Spatiotemporal Forecasting of Functionality States of Community Building Portfolios under Flood Evolution

Severe flooding induced by heavy rainfall presents a significant challenge in metropolitan areas worldwide. Despite the imperative for effective preflood emergency management, existing early warning systems rely primarily on meteorological or hydrological factors. These factors are often mapped with coarse spatial-temporal resolution and lack direct correlations to the immediate consequences of hazards, thus hindering fine-grained estimation of evacuation demand to support proactive preflood evacuation planning. To address this gap, our study makes a twofold contribution: (1) introduction of a flood-specific, evacuation-oriented, and multilayered building functionality metric (BFM-Flooding); this metric assesses building functionality states under flooding by aggregating building damage conditions at three distinct layers of granularity while accounting for disruptions in supporting utilities. (2) Establishment of a comprehensive building functionality state forecasting framework (BFS-Forecasting); the framework delves into the temporal transition dynamics between different functionality states for different building archetypes; it predicts the spatiotemporal progression of functionality states for building portfolios and the resulting evacuation demand within a community during flood evolution. Finally, an illustration using Suzhou Industrial Park (SIP) in Jiangsu Province, China, under severe rainstorm conditions demonstrates that the twofold solution can provide valuable support to policymakers in making well-informed preflood evacuation decisions at a fine spatiotemporal resolution.