C471 GEOHAZARDS

Seismic hazard:quake hazards & forecasting

C471 GEOHAZARDS

Structure of talk The range of seismic hazards

primary (e.g.ground shaking, surface rupture, landslides)

secondary (e.g. fire-following, toxic contamination)

Forecasting earthquakes definitions types of forecasting probabilistic forecasting

Earthquake prediction setting the scene

C471 GEOHAZARDS

Hazardous phenomena generated by earthquakes

Primary ground shaking ground lurching &

displacement ground settlement liquefaction landslides,

mudslides and avalanches

tsunami & seiches

Secondary structural

collapse fire-following falling material floods from dam

bursts & levée failures

C471 GEOHAZARDS

Ground shaking A number of aspects of

ground shaking affect severity of damage

peak ground acceleration

average ground acceleration

duration

Duration is critical 100 s quake would be

far more destructive than 30 s quake with similar shaking levels

Kobe (1995) - strong motion just 11 s long

lengthened to 100 s in areas of soft soilsKobe 1995

C471 GEOHAZARDS

More on duration of strong motion

Quake M affects duration much more than it affects max acceleration

Larger the M, the longer the fault rupture, and the larger the area from which seismic waves generated

Duration of shaking increases with distance from fault but intensity is less

Due to dispersion effects of seismic waves

Higher frequencies are attenuated more than lower ones

F

Intensity of shaking

Duration of shaking

Increasing distancefrom fault

C471 GEOHAZARDS

Duration and peak acceleration of strong motion

Magnitude Rupture length (km) Peak acceleration (%g) Duration (s) 5.0 9 2 5.5 5 – 10 15 6

6.0 10 – 15 22 12 6.5 15 – 30 29 18 7.0 30 – 60 37 24 7.5 60 – 100 45 30 8.0 100 – 200 50 34 8.5 200 – 400 50 37

C471 GEOHAZARDS

Local influences on ground shaking

Physical properties of soil and rock

competent bedrock may transmit peak accelerations > 2g

sands can transmit up to ~ 0.6g

gravels much higher clays only capable of

transmitting up to 0.15g

Geological structure focus waves

No simple correlation between M, acceleration,and distance from quake source

C471 GEOHAZARDS

More on ground conditions

Subsoil Average change in intensity Rock (e.g. granite, gneiss, basalt) - 1 Firm sediments 0 Loose sediments (e.g. sand, alluvium) + 1 We sediments, artificially filledground + 1.5

Ground conditions can also affectseismic intensity

C471 GEOHAZARDS

The clay problem (Mexico City 1985)

M s 7.9 quake

370 km away in Central American Trench

Little damage over intervening distance

High frequency ground motions attenuated out

Mexico City struck by low frequency ground motions close to natural vibration frequencies of underlying saturated lake-bed clays

Clays amplified motions up to 50 times compared to adjacent solid rock

Clay up to 40 m thick Damage correlated with

thicknessMexico City 1985

C471 GEOHAZARDS

Ground lurching & displacement

May be major problem close to source fault

Displacement may be vertical, horizontal or oblique

Movements may be very large

San Francisco 1906 in places horizontal

displacement of 6.5 m

Can be severely damaging for buildings close to faults

Particularly damaging for roads, railway lines, canals and pipelines that cross fault

San Francisco 1906

C471 GEOHAZARDS

Differential ground settlement

May involve uplift or subsidence

Often fault related Alaska 1964

shorelines uplifted by ~10m in places

subsided by ~2m in others

>250,000 sq km affected

May result in inundation by sea

Alaska; Izmit (1999)

Damage to ‘lifelines’

Izmit 1999

C471 GEOHAZARDS

Liquefaction

Sands and silts undergo temporary loss

of strength behave as viscous fluid

Seismic waves cause void collapse resulting in densification

Drainage of pore water cannot be achieved rapidly enough resulting in excessive pore pressures

