- © Seismological Society of America
Operating a seismic data center is like being in the army. Most of the time it is “hurry up and wait,” but it can quickly change to “the plan goes out the window once the first bullet is fired.”
So it goes at the U.S. Geological Survey (USGS)’s National Earthquake Information Center (NEIC), home to one of the world’s foremost seismic monitoring systems. Located in Golden, Colorado, just west of Denver, NEIC staff provides the latest, most accurate earthquake information available within minutes of the occurrence of a global event. That means NEIC is ready to respond at a moment’s notice.
NEIC determines, as rapidly and as accurately as possible, the location and size of all significant earthquakes worldwide, including estimates of economic loss and casualties, and swiftly disseminates that information to emergency responders, governments, the public, and the news media. NEIC maintains an extensive seismic database as well—it is the nation’s data center and archive for earthquake information. It also pursues an active research program to improve its ability to characterize earthquake sources and their likely impact.
NEIC’s job and its tools have evolved substantially over the years, and especially since the turn of the century. This is the tale of the evolution between 2000–2016, when NEIC dealt with 21 M 8 or larger earthquakes (including two M 9.1 events) and a number of other devastating earthquakes (see Fig. 1). This took place against the backdrop of an explosion of available real‐time seismic data, instantaneous global communication and societal awareness, and, consequently, higher expectations for information services.
THE EARLY YEARS
NEIC was established in 1966 and became part of USGS in the early 1970s. Its first 30 years coincided with the birth of seismology’s digital age, and we became adept at using Digital Equipment Corporation (DEC) AlphaServers for the heavy lifting within date acquisition, processing, and distribution. By the late 1990s, NEIC needed an overhaul. New staff members were loath to use the old DEC operating system, which was not flexible enough to meet changing requirements and opportunities. Modernization was necessary to support development of the Advanced National Seismic System (ANSS; USGS, 1999), which included higher levels of coordination and interoperability between NEIC and the regional seismic networks (RSNs) operated by state and federal agencies and public and private universities. Concurrently, open access to broadband and strong‐motion stations was becoming the norm worldwide, facilitating new real‐time approaches to improved source characterization of earthquakes.
When it reached a critical mass of technical, research, and management staff in the early 2000s, NEIC started developing a new monitoring system under the leadership of Ray Buland and Alex Bittenbinder. A reorganized NEIC staff established clear lines of responsibility among research, development, and operations. Researchers no longer maintained the operational algorithms they developed. Instead, researchers became subject‐matter experts for the developers who design and code the systems, and the IT staff who maintain them and make them operationally robust.
NEIC also began shedding responsibilities better handled by others. Seismic field operations including the National Seismic Network, the Intermountain West Network, and aftershock systems were transferred to the USGS Albuquerque Seismological Laboratory (ASL), guaranteeing professionally managed high‐quality data. Importantly, this realigned human resources so that NEIC and ASL could work together more efficiently on new processing and quality control systems, algorithms, and monitoring concepts.
NEIC acquires real‐time waveform data and associated metadata from more than 2000 stations contributed by 132 seismic networks worldwide.
To maximize access to available national and global waveform data, NEIC worked to be protocol‐agnostic, acquiring and distributing data in numerous open‐source protocols, and facilitated data sharing with RSNs and international partners (Patton et al., 2015). NEIC also adopted the established international source parameter description standard QuakeML and developed the Product Distribution Layer for internetwork communication of derived source parameters and related information (Guy et al., 2015). Recognizing the burden of routinely checking for availability of new data, NEIC invested in self-discovery and self-configuration procedures by checking all known repositories and services for updates to station metadata and new waveform data. Once waveform and station metadata are discovered and validated, acquisition and processing systems reconfigure themselves to use the new or updated data automatically. Presently, NEIC acquires real‐time waveform data and associated metadata from more than 2000 stations contributed by 132 seismic networks worldwide.
REAL‐TIME MONITORING AND THE 2004 SUMATRA EARTHQUAKE
A year before the 2004 M 9.1 Sumatra earthquake, the first new NEIC real‐time system was deployed. The system, Response Hydra, performed reasonably well for the 2004 Sumatra sequence, locating aftershocks with few processing bottlenecks and providing reliable mb (body‐wave), Mw (moment), and Ms (surface‐wave) magnitudes for many of the aftershocks, but the system still had serious deficiencies. For example, NEIC reported the mainshock as an Ms 8.5 about 80 min after origin time. That was the extent of NEIC’s capabilities—“we’d run out of ammo!” NEIC had no in‐house ability to further model the source size and rupture process. The longest and most anxious hours of my professional life were waiting for the Lamont–Doherty Earth Observatory Global Centroid Moment Tensor (CMT) solution approximately six hours after the Sumatra earthquake.
