How Google has been using AI to successfully predict floods 7 days before they hit
In some cases, Google was able to predict an area getting flooded by as much as seven days in advance.
These findings, far from being mere tech industry claims, have been validated and published in the reputable scientific journal Nature. Considering floods as the most prevalent natural disasters globally, the introduction of an early warning system marks a pivotal advancement.
Historically, predicting floods has posed considerable challenges due to the absence of streamflow gauges in most river systems.
Google overcame this hurdle by harnessing machine learning models trained on diverse datasets encompassing historical events, river level measurements, elevation data, and terrain readings, among other variables.
Leveraging this rich pool of information, the company generated localized maps and conducted numerous simulations, numbering in the hundreds of thousands per location. This multifaceted approach empowered the models to accurately anticipate impending floods.
While Google has effectively developed highly precise models tailored to specific locations, its aspirations extend to addressing the global flood prediction challenge.
Although the company achieved successful predictions up to seven days in advance in select instances, the average forecast horizon stands at approximately five days. Nonetheless, Google remains optimistic, asserting that it has expanded the “reliability of currently-available global nowcasts from zero to five days.”
Furthermore, the technology has notably enhanced forecasting accuracy in underrepresented regions such as certain areas of Africa and Asia.
The impact of this technology is extensive, with Google providing accurate flood forecasts for 80 countries, encompassing a collective population of 460 million individuals.
These forecasts are readily accessible through Google Search, Google Maps, and Android notifications. Additionally, users can access this information via the company’s dedicated Flood Hub web application, which was inaugurated in 2022.
Looking ahead, Google is committed to further exploring the potential of machine learning in refining flood forecasting models. Collaborating with academic researchers, the company aims to optimize its AI-driven approach, envisioning the development of a comprehensive “global end-to-end flood forecasting platform.”
Google’s pioneering efforts in leveraging AI for flood prediction not only signify a significant technological advancement but also hold promise in mitigating the impact of one of the world’s most devastating natural disasters.
As the company continues its pursuit of innovation in this realm, the potential for enhanced resilience and preparedness against floods worldwide becomes increasingly tangible.
