Abstract: “In large reservoirs, the water column alters in-situ stress state along an existing fault or fracture. The load of the water column is often so intense that it can significantly change the stress state leading to induced seismicity. Various empirical and calculative techniques are in place to predict the probable occurrence and magnitude of such seismic variations. This paper utilizes the Relevance Vector Machine (RVM) approach for prediction of Magnitude (M) of reservoir induced earthquake. RVM is developed in probabilistic framework. It produces sparse solution. RVM uses two input parameters namely depth of the reservoir and the other being a comprehensive parameter representing reservoir geometry for prediction of M.”

Keywords: reservoir–induced seismicity, relevance vector machine, radial basis function, Gaussian prior distribution, kernel function