Collaborative Research: ATD: Geospatial Modeling and Risk Mitigation for Human Movement Dynamics under Hurricane Threats

飓风对沿海各州构成重大威胁。近年来，飓风威胁频发，严重影响了受灾地区的民生和经济活动。当飓风逼近时，不同的利益相关者需要及时做出决定。例如，一家企业是否应该关闭几天，同时考虑安全风险和收入损失？或者，当飓风的轨迹仍然存在很大的不确定性时，一个社区是否应该疏散其居民？在需要大规模疏散的情况下，警务和商品供应等公共资源应分配在何处？本研究利用从移动电话地理定位和车辆交通监控中收集的人类动态数据中的信息，并开发最先进的统计和地理空间模型，以实现数据科学驱动的决策。这项研究的成果将为提高防灾和救灾工作的有效性提供有用的数据科学工具箱。在方法上，该项目旨在利用多个流数据集中的信息，并使用统计和地理空间建模来解决以下组成任务。第一部分侧重于飓风疏散期间交通流网络的需求预测和浪涌检测。该任务将建立一个时间流量网络模型，量化交通变化率，并确定影响需求的重要时空协变量。第二部分关注疏散流和需求分散的动态优化。本课题将结合疏散需求的统计预测能力、流量最大化的组合优化能力和分散过剩需求的最优运输能力，显著提高引导交通远离飓风影响地区的效率。第三部分侧重于在飓风准备和飓风后恢复期间使用手机跟踪数据量化对经济活动的影响。本任务将量化飓风威胁下的经济依赖关系图及其变化，并找出由于关闭而面临较大经济风险的亚人群。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。

Hurricanes pose a significant threat to coastal states. In recent years, the high frequency of hurricane threats has severely impacted the livelihood of people and the economic activities of impacted regions. As a hurricane approaches diverse stakeholders need to make timely decisions. For example, should a business shut down for a few days, accounting for both safety risks and revenue losses? Or should a community evacuate its residents when there is still large uncertainty about the hurricane's trajectory? Where should public resources such as policing and commodity supplies be allocated, in case of a need for mass evacuation? This research leverages the information in the human dynamics data, collected from mobile phone geopositioning and vehicle traffic monitoring, and develops state-of-the-art statistical and geospatial models to enable data science-driven decision-making. The output of this research will provide useful data science toolboxes for improving the effectiveness of disaster prevention and relief efforts.On the methodologies, this project aims to harness the information in multiple flow datasets, and use statistical and geospatial modeling to address the following component tasks. The first component focuses on demand prediction and surge detection in the traffic flow network during a hurricane evacuation. This task will develop a temporal flow network model, quantify the traffic change rate, and identify the important spatiotemporal covariates that influence the demand. The second component focuses on the dynamic optimization of evacuation flows and demand dispersion. This task will combine the power of statistical prediction for the evacuation demand, combinatorial optimization on flow maximization, and optimal transport to disperse excess demand to significantly improve the efficacy of directing traffics away from hurricane-impacted regions. The third component focuses on quantifying the impacts on economic activities using cellphone tracking data during the hurricane preparation and post-hurricane restoration periods. This task will quantify the economic dependency graphs and their changes under hurricane threats, and find out the sub-populations at large economic risk due to shutdown.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.