Probabilistic Modeling

You are here

Probabilistic Modeling
Probabilistic Modeling

Research Focus Area

Developing innovative approaches in probabilistic analysis and transferring methodologies from one field to others, including:

  • probabilistic regional hazard modeling
  • non-homogenous virus spreading
  • big-data analysis
  • infrastructure systems resilience and so on.

Lehigh's Probabilistic Modeling Group (PMG) conducts research in several fields of applied science and engineering. For instance, they are currently combining methodologies from probabilistic regional hazard modeling, non-homogeneous virus spreading, and big-data analysis to predict the probability of Ebola virus reaching various cities of a broad geographic region. Another interest is resilience, the ability of a system to withstand external perturbations and rapidly recover to a satisfactory level of performance. In this line of research they develop computational techniques for the quantification of the resilience of infrastructure networks with respect to seismic events, addressing all phases of the recovery (immediate emergency, mid-term infrastructure recovery, and long-term socio-economic recovery).

Research Focus Group Members

The team includes approximately ten core members, and 20 additional members. This collaborative effort has led to a number of successful interdisciplinary proposals. It has established a new graduate certificate across colleges. Other activities include symposia and retreats.

Probabilistic Modeling Group


New approach to the stochastic representation of hurricanes & storm surges

New Research Project: "New approach to the stochastic representation of hurricanes and storm surges"
New collaborative interdisciplinary project started across engineering and mathematics, supported by accelerator grant from Lehigh University. The risk from natural hazards for a specific region is usually assessed predicting the losses associated with a set of representative extreme event scenarios. The selection of these scenarios is extremely delicate, because computational resources typically constrain heavily the number of cases that can be investigated. Hence, the selected scenarios have to be as few as possible, and yet provide a comprehensive probabilistic description of the regional hazard. The investigators have developed a technique called "Hazard Quantization" (HQ) to achieve this goal, but it falls short when there is the need to describe multiple hazards occurring simultaneously, such as strong hurricane winds and storm surge. This project will investigate a new methodology that can capture simultaneously multiple hazard intensity measures spread over a region. The team includes experts in stochastic modeling of random functions and experts in analysis of regional natural hazards.
The project will result in a new approach grounded in solid mathematics, that will yield substantial enhancements in the computational efficiency of these analyses, enabling their applicability to practical-size problems and, ultimately, advancing disaster mitigation and response.
Currently a involved in this project is Paolo Bocchini and postdoc Vasileios Christou from RCEAS. And from CAS, Daniel Conus, Wei-Min Huang, plus an undergraduate student (and hopefully in September also a PhD).