HOW GOOD SHOULD WE EXPECT EARTHQUAKE HAZARD MAPS FOR CONTINENTAL INTERIORS TO BE?

The short earthquake record relative to the low seismicity rates in continental interiors like eastern North America make it challenging to make hazard maps and infer how well they should be expected to predict future shaking. We explore these questions by simulating an area's shaking history and comparing “observed” shaking to that predicted by a map generated for the same parameters. The simulations yield shaking distributions with mean consistent with the map, but individual shaking histories show large scatter. Infrequent large earthquakes cause shaking much stronger than mapped. An earthquake history can yield shaking higher or lower than that predicted while being consistent with the hazard map. The scatter decreases for longer observation times because the largest earthquakes and shaking are increasingly likely to have occurred. For the same reason, scatter is much greater for a continental interior than a more active plate boundary. For a continental interior, where the mapped hazard is low, even an M4 event produces exceedances at some sites. Larger earthquakes produce exceedances at more sites. Thus many exceedances result from small earthquakes, but infrequent large ones may cause very large exceedances. However, for a plate boundary, an M6 event produces exceedance at only a few sites, and M7 produces them in a larger, but still relatively small, portion of the area. A real map involves assumptions that are unlikely to be exactly correct and thus will contribute additional scatter. For example, it is unclear whether the largest possible earthquakes along the eastern continental margin of North America have occurred. Generating synthetic earthquake histories and sampling them over a few hundred years shows that the maximum magnitudes appearing most often in simulations are essentially those observed, and smaller than the simulation maxima. Future earthquakes along the margin may thus be significantly larger than those observed to date. As reality gives only one history, it is hard to assess whether misfit between actual shaking and a map — notably higher-than-mapped shaking — arises by chance or reflects biases in the map. Due to this problem, there are limits to how well we can expect hazard maps to predict future shaking, and to our ability to test the performance of a hazard map based on available observations.