Recognition of environmentally caused variations in radon time series

Four techniques (pattern recognition, Fourier transformation, cross-correlation, and multiple linear regression) have been used to recognize nontectonic environmental factors affecting groundwater radon data collected by the Caltech automated geochemical network. Several factors, including air temperature, rainfall, water level, carbon dioxide concentration in the water, and barometric pressure, were found to correlate with radon level at some stations in the network. For example, approximately 60% of the variance in radon signal at one site studied (Pacoima) could be accounted for by nontectonic influences. However, it was not possible to correlate all the observed changes in radon concentration with nontectonic environmental variables. Some of the observed radon anomalies did appear to be related to either regional changes in tectonic strain or some individual earthquakes. A model is developed to account for spike-like increases in radon caused by carbon dioxide emission.