End point is development of a QUICK condition

material behaves as heavy liquid with virtually no shear strength

Voidcollapse

Pore pressureincreasesresult

Originalsedimentstructure

C471 GEOHAZARDS

More about liquefaction When pore water pressures

exceed normal stress imposed by weight of sediment column the material behaves like a fluid

Water moves upwards from void spaces to surface forming sand boils

Particularly at risk loosely packed sands

and silts used for land fill & reclamation

Results of liquefaction foundering of buildings damage to utilities mass movements

Sand boils result fromexpelled water

C471 GEOHAZARDS

Examples of liquefaction

Niigata (Japan 1964) wholesale foundering

of apartment blocks

Kobe (Japan 1995) severe damage to

port facilities built on reclaimed land around Osaka Bay

Loma Prieta (Calif. 1989) liquefaction of sub-

surface wet sand layers in fill over 100y old caused serious damage in Marina District of SF

Niigata (Japan 1964)

C471 GEOHAZARDS

Maximum possible distances for liquefaction

M s 8.5

20 km

125 km

650 km

1500 km

M s 8

M s 7

M s 6

C471 GEOHAZARDS

Seismogenic mass movements Strong ground motion

commonly results in gravitational slides

Lituya Bay (Alaska 1958) large landslide triggered

60m surge with run-up of >500m

Liquefaction of material on a slope will trigger rapid flow failure

killed 200,000 in Kansu (China 1920) quake

Lateral spreads can be locally v.

damaging Alaska 1964; damaged

200 bridges

Lateral spreadSan Francisco 1906

C471 GEOHAZARDS

Lituya Bay (Alaska 1958)

C471 GEOHAZARDS

Other major seismogenic landslides

Huascaran (Peru 1970)

Sherman glacier(Alaska 1964)

C471 GEOHAZARDS

Seiches & tsunami

Seiches: oscillations set up in enclosed bodies of water by distant quakes

Observed in English lakes in 1950 due to Assam (India) quake

May cause damage to retaining walls of reservoirs and flooding

Tsunami: common feature of

submarine earthquakes; enhanced by submarine landslides

also from sub-aerial seismogenic landslides

Hilo (Hawaii) 1946

C471 GEOHAZARDS

Some notable seismogenic tsunami

Year Origin Run-up height (m) Damage to Comments 1755 Eastern Atlantic 5 - 10 Lisbon (Portugal) Reached Caribbean 1868 Peru & Chile > 18 South America

Hawaii Detected in New Zealand

1896 Honshu (Japan) 24 Sanriku coast (Japan) 26,000 killed 1908 Messina (Italy) 5 Messina & Reggio

Calabria >8,000 killed

1946 Aleutian Islands (Alaska)

17 Hawaii >150 killed

1964 Alaska 8.5 Crescent City (California)

>100 killed

1993 Hokkaido (Japan) 20 Okushiri Island >200 killed 1994 Java (Indonesia) 11 Malang >200 killed 1998 Offshore PNG >10 Aitape ~2,500 killed

submarine landslide

C471 GEOHAZARDS

Destructive capacity of tsunami High velocities

700 - 800 kph in deep water

impact velocities 50 kph or more

Long wavelengths 150 - 250 km

Wave period tens of minutes

Run-up heights in excess of 20m

Ocean-basin extent possible

Multiple waves in a tsunami ‘wave train’

Aitape (PNG) 1998

C471 GEOHAZARDS

Tsunamigenic earthquakes

<10% of submarine quakes generate tsunami

Magnitude 6.5 or greater Focal depth < 50km Most destructive have

depths of < 25 km Vertical uplift of large

area of sea bed Link between average

run-up and quake size May be strongly focused Submarine landslides

enhances tsunami potential

Submarine slide simulation(PNG 1998)

C471 GEOHAZARDS

Tsunami magnitude scale (Japan)

