Faust Project description

 

 
 
 
 
 
 

Summary of the project
 

Project workprogramme
 

Project milestones and deliverables
 

The partnership
 

Interaction with other projects


 
 

Summary of the project

Most of the national seismic hazard maps in Europe are based on Cornell-like approaches, and the seismotectonic contribution is limited to geometry of the seismogenic zones. Even recent trans-national approaches in the framework of GSHAP initiative, use this technique, without taking any advantage of the already available data on active faulting. The large availability of historical data and the widespread idea that surface faulting was not observable in Europe, led to methodologies based on the use of seismogenic faults being dismissed as unfeasible.

Now, on one hand those methodologies are more and more advanced, with the inclusion of GPS data and modelling of faults interaction. On the other hand, in recent years the search for seismogenic faults has become a widely accepted practice in Mediterranean Europe and very recent advances raise the problem also for Northern European countries.

With this background this project aims at the following:

Three main guidelines will be followed to guide the investigation.

As a first point, the proposers bear in mind that too often the main problem in transferring the information from geologists to seismologists and engineers suffered from the absence of a feedback about the usefulness of the data. For example, a geological map reporting all the surface cracks, breaks and settlements observed in the field may be useless if not misguiding for the seismologist that wants to start his simulations from the single seimogenic fault which is responsible for all the observed ground features: seismologists and engineers needs to select models according with the available geological and historical data and vice-versa. Thus a clear correlation between field observation and seismogenic structure is the prerequisite for the inclusion in the data base, in order to avoid the confusion between the so called "active faults" and those that are actually responsible for significant earthquakes. The reason for this misunderstanding may arise on the one hand from the fact that traditional structural geologists deal with observation of upper crustal complexity that may have no bearing on the potential complexity of a future earthquake. On the other hand, seismologists and engineers need to extract from the geologists' knowledge information that is compatible with the (usually limited) complexity and (normally long) scale length of their models.

The second point is that as any mature scientific technique, geology must not confuse the observations with its interpretations and should always attach an uncertainty estimate to the data it provides. This uncertainty may arise from two factor: one intrinsic to the data (observation technique used, like limited trenching of an extended fault), and the second one deriving from the conversion of observation in a model. Any model can be suitable, according with the data available, but it must be clear that their use by seismologist is restricted according to the characteristic of the models. E.g., a study on the dynamical interaction between faults cannot be performed using only fault's projected intersection with the surface, or simply the estimated length of a fault surface projection should not be used straightforward to infer maximum magnitude without any other consideration about reliability itself, seismogenic depth, rock mechanical characteristic and expected stress drop (end users tend sometime to forget that the fault length vs. magnitude relationship is obtained using a set of observed surface ruptures, but the faults able to produce ground breaking are not the whole family of seismogenic faults, that include also members that produce only limited coseismic surface ruptures or are positively blind: The inferred length of fault surface projection should be treated as a model with its uncertainty, not as a datum).

The third point is that the dimensional difference between faults and the representation of the world in which fault models are used should not exceed unity: a linear fault model is suitable only in a plane view of the studied area, a fault plane model is to be inserted in a 3-dimensional seismicity model with no time dependence, a fault plane with characteristics varying in time is suitable to study of 3-d fault interaction in time.

Project workprogramme

The steps (or subtask) of the project are the following:

1. Study of seismogenic sources

2. Comparison of hazard estimates

3. Building of an European active fault data-base on a WWW site

Study of seismogenic sources

As a preliminary step towards the goal of the project, a map of the largest historical earthquakes in Europe will be implemented in the WWW data base as a reference starting point to locate the earthquakes that must have produced observable faulting. Refinement of the available data (such as the method for deriving fault parameters from intensity data developed in the EC project SCENARIO) will lend further support to geological studies.

The proposed project aims at maximising the extent and the quality of the geological input in computing earthquake hazard. The ultimate goal of this first step is to provide the location, geometry, size, style, maximum expected earthquake and recurrence characteristics of all the seismogenic sources according with the principle that a different level of the information gathered on the field may lead to different fault models of increasing order of detail and complexity.

To accomplish the above mentioned goals the project will:

For the whole of Europe:

For the test areas of Calabria (Italy), SE Iberian Fault System (Spain) and Hellenic Arc (Greece): The results of the first step will be formatted in a way that is immediately suitable for obtaining forward models of the expected ground motion and for subsequent verifications against observed felt reports.

Comparison of hazard estimates

In order to assess how much different fault models change the image of seismic hazard, several techniques will be used and compared, and namely:

Reference Hazard Model

The statistical estimates performed to validate the subsequent results and to verify their reliability will incorporate observed damages due to past earthquakes directly into site hazard estimates. The technique that will be used for seismic hazard estimates using historical data was developed by one of the partner, and is has already been used for seismic hazard estimates on several sites in Europe. The main outcome of this analysis will be the probability that a given intensity may be felt during an assigned period of time. This will allow comparison with the occurrence ratio of earthquakes deduced by the seismogenic fault models.

