Authors: Zhi Li – University of Colorado Boulder
Title: Agentic AI in Hydrologic Modeling: Where do we stand?
Presentation Type: Lightning Talk
Abstract: The configuration and calibration of hydrologic models remain significant bottlenecks, often requiring exhaustive manual effort. We explore a pivotal question: Can AI agents execute hydrologic modeling workflows with human-level reasoning? This presentation demonstrates the capabilities of state-of-the-art Large Language Models (LLMs)—including Opus 4.7, Gemini 3.1 Pro, and GPT-5.5 Pro—within an autonomous agentic framework. Based on our findings, we propose a specialized, domain-specific LLM, fine-tuned to match the performance of frontier models while offering a more targeted solution for the hydrology community.