Aftershock sequences of earthquakes with magnitudes 5.0 and lower are difficult to detect and locate by sparse regional networks. Signals from aftershocks with magnitudes 2 to 3 are usually below detection thresholds of standard 3-C seismic stations at near regional distances. For seismic events close in space, the method waveform cross correlation (WCC) allows to reduce detection threshold by at least a unit of magnitude and to improve location precision to a few kilometres. Therefore, the WCC method is directly applicable to weak aftershock sequences. Here, we recover seismic activity after the earthquake near the town of Mariupol (Ukraine) occurred on August 7, 2016. The main shock was detected by many stations of the International monitoring system (IMS), including the closest primary IMS array stations AKASG (6.62 deg.) and BRTR (7.81), as well as 3-C station KBZ (5.00). The International data centre located this event (47.0013N, 37.5427E), estimated its origin time (08:15:4.1 UTC), magnitude (mb=4.5), and depth (6.8 km). This event was also detected by two array stations of the Institute for Dynamics of Geospheres (IDG) of the Russian Academy of Sciences (RAS): portable 3-C array RDON (3.28), which is the closest station, and MHVAR (7.96). Using signals from the main shock at five stations as waveform templates, we calculated continuous traces of cross correlation coefficient (CC) from the 7th to the 11th of August. We found that the best templates should include all regional phases, and thus, have the length from 80 s to 180 s. For detection, we used standard STA/LTA method with threshold depending on station. The accuracy of onset time estimation by the STA/LTA detector based on CC-traces is close to one sample, which varies from 0.05 s at BRTR to 0.005 s for RDON and MHVAR. Arrival times of all detected signals were reduced to origin times using the observed travel times from the main shock. Clusters of origin times are considered as event hypotheses in the phase association procedure. As a result, we found 12 aftershocks with magnitudes between 1.5 and 3.5. These small events were detected neither by the IDC nor by the near regional network of the Geophysical Survey of RAS, which has three closest 3-C stations at distances of 2.2 to 3.5 degrees from the studied earthquake. We also applied procedure of relative location and all aftershocks were found within a few km from the main shock.

1. Detection and location of small aftershocks using
waveform cross correlation
Ivan Kitov, Seismic-Acoustic Officer, IDC/SA/SM
Irina Sanina, Head of Laboratory, IDG RAS
Sergey Sergeev, Researcher, IDG RAS

2. International Data Centre Page 2
Outline
• Mariupol earthquake
• Seismic observations
• Waveform cross correlation: master events and
waveform templates
• Detection on cross correlation traces
• Local Association (LA) of cross correlation detections
• Relative location
• Recovery of the Mariupol aftershock sequence
• Seismotectonic interpretation

3. International Data Centre Page 3
Motivation
• Seismicity is mostly related to tectonic forces or
industrial activity
• Seismicity is confined to relatively small or narrow
arears
• Seismicity is highly repeatable in location and
magnitude as well as in similarity of signals
• Waveform cross correlation (scalar product) is an
optimal detector of similar signals (matched filter)
• There are thousands of digital seismic stations
• There are huge databases with digital waveforms

4. International Data Centre Page 4
7 Aug 2016 Mariupol earthquake
Lon = 47.20°N ± 0.07°; Lat = 37.51°E ± 0.06°;
H = 24 ± 6 км; OT = 08:15:07.5 ± 0.7 s
Kp = 12.2 ± 0.2 (5); mb = 4.4 (7); MSH = 4.6 (4);
Md = 4.3 (6); Mc = 5.0
Geophysical Survey of Russian Academy of Sciences
Regional solution (12 stations)
International Data Centre
Lon = 47.0013°N; Lat = 37.5427°E
H = 8 ± 5.3 км; OT = 8:15:04.09;
mb = 4.5 ± 0.1

5. International Data Centre Page 5
201608070815A UKRAINE-MOLDOVA-SW RUSSIA
Date: 2016/ 8/ 7 Centroid Time: 8:15: 9.3 GMT
Lat = 47.06°N; Lon = 37.59°E
Depth = 17.7 km;
Half duration = 0.5 s;
Mw = 4.6; Ms = 4.8; Scalar Moment = 1.05E+23
Fault plane: strike = 101°; dip = 73°; slip = 178°
Fault plane: strike = 192°; dip = 88°; slip = 17°
7 Aug 2016 Mariupol earthquake
Global CMT Catalog