Response Hydra was not the only NEIC function that struggled in the first week or so after the Sumatra earthquake. It had been more than 40 years since the previous mega‐earthquake (M 9+), so we had trouble grasping the enormity of the event. Aftershocks were scattered over a distance of 1300+km, causing us to question the quality of our locations. Embarrassingly, and perhaps out of shock, it took us about 24 hours to validate that our location code was just fine and that the aftershock zone was indeed 4–5 times larger than we expected. Compounding our inability to properly model the source processes of the event, we had few tools and procedures to develop the seismotectonics context that the public, the news media, and governments wanted. The first week after the mainshock was a time of making do with limited capabilities yet gaining valuable experience and insight about how to improve future response operations.
A congressional funding boost following the 2004 Sumatra earthquake enabled NEIC to move to 24/7 operations, develop robust web infrastructure and expand web content, improve other aspects of our processing and distribution system, and further develop the Prompt Assessment of Global Earthquakes for Response (PAGER) response product (Wald et al., 2008) and improve “Did You Feel It?” (DYFI?; Wald et al., 1999). We started immediately on Bulletin Hydra (Patton et al., 2016), which would improve the capabilities of Response Hydra and eliminate the Preliminary Determination of Epicenters bulletin productions from the antiquated DEC AlphaServers.
We recognized that we could not develop new source‐modeling algorithms alone, so we solicited help from the academic community. To better characterize the size of very large earthquakes, Jascha Polet at Cal Poly Pomona helped us implement a surface‐wave CMT method in 2005 (Polet et al., 2008; Polet and Thio, 2011). By 2007–2008, Chen Ji at the University of California, Santa Barbara, and Hiroo Kanamori at the California Institute of Technology worked with us to operationalize finite‐fault modeling (FFM; Ji et al., 2002) and W phase CMT algorithms (Kanamori and Rivera, 2008), respectively. In the same time frame, Bob Herrmann at Saint Louis University implemented regional moment tensor procedures, lowering the threshold for reporting moment magnitudes (Herrmann et al., 2011). By the time of the May 2008 M 7.9 Wenchuan earthquake, we operationally reported CMT, W phase, and FFM solutions for the mainshock. W phase was extended to smaller magnitudes and regional distances (e.g., Hayes et al., 2009), so by the time the 2011 M 9.1 Tohoku, Japan, 2014 M 8.2 Iquique, Chile, and 2015 M 8.3 Illapel, Chile earthquakes occurred, we were operationally aware of the final size of each event (to within ±0.2 magnitude units) within 8–10 min of occurrence (e.g., Hayes et al., 2011). We now rely heavily on W phase, but we found it prudent to continue using teleseismic body‐wave and CMT methods as well because there are certain situations in which one method works when others fail. Indeed, just like the army, NEIC has learned to hope for the best but prepare for the worst.
NEIC products satisfy a need for accessible information and seem to have resulted in significantly fewer postearthquake interviews and calls, which in the past stretched human resources needed for earthquake response.
Development of PAGER was and still is a significant catalyst for improving NEIC. PAGER is a remarkable product that uses country‐specific knowledge of past earthquakes, databases of building and population vulnerabilities, and seismological source parameters of new events to estimate casualties and economic losses from significant earthquakes. Highly valued by the emergency response community for providing situational awareness and response assessment and planning, it is also easily accessible and understood by the public and news media. The Sumatra earthquake made PAGER a key NEIC product, and it stimulated the development of Slab1.0 (Hayes et al., 2012), improved estimates of earthquake source depth, and implemented FFM and several methods for better‐estimating moment magnitude.
PUBLIC DEMAND FOR EARTHQUAKE INFORMATION
In the early 2000s, the public and the news media repeatedly asked for more seismotectonic information to put notable earthquakes into an understandable geological and historical context. To fill this need, we engaged experts from academia (e.g., Kevin Furlong at Pennsylvania State University) and a series of summer interns to develop a regional tectonic framework within which new earthquakes could be interpreted. This led to the production of a series of regional seismicity maps with accompanying tectonic summaries (e.g., Rhea et al., 2010). These summaries are now posted automatically for all related earthquakes located by NEIC. For earthquakes of broad public and news media interest, we add supplemental descriptive information about that specific earthquake in the first hours after the event occurs. Since 2000, we have published more than 250 of these earthquake‐specific write‐ups for events M 7 or larger and for many other smaller but impactful earthquakes. For all damaging earthquakes and widely felt events we produce earthquake summary posters, and for a few exceptional events (e.g., 2010 M 7.1 Haiti; 2010 M 8.8 Maule, Chile; 2015 M 8.3 Illapel, Chile), we have produced earthquake educational slides. These products satisfy a need for accessible information and seem to have resulted in significantly fewer postearthquake interviews and calls to NEIC, which in the past had often stretched our human resources just as they were needed for earthquake response. Major international news services routinely use these summaries in their own reporting, providing excellent earth science content to an audience that might not directly access our website.
