, October 1995,
Nice France. Ouest Editions Presses Academiques, pp 1067-1074.

Ohtsuki, A. & Harumi, K. (1983). Effect of topography and subsurface inhomogeneities on

seismic SV waves. Earthquake Engineering and Structural Dynamics, 11,
pp 441-462.
 94

Pedersen, H.A., Sanchez-Sesma, F.J. & Campillo, M. (1994a). Three-dimensional scattering

Table of Contents
by two-dimensional topographies. Bulletin of the Seismological Society of America,
84 (4), pp 1169-1183.

Pedersen, H., LeBrun, B., Hatzfield, D., Campillo, M. & Bard, P.Y. (1994b). Ground-motion
amplitude across ridges. Bulletin of the Seismological Society of America, 84 (6),
pp 1786-1800.

Potts, D.M. & Zdravkovic, L. (1999). Finite Element Analysis in Geotechnical Engineering:
Theory. London, Thomas Telford.

Restrepo, J.I. & Cowan, H.A. (2000). The 'Eje Cafetero' earthquake, Colombia of January
25 1999. Bulletin of the New Zealand Society for Earthquake Engineering, 33 (1),
pp 1-29.

Table of Contents
Rogers, A.M., Katz, L.J. & Bennett, T.J. (1974). Topographic effects on ground motion for
incident P waves: a model study. Bulletin of the Seismological Society of America,
64 (2), pp 437-456.

Sanchez-Sesma, F.J. (1990). Elementary solutions for response of a wedge-shaped medium

to incident SH and SV waves. Bulletin of the Seismological Society of America,
80 (3), pp 737-724.

SeismoSoft (2010). SeismoSignal v4.0. [Online] Available from URL: http://seismosoft.com.

[Accessed 25 June 2010].

Sitar, N. & Clough, G.W. (1983). Seismic response of steep slopes in cemented soils.
Journal of Geotechnical Engineering, ASCE, 109 (2), pp 210-227.

Table of Contents
Sitar, N., Nova-Roessig, L., Ashford, S.A. & Stewart, J.P. (1997). Seismic response of steep
natural slopes, structural fills and reinforced soil slopes and walls. In: Seco e Pinto
(ed.) Proceedings of the 14th International Conference on Soil Mechanics and
Foundation Engineering, 6-12 September 1997, Hamburg. Rotterdam, Balkema.

Spudich, P., Hellweg, M. & Lee, W.H.K. (1996). Directional topographic site response at
Tarzana observed in aftershocks of the 1994 Northridge, California, earthquake:
implications for mainshock motions. Bulletin of the Seismological Society of
America, 86 (1B), S193-S208.

Tripe, R. (2009). Topographic Effects on Seismic Motion. M.Sc. Dissertation, Imperial

College, London.
Table of Contents

Tucker, B.E., King, J.L., Hatzfeld, D. & Nersesov, I.L. (1984). Observations of hard-rock
site effects. Bulletin of the Seismological Society of America, 74 (1), pp 121-136.

University of Southern California (2000). Equivalent-linear Earthquake site Response Analysis

(EERA) version 2000. [Online] Available from URL: http://gees.usc.edu/GEES/.
[Accessed 25 June 2010].
 GEO PUBLICATIONS AND ORDERING INFORMATION

Table of Contents
土力工程處刊物及訂購資料

A selected list of major GEO publications is given in the next 部份土力工程處的主要刊物目錄刊載於下頁。而詳盡及最新的

page. An up-to-date full list of GEO publications can be found at 土力工程處刊物目錄，則登載於土木工程拓展署的互聯網網頁
the CEDD Website http://www.cedd.gov.hk on the Internet under http://www.cedd.gov.hk 的“刊物”版面之內。刊物的摘要及更新
“Publications”. Abstracts for the documents can also be found at
刊物內容的工程技術指引，亦可在這個網址找到。
the same website. Technical Guidance Notes are published on
the CEDD Website from time to time to provide updates to GEO
publications prior to their next revision.

Copies of GEO publications (except geological maps and other 讀者可採用以下方法購買土力工程處刊物(地質圖及免費刊物

publications which are free of charge) can be purchased either 除外):
by:

Writing to 書面訂購
Publications Sales Unit, 香港北角渣華道333號
Information Services Department, 北角政府合署6樓626室
Room 626, 6th Floor, 政府新聞處

Table of Contents
North Point Government Offices,
333 Java Road, North Point, Hong Kong. 刊物銷售組

or 或
 Calling the Publications Sales Section of Information Services  致電政府新聞處刊物銷售小組訂購 (電話：(852) 2537 1910)
Department (ISD) at (852) 2537 1910  進入網上「政府書店」選購，網址為
 Visiting the online Government Bookstore at http://www.bookstore.gov.hk
http:// www.bookstore.gov.hk  透過政府新聞處的網站 (http://www.isd.gov.hk) 於網上遞交
 Downloading the order form from the ISD website at 訂購表格，或將表格傳真至刊物銷售小組 (傳真：(852) 2523
http://www.isd.gov.hk and submitting the order online or by
7195)
fax to (852) 2523 7195
 以電郵方式訂購 (電郵地址：[email protected])
 Placing order with ISD by e-mail at [email protected]

