Several sharp-eyed followers have noticed that our new recent earthquake webpages have had several duplicate (and even triplicate!) earthquakes.  To understand why this happens, you need a little background about how the earthquake notification biz works. 

Many "ears" to the ground:

A number of seismic networks are listening for earthquakes simultaneously.  PNSN is the most sensitive, of course, so we will locate everything that happens, even small magnitudes, with high precision, in Washington and Oregon.  We'll also locate larger earthquakes around the edges of our network, out under the Pacific and in Idaho, in California and British Columbia, but with reduced accuracy.  Similarly, the California Integrated Seismic Network (CISN), Montana Earthquake Studies Office (MESO), the National Earthquake Information Center (NEIC in Golden, CO), and the Canadian Seismic Network, (Also the Tsunami Warning Centers for the Pacific and West Coast and Alaska Center) will locate earthquakes in Oregon and Washington, if they are large enough to be recorded by these networks.  All of these seismic networks also exchange seismograms, in real time, for enough seismometers to help each other "surround" earthquakes on the edges of each of our networks.   So there may be several agencies producing locations and magnitude estimates for earthquakes as soon as an earthquake occurs. 

The notification mess:

In addition to the realtime seismic waveform data exchange, we have a system of exchanging information about earthquake sources that we locate. This is now called EIDS (Earthquake Information Data System, or something like that).  EIDS recently took over from a system called QDDS/QDM (which worked very well for many years, but I doubt anybody around knows what those letters stand for anymore).  To be more complete, QDDS uses a terse data format called CUBE (or CUBIC depending on to whom you talk), in which all the information about an earthquake is packed into an 80-character line (remember punch cards?).

EIDS is much more flexible, theoretically allowing us much more room for information, BUT, the formats for doing this are still being developed!   We seismic networks are a chatty bunch, and we all (well, maybe the Canadians not yet--they're more reserved, eh?) send everything we locate into EIDS.  EIDS, then, is the system that is always listening to (and immediately sharing with all of the connected seismic networks) information about the earthquakes.  EIDS can get TONS of notifications about an earthquake. For example, our duplicate automatic systems can both submit solutions, as will NEIC and CISN, say, for an earthquake offshore of Oregon.  Then when our analyst reviews it within a couple of minutes and blesses it as a real earthquake, it'll get sent again with a different version number.  Within 10 minutes the event may be relocated after a careful culling to get the best location and magnitude possible.  Sent again into EIDS.  Meanwhile the same could be going on at CISN and NEIC.  How to keep up with this is a challenge.

Who's "authoritative"?

There is logic in place within the EIDS servers that automatically judges which networks solution is "authoritative".  Most of the time it's straightforward.  Earthquake in Oregon?  PNSN is authoritative.  Earthquake in Eureka, California?  CISN is authoritative.  And so on.  However it can get complicated, particularly for earthquakes on the "edge" of networks.  Preliminary locations for an earthquake may place it inside one networks' authoritative region (say PNSN), but subsequent relocations show it really occurred farther offshore, into the NEIC territory; authorities for solutions change.  CISN may submit a location of an Oregon event before us, thanks to their superior computers payed for by their vastly greater funding, but within seconds our solution is also submitted, and becomes authoritative; so authoritative solutions change.  Or ... CISN may locate an earthquake near the Oregon-California border on the California side of the line; PNSN locates it close by, but on the Oregon side of the line; who's is it?  Which solution trumps?  This is a fun game; you can think of your own cases. CISN locates and publishes to EIDS an earthquake in CISN territory, but then realizes it was really a teleseism that fooled the network into thinking it was a deep local earthquake.  They delete it from the system.  Authoritative solutions disappear. 

If you click on the symbols on our recent earthquakes map to see more information about the earthquake, you'll see, amongst valuable information such as the origin time, something called the EVID, which stands for "event id".  Here's where things get really sticky.  There is no uniform EVID system in place between the various networks ... just logic that tries to determine who's solution is authoritative.  So with no common EVIDs, different networks' locations can (will!) have different coordinates, depths, magnitudes, and evids. 

Our new webpages listen to EIDS, and try to make sense of the chatter, while EIDS is trying to make sense of the chatter it's hearing from all of us. Somewhere in that mess, signals can get crossed, and our webpages will show two or more solutions. They will have different evids, coordinates, and origin times, but generally too close together to really be two earthquakes, but our software is fooled nevertheless, and we find ourselves having to go in by hand to eliminate the non-authoritative ones. 

The solution? A new acronym, obviously

Right now our software engineer, Jon Connolly, is patching our code to deal with different edge cases. In a week, a summit meeting of seismic networks is going to take place at which we'll review the business rules for authoritativeness, and modify them as necessary.  Amusingly, we are moving forward beyond EIDS and into the NEXT generation of earthquake information exchange.  That will probably feature some sort of uniform event ID, but I don't know too much about it other than its acronym--PDL (for Product Development Layer) -- and isn't that what's really important -- another acronym?