Projecting climate resilience of urban building stocks: A data-augmented archetype approach for future Nordic climates

• Data-augmented framework complements archetype methods for climate modelling. • Augmented archetype approach captures the climate-sensitive attributes in modelling. • Residential stock is evaluated under near-term extreme heat and mid-term gradual warm. • Climate-vulnerable buildings and effective climate adaptation strategies are identified. Projecting building-stock climate resilience is essential for urban climate adaptation. The archetype-based approach, widely used for stock modelling, represents buildings with a small set of archetypes and scales the results to the full building population. However, building climate performance is highly sensitive to detailed building attributes, such as HVAC systems and controls, envelope thermal properties, and window/shading details, which greatly differentiate the buildings’ climate resilience. This sensitivity often conflicts with the core premise of classical archetype methods, which assume uniform attributes and rely on homogeneous modelling within same building groups to enable archetype’s scalability. The absence of climate-sensitive attributes may constrain the identification of resilient or vulnerable buildings and constrains the design of targeted, effective adaptation measures. This study aims to enhance the archetype methods by proposing a general data-augmented framework for building-stock climate modelling, enabling vulnerable buildings clustering and effective adaptation measures identification. The proposed approach is designed to complement archetype methods through integrating multi-source building data, augmenting archetype models, and performing data-driven analysis to support climate adaptation. It is applied to the residential building stock of Umeå, Sweden, under two Nordic future climate scenarios: a near-term extreme heat year (2030) and a mid-term gradual warm year (2050). The results indicate that the vulnerable buildings were successfully clustered and effective adaptation measures were identified. Building renovations such as adjusting to mechanical ventilation and behaviour adaptations like active curtain use were found to reduce overheating by 26% and 5%, respectively. Overall, this approach extends classical archetype methods for stock-level climate modelling, enabling targeted identification of at-risk buildings and selection of effective adaptation actions.