User login


You are here

Ph.D. Call Doctoral School in Civil, Environmental and Mechanical Engineering - University of Trento, Trento, Italy

oreste.bursi's picture

- Reference persons: Marco Broccardo (UNITN/DICAM), Oreste S. Bursi (UNITN/DICAM) ( -

C8 - scholarship on reserved topics

Funded by: MUR – Departments of Excellence

Title: From Natural to Anthropogenic Seismicity: A Holistic Risk Framework based on Probabilistic graphical models and Vine Copula theory



The primary goal of earthquake engineering is to ensure life safety and reduce the financial impact to acceptable levels. In addition to natural seismicity, the more recent anthropogenic seismicity is posing a societal quandary and an economic challenge for the feasibility of projects that rely on the use of deep-underground resources. This project tackle at the fundamental level the methodological framework for computing seismic risk in both cases, providing novel insights into static versus dynamic risk analysis.


Candidates can prepare proposals on the following themes:


1) A static vine copula approach for natural seismic risk analysis:


Since the late 60s, the computational methods to quantify seismic risk have been conceived as the convolution of three components: hazard, vulnerability, and exposure. The first attempts of computing anthropogenic-seismic risk are based on the same principles, although accounting for non-stationary seismicity. These classical approaches, which follow an apparent logical structure, had the advantage of decomposing seismic risk into different tasks, which have been handled—separately—by a different group of experts. Differently, this research proposal calls for a—holistic— approach for seismic hazard and risk quantification, which is based on merging the three components by using the most recent advancements in probabilistic graphical models, vine copula theory, and stochastic simulators. Rather than following the current research trends which constrain these methods within the current paradigm of seismic risk computation, we will use them to reformulate the foundations of the probabilistic framework. This will address the following open questions: 


·         What is the correct probabilistic structure among the seismic and engineering variables defining the natural seismic risk?

·         What is the role of the neglected dependencies in the seismic risk computation? 


We tackle this research brief using the vine copula theory. Precisely, we use the vine copula approach to formulate the multivariate probability distribution of all the variables defining the seismic hazard and risk framework. Moreover, we introduce a novel hierarchical stochastic model for simulating and forecasting statistically compatible ground motion time series. This enables the combination of computational science and engineering tools with the most recent advancements in uncertainty quantification and reliability analysis for estimating structural and infrastructural responses. Finally, we validate the framework with a series of benchmark analyses on complex structural and infrastructural systems.

2) A dynamic vine copula approach for natural seismic risk analysis:

In recent years, the increase in anthropogenic seismicity due to subsurface exploitation has posed a new scientific and technological challenge. Moreover, the trend is expected to continue in the next years as the exploitation of subsurface resources is an essential component of the green transition. Currently, the anthropogenic seismic risk has been tackled with the same approach of natural seismicity. However, its nature is fundamentally different. First, it is an anthropogenic and industrial hazard, which by definition should have different safety standards; second, it is a dynamic risk that depends on a time-varying stochastic feedback system. The problem is time-varying because the activity producing the hazard (usually fluid injections) is time-varying. The feedback is given by the monitoring system, which is, in general, present in these activities. Lastly, the problem is inherently stochastic has the uncertainties related to subsurface problems are simply vast. The key research questions read as follow

·         How should we merge the know-how of computational science and engineering with health monitoring at a fundamental level? 

·         Can we formulate a robust probabilistic forecast of the seismic activity and the associated risk in the case of anthropogenic seismicity? 


We tackle this research brief by combining vine copula theory and elements of probabilistic graphical models. Specifically, we define a Markovian time-variant probabilistic structure that accounts for all the dependencies governing the anthropogenic seismic risk. The framework is a dynamic extension of the static framework proposed for natural seismicity, fully integrated with the data acquisition scheme. Moreover, we extend the hierarchical stochastic model for simulating and forecasting anthropogenic ground motion time series and dynamic risk computations. Finally, we use the framework for developing a probabilistic forecast model of anthropogenic seismic activity and the associated risk.


Expected output (for both themes):

The outputs are summarized as follow:

i)              Three research papers

ii)             National and International conferences

iii)           Open-source MATLAB and/or Python package.

Subscribe to Comments for "Ph.D.  Call Doctoral School in Civil, Environmental and Mechanical Engineering - University of Trento, Trento, Italy"

Recent comments

More comments


Subscribe to Syndicate