Induced earthquake magnitudes are as large as (statistically) expected.
Nicholas van der Elst, U.S. Geological Survey Morgan Page, U.S. Geological Survey
Thomas Goebel, UC Santa Cruz and Caltech Deborah Weiser, USGS and UCLA
S. Mehran Hosseini, USC
Proto-OEF: one-year shaking forecast, including induced seismicity
Dallas
Denver
Washington
Atlanta
New
Orleans
Boston
Minneapolis
Raleigh
Lincoln Columbus
Oklahoma
City
Nashville
Austin
Phoenix
Sacramento
Denver
Seattle
Based on natural andinduced earthquakes
Based on the presumptionearthquakes occur naturally
Modified Mercalli Intensity
VIII+ Shaking severe, heavier damage
VII Shaking very strong, moderate damage
VI Shaking strong, felt by all, minor damage
V Shaking moderate, felt indoors by most, outdoors by many
IV Shaking light, felt indoors by many, outdoors by few
III Shaking weak, felt indoors by several
USGS Forecast for Ground Shaking Intensity from Natural and Induced Earthquakes in 2016
USGS map displaying intensity of potential ground shaking from natural and human-induced earthquakes. There is a small chance (one percent) thatground shaking intensity will occur at this level or higher. There is a greater chance (99 percent) that ground shaking will be lower than what is displayedin these maps.
Petersen et al., 2015
1 % probability of exceedence
Dependence on maximum magnitude
Mmax ≈ 7
Mmax = 6
• Are induced earthquake magnitudes restricted by injection volume?
• Or, are induced earthquake magnitudes determined by the pre-stress and connectivity of the fault network?
• are the observed sizes of induced earthquakes just a consequence of sampling the GR distribution?
Is there evidence that induced earthquakes are smaller than tectonic?
Tectonic earthquakes are not confined to the structure on which they initiate
2002 M7.9 Denali Earthquake
Other examples: El Mayor-Cucapah
Landers
Eberhart-Phillips… 2003
A classic question: Does Gutenberg-Richter reflect the
distribution of fault sizes, or the statistics of earthquake growth and
arrest on a complex, inter-connected fault network?
Cautionary Tale: Remote triggering
Velasco… 2008, Parsons… 2014
Denali remote triggering Global stack
3000 km
But then: 2012 M8.6 East-Indian Ocean
Pollitz…, 2012
Rate of M≥5.5 earthquakes
days since M8.6 event
remotely triggered M≥5.5 earthquakes
USGS
Could we make the same mistake with induced earthquakes?
Dallas
Denver
Washington
Atlanta
New
Orleans
Boston
Minneapolis
Raleigh
Lincoln Columbus
Oklahoma
City
Nashville
Austin
Phoenix
Sacramento
Denver
Seattle
Based on natural andinduced earthquakes
Based on the presumptionearthquakes occur naturally
Modified Mercalli Intensity
VIII+ Shaking severe, heavier damage
VII Shaking very strong, moderate damage
VI Shaking strong, felt by all, minor damage
V Shaking moderate, felt indoors by most, outdoors by many
IV Shaking light, felt indoors by many, outdoors by few
III Shaking weak, felt indoors by several
USGS Forecast for Ground Shaking Intensity from Natural and Induced Earthquakes in 2016
USGS map displaying intensity of potential ground shaking from natural and human-induced earthquakes. There is a small chance (one percent) thatground shaking intensity will occur at this level or higher. There is a greater chance (99 percent) that ground shaking will be lower than what is displayedin these maps.
• Are induced earthquake magnitudes restricted by injection volume?
• Or, are induced earthquake magnitudes determined by the pre-stress and connectivity of the fault network?
• are the observed sizes of induced earthquakes just a consequence of sampling the GR distribution?
Is there evidence that induced earthquakes are smaller than tectonic?
Re-evaluating the limit* on the magnitudes of induced earthquakes.
(*Other than the tectonic limit)McGarr, 2014
basic assumptions1. There are faults nearby
2. faults are stressed within one stress drop of failure
3. formation already saturated (∆P~∆V)
4. Gutenberg-Richter
5. Earthquakes are confined to the fluid perturbation
An analogy with aftershocks (triggered earthquakes)
• Suppose the total moment of an aftershock sequence were limited to the moment of the mainshock.
• Deterministic application of G-R implies that the largest aftershock should have 1/2 the moment of the mainshock.
• In magnitude:
1) ΔMas = 0.2 is not consistent with Båth’s law (too large in the expectation).
2) ΔMas = 0.2 is regularly exceeded (too small in the limit).
Mms −Mas(1) = 2
3log10
M 012M 0
⎛⎝⎜
⎞⎠⎟≈ 0.2
what went wrong?• Gutenberg-Richter is probabilistic. The number of aftershocks
nucleated scales with the mainshock stress perturbation
• But the sizes of the aftershocks are determined by the pre-stress and the tectonics (or whatever), not the size of the triggering perturbation.
