A probabilistic framework to construct tropical cyclone loss models for building portfolios

Tropical cyclones (TCs) evolve across time and space and can cause substantial damage to building portfolios. Therefore, timely and accurate cyclone damage assessment is crucial for effective risk management. One practical approach is to establish relationships between hazard intensity (e.g., wind speeds) and total damage across a region. However, when the study area is large, spatial heterogeneity, e.g. clustered building distributions, terrain variability, and differences in local wind intensity, can hinder accurate modelling of the regional hazard-damage relationship. To address this challenge, the present study employs a spatial clustering algorithm to divide the entire area into multiple sub-regions with relatively homogeneous internal characteristics. For each sub-region, a TC loss model is developed, defined as a function of wind speed at the sub-regional centroid and the corresponding building portfolio loss ratio. In application, the loss in all sub-regions is first assessed individually and then aggregated to estimate the total regional loss. This divide-and-aggregate approach significantly improves the accuracy and applicability of the TC loss models. The proposed framework provides a novel method for estimating cyclone-induced losses and can be applied, for example, to long-term risk management in large-scale communities. • A novel framework is proposed for developing cyclone damage models to rapidly estimate building portfolio damage using a small number of wind speeds as inputs. • The model accounts for uncertainties and correlations in cyclone wind fields and structural damage. • A spatial clustering algorithm is employed to partition the large-scale area into multiple sub-regions with relatively homogeneous internal characteristics. • The model estimates the mean and standard deviation of regional building portfolio damage by aggregating the damage of all sub-regions.