A multi-temporal Very High-Resolution optical satellite remote sensing framework for post-disaster reconstruction monitoring

Long-term monitoring of post-disaster reconstruction is essential for evaluating recovery processes, improving urban resilience, and reducing future vulnerability in regions and population exposed to recurrent extreme events. This contribution presents a remote sensing framework for assessing reconstruction dynamics using multi-temporal Very High-Resolution (VHR) optical satellite imagery, such as WorldView-2 and WorldView-3. The methodology is designed to provide detailed, building-level reconstruction assessment over multiple years, supporting quantitative analysis and trend evaluation. 

The framework that is proposed consists of manual photointerpretation with a structured classification schema to determine the reconstruction status of individual buildings. To enhance efficiency and consistency, the process is supported by AI-detected buildings which act as a foundational layer. Buildings are categorized into classes such as unchanged, removed, under reconstruction, reconstructed, newly built, or not applicable. These classes are presented as photointerpretation keys, that are developed based on observable indicators in VHR imagery, including structural integrity, roofing, facades, presence of construction materials, and signs of ongoing repair. It is widely acceptable that manual photointerpretation is time-consuming. However, for post-disaster monitoring, it is essential to accurately classify buildings, and the AI-supported workflow streamlines the initial building identification. 

On top of that, a proper temporal consistency was achieved with image acquisition in a necessary seasonal window across multiple years. In that way, interpretation bias caused by illumination, vegetation phenology, or coastal conditions, are minimized. Temporal analysis is performed by comparing building classifications year by year, allowing the identification of reconstruction progress, new construction, or abandonment. In terms of ambiguous building statuses are detected, multi-date interpretation is applied. 

Additionally, a multi-image manageable strategy is described, when uncertainty associated with illumination, cloud cover, atmospheric effects, or visual obstructions is raised. Thus, in cases where persistent cloud cover limits visibility, carefully selected alternative acquisition periods are used to maintain analytical continuity. 

To sum up, this framework demonstrates the value of VHR satellite-based, building-level reconstruction monitoring, combining methodological rigor with practical applicability for long-term recovery assessment and hazard-informed planning. Providing transferability and scalability, for long-term monitoring of post-disaster recovery, in which urban planning, resilience evaluation, and disaster risk reduction effort, can be supported.