Richter Magnitude Tsunami Magnitude Maximum run-up (m) 6 - 2 < 0.3 6.5 - 1 0.5 – 0.75 7 0 1 – 1.5 7.5 1 2 – 3 8 2 4 – 6 8.25 3 8 – 12 8.5 4 16 – 24 8.75 5 > 32

C471 GEOHAZARDS

Structural collapse

‘Buildings not earthquakes kill people’

Structural collapse depends upon

how well buildings constructed

how well maintained magnitude, duration &

acceleration of strong ground motion

the ground response (type & character of underlying soil & rock)

distance from epicentre

Pancake collapse of concretebuilding (Gujarat 2001)

C471 GEOHAZARDS

More on structural collapse Structural collapse primarily

due to inflexible response of buildings

low tensile and shear strength

high rigidity low ductility low capacity for

redistributing loads

Particularly poor adobe (mud bricks) and unreinforced masonry

Common problem: not ‘tying’ walls and floors together

Lack of enforcement of codes

Collapse of rock and cementbuildings (Gujarat 2001)

C471 GEOHAZARDS

Solutions to structural collapse

Design buildings to dissipate vibrational energy by inelastic (plastic) response at key locations (beam-column joints)

Aim is to have buildings that are strong and flexible; and that yield but don’t collapse

Ensure and enforce appropriate building codes

Continuing problem in developing countries

C471 GEOHAZARDS

Earthquake loss-susceptibility by construction type

VI VII VIII IX X Adobe 8 22 50 100 100 Unreinforced masonry (NSD) 3.5 14 40 80 100

Reinforced concrete frames (NSD) 2.5 11 33 70 100 Reinforced masonry (NSD) 1.5 5.5 16 38 66 Steel frames (seismic design) 0.4 2 7 20 40

Reinforced masonry (high-quality seismic design) 0.3 1.5 5 13 25

NSD = Non-seismic design

Modified Mercalli Intensity

Average damage (%)

C471 GEOHAZARDS

Fire-following quake

Major problem in 20th century quakes

San Francisco 1906 Tokyo 1923

More recently Kobe 1995 Factors contributing to fires

wooden buildings narrow streets inadequate water provision fractured gas mains &

power lines car fires discarded cigarettes overturned stoves and

water heaters chemical and petroleum

leaks and spills strong winds

San Francisco 1906

C471 GEOHAZARDS

Fire following Japanese quakes

Great Kanto quake (Tokyo-Yokohama) 1923

killed at least 140,000 tens of thousands burnt

to death

Great Hanshin quake (Kobe) 1995

fires started in old, cramped parts of city

many wooden buildings 146 fires started 23,000 homes destroyed

Tokyo today ~ 1 million wooden

homes

Kobe

Tokyo

C471 GEOHAZARDS

Floods & dam bursts

New Madrid area liquefaction during

1811-12 quakes caused banks of Mississippi to fail

Levee failure could be a problem in next quake

Lower San Fernando dam (California)

failed during quake in 1971

12 s of strong shaking peak acceleration of ~

0.5g upstream section of

dam collapsed but held - just!

Lower San FernandoDam (California) 1971

C471 GEOHAZARDS

Forecasting earthquakes Forecasting is not prediction

less precise based upon analysis of earthquake return periods rather than

identification of pre-cursor y signs

Active faults or fault segments do not rupture in a random manner

they have characteristic return periods (or at least return period envelopes)

these reflect strain accumulation along the fault and the capacity of the fault to resist strain up to a given characteristic point - for that fault or fault segment

There are complications: Rupture will not occur according to a rigid timetable - there is a

return period envelope rather than specific date Strain may be released by one large quake or a number of smaller

ones (e.g. Marmara Sea south of Istanbul) this has implications for risk assessment

C471 GEOHAZARDS

San Andreas example

Prior to 1906 M 8.25 San Francisco quake ~ 3.2m displacement across fault in 50 years

