ML weather forecasting success now targets workflow transformation

After machine learning models matched or surpassed traditional numerical weather prediction in skill, the focus shifts from forecast output to the working practices behind these systems. In a new arXiv paper (2606.25076), authors Dueben, Bauer, Fuhrer, Koldunov, and Kristiansen argue that ML and recent digital technologies will fundamentally reshape the forecasting value chain. They highlight six non-exhaustive areas where change is imminent: agentic software engineering (AI that writes and tests code), open and compressed data formats, shared verification workflows, interactive computing environments, and generative methods for creating forecasts and services. These innovations promise to accelerate model development, streamline data exploitation, and make weather prediction tools more accessible to a wider audience.

However, this transformation comes with significant requirements. Weather and climate centers will need to overhaul their infrastructures, update data stewardship practices, and adopt new trust and quality-assurance frameworks. Skills development becomes critical—teams must blend machine learning expertise with deep atmospheric physics knowledge. Service delivery will become more interactive and customizable, but operational reliability, human expertise, and the public-service mission must remain paramount. The paper serves as both a roadmap and a warning: embracing these changes is essential for staying relevant, but doing so without compromising scientific rigor and dependability is the real challenge ahead.