How can the human dimensions of disaster impacts be more accurately captured and represented in the analysis, modeling and simulation of disasters?
To enhance the disaster resilience of a nation; all the candidate recovery plans must be quantitatively compared considering their long-term economic performance. Due to the complex interplay among individual economic entities; and their dependence on infrastructures; such comparisons must be comprehensive considering the finest level of details such as repair state of each firm and access to lifelines. Standard post-disaster economic analysis methods such as DSGE are too coarse-grained to examine these complex interactions. Therefore; to ensure the fast recovery of the economy; which is one of the main human dimensions involved; we must develop tools to simulate a nation's economy as a collection of interacting individual economic entities including their dependence on each other and infrastructural components. Under the Post-K computer project; we developed an HPC-enhanced agent based model capable of simulating hundreds of millions of economic agents and their complex interactions utilizing high-performance computer hardware. We validated the model by reproducing the Japanese economy for the period of 2015 to 2020. The model is integrated with our team’s physics-based earthquake disaster simulator (IES) for seamless simulations of disaster to long-term economy. In order to conduct realistic simulations of post-disaster economies; the model has to be further extended by including detailed models of lifelines (e.g.; gas; water; sewer); road capacity estimation tool for evaluating the interruptions to supply chains according to the repair state of lifelines; limitations of imports and exports due to disruptions to ports and airports; etc.
What type of data and supporting research infrastructure would be necessary to enable novel, transdisciplinary approaches to answering these and other human-centered disaster questions?
In order to realize high resolution and comprehensive simulations of post-disaster economies; fine level data of economic entities of the nation; details of infrastructures and lifelines; observations of past disasters; information of consumer behavior following different disasters; etc. are required. The fine-level data of the economic entities are for initializing the individual economic agent; and the rest are for realistically modeling the post-disaster economy. While the commercially available historic balance sheets of firms can be used to initialize digital counterparts of the firms of a country; the data of government entities and data of various categories of households are required to be compiled. The hardest to find is the data to model how the firms and consumers adapt to the disaster hit economic conditions; and data for the validation of the model. Necessary data to set the consumer and firm behavior during the recovery period need to be collected in future studies by the experts in relevant research areas. Further; observations of past major disasters; their detailed recovery plans; and macro-economic indices of each industrial sector during the recovery periods have to be collected for validating the developed model. Due to the interdisciplinary nature of the project; researchers in disaster mitigation sciences; economy and data sciences have to work together to prepare this data.
In what ways can US-Japan collaborations advance these questions in new and important ways?
While we have developed a numerical tool capable of simulating millions of economic entities utilizing supercomputers and validated it; the proposed plan requires many other numerical tools and data for realistically simulating post disaster conditions. Specifically; the existing model needs to be extended by integrating with numerical tools for estimating the transportation capacity of the country according to the progress of the recovery of roads and other transportation infrastructures; estimating the level of damage and the repair time of affected infrastructures; estimating the damaged state of each pipe network and their supply capacity in the disaster hit area; etc. Such high-performance numerical tools have been developed by various research institutions and academic institutions of the US and Japan. Further; a large volume of data related to different types of disasters has been collected by both the countries. A US-Japan collaboration can significantly advance the development of the project by sharing related numerical tools; data; and expertise.