Next, social media (e.g., Facebook, Twitter) became the communication du jour. To take advantage of crowdsourcing information via Twitter, NEIC developed procedures for detecting potential earthquakes by triggering on surges in the use of the word “earthquake” or “quake” in multiple languages (Earle et al., 2010). NEIC routinely receives notification of an earthquake within 20–90 s of origin time in high‐risk areas of the world—often faster than we can associate detections on the closest stations in the area. NEIC also has the capability to use keyword indicators (e.g., “huge,” “fuerte”) to rapidly assess whether an event is likely damaging. Social media is also an important venue for distributing earthquake notifications by the USGS and contributing RSNs.
Following the April 2015 M 7.8 Gorkha, Nepal, earthquake, NEIC’s webpages were briefly the most‐visited U.S. government website, even beating out the Internal Revenue Service's “Where’s My Refund?” page.
One of the most significant changes in support of NEIC and ANSS operations was the re‐engineering of the Earthquake Hazards Program’s recent earthquakes website, including development of a curated comprehensive earthquake catalog (ComCAT) linked to the recent earthquakes pages. These products provide one‐stop shopping for authoritative earthquake information generated by the USGS‐supported RSNs and NEIC, which includes hosting historical earthquake catalogs from RSNs and international agencies, some dating back to 1900. The USGS recent earthquakes webpages are some of the most frequented earth science webpages in the world and serve as a primary reference for international news services. Following the April 2015 M 7.8 Gorkha, Nepal, earthquake, these webpages were briefly the most visited U.S. government website, even beating out the Internal Revenue Service website “Where’s My Refund?”
NEIC has a new adversary—injection‐induced seismicity—which requires new strategies and tools for monitoring.
Rapid characterization of the source parameters of the largest earthquakes is no longer the Achilles heel of NEIC operations. From a global‐monitoring perspective these events light up the planet like a Christmas tree, and the Global Seismic Network alone can be used to obtain an accurate Mw and rupture model. Yet NEIC has a new adversary—injection‐induced seismicity—which requires new strategies and tools for monitoring. These events have occurred mostly in Oklahoma and other states in the mid-continental United States since 2009, and they are believed to be triggered by injection of wastewater produced during oil and gas production (Ellsworth, 2013). To address systemic weakness in monitoring unexpected small‐magnitude seismicity in areas not routinely monitored by NEIC or RSNs, NEIC and ASL are improving configuration and instrumentation strategies to ensure timely integration of temporary stations into operations. Requirements for temporary deployments include sufficient station density to significantly improve location accuracy, lower the detection threshold, and capture near‐source high‐frequency ground motions (Yeck et al., 2016). To maximize the use of these stations, NEIC is developing procedures for using correlation detectors (Benz et al., 2015) and multiple frequency pickers (Lomax et al., 2012) that are suitable at local and regional distances, and they are also implementing self‐configuring associators to take advantage of changing station density. NEIC is experimenting with new single‐event and multiple‐event location procedures, updated crustal velocity models and phase statistics, and calibrated earthquake locations of event sequences (McNamara et al., 2015; Yeck et al., 2016) to produce a future catalog that has minimally biased locations with meaningful estimates of location error.
NEIC serves other roles as well. It acts as a backup to RSNs in the United States and as a conduit for real‐time data to the National Oceanic and Atmospheric Administration Tsunami Warning Centers, and it provides 24/7 infrastructure support to RSNs, ASL, and the Incorporated Research Institutions for Seismology (IRIS) for aftershock deployments and special studies (e.g., special deployments to record induced seismicity). NEIC participates with IRIS and the Air Force Technical Applications Center on coordinated data exchange, knowledge transfer, and training with other national seismic monitoring systems (e.g., the Afghan Geological Survey and the Chilean Seismic Network), a by‐product of which is access to often unique sources of high‐quality data and products.
Driven by need, expectations, heavy lifting by the academic community and RSNs, and a bit of luck, NEIC has evolved for the better. Like the army, NEIC has learned to adapt to meet changing needs—in our case, the changing needs of earthquake response—while maintaining the same familiar NEIC historic identity of a group that can be relied upon to provide accurate information in a timely manner. NEIC is a unique national resource that provides comprehensive analysis of global earthquakes of scientific and engineering interest. Because of the job we do and the rate at which large global earthquakes have been occurring, we are very experienced. The leadership at NEIC and ASL, combined with coordination and integration with the talented and committed staff at the ANSS RSNs, will ensure long and continued maturation of the ANSS. We are committed to serving the scientific community with the best scientific data available and the general public with timely and reliable information about seismic events.
Data and Resources
Advanced National Seismic System Comprehensive Earthquake Catalog is available at https://earthquake.usgs.gov/earthquakes/search/ (last accessed February 2017)