Research:

Problem definition, aim and motivation

The main thrust of the project is in the identification of relevant parameters that can be extracted from measured geophysical signals (gravimetric, seismic, electromagnetic, etc.), and which characterize various regimes of crustal activity (calm, pre-earthquake, earthquake, aftershocks, submission) in a given region. The ultimate goal is the reliable identification of earthquake precursors (if such precursors exist), with immediate application in early warning and regional disaster management efforts. The study is justified by the fact that Romania is home to one of Europe’s active seismic regions, with a history of major earthquakes, which can potentially affect the livelihoods of a significant population.

The methodology that we propose is innovative in several respects. First, we are advocating a data mining approach which effectively uses the complex nature of interactions within the Earth’s crust. In our vision of geodynamics, as introduced by F. Munteanu and collaborators at the Romanian Academy Institute of Geodynamics, geophysical signals reflect a complex interaction between the crust, mantle, oceans, atmosphere, ionosphere, and a non-negligible effect of human activity. Therefore, we will not only investigate one source of signals (e.g., seismic), but also signals from many other sources (gravimetric, electromagnetic, ULF, etc.). Second, the search for relevant parameters which describe geophysical signals is conducted within the range of nonlinear signal descriptors, which we think are better suited to grasp the highly nonlinear nature of physical processes at play inside the Earth’s crust. Third, we propose a novel monitoring approach which involves the construction of an abstract parameter space (the equivalent of a “phase space” in Physics), and the application of data mining and pattern recognition tools to the identification of regions within this space which correspond to “normal” and “abnormal” crustal activity.  The utility of such a parameter space extends well beyond the development of new earthquake early warning tools. One possible application is “fingerprinting” of various seismically active regions of the world, i.e., the identification of the region of phase space which is occupied by each fault line or active zone, in other words, its own “voice”. Another valuable application is the comparison within  phase space, between actual measured crustal events and patterns originating from various theoretical or computer models of crustal activity, which may ultimately contribute to model validation and to a better understanding of the complex Physics of plate motion.

The project is scheduled to unfold during a period of 3 years, (approx. Jan 2006 - Jan. 2009). Research is conducted by Stefan Gheorghiu, Ph. D., under the supervision of Dr. Florin Munteanu, the director of the Center for Complexity Studies in Romania.

