Building-Scale GeoAI for Short-Horizon Urban Inundation Prediction Under Sparse Labels and Distribution Shift: A Multi-Event Evaluation of Predictive Capability Boundaries in Taipei
Year: 2026
Type: article
Abstract: Urban inundation forecasting in high-density cities is complicated by sparse incident labels, nonstationary rainfall, and changing sensing systems. Using Taipei as a case study, we develop a building-scale GeoAI decision-support prototype that integrates 30,421 LOD1 building-centroid nodes, 41,636 graph edges, built-environment attributes, and eight extreme-rainfall events from 2015 to 2026. We evaluate a static inundation-susceptibility baseline, 60- and 120-min event-level forecasts, tree-ensemble models, a Temporal GNN, and ConvLSTM, and separately test temporal and spatial transfer under compound distribution shift. The analysis adopts a positive-unlabeled (PU) learning perspective and distinguishes event-level, pooled, temporal-transfer, spatial-transfer, and scale-sensitivity evaluation regimes. Across seven events with verifiable outputs, macro-averaged AUCs were 0.694 at 60 min and 0.677 at 120 min. XGBoost achieved the highest AUC (0.682) in a benchmark with non-equivalent spatial support, whereas forward temporal validation fell to 0.490. Physics-based synthetic augmentation increased the mean AUC for four historical gauge-era events from 0.4965 to 0.6768, although one event deteriorated. These results define the conditions under which building-scale GeoAI remains informative and where it fails, providing a more defensible basis for urban-resilience decision support, climate adaptation, and infrastructure continuity.
Source: Sustainability
DOI: 10.3390/su18199743
Authors: Ming-Chih Jason Wang, Chien-Min Chen
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,896
OpenAlex ID: W7214066646