California Institute of Technology

Engineering Applications

With only a few seconds to a minute of early warning time provided by earthquake early warning (EEW) system, what can we do to mitigate the potential earthquake loss? Furthermore, how can we optimize the potential benefit through various kinds of practical applications?

To answer the first question involves creativity and a lot of cross-disciplinary communication. Some good examples can be learned from Japan, a place with a lot of experience with earthquakes and EEW. Interesting applications have also appeared in research journals (see references below). The California Institute of Technology has a research group in civil engineering working on EEW engineering applications. Figure 1 shows some examples of potential or existing EEW applications.

To answer the second question, we have to first understand the limitations of EEW. Two fundamental problems of EEW that restrict its applications are: SHORT warning time and LARGE uncertainty. If we know for sure that an earthquake is going to soon strike a structure and induce a response of sufficient severity, then planned mitigation actions can be taken immediately in order to mitigate potential earthquake losses. In reality, the early warning information will involve some uncertainty and some EEW applications will produce a substantial economic loss if a false alarm occurs. Therefore, the decision of whether to take a mitigation action should involve sophisticated cost-benefit analysis. However, human intervention during this process would likely use up too much of the short warning time, preventing the mitigation actions from being activated in a timely manner. This motivates the development of an earthquake probability-based automated decision-making (ePAD) system, which is an on-going project of Prof. Jim Beck’s research group at Caltech. Figure 2 shows a brief introduction of ePAD’s fundamental concepts.

In addition, an earthquake early warning system in California is currently being tested through the California Integrated Seismic Network (CISN). The system aims to provide warnings in seconds or tens of seconds prior to the occurrence of ground shaking, depending on the distance to the epicenter of the earthquake. The estimated location and the magnitude of the earthquake will be updated in real time on a second by second basis. Similar to other earthquake early warning systems, the seismic intensity of the ground motion in a user's location will be provided. However, the shaking level experienced by a user in a tall building will be significantly different from that on the ground. Prof. Tom Heaton’s group is interested in the rapid estimation of seismic motion intensity in tall buildings during earthquakes. The building early warning system will send a message, including the expected shaking level and shaking duration, to the users in the building when it is about to be shaken by an earthquake; such information has shown to be capable of mitigating panic and confusion (Kubo et al, 2011).

Figure 1. Categorization of potential and existing EEW applications.

Figure 2. Basic structure of ePAD. DV stands for Decision Variable (Porter 2003). PBEE stands for Performance-Based Earthquake Engineering (Porter 2003), developed by PEER. RVM stands for Relevance Vector Machine (Tipping 2001), which is an advanced machine learning technique to create a simple surrogate model for complex model.


James Beck, Thomas Heaton, Ming Hei Cheng, Stephen Wu