ANALYSIS OF MULTIDIMENSIONAL DATA ON THE COMPOSITION OF GASES EMITTED FROM FAULTS IN THE EARTH'S SURFACE

The composition and intensity of gases recorded at fault location on the earth's surface are used to make an express forecast of the state of the earth's crust (including the forecast of earthquakes and volcanic eruptions) by. The method is based on unsupervised learning using a large amount of pre-collected data on the composition and concentration of gases released in the fault zone of the earth's crust. The composition and concentration of these gases contain information about the processes occurring in the depths of the earth, which makes it possible to predict earthquakes or other catastrophic events with some probability. The collected data serve to train the recognition system for newly received data by forming a system of clusters, each of which is a marker of a particular process in the earth's crust. The proximity of new data in the multidimensional space to the core of the cluster is a probabilistic measure of the event that caused the release of a gas mixture similar to a cluster.