Post-quake rebound on the fault was ~ 6.5m

Amount of time for strain released in quake to accumulate

(6.5/3.2) x 50 ~100 y

Return period until next comparable quake = 100y

Assumes uniform strain

accumulation quake did not alter fault properties

C471 GEOHAZARDS

Problems with forecasting

Forecasts only as good as the available catalogues

Historical catalogues good for well studied regions such as California, Japan, Europe, China

Poor for regions of low frequency-high magnitude seismicity

Cascadia subduction zone New Madrid Jamaica Western Europe

Catalogues need to go back further; requires geological studies Cascadia subd. zone

C471 GEOHAZARDS

The Seismic Gap concept

Defined as an area in an earthquake-prone region where there has been a below average level of seismic energy release

The 1989 Loma Prieta quake filled a gap that had been aseismic since 1906

Other gaps exist in Aleutian arc (Alaska) south of Istanbul Tokyo southern California

Istanbul seismic gap

C471 GEOHAZARDS

Seismic intensity forecasting

Other parameters can be usefully forecast than just timing of a quake

Forecasting seismic intensity at a particular site is vital for:

siting structures such as dams, schools, hospitals & emergency centres

constructing seismic hazard maps

Requires detailed information on geology, ground conditions

Seismic intensity forecast map - Tokai (Japan)

C471 GEOHAZARDS

Probabilistic forecasting

Most useful way of expressing a forecast of a future quake is in terms of probabilities

Most people are familiar with probabilities as a result of gambling

Example from San Francisco area (Bolt, 1999) 5 quakes > M = 6.75 in 155 y between 1836 & 1991 if events are random, another quake of 6.75 can be

expected in 155/5 y = 31 y with high probability

Problem: quakes not entirely random. On a particular fault system may be clustered (due to stress transfer) or follow certain trends

Alternative method of probabilistic forecasting is based on the ELASTIC-REBOUND model

Based upon estimates of strain accumulation across fault

C471 GEOHAZARDS

Strain measurement and forecasting

Geological mapping undertaken to define active fault segments

Assumption made that a discrete segment will rupture in one go

As Seismic moment links magnitude with rupture length this gives measure of maximum expected earthquake

Relationship between M s and fault rupture length L: M s = 6.10 + 0.70 log L

C471 GEOHAZARDS

Calculating probabilities

Next: determine slip history of each segment

Calculate strain accumulation rate for each segment

Slip history for fault segment can then be plotted against time

As slip is related to quake magnitude allows recurrence intervals between quakes greater than a given magnitude to be determined

Amountof slip

TimeMagnitude6

C471 GEOHAZARDS

The quake probability histogram

Quakefrequency

Recurrencetime

T1 T2

Construct histogram showing No. of quakes that occur with each specified recurrence time

Most probable recurrence interval is that which divides histogram into two equal areas

If time since last quake in the magnitude range is T1, the probability of the next quake occurring in T1 - T2 years = ratio of red area to yellow area

As recurrence time T2 increases ratio approaches 1 and a quake becomes virtually certain

The more consistent the recurrence time the better the forecast

T1 T2

C471 GEOHAZARDS

The quake probability histogram & the San Andreas

Suited to California & San Andreas fault system because active faults exposed at surface

Enables displacements to be measured easily and strain to be monitored

Method crucially depends on constraining well the number of potentially destructive quakes in historic time and their ages

For more discussion of problems see Bolt (1999) p228 - 229)

C471 GEOHAZARDS

Predicting earthquakes A highly controversial issue in seismology Involves giving a precise warning about the

timing and size of a future quake Reliant upon the occurrence of pre-cursory signs

in advance of a quake Method must be shown to be repeatable in order

to be of any use In a zone of high seismicity, any prediction is

going to have greater than chance than zero of being right

On the other hand - a prediction that is not fulfilled ensures that the method is invalid

C471 GEOHAZARDS

Proposed earthquake precursors

Changes in seismic velocities

Crustal deformation Groundwater changes Gas release Atmospheric effects Anomalous animal

behaviour Changes in magnetic

and electrical properties of the rocks

the so-called VAN method