Infrasound data processing for CTBT verification:

sourcesN. Brachet, P. Mialle, D. Brown, R. Le Bras

CTBTO PrepCom, Provisional Technical Secretariat, Vienna International Centre, Austria

Abstract

In its final configuration, the International Monitoring System (IMS) of the Comprehensive Nuclear-Test-BanTreaty Organization (CTBTO) Preparatory Commission will operate 60 infrasound stations distributeduniformly over the globe. The International Data Centre (IDC) in Vienna, Austria currently receives andprocesses in near real-time data from 39 of the 60 planned infrasound stations. Specialized software has beendeveloped to detect coherent infrasound signals, highlight the most significant detections as phases (as opposedto Noise), and subsequently group these phases to form events.The IDC produces timely, high quality Reviewed Event Bulletins (REBs) using the three waveformtechnologies: seismic, hydroacoustic and infrasound. At the present time, the contribution of infrasound data tothe REB has been intentionally limited as new software was developed, tested and adapted to the IDCoperational environment. Non-operational use of infrasound data has made significant progress inidentifyingandcharacterizinginfrasoundsourcesand indicatespotentialfor strongsynergywith other technologies(the

Table 1.Seed events association(I: Infrasound ; S: Seismic ; H:Hydroacoustic ; ( ) Seed event) (I*I)

(I*I)+S(I*I)+H(S*S)+I(H*H)+I

Valid Not valid

(I*S)(I*H)

Figure 2. Example of surface grid cells used by IDC networkprocessing in South East Europe. Each grid cell is 236 km inradius. There are over 7000 cells covering the globe.

Limiting the number of infrasound false association

Examples of IDC infrasound eventsVolcanic eruptions

The very active Colombian volcanoGaleras has had a busy year of it so far,with several explosive eruptions in 2009,the latest one being on 25 April.

The IMS Infrasound data may be used for civilapplication for detection of volcanic eruptions andassist in aviation safety. Monitoring volcaniceruptions in remote areas is still challenging and theinfrasound may contribute to the early warning

Galeras (Colombia)

andcharacterizinginfrasoundsourcesand indicatespotentialfor strongsynergywith other technologies(thenetwork processing stage).A large collection of infrasound reference events has been built by the IDC during the last five years, but only asmall fraction of them meet REB event definition criteria considering the Treaty verification mission of theOrganization. Candidate events types for the REB include atmospheric or surface explosions, explosivemeteors, rocket launches, large earthquakes and explosivevolcanoes.

Network processing

Station processing [Brachet, 2009] is followed by “network processing”, which combines all relevant non-noise infrasound detections along with detections from seismic and hydroacoustic technologies and attemptsto build and locate events from these associations. There events appear in the automatic Standard Event ListbulletinsSEL2 (produced4 hoursafterreal-time)andSEL3 (produced6 hoursafterrealtime).

Alti

tud

e (k

m)

120

80

160

It

303m/s330m/s

“Escaping rays”(high thermosphere, above 120 km)

295m/s295m/s

330m/s

303m/s

Limiting the number of infrasound false association

The number of candidate infrasound events grows exponentially with the number of operating IMS infrasoundstations. A number of criteria has hence been introduced to lower the number of false association . The criteriaare based on propagation distances, frequency of the detection or size of the PMCC families [Brachet, 2008].The association of infrasound automatic detections is reliant on achieving good detector performance and arealistic modeling of the wave propagation in the atmosphere. The same set of infrasound travel-time tables isused by the association and location algorithms. At the IDC,the modeling of infrasound wave propagation iscurrently done using three constant 330m/s, 295m/s and 303m/s celerity models which are range dependent(Figure 3a, b and c), and no azimuth correction due to atmospheric winds.

infrasound may contribute to the early warningprogramme for volcanic ash emission developed bythe International Civil Aviation Organization (ICAO).

I44RU

I45RU

I18DK

I53US

I10CA

Mount Redoubt (Alaska, USA)

The activity of Mt.Redoubt has beenvery intense andstrong from end ofMarch to mid-April.It has been pickedup by several IMSinfrasound stations.

http://www.avo.alaska.edu/

Interactive processing

bulletinsSEL2 (produced4 hoursafterreal-time)andSEL3 (produced6 hoursafterrealtime).

The events are located using an iterative non-linear least-squares inversion originally developed by Jordan andSverdrup [Jordan, 1981] which was later modified by Bratt and Bache [Bratt, 1988] to include time, azimuthand slowness observations. The IDC network processing doesnot currently make use of infrasound data indaily Operations, but mixed technology events using seismic, hydroacoustic and infrasound phases are beingbuilt in the IDC testing and development area.

An event with only infrasound observations will appear in anautomatic IDC bulletin if the event is seen by atleast two infrasound stations. At the current time, infrasound phases can be associated with events that are upto 60 degrees from a station. The experimental SEL3 bulletinproduces about 15-20 events per day that includearrivals from one or several of the 37 infrasound stations (Figure 1b). Infrasound data contributes to about11% to the total events formed with seismic and hydroacoustic technologies in the IDC testing anddevelopment area. Large efforts have been produced to decrease the number of false associations withinfrasound arrivals, as infrasound were still contributing up to 35% to SEL3 in January 2008 (Figure 1a).

