HoMer: Modelling Environments Affording Agent-Environment Interactions Leveraging GIS Data

Abstract

As many cities are increasingly threatened by natural disasters, the study of evacuation and population behaviour during those disasters is identified as a crucial point by many stakeholders and researchers. As the environment is very important in those studies, integrating Geographic Information Systems (GIS) data into an Agent-Based Social Simulation (ABSS) marks a significant step forward in crisis modelling. However, it also introduces a key challenge: adapting highly detailed spatial data to reduce environmental complexity while preserving the essential information necessary to approximate higher-resolution environments. Drawing on two disaster evacuation scenarios using different spatial data structures, namely graph-based and grid-based representations, this paper proposes an original methodology to address this challenge. The proposed methodology, HoMer (Homogeneous Merging), provides a structured approach to reducing spatial resolution while retaining crucial information and striking a balance in computational scale for environments that afford agent-environment interactions. It is composed of a three-step framework: (1) attribute reduction, (2) merging of spatial entities, and (3) evaluation of the output. Our results demonstrate that HoMer makes it possible to identify an appropriate ABSS resolution for a given project while maintaining maximum data integrity.