Collaborative Research: Using Precursor Information to Update Probabilistic Hazard Maps

在许多自然灾害情况下，在事件发生之前就可以获得前兆信息。对迫在眉睫的危险作出反应意味着时间有限；分析和决策必须加快进行。该项目开发了一种方法，用于响应来自各种数据源的信号，这些信号被解释为表明火山爆发受到威胁。先前的研究试图阐明物理过程如何预测火山喷发。例如，地震信号、气体排放和倾斜数据都被认为是火山爆发的前兆。但这些信号都没有被证明能做出可靠的预测。该项目采用了一种不同的方法，开发方法来整合前兆数据，确定喷发情景的可能演变，并迅速建立模拟研究和统计模拟器，以提供及时和可操作的信息，以决定行动方针。新方法将为快速构建基于概率的级联地球物理事件灾害预测图提供工具。预测和管理极端事件，从火山爆发到洪水再到股市崩盘，需要仔细分析危害事件、其输入因素和后果。必须将不同种类和不同保真度的数据纳入对即将发生的危险的详细分析中。调查人员将以他们过去对火山灾害特征的研究为基础。这项工作提供了长期的危害分析，并提供了从传入的前兆信息（如地震信号和气体排放）到喷发影响（如物质流动的可能路径）的桥梁。研究人员的目标是开发一种方法来更新物理模拟的输入分布，并将结果整合到新的自适应设计中，用于替代构建、有限模拟的快速评估和大规模并行仿真。该项目还将研究一种方法，该方法基于观测到的前兆信息与其增长之间的幂律关系，以估计喷发时间和不确定数据下的其他结果。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

In many natural hazard scenarios, precursory information becomes available before an event. Responding to an impending hazard means that time is limited; analysis and decision-making must proceed on an accelerated timetable. This project develops methodology for responding to signals from a variety of data sources that have been interpreted as suggesting that a volcanic eruption is threatened. Prior research has attempted to elucidate how physical processes predict an eruption. For example, seismic signals, gas emissions, and tilt data have all been implicated as volcanic eruption precursors. But none of these signals have been shown to make robust predictions. This project undertakes a different approach, developing methods to integrate precursor data, decide on the likely evolution of eruption scenarios, and rapidly build simulation studies and statistical emulators, to provide timely and actionable information on which to decide a course of action. The new methodology will provide tools to rapidly construct probability-based hazard forecast maps for cascading geophysical events.The prediction and management of extreme events, from volcanic eruptions to floods to stock market crashes, requires a careful analysis of the hazard event, its inputs, and its consequences. Data of different kinds, and of differing fidelity, must be incorporated into a detailed analysis of the impending hazard. The investigators will build upon their past research characterizing volcanic hazards. This work provides long-term hazard analysis and provides a bridge from incoming precursory information, such as seismic signals and gas emission, to eruption impacts, such as likely paths of mass flows. The investigators aim to develop methodology to update input distributions for physical simulations and to integrate outcomes into new adaptive designs for surrogate construction, rapid evaluations of limited simulations, and massive parallel emulation. The project will also investigate a methodology based on observed power-law relationships between precursory information and their growth to estimate the time to eruption and other outcomes under uncertain data.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.