Abstract: Component object model GIS (COMGIS) technique is adopted to develop the hazard evaluation system for earthquake-induced site liquefaction potential combined with BP model of artificial neural network (ANN). Firstly, the simulation of site earthquake influence field under scenario earthquake can be generated by calling SHAKE91, a commercial program to analyze seismic response of level layered site. Then, seismic hazard for sandy soil liquefaction can be evaluated by BP model of ANN which is modularized under VB by importing ANN toolbox from Matlab. Thus the analytical results from BP model of ANN can be simulated spatially by GIS; and the spatial distribution of site liquefaction potential also can be graphed. The research shows that: First, SHAKE91 software can be called directly from the system menu to simulate the site earthquake influence field. Second, the relatively perfect results can be obtained when the BP model of ANN is utilized to predict the latent hazard of earthquake-induced site liquefaction. Third, by GIS technique, the evaluated results can be perfectly matched with the site information spatially. Thus the critical site liquefaction hazard area can also be identified.

Key words: model of artificial neural network, COMGIS, earthquake-induced site liquefaction, SHAKE91, Matlab