1. The heterogeneity of the Earth’s crust plays a subtle, but crucial role in geodynamics and earthquake science, and ultimately influences the success of earthquake monitoring. As revealed by almost a century of documented data and theoretical study, the crusts is a complex dynamical system of plates and sub-plates, which continuously break up and re-assemble following local interactions. Early studies by Sadovsky et al. [1] showed that structuring by fragmentation leads to a power law distribution of sizes. The idea was further refined by Cheng and Redner [2], the group of Munteanu in Bucharest [3], and notably by Sornette et al. [4], who showed that the size distribution of crustal plates and subplates is a fractal-like power law extending over many orders of magnitude in size. This peculiar size distribution has many interesting consequences, one of which being that plate size may correlate with certain frequency ranges of recorded geophysical signals. Maybe more importantly, Sornette [5] argues that a major earthquake can be modelled as a critical phenomenon (like a phase transition), and that the knowledge of the discrete nature of the size distribution of plates/subplates may provide a promising handle on predicting the timing of such devastating events. Also, we should mention that the time scale of characteristic inter-plate stick-slip motion may also be associated with plate size. These observations may be important for the development of signal monitoring strategies which form the object of the current project.
2. The equivalent model of an active region of the crust is continuously changing, as the system undergoes internal re-organization to accommodate loading stresses. The dynamics of the system depend as much on the load, as on the rate of loading. Per Bak’s Self-Organized Criticality (SOC) model [6] proved very popular for the description of earthquake activity. The drawback of the classical SOC theory is that it does not allow any event prediction whatsoever, and this fact is often disproved by evidence in the field. Newer, more sophisticated models [7] allow a certain amount of predictability by allowing a temporary de-correlation of the stress on a large scale. The idea is that the system continuously goes in and out of a critical state, and that progression towards criticality is accompanied by internal reorganization and energy dissipation. The concept of internal re-organization is central to the theory of complex systems and also to the current project. We propose that one may get a handle on the accumulation of tectonic stresses by monitoring this re-organization process. In general, internal re-organization is reflected in changes in the degree of complexity (e.g., the number of degrees of freedom of an equivalent model) of signals output by such a system. This vision may also provide a handle on the “calm before the storm” phenomenon, in which major catastrophic events are often preceded by a period of unusual calm.
3. Clues to stress accumulation are to be found in the calm “noise” (e.g., microseismicity) associated to plate movement. As mentioned above, we believe there may be clues to internal reorganization and stress accumulation encoded in the complexity of geophysical signals measured in active seismic areas. A key assumption is therefore that one has to monitor the calm periods (microseismicity), and that the type of information we are looking for (patterns, changes in signal complexity, etc.) is not carried by the seismic traces of the major event themselves. Most measuring stations are geared towards recording major tremors, which are irrelevant to the current project.
   d)The nature of measured signals is essential to the success of earthquake monitoring. Simultaneous measurements of different kinds (microseismicity, electromagnetism, etc.) should be combined to form a unified picture of dynamics. We are advocating a multiparametric heuristic approach to earthquake monitoring, which does not assume any particular crust model. Still, any observations we make about the behaviour of a representative point in an abstract parameter space may help prove or disprove the validity of one or the other of existing crust models. 
4. The geographic position of measurement stations may play a crucial role in earthquake risk management. Recent theoretical developments hint at the possibility that an important aspect of seismic phenomena has been largely overlooked so far by traditional geophysical practice. More often than not, the placement of seismic measurement stations and observatories was chosen for either obvious reasons (e.g., on top of known fault lines), or for convenience (proximity to existing infrastructure, historical buildings and existing equipment, etc.). So far, to the best of our knowledge, nobody seriously considered exploring the optimal placement of measuring stations, both with respect to latitude/longitude, but also depth. The particular nature of the crust as a rather thin interface between the visco-solid Earth and the atmosphere/oceans makes for potentially very interesting “membrane” behaviour. Recent studies by the group of Sapoval [8] have shown that the vibration of membranes with fractal boundaries has some peculiarities, such as the capability to produce localized modes. Correlating this observation with the fact that the crust is a fractal-like aggregate of sub-plates (point a) above), one has to seriously consider the possibility that there may be preferential measurement positions, which may convey qualitatively different information than that picked up by neighboring observatories. Moreover, the membrane nature of the crust makes it an effective “noise amplifier”, so that again, surface probes may measure qualitatively different information from probes placed deep underground. Interestingly, through the Romanian Academy, our group has access to a measuring station unique in this part of the world, which is placed 800 m inside a mountain in Western Romania, on the site of an ancient mine. As part of the current project, we plan to investigate and qualify differences between deep and surface geophysical measurements, as to the information pertaining to earthquake risk management.

At the end of this stage, we will have a theoretical foundation for the proposed monitoring method, and an optimized abstract parameter space for seismic monitoring of the Vrancea active zone in Romania.

YEAR 3: Implementation of the seismic measurement protocol.

Details of this phase will be published later.

References

[1] M. A. Sadovsky, Geofizika 19, 69 (1983).
[2] Z. Cheng, S. Redner, Phys. Rev. Lett. 60, 2450 (1988)
[3] C. Suteanu, C. Ioana, F. Munteanu, D. Zugravescu, Revue Roumaine de Geophysique 42, 15 (1998)
[4] D. Sornette, V. Pisarenko, Geophysical Research Letters 30, 1105 (2003)
[5] D. Sornette, Critical Phenomena in Natural Sciences, 2^nd edition, Springer (2006)
[6] P. Bak, C. Tang, Journal of Geophysical Research 94, 15635 (1989)
[7] Y. Huang, H. Saleur, C. Sammis, D. Sornette, Europhysics Letters 41, 43 (1998)
[8] S. Felix, M. Asch, M. Filoche, B. Sapoval, Journal of Sound and Vibration 299, 965 (2007)