Unravelling information on impactful geo-hydrological hazard events with HazMiner, a multilingual text mining method developed through a global scale coverage application

Valkenborg, Bram; Dewitte, Olivier; Dananjaya, I Wayan Gita Krishna; Smets, Benoît

doi:10.5194/nhess-26-5507-2026

Articles | Volume 26, issue 11

https://doi.org/10.5194/nhess-26-5507-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-26-5507-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 26, issue 11

Research article

|

02 Nov 2026

Research article |

| 02 Nov 2026

Unravelling information on impactful geo-hydrological hazard events with HazMiner, a multilingual text mining method developed through a global scale coverage application

Bram Valkenborg, Olivier Dewitte, I Wayan Gita Krishna Dananjaya, and Benoît Smets

Data sets

HazMiner dataset Bram Valkenborg et al. https://doi.org/10.5281/zenodo.21965363

Model code and software

HazMiner code Bram Valkenborg et al. https://doi.org/10.5281/zenodo.21965339

HazMiner non-event NLP model Bram Valkenborg et al. https://doi.org/10.57967/hf/9762

Short summary

Data gaps in datasets on floods, landslides, and flash floods influence our understanding of these hazards. We present HazMiner, a new method to extract their location, timing, and impacts from online news articles. We applied HazMiner at the global scale to 6,366,905 news articles published from 2017 through 2024. This resulted in the detection of 15,313 floods, 6,836 landslides and 2,258 flash floods. HazMiner outperforms current hazard datasets, especially in data-scarce regions.