Over the next week, Southern California has only a 27% chance of experiencing a third earthquake greater than magnitude 6, but a 96% chance of going through a tremor of magnitude 5 or higher. Those precise probabilities were generated by scientists at the United States Geological Survey (USGS), using models based on longstanding principles of seismic behavior and decades of data on aftershocks from earthquakes. But the same predictive power does not extend to forecasting when and where earthquakes will strike in the first place, experts acknowledge. “Even if it’s a theoretical possibility, it may be a practical impossibility,” said Andrew Michael, a California-based geophysicist at the USGS. A powerful 7.1 magnitude quake shook the remote town of Ridgecrest in the Mojave Desert on Friday evening, a day and a half after a 6.4 magnitude temblor was recorded in the same region. On average, a quake is followed by an even stronger tremor only about 5% of the time, though it happens more frequently in areas with significant geothermal activity, like the Mojave, according to Michael. The USGS first began offering public aftershock forecasts in the 1980s, Michael said. The models rely on basic laws governing earthquake behavior that have been known for the better part of a century, experts said. The typical shallow earthquake creates a series of aftershocks that diminish exponentially, with each successive day bringing half as many tremors. In addition, the frequency of earthquakes drops as the magnitude increases – a region will have 10 times as many magnitude 6 quakes as magnitude 7, Thomas Heaton, a seismologist at the California Institute of Technology, said in an email. “There is now a catalog that stores the locations, sizes and times of millions of past earthquakes,” he said. “It is quite straightforward to characterize the statistical behavior of these events.” The current model used by USGS predicts the number and size of aftershocks based on the largest earthquake, or mainshock. But seismologists believe aftershocks behave more like the spread of a disease in an epidemic, Michael said. Just as each sick patient can infect others, so too can each aftershock in turn create its own subsequent aftershocks. Michael said the USGS is perhaps a year away from adopting a newer model that incorporates that theory in making public forecasts. Eventually, he said, the hope is that officials will be able to predict how much shaking a particular town might experience in the aftermath of a quake, so people can be better prepared. But predicting a big quake in the first place remains out of reach. The best seismologists have been able to do is to use historical data to predict how likely it is that an earthquake of a certain magnitude will hit a particular region over a period of time. That forecast is crucial for establishing safety standards in building codes, for instance. But it will not help warn a city’s residents that a quake is imminent. “The problem is the earth is a complex thing,” said Christopher Scholz, a professor emeritus at Columbia University’s Lamont-Doherty Earth Observatory. Scholz noted, for instance, that the 1999 Hector Mine quake in the Mojave Desert was thought to have been triggered by a 1992 quake. That seven-year gap is but a fleeting moment in geological terms. Federal and state officials hope a new system that alerts residents when shaking is detected might give people a few seconds to seek shelter, say, or stand in a doorway. “I work on the physics of the rupture process and I have become convinced that the physics is in a class of phenomena that are called ‘chaotic,'” Heaton said. “If this is true, then we will probably never predict earthquakes.”