A new method for determination of the relationship between radon and earthquakes as earthquake precursor: Regressive spatio temporal point cumulative semivariogram

Radon (222Rn) is a naturally occurring radioactive noble gas continuously generated in the uranium bearing rocks by alpha emitted from 226Ra. Because of itself noble gas character and radioactive decay, it is a unique component especially for determining temporally varying process in geological environments. For this reason, analysis of spatial and temporal changing of radon gas flux in the soil is very important for doing geo-dynamical assessment and determining radon characteristic of interested region. Occurring of geological phenomenon, distribution of rock and soil layers are neither isotropic nor homogeneous. All of these lead to showing nonhomogeneous distribution of 222Rn. Spatial analysis methods are very useful for modeling of variables exhibiting these type behaviors. In addition to this, 222Rn concentration changes depend on enviromental variable. For this reason, a spatial model that contains all these effects will be more useful in determining the character of the region. For this purpose, a new model (Regressive Spatio Temporal Point Cumulative Semivariogram) containing all these factors in the study was proposed and applied to the stations located on the East Anatolian Fault Zone and the characteristic of the region was tried to be extracted. In addition, modeling was performed using different learning algorithms of Artificial Neural Networks and reliability of the proposed method was tested. It has been tried to determine the source of the 222Rn anomalies with Time Series analysis. Key Words: Radon, Earthquake, Semivariogram, Geostatistic, Spatial Prediction, Regresive Spatiotemporal Point Cumulative Semivariogram