The ArtEmis project – from forecasting to prediction

by Prof. Ramon Wyss (Royal Institute of Technology Stokolm Sweden)

Wednesday 4 Feb 2026, 15:00 → 17:00 Europe/Rome

M. B. Ceolin meeting room

Prior to the Kobe earthquake 1995, a radon detector measuring in a mine
close to the epicenter, observed a pronounced peak one week before the
7.2 magnitude Hanshin earthquake that devastated the city of Kobe.
Other measurements could conclude that the groundwater chemistry had changed before the earthquake. Recent measurements of the chemical composition of groundwater in a borehole in Iceland show precursor phenomena. In this timespan of 30 years, we may ask ourselves why we have not come further. And what would be a roadmap to move from forecasting to prediction? When it comes to changes in radon concentration as a precursor phenomenon,  a literature search will generate more than 800 references. In spite of the tremendous effort made to study radon at various sites, there are no firm conclusions whether it can serve as a reliable precursor. The ArtEmis project, that is funded by Euratom, intends to clarify the longstanding topic of radon as earthquake precursor and outlines a pathway how science can move from forecasting to prediction. The project outlines three essential steps in order to advance the field of earthquake forecasting by using geochemical tracers like radon:
1) the geographical coverage of the sensor system needs to be such, that it is sensitive to different locations with a sufficient granularity. This means that one needs many sensors along fault zones, a massive sensor system. This will require that the sensor is cheap and easy to deploy.
2)radon concentration is subject to strong variations based on atmospheric changes like pressure, temperature, rain, sun cycle and more. Hence, the radon detector needs to be placed in groundwater, that provides stable pressure and temperature, and that is shielded from atmospheric influence to a large extent. The sensor can easily be equipped with additional instruments measuring temperature, sound wave etc.
3)the radon signal from many sensors require advanced analysis, where machine learning algorithms and eventually AI will need to be employed to establish the correlation pattern with seismic movements.
The speech will present The ArtEmis project that has developed a groundbreaking cheap sensor, that measures in real time with highest
sensitivity. The sensor system and measurements from the project in Greece and Switzerland will be presented. A road towards expansion and future development will be discussed.