You may have missed a big breakthrough in the ML weather forecasting revolution over the holidays: GenCast – Google DeepMind's new generative model! The importance of probabilistic weather forecasting in various critical domains such as flood forecasting, energy system planning, and transportation routes cannot be underestimated. Being able to accurately measure uncertainty in forecasts, especially as it relates to extreme events, is critical to making well-informed decisions that involve important cost-benefit considerations and effective mitigation strategies.
Traditionally, the probabilistic forecasting approach involves creating ensembles from physics-based models, which sample from a joint distribution over spatiotemporally coherent weather trajectories. However, this method can be computationally expensive. An attractive alternative is the use of machine learning (ML) forecasting models to generate ensembles. However, current state-of-the-art ML forecasting models for medium-term climate primarily focus on producing deterministic forecasts that minimize the mean squared error.
Despite the improved skill scores associated with these models, they face a limitation in terms of lack of physical consistency. This limitation becomes more pronounced the longer the timescales, affecting their ability to accurately characterize the joint distribution of weather events.
The paper presents a novel machine learning-based approach for probabilistic weather forecasting known as GenCast. This innovative method generates 15-day global ensemble forecasts that demonstrate superior accuracy compared to the leading operational ensemble forecast, namely the European Center for Medium-Range Weather Forecasts (ECMWF) ENS, while requiring additional calculation time. significantly lower. GenCast operates by implicitly modeling the joint probability distribution of the climate state in space and time. It operates on a 1° latitude and longitude grid, using 12-hour time steps, and represents six surface variables and six atmospheric variables at 13 vertical pressure levels.
Evaluation of GenCast forecasts shows that it maintains detailed patterns and consistency in weather predictions. Comparisons with ENS indicate that GenCast ensembles are just as reliable, if not more so. GenCast is efficient: you can create a 15-day forecast in about a minute using Cloud TPU v4. This means that it is possible to generate a large number of forecasts (𝑁 ensemble members) in a short time with multiple TPUs. This efficiency opens up the possibility of using much larger arrays in the future.
In a broader context, GenCast represents a significant advance in machine learning-based weather forecasting, demonstrating greater proficiency than the leading operational ensemble forecast with 1° resolution. This development marks a fundamental step toward ushering in a new era of machine learning-driven ensemble forecasting, expanding its relevance and utility across a wide range of domains. Additionally, looking ahead, GenCast offers a glimpse into the potential of adopting machine learning to revolutionize our understanding and prediction of complex weather patterns, with far-reaching implications for various industries and decision makers.
Review the Paper. All credit for this research goes to the researchers of this project. Also, don't forget to follow us on Twitter. Join our 36k+ ML SubReddit, 41k+ Facebook community, Discord channeland LinkedIn Grabove.
If you like our work, you will love our Newsletter..
Don't forget to join our Telegram channel
Janhavi Lande, Graduated in Engineering Physics from IIT Guwahati, Class of 2023. She is an upcoming data scientist and has been working in the world of ml/ai research for the last two years. What fascinates him most is this ever-changing world and its constant demand for humans to keep up. In her hobbies she likes to travel, read and write poems.
<!– ai CONTENT END 2 –>
