Data Fusion and change detection techniques based on optical and Synthetic Aperture Radar satellite imagery for damage mapping and multi-temporal assessment of the recovery and reconstruction process after natural disasters

This doctoral research aims to evaluate the potential of Earth Observation (EO) and Geographic Information Systems (GIS) for monitoring long-term recovery and resilience in disaster-prone environments. While remote sensing is a well-established means for damage assessment during emergencies and the scientific literature has focused on developing new algorithms and improving the accuracy of the existing ones, very few studies have shown how satellite imagery can be used by technical officers of affected countries to provide crucial, up-to-date information to monitor the reconstruction progress and natural restoration. Two case studies were developed in Haiti after Hurricane Matthew and in La Mojana, Colombia, to assess recovery after a series of catastrophic floods. By integrating freely available optical and Synthetic Aperture Radar (SAR) satellite imagery—primarily Sentinel-1 and Sentinel-2—and, in the case of La Mojana, machine learning techniques such as Random Forest classification, the study demonstrates cost-effective methodologies capable of operating under data and resource constraints. In Haiti, the analysis following Hurricane Matthew operationalized EO-based recovery monitoring at a national scale, enabling the detection of affected areas, reconstruction patterns, and settlement dynamics even in inaccessible regions. In La Mojana, the fusion of SAR and optical data enhanced flood impact assessments, revealing the spatial and temporal complexity of recovery processes. The comparative analysis highlights the need for methodological accessibility and scalability in developing and low-income countries, where technical capacity and data availability are limited. It underscores that recovery is neither linear nor homogeneous, requiring temporally adaptive monitoring frameworks aligned with local hazard cycles. The research also emphasizes the critical role of multi-sensor validation, demonstrating that analytical reliability depends on the integration of diverse datasets and ground-truth verification. Beyond methodological advances, the thesis argues that EO technologies are strategic enablers for resilience governance, bridging short-term response and long-term development planning. The findings reveal how open-access satellite data can democratize disaster monitoring, empower local institutions, and support risk-informed reconstruction policies. Ultimately, this work proposes a paradigm shift from reactive crisis management toward proactive, geospatially informed recovery systems, offering a scalable and inclusive framework to operationalize resilience in the face of escalating climate-related hazards.