### Inhalt des Dokuments

### zur Navigation

## Prediction of earthquakes: What can we learn from laboratory models?

Earthquakes and the resulting tsunamis are two of the most destructive natural hazards. The dynamics and the predictability of earthquakes has been the focus of intensive research for decades. Nevertheless, earthquake prediction is not possible yet. A correct warning has been achieved prior to a few single quakes, but still most large quakes occur without any measurable precursors.

A majority of the recent research on earthquake dynamics uses complex experimental set-ups, considering a variety of different effects. The research conducted at our chair pursues a new idea for the analysis of earthquake dynamics and prediction by following a strict bottom-up approach. As first step the complex and divers system of quaking tectonic plates was reduces to its underlying mechanical configuration: An unstable frictional contact under growing load. The goal was to find an elementary laboratory system with comparable dynamics and achieve a prediction of the ‘tiny earthquakes’ in this system. If the prediction of real earthquakes shall be possible one day, the prediction of a simple and well measurable laboratory system should be possible as well. The considered system contains of a dry frictional contact with a sliding body with dimensions in the centimetre range. Using this set-up extensive experimental series were performed with varied contacting bodies, applied load and used materials. In total 972 data sets were evaluated.

We were able to show that a prediction of the slip event, which translate to a quake in a large natural system, is possible in our system. In the PhD thesis by Dr. Birthe Grzemba two methods are presented to predict the time of the next slip event. The developed methods both use the slowly accelerating creep process taking place in between the slip events as a precursor, but proceed differently: The stand-alone prediction only needs the measured position data as input and no parameter, the fitting-based prediction uses an approximate solution of the mechanical model, fits it to the experimental data using parameters and then gains a prediction from the model solution. Both methods were tested for universality, accuracy and earliness and a comparison was performed. The fitting-based method can be applied earlier in the creep process, but the stand-alone method delivers more accurate predictions. The majority of all slip events can be predicted successfully using the proposed methods.

The success of these studies encourages a completely new approach to earthquake prediction. As next step the applicability of the introduced methods to larger and more complex systems has to be shown. These first promising results need to be brought closer to real systems step by step.

**References:**

1. Popov V.L., Grzemba B., Starcevic J., Popov M. **Rate and state dependent friction laws and the prediction of earthquakes: What can we learn from laboratory models?** – Tectonophysics, 2012, v. 532-535, pp. 291-300.

2. Grzemba B. **Predictability of Elementary Models for Earthquake Dynamics. **- ISBN: 9783737518550, 156 p.,Epubli Verlag, zgl. TU Berlin , 2014.