Ground Motion Prediction Under Uncertainty

这个项目的目的是预测旧金山湾金银岛的强烈地面运动，这是由海沃德和圣安德烈亚斯断层假定的地震引起的。近年来，各种工程师和地震学家组织进行了地震预报演习；例如，加利福尼亚的土耳其平原（1990年），日本的芦原谷（1992年），以及最近的日本神户/大阪地区（1998年）。基本的想法是试图模拟已测得的地面反应和/或预测没有记录的地点的反应，以便比较不同的预测方法。这些演习突出了本项目将试图避免的几个困难，重点放在一个非常有记录的地点——位于加利福尼亚州金银岛的国家岩土试验地点——并要求预测海沃德和圣安德烈亚斯断层的运动。美国地质调查局（United states Geological Survey）的研究人员最近汇编了旧金山湾区的地质数据，再加上加州交通局（CALTRANS）对新海湾大桥跨度的深入地质研究，大大减少了道路不确定性问题。该项目包括两部分预测练习。在第一部分（本项目）中，预测人员将开发金银岛下的基岩运动谱，用于两个断层上的情景地震。在第一个资格预审阶段之后，将选择预测者，在此阶段将选择一系列方法。一旦对这一记录良好的环境做出预测，在最后的研讨会上，不同方法中隐含的不确定性将更容易量化。来自不同大学的至少五个学生团队也将被邀请进行预测，在最后的研讨会上进行讨论，并纳入会议记录。一旦这项工作完成，第二阶段的计划是，预测者将在岛下获得一致的基岩运动，并预测90米深软土剖面中不同点以及岛上其他点的土壤反应。第二阶段将是一个后续项目，旨在通过最先进的方法研究源路径站点预测中涉及的最终累积不确定性。

The objective of this project is to predict the strong ground motions at Treasure Island, in San Francisco Bay, resulting from postulated earthquakes on the Hayward and the San Andreas faults.Earthquake prediction exercises have been carried by various organizations of engineers and seismologists in recent years; for example for Turkey Flat, California (1990), the Ashigara Valley, Japan (1992) and most recently for the Kobe/Osaka region in Japan (1998). The basic idea was to attempt to model measured ground response and/or to predict response for sites at which recordings had not been made, permitting different methods of prediction to be compared. These exercises highlighted several difficulties that this project will attempt to avoid by focusing on one very well documented site - the National Geotechnical Experimentation Site at Treasure Island, California - and asking for predictions of motion on the Hayward and the San Andreas faults. A recent compilation of geologic data for the San Francisco Bay Area by researchers at the United State Geological Survey, combined with the intensive geologic studies by CALTRANS (California department of transportation) for the new Bay Bridge span have drastically reduced path uncertainty problems.The project involves a two-part prediction exercise. In the first part (this project), predictors will develop bedrock motion spectra under Treasure Island, for scenario earthquakes on the two faults. Predictors will be selected after a first pre-qualification phase, in which a selection of a range of approaches will be made. Once predictions for this well-documented setting are made, uncertainties implicit in the different methods will be easier to quantify at a final workshop. At least five student teams from different universities will be invited to present predictions as well, to be discussed at the final workshop, and included in the proceedings. A second phase is planned once this work is completed, in which predictors would then take the consensus bedrock motions under the island and predict soil response at various points in the 90-meter-deep soft soil profile, as well as at other points on the island. This second phase will be a follow-on project, intended to study the final cumulative uncertainties involved in source-path-site predictions by state-of-the art methods.