You are on page 1of 56

Ground Motion Prediction Equations

Steve Kramer

Department of Civil and Environmental Engineering


University of Washington
Seattle, WA
Prediction of Prediction of
Ground Motions earthquakes

We can’t predict earthquakes


We can predict ground motions for different earthquake scenarios
Predictions are conditional upon occurrence of the scenario
Source Effects
Slip Distribution, Rise Time, and Slip Velocity
Slip on a fault is not uniform – varies spatially
Slip at a point occurs over time interval known as rise time
Rise time also varies spatially
Slip velocity = slip / rise time
Source Effects
2011 Mw = 9.0 Tohoku earthquake
Largest slip occurred near hypocenter
Shallow part of fault
Slow rise time (~90 sec)
Very low frequency (< 0.05 Hz)
Produced large rise in seafloor
Caused large tsunami
Strong ground motion affected by asperities
located closer to shore (red rectangles)
Less slip
Much shorter rise time – led to higher
frequencies
Path Effects
Geometric spreading

  R0  p1 Geometric spreading
   R  R1
  R of body waves
 p2
 R1 
Z  R    Z  R1    R1  R  R2
 R
 
p
R2  3
 Z  R2    R  R2 Geometric spreading
  R of surface waves

Amplitude decreases
as distance increases
Path Effects
Anelastic attenuation (damping)

  fR 
FP  R, f   Z  R  exp  
 Q  f V
 s ,crust 

“Quality factor –
1
Deff  f   accounts for material
2Q  f  damping and wave
scattering

With
saturation

At close distances, Rrup


may be very small, but
energy may be coming
from more distant part of
fault - saturation
Site Effects
Local ground response
Impedance effects
Stiffness decreases as waves approach ground surface
For constant energy transmission, lower stiffness requires greater
movement (displacement, velocity, acceleration)
Greater movement implies greater damping, which leads to lower
movement
Tradeoff between amplification effects of decreasing impedance and
de-amplification effects of increasing damping
Resonance effects
Sharp impedance contrasts can lead to resonance effects at specific
frequencies (periods)
Recall characteristic site period, Ts = 4H/Vs
Nonlinearity
Site Effects
Basin effects
Source below basin
Site Effects
Basin effects
Source outside basin
Site Effects
Basin effects Santa
Monica
Basin

Outside basin Inside basin


Site Effects
Basin effects

Outside
basin

Inside
basin

Outside
basin
Site Effects
Topographic effects

Scattering of waves

Different stratigraphies
Site Effects
Topographic effects
Site Effects
Topographic effects
Magnitude Scaling

Short-period
(high frequency)
Magnitude Scaling

Short-period Long-period
(high frequency) (low frequency)
Distance Measures
Distance Scaling

Short-period
(high frequency)
Distance Scaling

Short-period Long-period
(high frequency) (low frequency)
Hanging Wall Effects

Geometry
Hanging Wall Effects

Shaking is stronger
on hanging wall
Site Effects

FS  Flin  Fnl
Average shear
 VS 30  wave velocity
Flin  c ln 
 Vref 
in upper 30 m
 
Amount of amplification
varies with period

Softer
soils
amplify
more
Site Effects

FS  Flin  Fnl

 VS 30   xIMref  f 3 
Flin  c ln  Fnl  f 2 ln  
 Vref 
   f3 

High frequencies
decrease in
stronger shaking
Expectations

Larger magnitude leads to stronger motions

Larger distance leads to weaker motions

M Attenuation relationship
Prediction of ground motion
ln Y parameter for a particular earthquake
scenario

Ground Motion Prediction


Equation (GMPE)

log R
Ground motion prediction equations

GMPEs – treatment of uncertainty

Standard error - use to evaluate conditional probability


log lm

ln Y P[Y > Y*| M=M*, R=R*]

esln Y Y = Y*
ln Y
Epsilon (e) represents the number of
standard
M =deviations
M* a particular value of Y*
is above or below the median value of Y
ln Y* - ln Y
e sln Y
R = R* Me e > 0 if Y*
log> median
R Y
Mmax M
Ground motion prediction equations

GMPEs – treatment of uncertainty

Standard error - use to evaluate conditional probability

ln Y
P[Y > Y*| M=M*, R=R*]

ln Y
Y = Y*
e<0
M = M*

R = R* log R

M
Ground motion prediction equations

Presented in form of equation for median plus standard error term


Ground motion parameters assumed to be lognormally distributed
(logarithm is normally distributed; positively skewed; no probability
of negative value)
Equation produced by regression analysis – fitting curve(s)
through scattered data
Equation produces mean value of ln y (mln y)
mln y
Corresponding value of y is median value, yˆ  e
Predicted median value corresponds to “typical” conditions

Based on empirical data interpreted through regression analysis


Size of database has increased
Sophistication of regression procedures have increased
Model complexity has increased
Site Effects
GMPEs – Consideration of site effects

