"While humans have assembled a lot of weather data, flash floods are too short-lived and localized to be measured comprehensively, the way the temperature or even river flows are monitored over time. That data gap means that deep learning models, which are increasingly capable of forecasting the weather, aren't able to predict flash floods."
"Google researchers used Gemini - Google's large language model - to sort through 5 million news articles from around the world, isolating reports of 2.6 million different floods, and turning those reports into a geo-tagged time series dubbed "Groundsource." It's the first time that the company has used language models for this kind of work."
"Google's flash flood forecasting model is now highlighting risks for urban areas in 150 countries on the company's Flood Hub platform, and sharing its data with emergency response agencies around the world. António José Beleza, an emergency response official at the Southern African Development Community who trialed the forecasting model with Google, said it helped his organization respond to floods more quickly."
Flash floods kill over 5,000 people annually but remain difficult to predict due to their localized, short-lived nature and lack of comprehensive monitoring data. Google addressed this data gap by using Gemini to extract flood information from 5 million news articles worldwide, identifying 2.6 million distinct flood events and converting them into a geo-tagged time series dataset called Groundsource. This dataset trained an LSTM neural network model to forecast flash flood probabilities using global weather forecasts. The resulting model now operates on Google's Flood Hub platform, providing risk assessments for urban areas across 150 countries and sharing data with emergency response agencies. While the model has limitations including low resolution and reduced precision compared to specialized systems, it has demonstrated effectiveness in accelerating emergency response efforts.
#flash-flood-prediction #machine-learning #natural-language-processing #emergency-response #weather-forecasting
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]