Brightband is a Public Benefit Corporation with a mission to provide AI Weather forecasting tools to all. As part of our commitment to accelerating innovation in AI Weather, Brightband runs and compares the major AI Weather forecasting models. ☁️
NVIDIA just released its new Earth2 Medium Range Model. Brightband was excited to be asked to be amongst the first to test it. 🌍
Brightband is now generating an ensemble forecast using the Earth2 Medium Range Model daily, and will be performing a more complete evaluation and comparison of it alongside other AI models in its operational suite. Medium Range refers to the 15-day global forecast generated by the model.
The fact that NVIDIA has made this model open-source helps speed up innovation. Reducing friction for organizations in the weather enterprise to run, compare and improve AI models attracts innovators and makes weather technology better for all of us.
NVIDIA’s new model joins the ranks of efforts (including at Google DeepMind) aiming to produce well-calibrated ensemble forecasts. These sorts of forecasts are vital for meteorologists, who typically need to understand uncertainty in how the weather forecast might evolve. They’re also a key input to weather intelligence products aimed at supporting real-world decision-making. Ensemble forecasts can be extremely expensive to perform using traditional numerical weather prediction, but the speed and low cost of models such as Earth2 Medium Range make this forecasting technique accessible to a variety of users.
To improve that accessibility, NVIDIA is integrating Earth2 Medium Range into their earth2studio toolkit. This toolkit provides access to other NVIDIA and open source AI weather models, as well as tools for temporal interpolation, cyclone tracking, downscaling, and more.
In addition to NVIDIA’s Earth2 Medium-Range Model, Brightband also runs Aurora, GraphCast, Pangu-Weather, AIFS-Single and AIFS-ENS. We will continue to add models.