Felzer…, 2006
Num
ber o
f trig
gere
d M
3.5+
King…, 1994
the moment-cap hypothesis1. There are faults nearby
2. faults are stressed within one stress drop of failure
3. formation already saturated (∆P~∆V)
4. Gutenberg-Richter
5. Earthquakes are confined to the fluid perturbation
the sample-size hypothesis1. There are faults nearby
2. faults are stressed within one stress drop of failure
3. formation already saturated (∆P~∆V)
4. Gutenberg-Richter (probabilistic)
5. Earthquakes are confined to the fluid perturbation
Tests of the sample size hypothesis
1. The maximum observed magnitude is proportional to the log of the number of prior induced events.
2. The order of occurrence of the largest earthquake is random within the sequence.
3. Bonus test: injected volume controls number not moment of induced eathquakes.
Quick sanity check1. Magnitudes are distributed Gutenberg-Richter. 2. No upper limit. 3. Number is proportional
to volume injected.
M̂max =1b
Σ + log10V( )
Pawnee?
expected distribution of Mmax
1) Gutenberg Richter
N M( ) = N10−b M−Mc( )
2) N =1 for Mmax (the mode)
M̂max = Mc +1blog10 N
Probability densities (cdf)
Mmax
2 3 4 5 6 7 8
prob
abilit
y de
nsity
0
0.5
1
1.5 3 10 30 100 3001000
Mmax-Mmode
-2 -1 0 1 2 3 4
prob
abilit
y de
nsity
0
0.5
1
1.5 3 10 30 100 3001000
Mmax
2 3 4 5 6 7 8
prob
abilit
y de
nsity
0
0.5
1
1.5 3 10 30 100 3001000
Mmax-Mmode
-2 -1 0 1 2 3 4
prob
abilit
y de
nsity
0
0.5
1
1.5 3 10 30 100 3001000
1) Gutenberg Richter
N M( ) = N10−b M−Mc( )
2) N =1 for Mmax (the mode)
M̂max = Mc +1blog10 N
3) correction for using N prior to Mmax
Probability densities (cdf)
ˆ ′Mmax = M̂max + log10 e( )
expected distribution of Mmax
GR statistics of each sequence• Magnitude of
completeness Mc
• b-value (excluding largest event)
• Number above Mc
Mag - Mc-2 -1 0 1 2 3
Nor
mal
ized
Num
ber
10-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
101
b = 1
KTB
BUK
GOK
SZ1
SZ2
SZ3
SZ4
BAS
A87
CB1
A01
YOH
PDX
R11
GUY
POH
TTX
RMA
POK
HAR
M̂max = Mc +1blog10 N
Observed vs. expected magnitudesMexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Mmax vs. Nprior
Taking into account Mc only
Mexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Mmax vs. Mexpected
Taking into account Mc, b, Nprior
M̂max ∝ log10 Nref M̂max = Mc +1blog10 N
Mmax - Mml-1 0 1 2 3 4
Prob
abilit
y D
ensi
ty
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Mmax - Mml
-1 0 1 2
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.93
Nprior/Ntot0 0.2 0.4 0.6 0.8 1
Inst
ance
s
0
1
2
3
4
5
N/Ntot
0 0.5 1
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.3
Test #1: Mmax is consistent with Nprior
Mmax vs. Mpredicted
Taking into account Mc, b, Nprior
Naive prediction
Conditional prediction
Distribution of difference between Mobs and Mpredicted
Mexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Mmax vs. Mexpected
A+
(N-1)/Ntot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFWBASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOK HAR
Vol/Voltot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFW BASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOKHAR
Mmax - Mml-1 0 1 2 3 4
Prob
abilit
y D
ensi
ty
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Mmax - Mml
-1 0 1 2
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.93
Nprior/Ntot0 0.2 0.4 0.6 0.8 1
Inst
ance
s
0
1
2
3
4
5
N/Ntot
0 0.5 1
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.3
Test #2: order of occurrence is random
13 out of 21 in latter half (10.5 expected)
C-Nprior/Ntot
(N-1)/Ntot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFWBASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOK HAR
Vol/Voltot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFW BASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOKHAR
Mmax - Mml-1 0 1 2 3 4
Prob
abilit
y D
ensi
ty
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Mmax - Mml
-1 0 1 2
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.93
Nprior/Ntot0 0.2 0.4 0.6 0.8 1
Inst
ance
s
0
1
2
3
4
5
Unb
iase
d In
stan
ces
1
2
3
4
5
N/Ntot
0 0.5 1
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.3p = 0.68
13 out of 21 in latter half (10.5 expected) excluding cases shut down because of large quakes: 9 out of 16 in latter half (8 expected)
Test #2: order of occurrence is random
A-
C-Nprior/Ntot
Test #3: Number proportional to volumeMexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Number vs. Volume
Taking into account Mc
log10(Npred/Ntot)10-1 100 101
Cum
ulat
ive
Frac
tion
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1n = 0.5n = 1n = 1.5
Scaling power n0 0.5 1 1.5 2
Varia
nce
0.2
0.3
0.4
0.5
Ntot
N0
= VtotV0
⎛⎝⎜
⎞⎠⎟
n
n=1
A-
1. faults are distributed uniformly over one stress drop from failure.
