๐ ๐๐ง๐ญ๐ซ๐จ๐๐ฎ๐๐ข๐ง๐ [Google Researchโs] ๐๐ซ๐จ๐ฎ๐ง๐๐ฌ๐จ๐ฎ๐ซ๐๐ - ๐๐ง ๐จ๐ฉ๐๐ง ๐ฌ๐จ๐ฎ๐ซ๐๐ ๐๐๐ญ๐๐ฌ๐๐ญ ๐จ๐ ๐ก๐ข๐ฌ๐ญ๐จ๐ซ๐ข๐ ๐๐ฅ๐จ๐จ๐ ๐๐ฏ๐๐ง๐ญ๐ฌ ๐๐ซ๐จ๐ฆ ๐ง๐๐ฐ๐ฌ ๐๐ซ๐ญ๐ข๐๐ฅ๐๐ฌ.
--
https://doi.org/10.31223/X5RR2K / https://eartharxiv.org/repository/view/12083/ <-- shared paper
--
https://zenodo.org/records/18647054 <-- shared link to associated dataset
--
https://research.google/blog/protecting-cities-with-ai-driven-flash-flood-forecasting/ <-- shared Google Research flash flood blog
--
https://sites.research.google/gr/floodforecasting/ <-- shared link to Google Research flood forecasting effort entry page
--
https://research.google/blog/introducing-groundsource-turning-news-reports-into-data-with-gemini/ <-- shared blog article for Groundsource link to the paper and dataset above]
--
https://research.google/blog/protecting-cities-with-ai-driven-flash-flood-forecasting/ <-- shared blog article for the Flash Flood model [link to the paper above]
--
https://blog.google/innovation-and-ai/technology/research/gemini-help-communities-predict-crisis/ <-- shared note from [the Google teamโs] SVP of Research on how Groundsource is helping disaster resilience
--
H/T Frederik Kratzert | Senior Research Scientist @ Google
โ[This #Google #research team have] developed a methodology that leverages Gemini to turn unstructured global news reports into actionable, life-saving data. This was easily one of the most exciting projects in the recent past that [the team member] was directly involved in.
๐ ๐๐ก๐๐ญ ๐ข๐ฌ ๐ข๐ญ?
[The team] used Gemini to analyze public news reports across 80 languages, extracting georeferenced information and times of past floods. [They] are open-sourcing [their] first Groundsource dataset: an unprecedented archive of 2.6 million historical flash flood events spanning 150+ countries!
๐ ๐๐๐๐ฅ-๐๐จ๐ซ๐ฅ๐ ๐๐ฆ๐ฉ๐๐๐ญ
The dataset is already being put to work to train [their] new flash flood forecasting models. What is the problem with flash flood models: As always, data! There are very few datasets on reported flash flood events, especially if we consider a global context. However, such a dataset is needed not only for potentially training a data-driven model, but also just for validating any model or getting a better understanding of the global picture.
๐ฎ ๐๐ก๐๐ญ ๐๐จ๐ฆ๐๐ฌ ๐ง๐๐ฑ๐ญ
This Groundsource dataset is just the beginning and [they] are considering expanding it into multiple directions. First, by looking at more unstructured information beyond verified news outlets. Second, by expanding beyond just floods. There are various different natural hazards and events that lack good datasets. The methodology that [they] applied here to create a dataset of flood events can easily be extended to different use cases.
๐ ๐ ๐ฎ๐ซ๐ญ๐ก๐๐ซ ๐ข๐ง๐๐จ๐ซ๐ฆ๐๐ญ๐ข๐จ๐ง
[links aboveโฆ]
--
Stay curious about what else [they] have in the pipeline ๐ โฆโ
--
โHigh-quality historical flood data is critical for disaster risk management, infrastructural planning, and climate change attribution; however, existing global archives are constrained by sparse geographical coverage, coarse spatial resolution, or reliance on prolonged satellite observation. To address this gap, [the authors] introduce Groundsource, an open-access global dataset comprising 2.6million high-resolution historical flood events, curated from the automated processing of over 5 million news articles across more than 150 countries. [Their] methodology leverages Gemini large language models (LLMs) to systematically extract structured spatial and temporal data from unstructured journalistic text. Comprehensive technical validation demonstrates that the pipeline achieves an 82% practical precision rate in manual evaluations. Furthermore, spatiotemporal matching against established external databases reveals recall capturing 85% to 100% of severe flood events recorded in the Global Disaster Alert and Coordination System (GDACS) between 2020 and 2026. By transforming unstructured global news media into a structured, localized event archive, Groundsource provides a massive-scale, extensible resource to support the training of predictive hydrological models, quantify historical exposure, and advance global disaster researchโฆโ
#GoogleResearch #Gemini #AI #ClimateTech #MachineLearning #DataScience #FloodForecasting #Sustainability #TechForGood #aggregation #curated #newsarticles #news #media #article #harvesting #reports #reporting #global #world #historic #naturalhazards #naturaldisaster #floods #flooding #flashflood #water #hydrology #extremeweather #climatechange #GDACS #Groundsource #GIS #spatial #mapping #spatialanalysis #spatiotemporal #geographic #openaccess #openscience #opendata #floodevents #LLM #gemini #largelanguagemodel #deeplearning #AI #precision #metrics #historicresource #model #modeling #forecasting #opensource
#Google | #GoogleResearch | #GoogleDeepMind | #GoogleGemini