Google has just unveiled an enhanced version of its flood forecasting system, powered by artificial intelligence (AI). The search giant will expand its coverage to 100 countries, providing riverine flood forecasting services to an audience of 700 million worldwide. The company is providing its datasets to researchers and partners to enhance their understanding of its systems and maximize the benefits of the forecasting system. Moreover, Google has created a new application programming interface (API) allowing them to effortlessly access the data.
Google has enhanced its forecasting models and expanded their implementation.
The tech giant outlined the expansion plans for the flood forecasting system in a recent blog post. The model currently includes 100 countries and spans a population of 700 million, an increase from the previous coverage of 80 countries and 400 million people.
The company attributes its successful expansion to the progress made in research, which enabled them to acquire a larger amount of labeled data for training the AI models. Additionally, they developed a new forecasting model to feed into their cutting-edge forecasting system, along with a new model architecture.
Furthermore, Google has also revealed its intentions to grant access to its flood forecasting model predictions to researchers and collaborators. This task will be accomplished using two methods: leveraging its current datasets from the Google Runoff Reanalysis and Reforecast (GRRR), alongside a newly developed API that is currently in progress.
Users are able to utilize the API to view the company’s hydrology predictions and projected flood conditions in urban and data-scarce regions. Partners and researchers affiliated with Google now have the opportunity to sign up and join the waitlist to indicate their interest in the AI-driven model.
The Flood Hub’s forecasting system has been enhanced with an extra data layer, incorporating 250,000 forecast points through the use of “virtual gauges”. Virtual gauges represent Google’s advanced simulation model that leverages geological and atmospheric elements to forecast the likelihood of river floods. The company has the ability to utilize the system for global discharge prediction. However, it will exclusively display areas for which the data can be confirmed with historical data collected from dependable sensors.
Moreover, the company has asserted that the new system surpasses its predecessors by providing flood forecasts for up to seven days ahead, showcasing enhanced accuracy compared to the previous five-day predictions. Google is confident that its AI-powered forecasting system can assist public authorities in impacted areas to enhance crisis response strategies and effectively oversee the well-being of individuals and infrastructure.
