China’s Fengwu AI Forecasted Typhoon Landfall Five Days Ahead

China is increasingly utilizing artificial intelligence (AI) in weather forecasting, including its newest system named Fengwu. This system successfully predicted the landfall location and timing of a typhoon five days ahead with remarkable accuracy, highlighting the potential of AI-based forecasting models. Fengwu, alongside Huawei’s Pangu and Fudan University’s Fuxi, is part of a growing array of AI weather systems being developed in China to test their effectiveness in improving forecasting capabilities.

Unlike traditional numerical weather prediction, which relies on supercomputers to solve complex equations representing atmospheric physics, AI models are trained on massive datasets of historical weather data. This training allows them to learn patterns and predict atmospheric changes with significantly less computational power.

While Fengwu reportedly matched the performance of Google’s GraphCast on 80% of evaluated weather variables, it is essential to note that this metric refers specifically to the proportion of variables where the system outperformed the benchmark. The development of these AI-based forecasting systems aims to complement conventional weather models by delivering quicker and less computationally intensive predictions.

However, AI systems are not replacing human meteorologists, as they bring their unique expertise and experience to assess uncertainty and interpret forecasts. The integration of AI forecasting systems in China, alongside Pangu and Fuxi, underscores the growing importance of AI in various domains, such as wildfire risk assessment and industrial equipment failure prediction.

While AI can learn atmospheric patterns from historical data and generate forecasts more efficiently than conventional methods, the chaotic nature of weather systems and the presence of edge cases pose challenges to accurate predictions. Nevertheless, AI systems offer valuable speed advantages, enabling emergency officials to issue warnings and prepare infrastructure with more time at their disposal.

In summary, AI forecasting is emerging as a valuable tool alongside traditional physics-based models and human meteorologists, poised to enhance weather forecasting capabilities in the years to come.