No abstract

SummaryIn many parts of the world, earthquakes threaten regional infrastructure systems. For modeling risk using stochastic earthquake catalogs, random variables include rupture location and the damage state of different components. Thus, there is an infinite set of possible damage maps that a risk modeler could evaluate in an event‐based probabilistic loss model. Even a finite but large number of damage maps may not be practical, because many network performance measures are computationally expensive. Here, we show a computationally efficient method for selecting a subset of damage maps, corresponding ground‐motion intensity maps, and associated occurrence rates that reasonably estimates the full distribution of the ground‐motion intensity and a target performance measure using optimization. The method chooses a subset of maps and associated annual rates of occurrence that minimizes the error in estimating the distribution of a network performance measure as well as the marginal distributions of ground‐motion intensity exceedance. The joint distribution of the ground‐motion intensity is implicitly included in the objective function of the optimization problem via the network performance measure. We then show how to tune the optimization parameters based on consistency checks related to the network performance measure and the ground‐motion hazard. We illustrate the proposed method with a case study of the San Francisco Bay Area road network to estimate the exceedance curve of the average percentage change in morning commute trip time. This work facilitates expanded and risk‐consistent studies of the impacts of infrastructure networks on regional seismic risk and resiliency. Copyright © 2014 John Wiley & Sons, Ltd.

Using sensor data from fixed highway traffic detectors, as well as data from highway patrol logs and local weather stations, we aim to answer the domain problem: "A traffic incident just occurred. How severe will its impact be?" In this paper we show a practical system for predicting the cost and impact of highway incidents using classification models trained on sensor data and police reports. Our models are built on an understanding of the spatial and temporal patterns of the expected state of traffic at different times of day and locations and past incidents. With high accuracy, our model can predict false reports of incidents that are made to the highway patrol and classify the duration of the incidentinduced delays and the magnitude of the incident impact, measured as a function of vehicles delayed, the spatial and temporal extent of the incident. Equipped with our predictions of traffic incident costs and relative impacts, highway operators and first responders will be able to more effectively respond to reports of highway incidents, ultimately improving drivers' welfare and reducing urban congestion.

No abstract