Radon Content of Groundwater as an Earthquake Precursor: Evaluation of Worldwide Data and Physical Basis

The properties of a worldwide data set of 91 radon (222Rn) anomalies (the frequency of occurrence, the precursor time interval, and the distribution of peak amplitudes) are correlated with earthquake data such as the respective magnitude and epicentral distance. These anomalies were reported as precursors to earthquakes in the United States, USSR, China, Japan, and Iceland. Although the data set is incomplete and limited by experimental deficiencies, several consistent patterns emerge. Radon anomalies from different tectonic regions show similar patterns. The radon anomalies occur at greater epicentral distances for earthquakes of the larger magnitude. Anomalies preceding large earthquakes (M > 6) are frequently observed at a distance of 100 to 500 Ian. These distances are larger than several times the rupture dimensions of the future earthquakes. The time from the onset of an anomaly to the time of the earthquake (the precursor time) increases with magnitude but decreases with distance between epicenter and radon station. In addition, radon anomalies are observed more frequently prior to large earthquakes than prior to small ones, indicating that the preparation zone increases in size as magnitude increases. The peak amplitude does not scale with magnitude but forms a consistent pattern with epicentral distance in that the larger the earthquake magnitude, the farther away the largest amplitudes tend to occur. The preparation zone of the earthquake where the anomalies occur forms an almost continuous annulus that expands with time away from the future rupture zone. The outer radius of this annulus scales with the earthquake magnitude. Model calculations indicate that strain fields of at most lo-6 to lo-8 strain caused the radon anoma 1 ies. If these strains are divided by the appropriate precursor time, m1n1mum strain rates from lo-7 day-1 to 10-10 day-1 are obtained. Such small strains and strain rates suggest that in most cases neither mechanical crack growth induced by dilatancy nor mechanical coupling between pore pressure and the rock matrix caused the anomalies. Large changes in the orient at ion of the local strain field, however, could occur and affect the local stress intensity factor. Since changes in the stress intensity factor can result in stress corrosion, the occurrence of radon anomalies is attributed to slow crack growth controlled by stress corrosion in a rock matrix saturated by groundwater. Variations in groundwater that Introduction radon (222Rn) content of have been observed prior to Copyright 1981 by the American Geophysical Union. Paper number 181247. 0148-0227/81/0018-1247$01.00 some earthquakes constitute an important nonseismic precursor to earthquakes. Active national programs for radon monitoring in the United States, USSR, People's Republic of China, Japan, and Iceland have reported at least 91 anomalies associated with some 46 different earthquakes during the last 15 years. The basic features of the radon precursor are poorly understood, and currently, only Chinese scientists actually use local radon data in conjunction with other precursory data as a basis for issuing earthquake predictions. To establish the properties of changes in radon content as an earthquake precursor, this paper evaluates the main features of the worldwide data set of radon anomalies and proposes growth of small tensile cracks, controlled by stress corrosion, as a physical basis for the anomalies. Since there is considerable scatter in the data set, only qualitative analysis and interpretation of the main features are attempted. The main features of the radon data set (the frequency of occurrence, the precursor time interval, and the distribution of peak amplitudes) are correlated with earthquake data such as the respective magnitude and epicentral distance. It is likely that the results of the analysis will be influenced by experimental deficiencies. Such deficiencies could consist of not looking for anomalies prior to small earthquakes, large spacing between radon stations compared with the rupture dimensions of small earthquakes, and infrequent sampling. Nonetheless, a closer analysis of the data is warranted if it can aid in identifying possible consistent patterns in this data set. This paper does not address such features as false alarm rate or nonoccurrence of radon anomalies because those data are generally not available in the literature. A comparison of the properties of radon anomalies with the properties of other precursors is also not considered. Previous syntheses of radon anomalies and other types of earthquake precursors were all based on a much smaller data set than is treated here. Scholz et al. [1973] interpreted two radon anomalies and numerous other precursors in terms of the dilatancy diffusion model. They found a log-linear relationship between the precursor time and the earthquake magnitude, Rikitake [1975, 1979] used worldwide data to correlate the precursor time interval and earthquake magnitude. He found that a systematic correlation between precursor time and magnitude is more often the exception than the rule for earthquake precursors. A summary of recent radon data from the People's Republic of China is presented by Wakita [1978] and Teng [1980b], Teng [1980b] emphasizes the importance of