N ∝ ∆P
Cochran…, 2004
Consistent with tidal triggering of earthquakes
Mexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Why should number of triggered earthquakes scale with volume?
Why should number of triggered earthquakes scale with volume?
1. faults are distributed ~uniformly over one stress drop from failure.
N ∝ ∆P
2. nucleation sites are distributed ~uniformly over the volume.
N ∝ ∆PV
3. Pressure proportional to injection volume over reservoir volume.
∆P ∝ ∆V/V
N ∝ ∆V
Shapiro…, 2007,2010,2011
Cochran…, 2004
Consistent with tidal triggering of earthquakes
Mexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s
-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Mexp
0 1 2 3 4 5 6 7
Mob
s
0
1
2
3
4
5
6
7
0.05
0.95
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDX
R01
R11
GUYPOH
TTX
RMAPOK
HAR
Volume (m3)102 103 104 105 106 107 108
Mob
s
0
1
2
3
4
5
6
7
-3
-2
-1
0
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBAS A87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
CB2
CB3
CB4
m310-6 10-2 102 106
Mob
s-6
-2
2
6
GDF
Volume (m3)102 103 104 105 106 107 108
Num
ber a
bove
Mre
f
100
101
102
103
104
105
106
-3
-2
-1
0
KTB
BUK
GOK
SZ1SZ2
SZ3
SZ4
DFW
BAS
A87
CB1
A01
YOH
PDXR01
R11GUY
POHTTX
RMA
POK
HAR
Number above Mref
100 101 102 103 104 105 106 107
Mob
s
0
1
2
3
4
5
6
7
KTB
BUK
GOKSZ1
SZ2
SZ3SZ4
DFWBASA87CB1
A01YOHPDXR01
R11
GUYPOH
TTX
RMAPOK
HAR
Mmax - Mml-1 0 1 2 3 4
Prob
abilit
y D
ensi
ty
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Mmax - Mml
-1 0 1 2
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.93
Nprior/Ntot0 0.2 0.4 0.6 0.8 1
Inst
ance
s
0
1
2
3
4
5
Unb
iase
d In
stan
ces
1
2
3
4
5
N/Ntot
0 0.5 1
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.3p = 0.68Mmax - Mml
-1 0 1 2 3 4
Prob
abilit
y D
ensi
ty
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Mmax - Mml
-1 0 1 2
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.93
Nprior/Ntot0 0.2 0.4 0.6 0.8 1
Inst
ance
s
0
1
2
3
4
5
N/Ntot
0 0.5 1
Cum
ulat
ive
Den
sity
0
0.2
0.4
0.6
0.8
1p = 0.3
Confirmation of the sample size hypothesis
Num
ber
Volume
1. The maximum observed magnitude is predicted by the number of prior events
2. The order of occurrence of the largest earthquake is
random within the sequence.
3. The number of induced earthquakes is proportional to injected volume.
A+
A-A-
m310-6 10-2 102 106
Mobs
-6
-2
2
6
KTB
BUK
GOKSZ1SZ2
SZ3SZ4
DFWBAS A87CB1 A01YOHPDXR01
R11
GUYPOHTTX
RMAPOK
HAR
CB2
CB3
CB4
GDF
Goodfellow…, 2015
evidence from lab triggering• Need a wider range of volumes • All Cooper Basin stims • Lab hydrofrack: (4 ml, Mmax -7)
Mmax vs. log10(V)
obse
rved
Mm
ax
Volume (m3)104
6
2
-2
4
0106102100 108
lon-99.5 -99 -98.5 -98 -97.5 -97 -96.5
lat
35
35.5
36
36.5
37
37.5
Day
s si
nce
Jan
1. 2
014
0
100
200
300
400
500
600
700
800
900
Magnitude3 4 5 6 7
Cum
ulat
ive
Num
ber
100
101
102
103
104
b = 1.6 ± 0.07
3 4 5 6 7
Mm
ax P
roba
bilit
y D
ensi
ty
10-2
10-1
100
101
Was the M5.8 Pawnee earthquake expected?
P(Mmax ≥ 5.8) ~25%
KansasOklahoma
Conclusion• Induced earthquakes are as large as expected (they are triggered
tectonic earthquakes).
• The NSHM map should probably use the same magnitude limit for induced and tectonic earthquakes.
• Good news: the probability of a large earthquake can be estimated. Best predictor: Mmax ~ Σ log10V
• Bad news: there are no absolutes.
5 of 8 frack and EGS wells
Are EGS wells different?
(N-1)/Ntot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFWBASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOK HAR
Vol/Voltot0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b-value
0.5
1
1.5
2
2.5
KTB
BUK
GOKSZ1
SZ2
SZ3
SZ4 DFW BASA87 CB1 A01
YOHPDXR01 R11GUY
POH
TTX
RMAPOKHAR
Magnitude3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5
Num
ber
100
101
102
103
b = 1.7
1 year M90 = 5.8
Northern OK post-2014