Abstract

Hydrologic data are the cornerstone of many complex hydrologic research and operational workflows that produce useful hydrologic information for water resources decision-making, flood risk communication, and water availability implications. The National Water Model (NWM) is a valuable source of hydrologic data, particularly for its extended coverage of unmonitored streams and the large-scale continuity of hydrologic processes. Precomputed secondary datasets derived from this simulated data, such as streamflow indices, can improve operational efficiency through thresholding. Disseminating the data through modern technologies, such as application programming interfaces, is required for ready integration into applications. However, if extraction requires an intermediate service, there is additional latency. Large-language models can directly interpret patterns within hydrologic data or even translate results for a general audience. With connections to hydroinformatics tools and collaboration among LLMs, an agentic system can even automate the generation of complex hydrologic information. Therefore, in this study, we computed a streamflow index dataset covering up to 97.6% of NWM reaches, making flow metrics and long-term regime information readily available for virtually any US reach. To improve the service, we enriched the existing API with geospatial filtering and metadata features with no measurable latency penalty for tabular responses. We also developed a Model Context Protocol (MCP) server to serve NWM data in LLM applications. Finally, to smooth the hydrologic data-to-information lifecycle with minimal guidance from expert personnel, we developed a multi-agentic LLM system leveraging the NWM and other hydrologic data.

Degree

MS

College and Department

Ira A. Fulton College of Engineering; Civil and Construction Engineering

Rights

https://lib.byu.edu/about/copyright/

Date Submitted

2026-08-12

Document Type

Thesis

Keywords

National Water Model, Large-language model, Hydrologic Data Services, Multi-agent systems, Hydrologic Indicators, Artificial Intelligence

Language

english

Download

Included in

Engineering Commons

Share

COinS