Linking fossil reefs with earthquakes: Geologic insight to where

Linking fossil reefs with earthquakes:
Geologic insight to where induced seismicity occurs in Alberta
Ryan Schultz, Hilary Corlett, Kristine Haug,
Kelsey MacCormack, Virginia Stern, Todd Shipman
Seismicity in the WCSB

Seismicity in the WCSB is sparse and
relatively quiescent.

Long-lasting clusters have been
recognized.

Three clusters account for the majority of
Albertan seismicity: RMHSZ, BrC, CLS.
After Stern et. al., 2013
2
Induced Clusters

Majority of earthquake clusters have been linked
with petroleum resource development activities.

RMHSZ associated with conventional gas
production [Baranova et al., 1999].

Brazeau cluster (Cordel Field) link to waste-water
disposal [Schultz et al., 2014].

Cardston swarm associated with hydraulic
fracturing [Schultz et al., 2015a].

Crooked Lake (Fox Creek) region has multiple
distinct clusters associated with hydraulic
fracturing wells [Schultz et al., 2015b].
3
Few earthquakes?

Despite ubiquitous oil & gas development there
are only a few locations where EQs are reported
as induced.

Associations of wells & EQs estimates < 2% of
disposal and < 0.5% of HF wells [Atkinson et al.,
2016].

What are the controlling operational, geological,
and tectonic variables?
4
Double-Difference & Waveform Modelling

Use double difference location algorithm

to determine event geometry.

Epicentral errors: 10s km -> 1s km.
Beam average stacks of repeated
earthquakes.

Fit waveforms to constrain earthquake depth.
5
Induced EQ Depths

Similar findings are reported from DD,

Induced EQs are at or deeper than active formation.

Suggests reactivation of basal sedimentary or
waveform modelling, & other studies in the
WCSB [e.g., Zhang et al., 2016].
basement controlled faults?
6
Spatial Association

Robust epicentral locations coincide with the
margins of the Swan Hills Formation.

Geospatial correlation is decent: R2 0.85, and EQs
are no further than 20 km from margin.

Coincidence?
7
Coincidence?

Correlation =/=> Causation.

Spurious correlations are possible, what is the
statistical confidence of this spatial association?

Causal relationships require a physical rationale to
relate cause and effect via a correlation.
R2 0.80
8
Monte Carlo

Randomly guess 19 EQ locations, build distribution
from 106 trials, compare to observation.

<< 10-6 odds of occurring randomly i.e., similar to
flipping 20+ heads in a row.
9
Physical Rationale?

Highly confident Swan Hills & Induced EQs are related.

Obvious first guess: wells target nearby this formation.

Repeat Monte Carlo tests with skewed “dart throwing.”

Still recover significance. Same for deeper wells?
Rogers, 2014
10
Monte Carlo w/ Depth Bias
11
Monte Carlo w/ Depth Bias
All EQ data
>5σ
~All EQ data
>3σ
Most EQ data
90%
~No EQ data
2σ
12
Monte Carlo w/ Depth Bias
>3σ
All EQ data
>5σ
~All EQ data
~3σ
Most EQ data
•
•
•
17 Duvernay MS-HF
1 Disposal
1 Production
90%
6,1,1
~No EQ data
2σ
13
Reason for
Association?

Swan Hills is the first/deepest carbonate unit in study area, a
reef which grew in warm epeiric sea of the Devonian Period.
14
Reason for
Association?

Swan Hills is the first/deepest carbonate unit in study area, a
Berger & Davies, 1999
reef which grew in warm epeiric sea of the Devonian Period.

Reefs prefer to grow on bathymetric highs.

Spatial correspondence may be the result of reef nucleation
preference for faulting related paleobathymetric highs.
Purdy et al., 2003
15
Geological Rationale: Reef Nucleation

Swan Hills Formation is the first/deepest carbonate unit in study
area, a reef which grew in warm epeiric sea of the Devonian Period.

Spatial correspondence may be the result of reef nucleation
preference for faulting related paleobathymetric highs.
Purdy et al. 2003
Eaton et al., 1995
16
Geological Rationale: Communication
After Green & Mountjoy, 2005

Pervasive dolomite at Swan Hills build up margin
decreases significantly towards build up interiors.

Regional dolomitizing fluids could have flowed up vertical
faults and fracture systems, enhancing
porosity (2-5x) and permeability (2-63x).

Plausible hydraulic communication w/
basement, or increased hydraulic
diffusivity laterally.
After Eccles & Berhane, 2011
17
Implications*

Geographically biased activation potential observed
for induced seismicity.

Swan Hills Formation reef margin settings may
provide a geological proxy for geological features
which have a propensity for the initiation of induced
seismicity.

Detailed study of the geological, geomechanical, and
hydraulic conditions along the margins of this fossil
reef may provide criteria to prioritize regions with
increased seismogenic potential.
18
*Caveats

Two rationales conjectured, likely more, reasons for
association would need verification.

Despite correspondence, induced EQs have been
detected only in a few margin locations:

Geomechanical considerations have not been included, other
faults may be poorly oriented.

Reef nucleation is a complex process with many controlling
factors.

Other faults likely exist outside of reef-building
conditions i.e., induced EQs may still be observed
away from the reef margin.
19
Summary

Only a small number of wells actually induce EQs in the central WCSB.

EQs tend to occur at, or deeper than the active formation and into the shallow basement.

Induced EQs in the central WCSB are spatially associated with Devonian fossil reefs

Unlikely to simply be the result of (deep) wells targeting nearby the Swan Hills Formation.

The Swan Hills Formation reef margin settings may provide a geological proxy for geological
features which have a increased seismogenic potential.

Correspondence interpreted as reef nucleation preference to regions of paleobathymetric
highs (faults?), or enhanced hydraulic diffusivity.
20
Schultz, R., Corlett, H., Haug, K., MacCormack, K., Stern, V., Shipman, T., (2016). Linking
fossil reefs with earthquakes: Geologic insight to where induced seismicity occurs in
Alberta, Geophysical Research Letters 43, doi: 10.1002/2015GL067514.