Residents of Southern California are arguably more earthquake-savvy than citizens of any other region of the United States; we've had to be. The region also boasts an impressive depth of earth science expertise among its local media professionals, print and broadcast journalists alike. Yet one aspect of earthquake sequences seems to retain its capacity to surprise us: the late, large aftershocks that can occur months or even years after a significant earthquake. The twin weekend temblors of late April were the latest example of this phenomenon, coming on the heels of a late, large aftershock of the 1992 Landers earthquake that struck near Barstow on March 18 and was large enough to be widely felt in the greater LA region. How can these events be aftershocks? These mainshocks occurred 3 and 5 years ago, and it's been months, if not years, since aftershock rattling has been a concern.
The good news is that, compared to other earthquakes, aftershocks are usually well-behaved: they are defined to be events that occur within close proximity to the mainshock (no further from the mainshock rupture plane than the rupture plane was long). Events are classified as aftershocks as long as it takes for the regional seismicity to return to the pre-mainshock levels.
These definitions will obviously blur around the edges; the late April Northridge events were somewhat larger and later than expectations, based on average statistical models. It is fair to say that they were not expected, but that is the nature of probabilities and random processes: sometimes you flip a coin five times and get five heads. You wouldn't predict it, but neither would you faint from shock if it occurred once in a while. In fact, the nature of random processes is that the low-probability outcomes are guaranteed to happen, a small percentage of the time.
The bad news is that, by ever criteria used, an individual aftershock does not differ from an individual earthquake: the rupture processes and the ground motions are indistinguishable. As residents of the Newhall region discovered, a M5 aftershock can wreak havoc in a grocery store as well as any other M5 event. The oft-heard phrase, 'just an aftershock' carries a connotation that aftershocks are less worrisome than other earthquakes. But that is not necessarily so.
Within aftershock sequences, it is clear that subsequences occur all the time: a distinct burst of activity might poke its head up above the background aftershock rate, and then die off on its own timetable. Thus, aftershocks can have their own aftershocks, and some of them are also foreshocks to bigger events that follow (that is, bigger than the aftershock, although not the initial mainshock). A full statistical analysis of aftershock sequences has never been done; we do not consider probability increases caused by events that are themselves aftershocks in any of the existing models. At the very least, however, it is clear that subsequences do occur.
The complexity in aftershock sequences arises largely from the three-dimensional complexity of fault systems. Some aftershocks do occur on the same segment of fault that generated the mainshock, but, very typically, a large number of nearby related faults will generate substantial aftershock activity. Angelenoes may remember the M6.5 Big Bear aftershock to the 1992 Landers earthquake; Big Bear is considered an aftershock, but occurred in a region and on faults not even connected to those that produced the mainshock. The aftershocks near Barstow, including the this years' M5.1 event, were similarly detached.
But can a so-called aftershock be a foreshock to an event even bigger than the mainshock? The answer, if the question pointedly includes the word, 'can', is yes; one has only to look back to 1992 to see an example, when the April 1992 M6.1 Joshua Tree earthquake generated its own aftershock sequence, including events to the north, from where the June M7.3 Landers earthquake eventually initiated. But the question we really want answered replaces 'can' with 'will', and this one we can't answer. No earth scientist would be shocked if another Northridge-type earthquake occurred towards Simi Valley tomorrow; neither would we be shocked if nothing else big happened for another 50 years. The best answer we have comes back to statistics of observed earthquake sequences: if you want numbers, 1-in-20 are our best assessment of the odds than any earthquake, aftershock or not, will be followed by something bigger than itself. Two things can be noted: 1-in-20 is not zero, and the odds of a M6 or greater event have to be considered at least higher than they were previously, in the aftermath of a M5 earthquake (or two). But neither can 1-in-20 be considered bad betting odds: facing that chance of rain, you probably would not choose to lug an umbrella around all day. Since, with earthquakes, somewhat more is at stake than getting wet, the very best advice an earth science professional can give you is to use these early morning wake-up calls as just that. Go do all of those things that you know you should, but maybe haven't, to make sure you and your loved ones will be prepared, the next time 5 heads in a row do come up.