6. Geological and structural conditions
of the Mariupol earthquake region
1 – reverse fault; 2-4 – fractures of different ranks: 2 – basement, 3 – sole of the plate complex, 4 – neo-tectonic
joints (arrows are corresponded with dip plane; identified fault (solid line), supposed fault (dashed line); 5 –
subalkaline intrusion; 6 – epicenter of the earthquake (а) and aftershock (b); 7 - wedge-shaped block
Main characteristics:
1. There is highly graded joint zone
between two megablocks – slowly rising
Azov megablock and descending Black
Sea Sag. The boundary is reverse fault of
South-West – North-East strike.
2. Disjunctive node is formed by deep
different age and rank faults.
3. Proterozoic sub-alkaline intrusion is
connected with the disjunctive node.
4. The lock structure consists of wedge-
shaped blocks, which accumulate
excessive stress.
5. There are several historical earthquake
centers.

7. International Data Centre Page 7
Seismic network
Seismic observationsN
RDON BRTR
KBZ AKASG
E

8. International Data Centre Page 8
Waveform templates
RDON BRTR KBZ
AKASG MHVAR
Waveform templates for P-, S-, Lg-waves and the whole record
are tested in various frequency bands

9. International Data Centre Page 9
Waveform template
– high quality signal
Searching for similar (repeating) signals in
continuous waveforms
Cross correlation for arrays
Continuous multichannel cross correlation. For arrays, only vertical (e.g., AKASG)
or 3-C channels (e.g., RDON) are used. For 3-C stations, all three channels can be
used.

10. International Data Centre Page 10
Cross correlation detection
An example of cross-correlation detection: Station AKASG,
BP filter between 0.8 Hz and 2.0 Hz
STA
LTA
Threshold: SNR=STA/LTA >3.0
Cross Correlation Coefficient, CC
Valid signal

11. International Data Centre Page 11
Detection of aftershocks

12. International Data Centre Page 12
Local Phase Association (LA)
For all valid arrivals at primary stations, which are found with a
given master event, origin times, OTij, are calculated. The
empirical travel times from the master event to the relevant
primary stations, TTij, are subtracted from the arrival times, ATij.
OTij = ATij – TTij
where i is arrival index at station j.
TTij = TTj !
Empirical travel times from a master event to seismic stations are
characterized by ZERO modelling errors and very low measurement
errors. These conditions allow extremely accurate relative location

13. International Data Centre Page 13
Additional association grid:
relative location
-10
-5
0
5
10
-10 -5 0 5 10
N
E
MASTER EVENT
dtk = S · dk - travel time correction
OTk
ij = ATij - TTj + dtjk – corrected origin time
From the origin time residuals to
relative location
• k nodes, rectangular or circles;
• grid size from 1 to 100 km;
• spacing from meters to 10-15
km
• Average OT and RMS OT
residual are calculated in each
node
s

14. Aftershocks of the Mariupol earthquake
Stress relaxation of the Mariupol earthquake occurs mainly along the fault
of the sub-meridional strike and is consistent with the general geodynamic
situation and geological and structural conditions of the region
N/n.
station
Origin time М
0/6 2016.08.07/08:15 4.6
1/4 2016.08.07/08:18 3.9
2/2 2016.08.07/08:22 2.9
3/5 2016.08.07/08:24 3.0
4/2 2016.08.07/08:53 2.5
5/2 2016.08.07/09:29 2.5
6/2 2016.08.07/14:49 2.2
7/5 2016.08.07/16:14 2.8
8/3 2016.08.07/16:40 2.4
9/3 2016.08.09/13:12 2.4
10/2 2016.08.11/11:46 2.5
11/2 2016.08.11/14:15 2.8
12/5 2016.08.11/21:59 2.9

15. Discussion
• Aftershocks of the AUG-7-2016 Mariupol earthquake are
accurately located near major faults and discontinuities
• With waveform cross correlation, regional networks are able to
detect and accurately locate weak aftershocks (mb~2-2.5) of
intermediate size earthquakes (mb~4.5-5).
• Joint use of various networks enhances detection and improves
location accuracy
• Processing is practically automatic

16. Some more from cross correlation
1. Relative magnitude
2. Relative moment tensor
3. Depth estimate
4. Continuous automatic monitoring

17. Aftershock sequence of ML6.1
earthquake in Sakhalin
Catalogue EQalert.ru
Cross correlation catalogue