Time-independent hazard model

For the test areas considered there are available seismic hazard estimates based on Cornell's approach. For that areas the incorporation of active faulting will have a first step aimed to verify how much seismic hazard estimates are sensitive to the geometry of the fault zones. In practice, a first attempt will be made assigning all the seismicity relevant to a seismogenic area to the active fault(s) present in that area. To this scope it will be possible to use all the identified faults, since the only parameters required are geometrical. For selected faults having data on slip rates or recurrence times it will be possible to include this information in estimate of activity rates as well as in frequency-magnitude distribution.

A side benefit of this analysis will be a contribution to the widespread debate on the distribution underlying extreme events. According to several authors characteristic earthquakes do not follow Gutemberg-Richter distribution and their behaviour is separate from background seismicity. To others this fact is apparent and due to incompleteness in the catalogues that are too short if compared with the length of a seismic cycle. The most active areas of Europe may provide a very important test to both hypotheses, since the very long duration of the historical catalogues may match the duration of seismic cycles deduced from active faulting studies.

Deterministic ground motion estimate

The ground motion waveforms radiated by extended faults can be modelled by means of deterministic approaches based on a kinematic representation of the seismic source. While a dislocation model is sufficient to simulate the seismic radiation at low frequencies (f < 1 Hz), it is well known that the high frequency radiation is associated to the details of the rupture front propagation on the assumed fault plane and to the details of the medium where seismic waves are propagating. In recent years the attention of the investigators has been focused on the description of these two effects: the propagation in complex media and the complexity of the earthquake ruptures. The possibility of proposing reliable rupture scenarios for sites of engineering interest depends on the knowledge of the dimensions, of the geometry and of the rupture history on extended faults. The complexity of the seismic source is manifested in rupture models obtained by the inversion of ground motion waveforms, as well as the crustal heterogeneity results from tomographic images. Thus, there exists an evident limitation in proposing deterministic source models for simulating the high frequency radiation. The analyses of ground motion time histories indicate that the high frequency accelerations recorded near the seismic source are generally incoherent. Although, this incoherence is in part the result of scattering and multipathing, it also manifests the dynamic heterogeneity of the rupture process. This makes the modelling of ground accelerations in the time domain extremely difficult. For these reasons ground acceleration has only been analysed in time domain after low-pass filtering or in the frequency domain after smoothing the high frequency spectral amplitudes. In order to overcome the difficulties resulting from the incoherence of recorded ground accelerations, the stochastic simulation methods were proposed in the earthquake engineering practice. In this approach, the ground accelerations are simulated by generating random sequences whose amplitudes and durations are fixed by an adopted spectral model to account for source and attenuation contributions. These methodologies allow to reproduce, in a statistical sense, the peak values predicted by empirical regressions on experimental data. In particular, these methodologies are useful to extend the scaling of peak ground motions and response spectra outside the interval of magnitude and distances where the regressions were computed. However, the stochastic approaches have the limitation that they do not take into account for the amplitude variations of the ground motions simulated at observers having different azimuths. Moreover, the duration of the ground motion waveforms simulated by the stochastic methods depends mostly on the value adopted for the corner frequency. Based on the recent experience gained by two partners, we propose to use an original hybrid method, that allows to join the advantages of the deterministic simulation procedures (such as the correct source-to-observer geometry, the true duration of the rupture process modulated by directivity) to the stochastic approach that guarantees a random phase. The strategy of simulation is to compute the envelope of ground acceleration by using deterministic coherent source models and to use this envelope function in the stochastic simulation scheme. This hybrid method allows to compute incoherent ground accelerations, whose amplitudes and durations change as a function of the observer position (distance and azimuth) and of the rupture model (rupture velocity). This approach seems to be a promising and useful tool to predict the ground shaking at sites where no strong motion recordings are available. Even if there exist numerical procedures that allows to simulate the ground motions by means of incoherent source models, where the slip distribution is heterogeneous and the rupture front propagates at variable rupture velocity, we believe that such a model cannot be used to make predictions of ground motions for engineering purposes. This is the reason that motivated us to compute the envelope of ground motion, to be used in the stochastic procedure, by means of coherent rupture models whose parameters can be constrained by independent available seismological information. We aim to simulate the acceleration time histories for different rupture scenarios for municipalities included in the case study area.

Full scale fault interaction

Over the last 5 years a number of research groups have demonstrated that earthquakes occur in places where previous events have increased the "Coulomb Stress". This has been shown to apply both to aftershock sequences over time scales of days to weeks and to the interactions of large events over 10s of years. An example of such interactions is shown for 5 events in the Aegean trough (figure in the following page) and has been extended to cover 19 events between the 1912 and the present for the Aegean and for the events along North Anatolian fault since 1943. Of the all events in the these two studies only one does not clearly fall into a region where coulomb stresses have been increased by earlier events. Such consistency strongly suggests that the likely sites of future events can be effectively constrained by these methods.

Until now Coulomb modelling has exclusively employed information derived from earthquakes that have been studied instrumentally and in many cases events for which surface faulting was mapped after the earthquake. The start time of such studies must therefore be relatively recent and uncertainty exists because of the Coulomb stresses due earlier events have not been included. This is particularly important for areas of relatively low seismicity where seismic cycles are long and hence few previous events have enjoyed detailed study.

