A new view of earthquake and volcano precursors

A close coupling between monitoring and modeling is needed for a quantitative prediction of volcanic eruption and earthquake occurrence. The need is demonstrated for the volcano, Piton de la Fournaise, for which a relatively simple model can be constructed because of the isolated tectonic setting and can be tested in a short time because of the high rate of eruptions. Our conceptual model (Fig. 2) explains various observations for each of the 49 eruptions since 1972 in terms of the varying conditions of the model elements, allowing inferences on causal relations between consecutive eruptive periods, recognized as alternating active and quiet ones, as well as predictions on the possible eruption scenarios for the future. Lessons learned from the volcano are applied to earthquake prediction in California, and we found a model of earthquake loading process by plate-driving forces that can be effectively constrained by data from the existing seismic network. The model identified that the periods of several years preceding the 1952 Kern County earthquake and the 1989 Loma Prieta earthquake are in the anomalous phase of the loading process. A companion paper by Jin et al. (2004) in the present issue shows that the 1992 Landers earthquake and the 1999 Hector Mine earthquake were also preceded by the period of anomalous loading process.