Source StationDistance (km) 0 400200

Alti

tud

e (k

m)

40

0

Is

Is

IwIw

Is+ItIs+ItIw Is

Figure 3a. Travel time table with time and modeling errors.

The IDC InfrasoundReferenceEventDatabasecontainsa

IDC INFRASOUND REFERENCE EVENT DATABASE (IRED)

Figure 3b. Travel time tables with celerity.

Figure 3c. Example of propagation path ofinfrasound wave withWASP-3D ray tracing. Accidental explosions

Mt. Redoubt

REB

I56US

I10CA

Pipeline explosion and fire, 2009-01-13 (Ural, Russia)Accidental explosions as well asmining events are interestingsources to test the detectioncapability of the IMS network, theaccuracyof the locationandwhen

(It: Themospheric, Is: Stratospheric, Iw: Tropospheric)

References

Figure 1a. Map of automatic events built withinfrasound arrivals in the experimental SEL3bulletin of January 2008.

Seed events and fusion

January 2009

Figure 1b. Map of automatic events built withinfrasound arrivals in the experimental SEL3bulletin of January 2009.

The IDC InfrasoundReferenceEventDatabasecontainsalarge collection of identified events for which infrasoundsignals have been detected and reviewed at IMS stations.The IRED is in constant evolution as it is regularlyupdated with new reference events corresponding tovarious types of infrasound sources, in particularatmospheric or surface explosions, meteors, rocketlaunches and re-entries, large earthquakes, and volcaniceruptions. As of January 2009, 478 events have been savedin the IRED (Figure 4). The IRED is being used as abenchmark for tuning the automatic system, in particularto determine/refine the empirical frequency-distanceattenuation law, and to help discover and confirmadditional criteria which would enhance the globalassociationof infrasounddataat theIDC.

Inventory IRED contains 478events grouped in 11categories (as of January 2009).

IRED is available for NDCs (by request from IDC Services)

– Mine and quarry blasts (199)– Earthquakes (75)– Volcano eruptions (56)– Rocket launches/re-entries (48)– Explosions (45)– Meteorites and Bolides (23)– Aircraft (13)

– Military exercise (11)– Cultural noise (6)– Avalanches and landslides (1)– Synthetic data (1)

mreporter.ru

accuracyof the locationandwhenapplicable the fusion oftechnologies.

Conclusion

The interactive review of the automatic bulletin has demonstrated that the automatic system is functioning at alevel where infrasound or mixed-technology events can be reliably formed. Various types of infrasound sourceshave been identified, some of them are energetic enough to bedetected by several stations of the IMS network,in particular atmospheric or surface explosions, exploding meteors, rocket launches and re-entries, largeearthquakes, and volcanic eruptions.

Figure 4. Overview of the IRED.

INFRASOUND – Detection, Location, and Identification of Atmospheric Events

References

[Brachet, 2008] Brachet, N. et al., Reference Events Used for Developing andTesting Criteria in the Global Association of Infrasound Data at the IDC,EGU, Vienna, 2008.

[Brachet, 2009] Brachet N. Et al., Infrasound data processing for CTBT verification – station processing, ISS Conference , 10-12 June 2009, Vienna.

[Bratt, 1988] Bratt, S.R., and Bache, T.C., Locating Events with a Sparse Network of Regional Arrays, Bull. Seism. Soc. Am., 78, 780-798, 1988.

[Jordan, 1981] Jordan, T., and Sverdrup, Teleseismic Location Techniques and theirApplication to Earthquake Clusters in the South-central Pacific,Bull. Seism. Soc. Am., 71, 1105-1130, 1981.

[LeBras, 1999] Le Bras, Ronan et al., Integration of Seismic, Hydroacoustic, Infrasound and Radionuclide PIDC, 1999, 21st Seismic ResearchSymposium, Las Vegas, Nevada.

DisclaimerThe views expressed herein are those of the authors and do not necessarily reflect the views of the CTBTO Preparatory Commission.

Seed events and fusion

The association of arrivals is done with a grid search of hypothetical seed events (Table 1). The grid cells areuniformly distributed at the surface of the earth and also cover depth zones in geographic areas known forhaving deep seismicity (Figure 2).The arrival detected at the nearest station to the grid cell is called a driver, which is used to predict time at theother IMS stations. Stations which are consistent with the driver are added to the seed event. Arrivals can beassociated to multiple events, and conflicts are resolved by selecting the best quality events (i.e. based on thenumber of associated defined arrivals, the size of error ellipse, the distance to the nearest station, and theprobability of detection).

associationof infrasounddataat theIDC.

Interactive Review

Interactive analysis is the stage of IDC processing where results produced by automatic processing are reviewedand refined by analysts. This stage is concluded by the publication of the Reviewed Event Bulletin (REB).Specialized software (ARS-Geotool-PMCC) has being developed for infrasound interactive review to allowvisualizing the automatically calculated and stored PMCC results. These results are presented as plots of azimuthand speed versus time and frequency, which provides a comprehensive picture of the signal properties as theyevolve with time, and efficiently identifies signal detections regardless of the signal to noise ratio.

Figure 4. Overview of the IRED.