#China #Shanghai #Envision #Tianji #AI Weather Quest

Envision's Tianji Leads the Way in AI Weather Forecasting

In a groundbreaking achievement, Envision, a global leader in green technology, has announced that its innovative weather model, known as Tianji, has triumphed in the latest edition of AI Weather Quest, organized by the European Centre for Medium-Range Weather Forecasts (ECMWF). This competition, which gathers some of the most talented teams globally, is centered on the application of artificial intelligence in meteorology. Envision's Tianji model outperformed approximately 20 elite teams from around the world, excelling in all evaluated metrics including near-surface temperature, mean sea level pressure, and precipitation forecasting for periods spanning 19 to 25 days.

The AI Weather Quest

The AI Weather Quest evaluates participants through continuous real-time forecasts and verification, emphasizing sustained performance across various weather conditions. Unlike traditional forecasting, which typically looks at the immediate future, this contest focuses on probabilistic forecasts related to weekly average temperatures, mean sea level pressure, and accumulated precipitation. These metrics are assessed for predictions made 19 to 25 days in advance, which is particularly challenging due to the inherent uncertainties of long-range weather forecasting.

Teams submit their forecasts weekly, which are subsequently benchmarked against the actual weather that unfolds. The competition updates its standings regularly, providing a dynamic view of the leading models. This victory represents another significant milestone for Tianji, following its commendable ranking in the global assessment of WeatherBench 2, placing it alongside Google's WeatherNext2 model.

Tianji's Growing Reputation

The Tianji model has not only made waves in forecasting competitions but has also demonstrated its value in practical applications. For example, it achieved the top position in the AI Load Forecasting Competition hosted by the State Grid, as well as in the commercial competition organized by China Southern Power Grid. These accomplishments affirm the model's capabilities across a spectrum of forecasting challenges, showcasing its effectiveness from short-term predictions to sub-seasonal outlooks.

Robert Zheng, head of Envision's Tianji Weather Foundation Model, emphasized the challenges of sub-seasonal forecasting, remarking, "Longer forecasting horizons present greater uncertainty about atmospheric changes and increase the demand for a model's ability to identify evolving climate trends."

A Technological Marvel

Leveraging Envision's extensive experience in renewable energy, the Tianji model stands out as the world's first hybrid physics-AI weather model. It utilizes vast data sets, measuring petabytes of high-resolution temporal and spatial data, alongside a comprehensive three-dimensional data network that encompasses 'space-air-ground' variables. This allows for reliable forecasts across all time scales.

Enabled by high-performance computing, Tianji delivers swift, accurate short-term forecasts that significantly enhance predictions for wind speed and irradiance in critical regions. Covering everything from immediate minute-level predictions to sub-seasonal trend assessments, it provides comprehensive meteorological intelligence for both real-time monitoring and long-term planning.

In conclusion, Envision's Tianji model not only sets a new benchmark for weather forecasting through its application of AI but also underscores the potential for technology to enhance our understanding and predictions of climatic patterns. As the need for accurate weather predictions grows amid global climate change concerns, the advancements showcased by Tianji could transform how industries approach weather-dependent operations.