Bolland, Jill; Shanker, Neeraj; Reinen-Hamill, Richard; Fakhruddin, Bapon
Disasters are inherently complex with wide-ranging and cascading impacts. The exponential growth in data generated daily, coupled with the complex nature of disasters, means we are hitting the limits of humans’ capacity to fully exploit all the data available for disaster risk reduction (DRR). This can be addressed with well-designed, pretrained Artificial Intelligence (AI) algorithms that can analyse large, complex datasets and fuse heterogeneous data. However, machine-readable, semantically linked data is a precursor for the use of AI in DRR.
Nations possessing ample resources and technical proficiency are better positioned to leverage DRR data effectively, thereby potentially creating disparities in the accessibility and application of DRR data. Recent advances in technology – particularly remote sensing data, which is income-agnostic and provides global coverage – provide an opportunity to reduce DRR data gaps. Global DRR institutions should collaborate proactively with countries and regional institutions to ensure the provision of Findable, Accessible, Interoperable, and Reusable (FAIR) and open DRR data. This could help bridge any historical or emergent DRR data inequalities.
This deliverable explores the use of vocabularies in the DRR domain and how controlled vocabularies coupled with ontologies can enhance the semantic value of DRR data thereby improving interoperability. Enhancing semantic interoperability would result in improved collaboration and communication within the DRR domain and facilitate collaborations with other scientific domains. The final sections of this report provide examples of the use of remote sensing data and AI for DRR. We hope that the ideas and suggested actions in this report can be used to transform raw DRR data to valuable insights and decisions that produce tangible reductions in the impact of disasters worldwide.
The report is available on Zenodo.
