Earthquake Bulletins and Catalogs at the USGS National Earthquake Information Center
Stuart A. Sipkin, Waverly J. Person, and Bruce W. Presgrave US Geological Survey National Earthquake Information Center
This year marks the 75th anniversary of the National Earthquake Information Center (NEIC) and its predecessors in the Departments of Interior and Commerce, and the 60th anniversary of the first issue of the Preliminary Determination of Epicenters (PDE). The PDE for 1949 contained the locations of 120 earthquakes; by 1998, the number had grown to 21,688.
Prior to 1960, the year electronic computer programs for epicentral determination became operational, all earthquakes were located by drawing arcs on a globe with a compass. The photos (opposite) compare analysts locating an earthquake ca. 1965 (when the globe was still used for "rapid" response purposes) to the methods used today. Other milestones in the history of the NEIC are listed on the below.
NEIC Bulletins and Catalogs
The NEIC produces a number of bulletins and catalogs on different time scales:
Each succeeding product contains solutions reworked with more data. In addition to providing earthquake locations and magnitudes, these listings also contain information such as moment tensor solutions and descriptive commentary.
The "finger list" is available via the internet, e-mail, and pager.
Within two hours, this list includes at minimum, most earthquakes in the conterminous United States, Hawaii, and the more populated areas of Alaska of magnitude 4.5 or greater, most other earthquakes in the United States of magnitude 5.5 or greater, and most foreign earthquakes of magnitude 6.5 or greater. For most earthquakes fitting these criteria, a press release is also issued. In practice, many foreign earthquakes smaller than a magnitude of 6.5 are also included on the "finger list" within two hours, especially during the work week. Typically, such events include earthquakes in Mexico, central America, and northern South America of magnitude 5.0 or larger, earthquakes in central and southern South America, Japan, and the Kuril Islands of magnitude 5.5 or larger, and most other foreign earthquakes of magnitude 6.0 or larger. Within 24 hours, during the work week, most earthquakes in the conterminous United States with a magnitude of 3.5 or larger, in the populated parts of Alaska with a magnitude of 4.0 or larger, in Hawaii
with a magnitude of 4.5 or larger, in the Aleutian Islands with a magnitude of 5.0 or larger, and elsewhere in the world with a magnitude of 5.5 or larger will be provided on the "finger list". Most earthquakes in the United States that are felt will be provided on the "finger list" list on a 24-hour basis even if they are smaller than the criteria above. Events in energetic aftershock sequences that meet the 24-hour criteria, but not the 2-hour criteria will be included on the "finger list" at the discretion of the analyst. On weekends and holidays, posting events that meet the 24-hour criteria is on a best effort basis by the on-call analysts.
The Quick Epicenter Determinations (QED) is a daily publication of earthquakes that is available by e-mail and the internet, and reports 15-20 events per day. Events on the QED are reworked until 7 days after the date of the event. The on-line listing contains events from the current date back through the latest date published in the Preliminary Determination of Epicenters (PDE). The PDE is a weekly publication that follows the QED by about 5 weeks, listing the location, magnitude, and felt effects of 450-500 earthquakes. All earthquakes published in the PDE are recomputed using considerably more data than was available at the time of publication in the QED. Many additional earthquakes, which had insufficient data to be published in the QED, are located and included in the PDE. The final catalogs, published about 4 months later, are the PDE Monthly Listing and the Earthquake Data Report (EDR). Approximately 1,800 to 2,000 events are published in each PDE Monthly Listing and EDR. The EDR, which is available electronically, shows all data used to compute the earthquake hypocenters published in the PDE Monthly Listing.
Because of improvements in communications, leading to a much greater amount of data available prior to the publication of the PDE, the PDE Monthly Listing is now simply a compilation of all earthquakes listed in the PDEs for that month, with the addition of source parameter data and data from contributed local earthquake bulletins that have arrived in the interim. Significant Earthquakes of the World is a publication, updated weekly, that gives information on earthquakes that had a magnitude of 6.5 or greater, or that caused casualties or substantial damage.
In an effort to improve the effectiveness of both international and US monitoring of the Comprehensive Nuclear Test-Ban Treaty (CTBT), we began publishing a catalog of hypocenters and magnitudes of suspected US mine explosions in May 1997. This catalog is directed towards the "confidence building" element of the CTBT. The Mine Explosion Bulletin is only available electronically.
As World Data Center A for Seismology, the NEIC maintains a database of seismograph stations throughout the world and their international codes. As of 1 January 2000 this database contained information for 10,678 codes, of which 5,408 are listed as active.
