The territory of Romania is characterized by moderate to high seismicity. Notably, the Vrancea zone is known for producing relatively large intermediate-depth earthquakes that pose a high risk for the capital Bucharest and other Romanian cities. Other regions, like Galati, have also experienced recently shallow significant seismic activity that impacted the Galati city and neighbouring areas. Here we propose to build a comprehensive analysis and forecasting system for Romanian's seismicity, based on state-of-the-art forecasting models that are currently in use in seismic countries like US, Japan, Italy and Greece. We are planning to use advanced statistical and machine-learning tools to quantify the spatial and temporal features of Romanian seismicity that can be used for short (days to weeks) and medium (several months to one year) earthquake forecasts. The scientific results of the project will be published in international journals and on a dedicated website that finally aims to present real time forecasting information.
AFROS project aims to create an integrated framework for seismicity monitoring, analysis and forecasting for the Romanian territory. Besides of the more traditional techniques used for seismicity analysis and forecasting (e.g., temporal changes of earthquakes’ rates and/or frequency-magnitude distribution), the project will explore new machine learning algorithms that may better reveal “hidden” characteristics of earthquake occurrence complexity. Rigorous statistical testing will be done to discriminate between spurious seismicity changes and those that may have some geophysical background, with potential precursory character.
(1) The quantification of seismicity patterns for the intermediate-depth Vrancea earthquakes and the crustal, shallow seismicity that occurs in several seismogenic regions in Romania. This quantification implies the use of both traditional approaches, based on the space-time frequency-magnitude distribution of earthquakes, space-time clustering properties and Coulomb static stress changes, as well as new, exploratory approaches that rely on Machine Learning.
(2) Forecasting of seismicity, based on the patterns defined at (1). The forecasting will be attempted for various time-windows, defined here as medium (one year, three months) and short (one month, one week), following in general the CSEP (Collaboratory for the Study of Earthquake Predictability) forecasting framework. This step requires also choosing appropriate statistical tests for evaluating the performance of forecasts. We are planning to design a dedicated internet platform, where data and algorithms will be embedded with the purpose of generating semi-automatic forecasts.
(3) Incorporation of other geophysical data, besides seismicity, into the monitoring, analysis and forecasting. We are referring here to the carbon dioxide, radon emissions, air ionization and electromagnetic data that are already recorded by a multidisciplinary network of sensors in the Vrancea region. We will analyse possible correlations of these observables with seismicity and the probability gain that such geophysical data may bring for the forecasting.
We propose the elaboration, for the first time for the Romanian territory, of seismic forecasting algorithms that will be able to be applied in real time (or quasi-real time). This effort is part of a broader international framework for understanding different seismic regimes, as well as the space-time interaction between earthquakes, and for monitoring and forecasting seismicity.
In particular, we will focus our effort on estimating and forecasting the parameters of intermediate-depth seismicity in the Vrancea region. The algorithms used will also benefit from the use of ML (machine learning) techniques. The two PhD students participating in the project will be specially trained to acquire the appropriate expertise.
With the necessary changes to take into account the differences in the seismic regime, we will also develop algorithms for estimating and forecasting the parameters of crustal seismicity in Romania. The project will also investigate possible correlations between geophysical observations (e.g., variations in radon emissions or electromagnetic field) and seismicity.
The activities mentioned above will include the elaboration of research reports or regarding the preparation of the two doctoral students, the creation of a web page reflecting the development and activities of the project, the presentation of the project results at national/international conferences, the publication of papers in ISI journals and the creation of some informative flyers. We will also create a functional platform for data visualization and detection of seismic anomalies.