While significant progress toward earthquake prediction has occurred in the past four years, the current guarded optimism for continued progress reflects in part revised goals and new definitions. First, recognition [e.g., Evcrndcn, 1982] of the failures of the search during the 1970s for a reliable readilyobservable earthquake precursor has reemphasized the need to understand the fundamental physics of the earthquake generation process [e.g., Stuart, 1984/85] and brought acceptance of a probabilistic rather than a deterministic approach to earthquake prediction. That is, we now accept the notion that earthquake predictions must be couched in terms of probability gain and likelihood estimates. Second, a prediction terminology [Wallace et al., 1984] has been largely adopted that separates and expands •he tasks into long-term earthquake potential (no specific time window) and long-term prediction (time windows. of a few years to a few decades}, where progress has been significant, and intermediate-term prediction (time windows of a few weeks to a few years} and short-term prediction (time windows up to a few weeks}, where progress has lagged. In the 1970s, earthquake prediction usually was taken to mean a deterministic short-term or intermediate-term warning; now, probabilistic estimates of earthquake potential [e.g., Lindh, 1953; Sykes and N&henko, 1984] are accepted as the most reasonable and valid form for expressing the likelihood of a future earthquake and its hazards, as well as the uncertainties. Long-term earthquake potential and long-term earthquake prediction. Although there are differences in the use of the term 'seismic gap' [e.g., Habermann et al., 1983], for our purposes a seismic gap is a segment of a plate boundary that has not re-centl¾ ruptured so that a future shock is anticipated there. The continued successful use of the notion of seismic gaps to anticipate the size and location of large plate boundary-rupturing shocks (e.g., see Nithenko [1985] and the gap-filling 1985 central Chile M = 7.8 shock} is evidence that a methodology for the long-term prediction of at least some large interplate shocks is available. In recent years, attention has been focused on a number of as-yet-unfilled seismic gaps in the United States. On the Pacific-North American plate boundary in Alaska, future large shocks are considered likely along the Shumagin, Yakataga, and Adak sections [e.g., Shearer, 1986; Hags and Gori, 1986], although there is some evidence that little shear strain is accumulating in the crust near the Shumagin section so that a great thrust shock may not occur there [Savage and oeisowtki, 1986a]. Although no large historic shocks have occurred alongthe Washington and Oregon coasts, there is concern that the subducting Juan de Fuca plate could be the source of a great earthquake in the Pacific Northwest [e.g., Heston and Kanamori, 1984]. Similarly, there is concern that a magnitude 7 shock could rupture the historically quiet White Mountain This paper is not subject to U.S. copyright.Published in...