In 1998, 21,688 events were published in the PDE Monthly Listings. Of these, 2,779 (12.8%) and 1,076 (5.0%) were events within the 50 United States and 48 conterminous United States, respectively. These events were located using data from 2,384 stations, approximately 2,000 of which are outside the US In 1998 over one-million readings were received and processed from these stations. A "reading" indicates all data from one station for one event, sometimes known as a "reading group".
The magnitude threshold at which coverage is complete varies with geographic region. The reasons for this are two-fold. First is that the global network of seismograph stations is not spatially uniform. The second is that not all seismically active regions have a regional or local seismograph network that reports waveform and/or parametric data to the NEIC. The histogram on p. 3 is a logarithmic plot of number of globally located earthquakes versus magnitude for the year 1998. In order to get a rough idea of how magnitude completeness thresholds vary by region, we examined the data from each of the 50 regions first proposed by Gutenberg and Richter (1954) and formalized by Flinn and Engdahl (1965). Examining each region individually, we find magnitude completeness thresholds varying from 2.7 (Oregon, California, and Nevada) to 5.3 (Southeastern and Antarctic Pacific), with the majority of regions in the range of 4.3 to 4.4. Of course, many smaller regions covered by regional and local networks have a much smaller threshold.
What is in store for the future?
The software package used in the routine analysis is currently undergoing a complete revamping. Several major changes are being introduced and include both a new, more flexible database that will support a full range of data types, a new analyst interface, and a new location algorithm developed by Engdahl et al. (1998). The new location package includes a more robust regression scheme, a modern Earth model such as ak135 (Kennett et al., 1995) (or a closely-related model) that will be adopted as the one-dimensional reference model, and a statistical phase identification algorithm (Buland, 1986) that will be applied to later arriving phases, permitting these later phases to be confidently included with first arriving P waves in all hypocenter determinations.
History of the NEIC
1925 - Seismology Program established by US Coast and Geodetic Survey/Department of Commerce
1928 - Annual US Earthquakes series initiated
1940 - Preliminary Determinations of Epicenters first published
1960 - Electronic computer programs for epicentral determination became operational
1963 - Routine computation of mb magnitudes began
1966 - NEIC established by the Environmental Science Services Administration (ESSA)/Department of Commerce in Rockville, MD
1968 - Routine computation of MS magnitudes began
1970 - NEIC transferred to National Oceanic and Atmospheric Administration (NOAA)/ Department of Commerce
1972 - NEIC moved to Boulder, CO
1973 - NEIC transferred to US Geological Survey/ Department of Interior and moves to Golden, CO
1981 - Routine computation of moment tensor began
Analysts locating an earthquake ca. 1965, and today. Note that the compass in the 2000 photo is the same as the one being used in 1965 and was an integral part of the NEIC Y2K Contingency Plan.
of Earthquakes Published in 1998
The error bars of the computed hypocentral coordinates are 90% marginal confidence intervals incorporating Baysian information to stabilize estimates derived from small samples (Jordan and Sverdrup, 1981). It is assumed that the travel-time errors of the data used are independent, unbiased, and have an expected standard deviation of 1 sec. Monte Carlo experiments suggest that the error bars give reasonable estimates of hypocentral precision for events constrained by more than about 30 data. The 90% error ellipsoid for each event with a NEIC solution (i.e. not contributed) is published in the EDR, which is available on the NEIC ftp site (gldfs.cr.usgs.gov). However, care should be exercised in interpreting these numbers in terms of absolute location accuracy because of unmodeled biases. Analysis of events with independently known coordinates indicates that most PDE determinations are accurate to a few tenths of a degree in epicentral position and 25 km in depth. Because of these uncertainties, verbal descriptions of locations in news releases are rounded to the nearest 5 mi and/or 10 km.
Since August 1983, NEIC mb and MS average magnitudes have been computed using a 25% trimmed mean because studies (i.e. Buland, 1986) have shown that these distributions are non-Gaussian; they are too heavy-tailed or contain too many data points outside of what would be a normal standard deviation. This is reflected by the observation that individual station magnitudes vary by one order of magnitude or more from the mean. We believe that this is largely due to focal mechanism (i.e. lobe versus node) and to geologic structure, rather than to errors in the data themselves.
Contributed local magnitudes and locations are edited for gross errors, but are otherwise published as received. We usually have no information about the uncertainties of these solutions.