عنوان مقاله [English]
نویسندگان [English]چکیده [English]
Temporal variations of radon concentration in soil and groundwater might be one of the few promising precursors for earthquake prediction. In this study, the relation between radon concentration and aftershocks of Bam Earthquake (26/12/2003, Ms=6.8) has been investigated. The radon monitoring station was located at 29°N and 58.4°E, precisely on Bam Fault where there have been high occurrences of seismic activities. The study was carried out using an active method involving an Alpha Gurad PQR2000, Alpha Pump and relative accessories which is a device capable of accurately measuring radon concentrations every 10 minutes. Air was being pumped from ground to the measuring system with a flux of 1 L/min. Forced air suction was chosen in order to avoid stratification effects, very common for radon, due to its elevated weight. Radon-monitoring sites are usually chosen in the areas where higher concentrations of radon in the surface soil layer can be expected. For this propose, the radon monitoring site was placed exactly on Bam Fault, which was placed between Bam and Baravat Cities. Radon concentration monitoring data was collected in soil at 90cm depth exposed for a period of 90 days, every 10 minutes. Radon concentration changes are not only controlled by an earthquake, but they are also controlled by meteorological parameters at the radon monitoring site such as rainfall, soil moisture, temperature and atmospheric pressure. Therefore, in order to use radon variations as a reliable earthquake precursor, we must be able to differentiate changes that are due to earthquake from those which are not.
In recent years, artificial neural networks have become very powerful, intelligent tools, used widely in signal processing, pattern recognition and other applications. The main advantages of the method are the learning capability for developing new solutions to problems that are not well defined, an ability to deal with computational complexities, a facility of carrying out quick interpolative reasoning, and finding functional relationships between sets of data.
We have used a modified Adaline structure to estimate the temporal variation of radon concentration related to environmental parameters. This enables us to differentiate the changes due to phenomena in the earth such as earthquakes from those of environmental parameters. Radon concentration data obtained from our site and meteorological parameters measured in meteorological station of Bam were processed by the adaptive linear neural network, Adaline. It was indicated that the linear neural network was able to differentiate linear variations of radon concentration caused by the meteorological parameters from those arose from anomaly phenomena due to the aftershocks.