This problem can be alleviated using data from active fault mapping and the data collated in this project will be used to define zones of particular future hazard by including them in Coulomb modelling.

Building an European active fault data-base on a WWW site

The projects will exploit the potential of Internet and of the World Wide Web both in the development phase, when data will be defined and collected, as well as in the dissemination phase, when the results of the research will be made available to the world at large.

After an initial step, in which exchange formats and operating procedures will be defined and agreed upon, project participants will exchange project documents, reports, images, memos, personal communications as well as georeferenced data (maps, measurements, etc.) in an electronic fashion. In addition and in alternative to the usual point-to-point connections, such as electronic mail and file transfers, the information flow will be supported by a central site which will make available all the documents through an easy-to-use WWW interface. It will be therefore possible to share discussions and results among all partners in a fast, convenient and cost-effective fashion, minimising the need for travel. Moreover, the electronic availability of results will make it easier for the Community officers to monitor the progress of the project. When eventually the database will be available and released, the general public will easily access it through a World-Wide-Web interface.

Project development phase

The project development phase will entail defining and using a set of conventions for exchanging information among partners, through the support of a Internet site. Using the project planning documents and standard modelling tools, such as Petri Nets, a formal model of the information flow will be created. The next step will be to build a central Internet site maintaining the following main types of information: Specifically, the model will be implemented through a set of Common Gateway Interface (CGI) scripts on a World-Wide-Web site, so that each user will be recognised and will be presented with a menu containing only the options which are available at that specific time, depending on the current development phase of the project. For instance, it will be possible to make comments on a given document only after the draft version of that document will have been circulated.

The main functionalities available in this phase will include:

We plan of creating the site using a UNIX workstation with an industry-standard web server (e.g. Apache); the scripts necessary to handle the interaction will be developed using the Perl and C languages. For the creation of maps and for performing on-line calculations a choice will be performed among several commercial and public-domain packages which provide Geographic Information Systems (GIS) functionalities through a Web interface, notably those from Mapinfo Corporation and ESRI Inc.

In the deployment phase, the project development capabilities of the FAUST system will be of course removed, while the general public will access the main functionalities of FAUST through the World-Wide-Web interface.

The main functionalities to be implemented will take into account similar projects developed at a national scale and include:

The technology used will be essentially the same as in the former paragraph. The public might have access during the preparatory phase and see how the final results (e.g., hazard maps) are updated by the ongoing studies.

PROJECT MILESTONES AND DELIVERABLES

The structure of the project advancement monitoring will follow the scheme adopted by the coordinator in previous successful projects in the ENVIRONMENT and TACIS frameworks.

The basic principle is that there is a coincidence among milestones, deliverables and project tasks. This means that each project task must provide a deliverable (a stand alone one or as an input to subsequent tasks) and the fulfillment of the task with the issuing of the associate deliverable constitutes by itself a milestone to judge the advancement status of the project.

The proposed time schedule is reported in the following scheme:

At the completion of each one of the 12 sub-task listed above, a detailed technical report concerning the performed activities will be issued. Those reports will also constitute the backbone of official reporting toward EC (progress and yearly reports). Again, it is to be stressed that during the project lifetime the development of a WWW site (with access initially restricted to participants and EC officers) will be the most transparent possible mean of monitoring the project achievements. On this WWW site, the above mentioned reports (deliverables) will be available to participants and EC officers since their draft stage, and after that they will be released on the public-accessible domain.

The WWW public site that constitutes the main final deliverable of the project will be accessible to demonstrate in real time its advancement status, as well as the status of the products it will contain (hazard maps at various stages, fault maps, fault data-base, etc.)

The proposers believe that the possibility offered to EC officers to connect whenever they want to the restricted WWW site is, at the present state of the art, the most flexible and inexpensive option possible to follow at maximum detail the project advancements. Of course, this will not substitute the usual procedure of official reporting and accounting, but will permit to have a continuous check of the project without having to wait for the fixed milestones.

Four plenary meetings will be organized: one at the start, one before the end of first year and the sumbission of the annual report, one before the end of second year and the sumbission of the annual report, one before the end of third year and the sumbission of the final report. They will be in form of workshop open to the contractors, scientist working on the project. During the mid-term review meeting (in the middle of second year) and the final meeting, two independent observers will be invited, and their names will be decided in consultation with the scientific officier.

THE PARTNERSHIP

The role of the partners is summarised in the following scheme:

ISMES activity in the project is divided in the following:

IPG activity in the project will be focused on: ING will: The main goals of the IGN are: to implement, in cooperation with the University of Barcellona (UB), the data about faults that exhibit Quaternary activity and the seismogenic parameters of specific faults.

The main goals of the UB group will be:

IRRS (that will take as sub-contractor the Imperial College) will: The N.O.A. will undertake the following research tasks:

INTERACTION WITH OTHER PROJECTS

Among the project shortlisted for fundings for the Environment IV Framework, some intercation is envisaged with the following two:
back to home page