2005 Salt Lake City Annual Meeting (October 16–19, 2005)

Paper No. 11
Presentation Time: 1:30 PM-5:30 PM


YE, Hong1, ZHOU, Qing1 and LEE, C.F.2, (1)Engineering Seismology, Institute of Geology, China Earthquake Admministration, Beijing, 100029, China, (2)Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China, yehhong@public.bta.net.cn

The work described in this paper is aimed at developing a new method for identifying areas of potential seismic sources. New techniques of Artificial Neural Networks (ANN) and Geographic Information System (GIS) are applied to achieve a highly computerized, efficient and objective system. A distinct feature of the present work is to implement an ANN system on a GIS platform. The integrated system analyses data extracted from the GIS and returns results to the GIS. We describe the working procedure including the construction of the GIS database, pre-treatment of input data, and training of the network classifier. This method has been applied to the coastal region of South China as a pilot study. The results of the pilot study indicate that the ANNGIS has the capability of properly dealing with complex non-linear relationships between earthquake occurrence and various seismotectonic features. Thus, it proves to be an ideal mathematical tool for the identification of potential seismic sources. The methodology deserves further consideration for use in the seismic hazard assessment of critical engineering facilities and regional seismic zoonation.