Utilizing Large Databases for Nuclear Explosion Monitoring: the Knowledge Base (KB) Aaron A. Velasco University of Texas at El Paso

Overview Motivation Nuclear explosion monitoring (NEM) Databases and the Knowledge Base (KB) Using the KB for research and for monitoring: an example Summary

What is Nuclear Explosion Monitoring (NEM)? Monitor the globe for covert nuclear weapons tests utilizing key technologies Goal of global monitoring systems is to find the “needle in the haystack” –Identify nuclear explosions in the “noise” of earthquakes Zero yield monitoring means we must detect, locate, and identify low magnitude “events” –No longer just a few countries with large weapon tests –Requires regional (<2000 km from source) monitoring (collecting data from in-close to a test site Must have access to regional data Must account for regional propagation

Earthquakes around the world

Why is it important?: False alarms can create international problems 1997 Novaya Zemlya Event Washington Times –Russia conducted a clandestine nuclear test on August 16, 1997 (Figure from M. Tinker) Created tension between governments Claim was wrong: event occurred in the Kara Sea

Current political climate: CTBT In 1996 President Clinton signs Comprehensive Test Ban Treaty (CTBT) –Continues to observe moratorium on underground nuclear testing –In 1999 U.S. Senate did not ratify treaty –Current administration still observing moratorium on testing Comprehensive Nuclear-Test-Ban Treaty Organization still active in Vienna, Austria –165 Member States –89 Ratifications U.S. continues its monitoring operation (since 1960s) External and internal research program continues

Science goals Use existing and developing technologies to improve the capability to monitor low yield nuclear weapons tests Department of Energy (National Nuclear Security Administration) and the Department of Defense wish to improve U.S. ability Monitoring performed my Air Force Tactical Air Command (AFTAC) Capability must improve for many fields –Seismic, Infrasound, Hydroacoustic, Satellite, Radionuclide

Global monitoring Large automated systems to monitor for nuclear explosions rely on teleseismic data (recordings from events > 3000 km) Large amount of funding (~$60-70M/yr) to help improve capability of these automated systems Seismic Infrasound Hydoacoustic Radionuclide

Nuclear explosions vs. earthquakes Physics are different –Explosions are compressional sources Generates strong P-waves, little shear energy (S-waves, Surface waves) –Earthquakes are shear sources Generate all wave types, but dependent on radiation pattern Empirical methods are preferred for monitoring –Easy to implement –Quick (no heavy computations) Must be able to record and understand “regional” recordings –Waves that travel through crust are much more complex than those traveling through body of the earth (mantle)

Seismic recordings of earthquakes and explosions

Current seismic monitoring challenges Old Regime -- Teleseismic Distances > 2000 km Large bombs, few countries Simple earth structure Simple seismograms Magnitude > 4 Distances < 2000 km Region-dependent complicated earth structure Complicated seismograms Magnitude < 4

How an event is identified Rules are applied to events, which usually rely on teleseismic data –Rely on fundamental differences between earthquakes and explosions –Deeper than 10 km event indicate earthquakes –Offshore events are usually ignored If an event is near a region of interest, it is usually flagged for further inspection To monitor smaller events, regional discrimination is the key (no teleseismic signals)

DOE/NNSA Knowledge Base The DOE/NNSA Knowledge Base is a combination of the information content, database storage framework, and interface applications needed to provide research products in an integrated form that will allow the United States National Data Center to meet U.S. nuclear explosion monitoring objectives.

What is a Knowledge Base? Set of data and/or databases (data warehouse) with a specific goal: Nuclear explosion monitoring DOE KB is comprised of information products (IPs) Each IP is comprised of data sets and research products that have a set focus (e.g., nuclear test sites, magnitude, location, event identification,etc.)

Three broad categories of knowledge Parametric Grids – Irregular grids – Station referenced corrections by phase for each technology – Corrections for travel time, amplitudes, azimuth, slowness Event Data – Details of events for reference – Includes waveforms, measurements, comments Contextual Data – Geophysical seismicity, gravity, attenuation, etc. – Geological rock types, faults, volcanoes, etc. – Geographical borders, facilities, cities, etc.

