CMS-9710953, Ronald Harichandran, Michigan State U. Observations from closely-spaced seismograph arrays since the late 1970's have shown that earthquake ground accelerograms measured at different locations within the dimensions of typical large engineered structures are significantly different. This has resulted in considerable attention being given to the modeling of spatially varying earthquake ground motions, and on determining its effect on the seismic response of large structures such as bridges, pipelines, and dams. Studies on various structural systems have shown that including the effect of spatially varying earthquake ground motions often increases critical responses, and hence large structures designed without consideration for this phenomenon may be unsafe. A key component in the characterization of spatially varying earthquake ground motions is the coherency function, which describes the incoherence between accelerations recorded at different spatial locations. Different functional forms have been proposed by different investigators for coherency functions, but studies on how these different forms affect the computed responses have been few. Also, since the low- frequency variation of coherency is significantly different among some of the commonly used models, it is critical to resolve contradictions and provide guidance to engineers. The objective of this research program is to use the high- resolution maximum entropy spectral estimation method and long-duration accelerogram records to resolve these contradictions, using records from the Valley of Mexico, the SMART-1 array in Taiwan, the Chiba array in Japan, and the Northridge earthquake in the U. S. This project is a cooperative research program with the Universidad Nacional Autonoma de Mexico (UNAM), involving a joint collaboration between a U. S. engineer, a Mexican engineer, and a Mexican seismologist.
