Abstract The purpose of this paper is to investigate whether the quasi-site-specific model established based on a municipal or regional database can be more effective in supporting site-specific predictions than one based on a global database. To do this, a Bayesian updating approach that combines site-specific data and a municipal database under realistic MUSIC (Multivariate, Uncertain and Unique, Sparse, and InComplete) attributes for the site data is adopted. This approach assumes that the site data and the municipal database follow the same distribution, which can be partially verified using standard correlation plots. The quasi-site-specific model can then be adopted to make site-specific predictions for the relevant design properties. Because real soil data follow MUSIC attributes, this Bayesian updating is only feasible in the presence of a probability distribution construction method that can handle such data. A real case study for Shanghai shows that a municipal database is more effective in supporting site-specific predictions than a global database. In contrast, another real case study shows that a regional database that covers multiple countries in the Scandinavian region is not necessarily more effective than a global database.

中文翻译：

---

市政数据库在构建特定地点多元概率分布中的作用

摘要 本文的目的是研究基于城市或区域数据库建立的准特定地点模型是否比基于全球数据库的模型更有效地支持特定地点的预测。为此，采用了贝叶斯更新方法，该方法将特定于站点的数据和在真实 MUSIC（多元、不确定和唯一、稀疏和不完整）属性下的市政数据库结合起来用于站点数据。这种方法假设站点数据和市政数据库遵循相同的分布，可以使用标准相关图进行部分验证。然后可以采用准场地特定模型对相关设计属性进行场地特定预测。因为真实土壤数据遵循 MUSIC 属性，这种贝叶斯更新只有在存在可以处理此类数据的概率分布构建方法的情况下才可行。上海的一个真实案例研究表明，与全球数据库相比，市政数据库在支持特定地点的预测方面更有效。相比之下，另一个真实案例研究表明，覆盖斯堪的纳维亚地区多个国家的区域数据库不一定比全球数据库更有效。