1:100 000, 1:20 000 and 1:5 000 geological maps can be 讀者可於下列地點購買1:100 000、1:20 000及1:5 000地質圖：
purchased from:

Map Publications Centre/HK, 香港北角渣華道333號

Survey & Mapping Office, Lands Department, 北角政府合署23樓
23th Floor, North Point Government Offices, 地政總署測繪處

Table of Contents
333 Java Road, North Point, Hong Kong. 電話: (852) 2231 3187
Tel: (852) 2231 3187
Fax: (852) 2116 0774 傳真: (852) 2116 0774

Requests for copies of Geological Survey Sheet Reports and 如欲索取地質調查報告及其他免費刊物，請致函：

other publications which are free of charge should be directed to:

For Geological Survey Sheet Reports which are free of charge: 免費地質調查報告:
Chief Geotechnical Engineer/Planning, 香港九龍何文田公主道101號
(Attn: Hong Kong Geological Survey Section) 土木工程拓展署大樓
Geotechnical Engineering Office, 土木工程拓展署
Civil Engineering and Development Department,
Civil Engineering and Development Building, 土力工程處
101 Princess Margaret Road, 規劃部總土力工程師
Homantin, Kowloon, Hong Kong. (請交:香港地質調查組)
Tel: (852) 2762 5380 電話: (852) 2762 5380
Fax: (852) 2714 0247 傳真: (852) 2714 0247
E-mail: [email protected] 電子郵件: [email protected]

For other publications which are free of charge: 其他免費刊物:

Table of Contents

Chief Geotechnical Engineer/Standards and Testing, 香港九龍何文田公主道101號

Geotechnical Engineering Office, 土木工程拓展署大樓
Civil Engineering and Development Department, 土木工程拓展署
Civil Engineering and Development Building,
101 Princess Margaret Road, 土力工程處
Homantin, Kowloon, Hong Kong. 標準及測試部總土力工程師
Tel: (852) 2762 5346 電話: (852) 2762 5346
Fax: (852) 2714 0275 傳真: (852) 2714 0275
E-mail: [email protected] 電子郵件: [email protected]
 MAJOR GEOTECHNICAL ENGINEERING OFFICE PUBLICATIONS

Table of Contents
土力工程處之主要刊物

GEOTECHNICAL MANUALS
Geotechnical Manual for Slopes, 2nd Edition (1984), 302 p. (English Version), (Reprinted, 2011).
斜坡岩土工程手冊(1998)，308頁(1984年英文版的中文譯本)。
Highway Slope Manual (2000), 114 p.

GEOGUIDES
Geoguide 1 Guide to Retaining Wall Design, 2nd Edition (1993), 258 p. (Reprinted, 2007).
Geoguide 2 Guide to Site Investigation (1987), 359 p. (Reprinted, 2000).
Geoguide 3 Guide to Rock and Soil Descriptions (1988), 186 p. (Reprinted, 2000).
Geoguide 4 Guide to Cavern Engineering (1992), 148 p. (Reprinted, 1998).

Table of Contents
Geoguide 5 Guide to Slope Maintenance, 3rd Edition (2003), 132 p. (English Version).
岩土指南第五冊 斜坡維修指南，第三版(2003)，120頁(中文版)。
Geoguide 6 Guide to Reinforced Fill Structure and Slope Design (2002), 236 p.
Geoguide 7 Guide to Soil Nail Design and Construction (2008), 97 p.

GEOSPECS
Geospec 1 Model Specification for Prestressed Ground Anchors, 2nd Edition (1989), 164 p. (Reprinted,
1997).
Geospec 3 Model Specification for Soil Testing (2001), 340 p.

GEO PUBLICATIONS

Table of Contents
GCO Publication Review of Design Methods for Excavations (1990), 187 p. (Reprinted, 2002).
No. 1/90
GEO Publication Review of Granular and Geotextile Filters (1993), 141 p.
No. 1/93
GEO Publication Foundation Design and Construction (2006), 376 p.
No. 1/2006
GEO Publication Engineering Geological Practice in Hong Kong (2007), 278 p.
No. 1/2007
GEO Publication Prescriptive Measures for Man-Made Slopes and Retaining Walls (2009), 76 p.
No. 1/2009
GEO Publication Technical Guidelines on Landscape Treatment for Slopes (2011), 217 p.
No. 1/2011
Table of Contents

GEOLOGICAL PUBLICATIONS
The Quaternary Geology of Hong Kong, by J.A. Fyfe, R. Shaw, S.D.G. Campbell, K.W. Lai & P.A. Kirk (2000),
210 p. plus 6 maps.
The Pre-Quaternary Geology of Hong Kong, by R.J. Sewell, S.D.G. Campbell, C.J.N. Fletcher, K.W. Lai & P.A.
Kirk (2000), 181 p. plus 4 maps.

TECHNICAL GUIDANCE NOTES

TGN 1 Technical Guidance Documents