From Natural Language to Reproducible Climate Analysis: FrevaGPT in the Geosciences

Large language models (LLMs) have the potential to transform how climate scientists interact with data by lowering technical barriers and enabling more intuitive analysis workflows. Building on previous demonstrations of LLM-assisted climate analysis, we present how FrevaGPT, an LLM-powered scientific assistant integrated into Freva - a climate data search and analysis platform- , supports climate scientists in their day-to-day data exploration and analysis. FrevaGPT interprets natural language queries and automatically generates traceable, editable, and reusable analysis scripts that can be executed within established scientific environments. It retrieves relevant datasets and literature, performs analyses, and visualises results, therefore allowing researchers to focus on scientific interpretation rather than coding intricacies. By leveraging a broad repository of climate observations and model output, FrevaGPT ensures transparent and reproducible workflows that adhere to best practices in climate research. It also integrates seamlessly into Jupyter-AI and, by making use of the Freva library, combines the code-generating capabilities of LLMs with contextual understanding of how to access relevant datasets on the HPC cluster. As a “co-pilot” for geoscientists, the system not only responds to explicit requests but also proactively suggests relevant climate modes, events, and next analytical steps, helping to uncover insights that might otherwise be overlooked. Practical use cases demonstrate how FrevaGPT assists with interactive exploratory analysis and hypothesis refinement across climate datasets of varying complexity. By embedding LLM-assisted natural language interaction into real-world climate research workflows, this work highlights methodological considerations and opportunities for enhancing scientific productivity, promoting broader adoption of NLP and AI tools among Earth system scientists. We provide scientific evaluation of FrevaGPT’s capability through a benchmark suite. A live demo will be presented and can be used by the audience to do real climate analysis on a high-performance computer with access to petabytes of Earth system data - starting with a simple prompt.