Based on large databases


of recorded motions –
many sites, many events

Relationships based on

Magnitude
regression analyses using
all data

Results correspond to
“average” conditions as
characterized by site
characteristic such as Vs30
Site Effects
GMPEs – Historical Evolution

Campbell (1981)
ln PGA  4.141  0.868 M  1.09 ln R  0.0606 exp( 0.7 M )
s lnPGA  0.37

Campbell and Bozorgnia (1994)


ln PGA  3.512  0.904 M  1.328 ln R 2SEIS  [0.149 exp( 0.647 M )]2
 1.125  0.112 ln RSEIS   0.0957 M F Style of faulting

 0.440  0.171ln RSEIS SSR  0.405  0.222 ln RSEIS SHR

Site factors
0.889  0.0691M M  7.4
s 
 0.38 M  7.4
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2009)


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2009)

Magnitude
term

Distance
term

Style of
faulting term
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2009)

Buried/surface
rupture term

Hanging wall/foot
wall term
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2009)

Site conditions term

Basin term

Required inputs:
M, R, style of faulting, rake angle, depth to top of rupture, dip angle of rupture
plane, hanging wall indicator, Vs30, depth to 2.5 km/sec shear wave velocity,
PGA on reference rock outcrop (Vs30 = 1,100 m/sec)
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (20014)

Magnitude term
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Geometric attenuation term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Style of faulting term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Hanging wall term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Shallow site response term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Basin response term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Hypocentral depth term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Rupture dip term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (20014)

Anelastic attenuation term


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Uncertainty (linear)
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Uncertainty (including nonlinearity)


Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

Total aleatory uncertainty

Between-event Within-event
uncertainty uncertainty
Uncertainty
Between-event and within-event components

Campbell and Bozorgnia NGA (2014)

ln IM Within-event
residual for
Event 1 at
site j, f1j

Event 1
median

ln R
Uncertainty
Between-event and within-event components

Campbell and Bozorgnia NGA (2014)

ln IM Within-event
residual for
Event 1 at
site j, f1j

Event 1
median
Event 2
median
Within-event
residual for
Event 2 at
site k, f2k

ln R
Uncertainty
Between-event and within-event components

Campbell and Bozorgnia NGA (2014)

Between-event residual
for Event 1, t1

ln IM Within-event
residual for
Event 1 at
site j, f1j

Between-event
residual for Event 2, t2 Event 1
median
Event 2 GMPE
median median
Within-event
residual for
Event 2 at
site k, f2k

ln R
Site Effects
GMPEs – Historical Evolution

Campbell and Bozorgnia NGA (2014)

0.1 g 0.3 g

0.5 g 0.7 g
Site Effects
GMPEs – Current Models

Active Crustal Regions (ACRs) – NGA West 2 models


Abrahamson, Silva, and Kamai (ASK) (2014)
Boore, Stewart, Seyhan, and Atkinson (BSSA) (2014)
Campbell and Bozorgnia NGA (2014)
Chiou and Youngs (2014)

All available at:

http://peer.berkeley.edu/ngawest2/final-products/
Site Effects
GMPEs – Current Models

Stable Continental Regions (SCRs) – NGA East model


Earthquakes occur in areas Central and Eastern United States
that are not on plate boundaries. Earth’s crust is harder and
more intact, so waves attenuate more slowly with distance
than in active crustal regions.

Available at:

http://peer.berkeley.edu/publications/peer_reports/re
ports_2017/2017_03_final_goulet_4.2.17.pdf
Site Effects
GMPEs – Current Models

Stable Continental Regions (SCRs) – NGA East model


Earthquakes occur in areas Central and Eastern United States
that are not on plate boundaries. Earth’s crust is harder and
more intact, so waves attenuate more slowly with distance
than in active crustal regions.
Site Effects
GMPEs – Current Models

Subduction Regions (ACRs) – NGA Subduction models


Under development

Current practice: Use BC Hydro GMPEs


Site Effects
GMPEs – Current Models

Subduction Regions (ACRs) – NGA Subduction models


Under development

Current practice: Use BC Hydro GMPEs

PEER NGA Subduction project is underway


Look for new subduction GMPEs in near future
Summary
• GMPEs make use of ground motion recordings to enable prediction of
ground motion intensity measures for specific earthquake scenarios
• Regression procedures are used to develop an expression for the median
intensity measure, and residuals with respect to the median are used to
characterize the uncertainty in the intensity measure
• The form of a GMPE is logically based on fundamental physical principles
in order to provide the most reliable prediction with the smallest number
of required coefficients in the regression analysis
• Ground motions are influenced by source, path, and site effects, and
GMPEs include terms to account for each
• Ground motion intensity measures are almost always lognormally
distributed, and the standard error, sln IM, can be broken down into within-
event and between-event components
• Different GMPEs are available for different tectonic environments

You might also like