WHY DO EARTHQUAKE HAZARD MAPS PREDICT HIGHER SHAKING THAN OBSERVED?

One of seismology’s most important contributions to society are hazard maps that predict levels of earthquake-caused ground shaking that should be exceeded with only a certain probability over a given period of time. These maps are used worldwide to develop building codes for earthquake resilient structures. Billions of dollars-worth of property – and countless lives – rely on the maps’ accuracy. Despite their importance, a major problem has persisted: until recently, no one tested how well these maps predict the shaking that actually occurs.

Using the California Historical Intensity Mapping Project (CHIMP) dataset of shaking intensities from the largest Californian earthquakes over the past 162 years, we compare historical shaking observations to hazard maps for California. We find that the maps overpredict relative to observed historical shaking. We find similar results for Italy, Japan, France, and Nepal. The consistency of this overprediction for different regions suggests a possible fundamental issue with the methodology of the current maps.

We explore what effects – or combinations – cause this problem and how it can be fixed. The question is challenging because maps use the limited observed historical earthquake record and geological faulting data to predict where and when earthquakes will occur, how large they will be, and how much shaking will result. We explore multiple possible effects – variations in near surface geology, inconsistent minimum earthquake magnitudes between data and maps, uncertainties in the shaking measurements, mean vs. median predicted hazard, and conversion equations between ground motion parameters and shaking intensities. The last seem to be the most significant contributor to the overpredictions, which can be reduced substantially by new conversion equations.