Complex stochastic approach for prediction of natural catastrophic events: earthquakes and volcanic eruptions

In this paper proposed new stochastic approach for prediction of earthquakes and volcanic eruptions. The idea of approach is adjustment of heterogeneous observations for reliable prediction of hazardous events. Methods of data acquisition, fusion and processing are generally described. Common physical models of earthquakes' and eruptions' monitoring are mentioned. The unified (generalized) scheme of assemblage heterogeneous measurements to combined solution is suggested. These observations upon various precursor events and characteristics produce sets of regression data. Correlation and covariance dependencies of data are researched. The key point of approach is applicability of stochastic properties of mixed sources data for improvement of reliability and quickness of prediction and notification. Applications of Kalman filtering for data and sensor fusion are investigated. Technique of information mining for collected data is elaborated. Prediction information system designed to use both recent measurement data and long-term observations, which accumulated in data storage. Short-term prediction is based on both remote (satellite) sensing and distributed network of in situ sensors. The goal of this research is to propose high reliable methods of complex sensing and early stochastic prediction of natural hazards.