Relational databases

Nuclear explosion

Elements of a KB: Development of Research Products What Researchers Develop –Waveforms and Catalogs –Ground truth –Discrimination recommendations –TT tables and corrections surfaces –Regional magnitudes –Scripts and algorithms –Station information –Detection thresholds, –MDAC parameters –Group velocities –1-D path specific velocity models –1 Hz Lg Q models –Lop Nor circle characterization

What can we use the KB for? Reference or library for information Data mining for identifying unique processes that were overlooked Well-vetted software for established techniques Develop new techniques to improve discriminating between earthquakes and explosions Manipulate large amounts of information with proper referencing (metadata) Improve ability to do research!

An example of using a KB An event that might be flagged as suspicious

What makes this event interesting? Exhibits explosion-like characteristics. Occurred within China Regional discrimination would classify as explosion Traditional discriminations (M S,vs mb) would classify event as earthquake Occurred on the eastern most edge of the Tibetan plateau

Can we find an answer using the KB? Use database of waveforms to see if this type of event is typical for the region Characterize geology and wave propagation in region Perform traditional techniques and using KB software on the event Apply new techniques on waveforms that have been developed in KB

Propagation effects Zone of Sn attenuation mapped by McNamara (1995)

The answer: This event was an earthquake This event was an earthquake Source modeling indicates that this event occurred at km depth and was a strike-slip event Propagation effects –Near zone of S-attenuation –Moho topography (70 km crust near source to 45 km crust near station can cause focusing) Source effects –Rupture directivity

Summary The NEM R&E program is developing research products for operational monitoring National Laboratories develop information products that can be used for the DOE KB DOE Knowledge Base serves to get basic research into operational context KB use for improving research Information technology key to the future of science

Comparison to Near Events Raw waveforms

Comparison to Near Events High pass at 1 Hz

Comparison to Near Events P waves with HP at 1 Hz

Apply Traditional Techniques Locate the event –Determine depth Teleseismic discriminants Regional discriminants

Relocations Added regional picks Used TT correction surfaces Locates near copper mines Area of moderate seismicity Obtained stable solution with moderate error ellipse

Location Method

Depth Phases? Depth phases indicate deeper than surface explosion

Teleseismic Discrimination M S vs m b –Body wave excitation much greater than surface wave excitation for explosions (except Rg) Calibrated magnitude based on previous studies Both M S values place event in earthquake population

Regional Discrimination Spectral ratios of regional phases using only distance correction High frequency Pn/Sn classifies event as explosions Cross spectral Pn ratio within

Still Ambiguous: What next? Contact national authority and say that we have a “suspicious” event? –NO!! Source modeling using regional techniques –Event too small to use typical teleseismic methods –Utilize longer period information that is not as susceptible to propagation effects Surface waves

Longer Period Observations

Inversion for Focal Mechanism and Depth Too small to be modeled by global catalogs Created reflectivity synthetic seismograms for quite of velocity models Applied phase match filter as obtained from data Matched dispersion characteristics at a station to models Inverts for depth and mechanism

Grid Search for Focal Mechanism

P-wave Spectral Characteristics Comparison to other events High P-wave energy Signs of directivity? Triplications due to structure?

Implications Current nuclear explosion monitoring systems do not perform source modeling The Harvard Seismology Group routinely models all earthquakes greater than about a magnitude 5.0. Implementation of source modeling remains key for small magnitude events, but is often ignored in the community False alarms can cause international incidents –Recent Novaya Zemlya event was in Kara Sea

Summary This event was an earthquake Regional discrimination failed because of unmodeled propagation effects Source directivity may have also contributed Source modeling is critical to prevent false alarms

Department of Energy NEM R&E Goals and Objectives Detect, locate, identify, characterize, and enable attribution of nuclear detonations: –Develop and deliver satellite-based sensor systems with improved capabilities for monitoring evasively conducted explosions in the atmosphere and in space, including tests by emerging nuclear states. –Deliver software components, prototype hardware, and an integrated knowledge base for ground-based monitoring for compliance with test bans and moratoria and for operating the United States NDC.

Ground-based systems Radionuclide Seismic Hydroacoustic Infrasound

Event Discrimination Exploit differences in source processes Traditional methods –mb vs Ms (body wave magnitude versus surface wave magnitude) –Depth –Onshore vs offshore Regional methods –Empirical phase ratios (P to S ratio) –Magnitude and distance amplitude corrections (uses an earthquake model; does not work for explosions)