# Estimating Ground Motions In The Largest Crustal Earthquakes

Earthquakes cause many deaths and injuries worldwide. For example, the 2015 Nepal earthquake is thought to have killed nearly 9,000 people. This event was only one of many events contributing to the 750,000 deaths attributable to earthquakes during the period 1996 to 2015 according to a recent report by the United Nations.

This is a higher total than deaths caused by all other natural disasters (e.g. storms, heatwaves and floods) combined. Earthquakes also cause considerable economic losses. For example, the 2016 Kumamoto (Japan) earthquake is thought to have caused \$42 billion-worth of damage according to the Japanese government. The largest earthquakes can also have dramatic and permanent effects on the landscape, e.g. roughly 180km of surface faulting occurred in the 2016 Kaikoura (New Zealand) earthquake.

The vast majority of earthquake-related deaths and injuries and economic losses are caused by damage (including total collapse) to buildings or their contents. When designing buildings and infrastructure (e.g. bridges and power plants) to withstand earthquakes, engineers require estimates of the ground motions that their structure could be subjected to in future earthquakes. These estimates are computed by engineering seismologists using two sets of mathematical models.

The first of these provides an assessment of the rate of occurrence of earthquakes of different sizes within a few hundred kilometers of the structure, e.g. how often does a magnitude 6 earthquake occur on a nearby fault? The second model predicts what the ground motion would be given the occurrence of different earthquakes, e.g. if the magnitude 6 earthquake occurred what is the expected ground shaking at the base of the structure? Many hundreds of these so-called ground-motion models have been published.

One of the greatest challenges in developing ground-motion models is a lack of recordings of previous earthquakes, particularly those within the region of interest. This challenge is particularly acute when considering the largest earthquakes because they happen so rarely. Generally, less than one earthquake of magnitude 7.2 or larger occurs in the Earth’s crust every year and is recorded at close distances by modern seismographic networks.

In our recent study, we collected data from 38 crustal earthquakes of magnitudes 7.2 or larger that had occurred worldwide over the past 60 years. Many of these data had been forgotten about and never collated before and hence they had not been used to derive the most recent ground-motion models. Because of its importance for seismic design and since it is often the only information available concerning ground motions in past earthquakes in our study we considered the maximum horizontal acceleration measured from each of the available seismograms. The newly-collated data were compared to predictions from eight ground-motion models that are routinely used to assess the earthquake shaking that a structure may suffer during its lifetime.

We found that these eight models provide, on average, a good match to the maximum horizontal accelerations observed in the largest crustal earthquakes. This is reassuring for earthquake engineering. The study, however, did show that the ground motions in some earthquakes (e.g. the 2001 Bhuj earthquake in India) were much higher than those that would be expected given the magnitude.

The data collated for this study will be invaluable for the derivation of new ground-motion models for use in the design of safer structures and consequently for the reduction of earthquake risk.