What’s the difference between predicting and forecasting earthquakes?
A prediction of an earthquake needs to state exactly where and when the event will happen, with enough specifics to be useful for response planning purposes. For example, the statement "there will be an earthquake tomorrow at 7:45 AM" is almost certainly going to be correct somewhere in the world, but it has no value as a prediction. Similarly, saying that there will eventually be a large earthquake on a very active fault is useless; while it is true, almost nothing can be done with this information. Without a specific date or location, a statement cannot be a prediction.
Currently, no one can predict where or when big earthquakes will occur to meet these criteria. However, seismologists have gotten much better at forecasting earthquakes. On a sunny Monday afternoon, meteorologists may forecast a wet weekend ahead, and can add an estimate of how certain they are that their forecast will be accurate; i.e. “we forecast a 60% chance of rain on Saturday.” That can also be expressed as “there is a 40% chance that it will not rain on Saturday.” Earthquake forecasting is very similar, and comes in different flavors that inform how to build appropriately in a given area. The USGS National Seismic Hazard Maps provide us with information on how strongly the ground is likely to shake in any given area within a 50 year period.
The probability of a specific fault producing an earthquake of a certain size can also be calculated, but these estimates are much rougher; in many cases, we only know about a few (or maybe only one) earthquakes that occurred there in the past. Inevitably, some guesswork goes into estimating how likely it is that another earthquake will occur in the next 50 years. In a region such as the Puget Sound area, we can collect additional information to help keep forecasts reasonable. By using GPS instrumentation to track how much strain accumulates as the Puget Sound region is squeezed up against British Columbia, it’s possible to calculate a strain or deformation “budget”. The regional strain accumulation can only justify so many earthquakes over a 10,000 year period, and tracking it can help to avoid forecasting unrealistically high or low numbers of earthquakes on faults in the region.