The cross-correlation (CC) and master event technique is efficient in Comprehensive Nuclear-Test Ban Treaty (CTBT) monitoring. Two primary goals of CTBT monitoring are detection and location of nuclear explosions. Therefore, the CC monitoring should be focused on finding such events. The use of physically adequate masters may increase the number of valid events in the Reviewed Event Bulletin (REB) of the International Data Centre by a factor of 2. Inadequate master events may increase the number of irrelevant events in REB and reduce the sensitivity of the CC technique to valid events. In order to cover the entire earth, including vast aseismic territories, with the CC based nuclear test monitoring we conducted a thorough research and defined the most appropriate real and synthetic master events representing underground explosion sources. A procedure was developed on optimizing the master event simulation based on principal component analysis with bootstrap aggregation as a dimension reduction technique narrowing the classes of CC templates used in detection and location process. Actual waveforms and metadata from the DTRA Verification Database were used to validate our approach. The detection and location results based on real and synthetic master events were compared.

1. Performance of waveform cross correlation
using a global and regular grid of master events
Bobrov, D., I. Kitov, and M. Rozhkov
International Data Centre
Preparatory Commission for the Comprehensive
Nuclear-Test-Ban Treaty Organization
Provisional Technical Secretariat
Vienna International Centre
P.O. Box 1200
A-1400 Vienna
AUSTRIA
Mikhail.Rozhkov@ctbto.org
International Data Centre
Page 1

2. Objectives
Building a global grid of master events for waveform cross
correlation.
Assessing the performance of waveform cross correlation as a
technique of seismic monitoring using the global grid of
master events.
International Data Centre
Page 2

3. Outline
1. Motivation
2. Global seismic monitoring: IMS
3. Global seismicity: IDC view
4. Global cross correlation grid: a design
5. Cross correlation at teleseismic distances
6. Underground nuclear explosions as master events
7. Synthetic master events
8. Principal and Independent Component Analysis
9. Testing with world seismicity of February 12, 2013
10. DPRK 2013 of February 12, 2013
International Data Centre
Page 3

4. Cross correlation as an IDC technique
Motivation
•
Regional studies demonstrate significant improvement in detection,
location, and magnitude estimation.
At least an order of magnitude!
•
•
•
•
•
Many IMS primary stations are arrays enhancing the capability of
cross correlation analysis.
For arrays, correlation distance depends on phase and its slowness.
At teleseismic distances, high level of cross correlation is observed
for signals from events spaced by 100 km and even more.
Remote events may have similar signals.
Small events can be considered as point sources emitting signals
identical in shape when co-located.
International Data Centre
Page 4

5. IMS, seismic network
The primary network includes 25 arrays
Blue circles – primary arrays, blue triangles – primary 3-C stations.
Yellow circles – auxiliary arrays, yellow triangles – auxiliary 3-C stations.
Red stars – underground nuclear explosions.
International Data Centre
Page 5

6. Global seismicity: the IDC view
Waveform cross correlation relies on high quality master events
International Data Centre
Monitoring is global.
How to populate the aseismic area with quality master event?
Page 6

7. Global Cross Correlation Grid
What is Grid?
•
Grid is a set of loci of hypothetic
master events.
• Master is a set of waveform
templates linking array station and
the locus.
• Spacing between masters ~140 km.
• P-wave templates from three to ten
IMS primary arrays per master.
• Distance for P-phase from 6 to 90
degrees.
• At least three IMS stations to create
an REB event.
Possible templates:
1.
2.
3.
Real waveforms
Grand masters
Synthetic waveforms
International Data Centre
Page 7

8. Global Cross Correlation Grid
Segment
The segment is a set of grid cells with R = 100 km
International Data Centre
Page 8

9. Multichannel waveform cross correlation
6s
6s
CC essentials
CC
Multichannel waveform
template
Four frequency bands
Adjusted template length
Waveform quality check
CC for individual channels
Averaged CC trace
Detection
STA
Detection rule:
LTA
SNR=STA/LTA≥3.0
CC > CCtr
SNR_CC > SNRtr
Multichannel CC-detector better sees signals from slave events close to the master
International Data Centre
Page 9

10. GCCG: Local Association
Location mesh
1.
Five circles with ~25 km
increment in radius.
2.
91 nodes for origin time
calculation.
3.
All hypotheses at the
outer circle are neglected
since they have to be
created by neighboring
masters.
International Data Centre
Page 10

11. GCCG:
resolution of conflicts between masters
Detection of the same event by same stations but with different masters
(case study: DPRK-2013)
For given arrays the number of
detecting masters depends on the
distance (slowness), azimuth, and
aperture.
International Data Centre
The number of detections by
each master with nine stations
(templates).
Page 11

