Hurricane evacuation has long been a difficult problem perplexing local government. Hurricane Irma in 2017 created the most extensive scale of evacuation in Florida’s history, involving about 6.5 million people in a mandatory evacuation order and an estimated 4 million evacuation vehicles. Traffic jams emerged in mid-Florida and rapidly spread to involve the entire state. To understand the hurricane evacuation process, the spatial and temporal evolution of the traffic flow is a critical piece of information, but it is usually not fully observed. Based on game theory, this paper employs the available traffic observation of main highways to reconstruct the traffic flow on all highways in Florida during Irma. The reconstructed traffic conditions compare well with those simulated by dynamic models while the reconstruction model is computationally much cheaper to use. Validation with smartphone data further confirms that the reconstruction model captures the traffic conditions for real evacuation processes. The reconstructed data show that the evacuation rates for 5 representative cities -- Key West, Miami, Tampa, Orlando, and Jacksonville-- in Florida were about 90.1%, 38.7%, 52.6%, 22.1%, and 7%, respectively. The peak evacuation traffic flows from Tampa and Miami arrived in the Orlando region at almost the same time, triggering the catastrophic congestion through the entire state. Also, the evacuation for Hurricane Irma was greater than that predicted by an evacuation demand model developed based on previous event and survey data. The detailed evacuation traffic flow reanalysis accomplished in this article lays a foundation for studying evacuation demand as well as developing evacuation management policies.

长期以来，飓风撤离一直是困扰地方政府的难题。2017年的艾尔玛飓风创造了佛罗里达历史上规模最大的疏散规模，约有650万人被强制撤离，并估计有400万人撤离车辆。佛罗里达州中部出现了交通堵塞，并迅速蔓延到整个州。为了了解飓风疏散过程，交通流的时空演变是至关重要的信息，但通常没有完全观察到。基于博弈论，本文利用可用的主要高速公路交通观察来重建Irma期间佛罗里达州所有高速公路上的交通流量。重建后的交通状况与动态模型所模拟的交通状况相比非常好，而重建模型在计算上则便宜得多。使用智能手机数据进行的验证进一步确认了重建模型能够捕获实际疏散过程的交通状况。重建的数据显示，佛罗里达州的五个主要城市（基韦斯特，迈阿密，坦帕，奥兰多和杰克逊维尔）的疏散率分别约为90.1％，38.7％，52.6％，22.1％和7％。来自坦帕和迈阿密的疏散交通高峰几乎同时到达奥兰多地区，引发了整个州的灾难性交通拥堵。同样，飓风“艾尔玛”的疏散大于根据先前事件和调查数据开发的疏散需求模型所预测的疏散。