CyberGIS-enabled reproducible agent-based modeling for scalable emergency evacuation

Agent-based models represent an effective methodology for studying the complexity of emergency evacuation. However, due to the high computational intensity that increases dramatically with regard to evacuation area size and the number of people to be evacuated, agent-based evacuation models are typically applied to relatively small areas and populations. In order to make agent-based models scalable to large evacuation areas and population sizes for emergency decision support, it is important to not only effectively harness advanced cyberinfrastructure and geospatial big data, but also make modeling workflows accessible and reproducible by researchers and decision makers. In this dissertation research, a novel cyberGIS-based approach to reproducible and scalable modeling of emergency evacuation is developed to encompass 1) systematic design of the approach for examining the reproducibility of scalable modeling scenarios for researchers and decision makers; 2) algorithmic innovation for achieving desirable computational scalability of agent-based evacuation modeling; and 3) novel geospatial big data analytics for modeling fine-scale population distribution that is important to agent-based evacuation modeling. An agent-based evacuation model is developed based on a reproducible cyberGIS science gateway framework named CyberGIS-Jupyter; enhanced by a novel network-partition algorithm for computational scalability; and improved using fine-scale population distributions derived from location-based social media data. The central contribution of this dissertation research is to achieve computational scalability and reproducibility for spatially explicit agent-based modeling to gain new fundamental knowledge of mass emergency evacuation.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Geography

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2020

Author dc:creator

• Yin, Dandong

Contributors dc:contributor

• Wang, Shaowen
• Cidell, Julie
• Chang, Kevin
• Ouyang, Yanfeng

• cyberGIS
• agent-based modeling
• computational reproducibility
• network partition
• population modeling

Statement dc:rights

• Copyright 2019 Dandong Yin

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/106365

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/106365