Você está na página 1de 27

# Logic Tree Approach Used In

PSHA

Presented By

RAVINDRA K. GOLIYA

Introduction

manner.

problems.

## First introduced into probabilistic seismic hazard

analysis (PSHA) by Kulkarni et al. (1984)

## Seismic Hazard Analysis

Quantitative estimation of ground shaking hazards at a
particular site.

parameters.

## Ground motion parameters are determined using

predictive relationships.

## Probability of exceedence of ground motion

parameter during a particular time period is
found.

## Types of uncertainties involved in

SHA
There are two types of uncertainties that are
dealt with in SHA :

Aleatory uncertainty

Epistemic uncertainty

Aleatory Uncertainty
Uncertainties that are related to an apparent
randomness in nature.

## It is easy to incorporate such variability directly

into the hazard calculations.

## Ground motion prediction equations are the

most important aleatory uncertainty in SHA.

Epistemic Uncertainty
Uncertainties due to insufficient knowledge are
known as epistemic uncertainties.

## The main epistemic uncertainties involved in any

SHA:
Characteristics of the seismic source zones
The model for the recurrence relationship.
The maximum earthquake magnitude.

Logic Tree
Logic tree contains a series of nodes and

branches.
Each node represents an assessment of a state
of nature or an input parameter.

## Branch represents one possible discrete

alternative for the state of nature.

## Logic Tree contd

branches are placed
parameter estimation.

a sufficient number of
at a given node to
the uncertainty in the

## Logic Tree Weights

Probabilities that represent the relative
likelihood or degree of belief that the branch
represents are known as weights.

judgments.

## Logic Tree Weights contd

Vick (2002) defines :
Probability as a measure of stable
frequency.

An expression of belief.

## Features of Logic Tree Weights based on

frequency-based probabilities
1. Weights at a node must sum to unity.
2. Events for which they are defined must be
exhaustive.
3. Events for which they are defined must be
mutually exclusive.
4. Must represent the relative rate at which
alternative events are occurring.

degree of belief

1.

2.

alternatives.

3.

## Options on the branches of a logic tree will not

be mutually exclusive.

4.

## Weighted average of the hazard values does

not give the expected value.

## How to assign weights?

General considerations

## The degree of dependence between branches of

the logic tree should also be considered.

## Candidate equations should be grouped into two

categories intrinsic and application specific.

## Intrinsic factors - confidence of the user in a

particular equation.

## The application-specific characteristics can be

divided into those

## related to the conventions adopted for the hazard

calculations.
related to the specific environment for which the SHA
is performed.

## Ranking and Combining Criteria for Logic

Tree Weights
a hierarchy for the influence of each factor on the
final weighting.

## a transparent system for combining the grading

for the different factors into a single weight.

## Logic Tree for New Madrid Seismic

Zone: A Case Study
The area considered
in the study is, as
figure, between 35oN
to 38oN latitude and
88oW
to
92oW
longitude.
Different
faults and position of
past earthquakes are
also shown in the
same figure.

Seismic Zone.

## Using a logic tree to incorporate model

alternatives into a seismic hazard analysis
leads to a distribution of possible seismic
hazard results. This distribution can be
represented by a mean seismic hazard map
and a coefficient of variation (COV) map as
shown in the figures below.

relations.

## Multiple opinions are not uniform data sets and

cannot be averaged.

## Logic trees make sense

earthquake risk analysis.

when

used

in

## Problems with Logic Tree approach

Problems with Logic Tree approach as listed by E.
L. Krinitzsky (1995):
Other attenuations could have been used in logic
tree.

than guesses.

## Maximum magnitudes with the percentage

weightings could have been made into an almost
infinite series of weighted values.

## Recurrence rate and b-values are totally

unsuitable for generating design values.

Conclusion

## Best way to address procedure related, or

epistemic uncertainties, is using Logic Trees.

## Ground motion models used should very closely

represent the future ground motion.

## Importance of the weights decreases as the

number of branches used in Logic Tree
increases.

## Issues rose by different researchers in their

papers should also be kept in mind.