Google's artificial intelligence (AI) flood prediction system, Flood Hub, has been enhanced to forecast floods four days in advance in both data-poor and data-rich areas. The system can accurately predict floods even in regions with limited water flow data, like South America and Africa, and matches the accuracy of conventional systems on the same day.
AI system predicting floods in data-poor regions
In a major advance in artificial intelligence, Google's Flood Hub has tremendously improved its functionality. The system can now accurately predict floods four days in advance in regions with limited data on water flow, such as South America and Africa. This is a significant step forward given that these regions typically have less accurate measurements for water flow, thereby making flood prediction challenging. Furthermore, the system's predictions are as accurate as conventional systems even in data-rich areas like Europe and the US.
Data scarcity affecting flood prediction in lower-income countries
A key issue in flood prediction worldwide is the lack of accurate water flow measurements. In numerous waterways across the globe, data is sparse or non-existent, making flood prediction quite arduous. This is particularly prevalent in lower-income countries, which are disproportionately affected by this data deficiency. These countries often lack the resources necessary to collect extensive water flow data, thereby affecting the accuracy of flood predictions.
On the other hand, higher-income countries generally have more accurate flood predictions due to better data collection methods. These countries typically have well-measured rivers and lakes, thus providing a more complete picture for forecasting floods. However, the achievements of Google's AI flood prediction system highlight that accurate flood prediction is possible even in data-poor areas, thereby leveling the playing field.
