You are on page 1of 11

Gutenberg-Richter Law

logN(M)=a-bM
Introduction

Earthquake magnitude number/per year


>9.0 0
8~8.9 1
7~7.9 18
6~6.9 108
5~5.9 800
4~4.9 6200
3~3.9 49000
2~2.9 300000
Dashed line: N is the number of earthquake
per year of magnitude= M M / 2
Solid line: N is the number of earthquake per
year of magnitude M
Gutenberg - Richter Relation

logN vs. M are typically modeled with a


Gutenberg - Richter relation:
logN(M)=a-bM
M is the earthquake magnitude
N is the number of earthquake per
year of magnitude.
a is called the productivity
b is called the b-value, and is typically
in the range of 0.8-1.1.
The b-value has served as a kind of
tectonic parameter.
1.8-1.0 oceanic ridge
1.0-0.7 interplate
0.7-0.4 intraplate
Using the GR Relation

Suppose b=1, and you are given that there is


one M 5.0+ earthquake per year in the
region. How often does an M 7.0+ occur?
logN(M)=a-bM
log(1)=a-(1)(5) --> a=5.0
logN(7)=5-(1)(7)=-2
N(7)=10-2=0.01/year
Some subtleties with GR
Need to distinguish between N(M) and n(M)
N(M) is a cumulative curve, giving the number of
earthquakes of magnitude M or larger per year.
n(M) is incremental, such that n(M)dM gives the
number of earthquakes in a magnitude range of
width dM, centered on M
dN ( M )
They are related: n ( M ) =
dM
Watch notation: some authors are not careful
with which symbol they use.
How to find the b-value?
Two ways.
Count n(M) or N(M), then fit a least-squares line
to logN or logn vs M.
The alternative is the maximum likelihood
method: log10 e 0.434
b= =
(M M min ) (M M min )
Mmin is the smallest earthquake in the catalog.
M is the average.
Minimum magnitude is 4.8.
Average magnitude: 5.15
Application to find the b-value:
1966 data
Least squares: We had b=0.84
Maximum likelihood:
0.434 0.434
b= = =1.24
(M Mmin) (5.15 4.80)

Dont be surprised if the two methods disagree.


Maximum likelihood puts more weight on smaller
magnitudes.
But the two approaches usually give closer results
for larger data sets.

You might also like