September 20, 2024
Artificial Intelligence (AI) is emerging as a viable solution as the amount of weather and climate data surpasses what traditional forecasting methods can handle. The recently funded startup Brightband aims to revolutionize weather forecasting by integrating machine learning into both open-source and proprietary platforms.
Currently, weather predictions rely on decades-old statistical and numerical models. While these physics-based models are reliable, they often require supercomputers and extensive processing time, making them less efficient. In contrast, AI has shown a remarkable capacity to predict weather events by learning from global weather patterns and observations. It excels at identifying patterns within large datasets.
So, why isn't AI the standard in weather forecasting yet? According to Julian Green, CEO and co-founder of Brightband, one of the main reasons is that weather companies and government agencies often struggle with attracting top AI talent. At the same time, tech companies, while skilled in AI, don’t prioritize weather as a main focus of their operations. He adds that Brightband is uniting experts in meteorology, data, and artificial intelligence to implement AI in forecasting and make it accessible to everyone.
Based on extensive meteorological data, the company is currently developing its own AI-driven forecasting algorithm. Daniel Rothenberg, co-founder and head of data and weather, explains that they are building the foundation of the physics-based models, AI allowing them to extend and improve upon them, acknowledging the legacy of earlier models.
One of the key advantages of Brightband's approach is speed. AI-driven models are particularly beneficial for industries with specific needs, such as energy, transportation, and agriculture, as they can provide forecasts much more quickly and at a lower cost compared to traditional methods.
Notably, Brightband has committed to making its core forecasting models publicly accessible. The company's strategy is to provide the model, the training data, and the evaluation metric, all without a cost, while focusing on delivering premium services for more specialized capabilities.
According to Rothenberg, the challenge of working with the vast amounts of historical weather data—like that collected from weather balloons and satellites—often leads to it being overlooked. To enhance the accuracy and scope of its forecasts, Brightband plans to incorporate this untapped data into its models.
Although Brightband is still a young company, it aims to have a working model by the end of 2024. Prelude Ventures spearheaded the startup's $10 million Series A funding round. Even though it operates as a for-profit entity, Brightband is structured as a public benefit corporation (PBC), reflecting its commitment to transparency and mission-driven goals.
Before implementing any climate-specific solutions, anticipate a weather forecasting product; further enhancements are likely to be introduced by the end of the year. Utilizing an open-source approach and cutting-edge AI technology, Brightband has the potential to revolutionize weather forecasting, making it more cost-effective, faster, and accessible to a broader audience.