12. GCCG:
resolution of conflicts between masters
Creation of the same event by different masters
# of detections relevant to the slave
event
# of all detections in the events with at
least one detection from the slave event
Only two master events have nine detections from the slave
International Data Centre
Page 12

13. Cross correlation: explosion signals
Towards seismic monitoring of underground nuclear explosions
• 100 waveforms
• 25 underground
nuclear explosions
• 6.2 > mb > 4.5
• 2015 m > H > 150 m
• 60 stations
• 16º > Δ > 100º
International Data Centre
Page 13

14. Cross correlation of explosion signals
Synthetic
seismograms:
Δ = 30º, 45º, 60º, 90º
H = 0.1, 0.3, 0.6, 1.0,
2.0 km
Fc = 0.8 Hz to 4.8 Hz
International Data Centre
Page 14

15. Cross correlation: applying PCA and ICA
CC of first 5 Real Principal and Independent Components with
106 Real UNE records with cumulative ICs on right
CC of first 5 Synthetic Principal and Independent Components
with 106 Real UNE records with cumulative ICs on right
International Data Centre
Good performance in both cases
Page 15

16. Station cases for 4 templates: AKASG
DPRK-2013 source corresponds to the purple circle
International Data Centre
Page 16

17. Station case: ILAR
DPRK-2013 source corresponds to the purple circle
International Data Centre
Page 17

18. Station case: WRA
DPRK-2013 source corresponds to the purple circle
International Data Centre
Page 18

19. Global Cross Correlation Grid
Testing, February 12, 2013
Grid: 25000 nodes
Group 1 = WRA, TORD, MKAR, ILAR, GERES, PDAR, CMAR, SONM, AKASG, BRTR, GEYT
Group 2 = ASAR, ZALV, YKA, ARCES, TXAR, KSRS
Group 3 = USRK, FINES, NVAR, NOA, MJAR
GG thresholds:
Detections: SNRmin = 0.5; SNR_Ccmin = 2.5; CCmin = 0.2; FKSTATmin = 2.5; AZRESmax = 20.0º;
SLORESmax = 2.0 s/º;
Events: dTorigin = 6s; NSTAmin = 3; AZGAPmax = 330º
xgrid1
xgrid2
xgrid3
xgrid4
arrival
22900402
24150132
21600999
34531530
origin
40396
48686
38975
95219
assoc
126595
152641
121775
297647
origin_cnf
5151
5773
5079
6542
assoc_cnf
16421
18488
16154
21488
International Data Centre
Page 19

20. Global Cross Correlation Grid
Global Grid DPRK-2013, locations for 4 cases of templates:
(a) AK135 synthetics master,
(b) PCA synthetic master,
(c) PCA real master, and
(d) DPRK-2013 master.
All masters except for case 4 were produced by replicating single template at each array
station element implementing predicted time delays for given master geographical position.
a
International Data Centre
b
c
d
Page 20

21. Global Cross Correlation Grid
REB DPRK 2013
Cross Correlation
Location
International Data Centre
Page 21

22. Global Cross Correlation Grid
Global Grid Location with constructed templates
Location results
•
•
•
•
•
REB – location by IDC,
xgrid1 – AK135 synthetics as template,
xgrid2 – first PC of synthetic record set,
xgrid3 – first PC of UNE set,
xgrid4 – DPRK-2013 genuine records at elements of all arrays is a template
set.
Location with synthetic PC is the same as location with genuine DPRK
records.
International Data Centre
Page 22

23. Global Cross Correlation Grid
•
•
•
V0.1: All master templates are synthetics same at all stations
V0.2: Master templates are station/master specific synthetics in 1D velocity
model
V0.3: Master templates are station/master/source (e.g. explosion) specific
synthetics calculated for 2D velocity structure (e.g. ak135+CRUST 2.0)
• V1.1: Real master templates are used where possible
• V1.2: Replicated master templates are applied where possible
• V2.0: The set of principal components are optimized where possible as
obtained by the PCA or ICA applied to the complete set of actual and
historical data
•
V3.0: Synthetic + real master templates based on principal components with
classification algorithms trained on actual data
International Data Centre
Page 23

24. Discussion
•
IMS array stations make possible automatic processing based on
waveform cross correlation.
•
Cross correlation is a powerful technique allowing to reduce the
detection threshold and relative location accuracy by an order of
magnitude, i.e. to find by 50% to 100% more (smaller) REB
events.
•
Real and synthetic master events may reduce the magnitude
threshold of seismic monitoring by 0.4 units of magnitude.
•
The Global Cross Correlation Grid is flexible (e.g. master density,
templates, number of stations, thresholds, etc.) to fulfill various
tasks including effective monitoring of UNEs.
International Data Centre
Page 24

25. Thank You!
International Data Centre
Page 25