Towards HydroLLM: approaches for building a domain-specific language model for hydrology

Dilara Kizilkaya; Yusuf Sermet; Ibrahim Demir

doi:10.2166/hydro.2025.100

Back

Towards HydroLLM: approaches for building a domain-specific language model for hydrology

Journal article

Open access

Peer reviewed

Towards HydroLLM: approaches for building a domain-specific language model for hydrology

Dilara Kizilkaya, Yusuf Sermet and Ibrahim Demir

Journal of hydroinformatics, Vol.27(10), pp.1652-1666

10/01/2025

DOI: 10.2166/hydro.2025.100

Files and links (1)

url

https://doi.org/10.2166/hydro.2025.100View

Published (Version of record) Open Access

Abstract

As large language models (LLMs) continue to expand, their effective adaptation to specialized fields remains a critical challenge. This work presents an initial step toward the development of HydroLLM, a domain-specific LLM for hydrology. We construct a dataset of approximately 8,800 hydrology-focused question–answer pairs, each with a supporting context passage drawn from textbooks and scientific articles. The dataset includes four instructional formats: multiple-choice, true/false, fill-in-the-blank, and open-ended. Using this corpus, we fine-tune several LLMs of varying type and scale – from compact (1.5B) to large (32B) parameter counts using parameter-efficient LoRA (low-rank adaptation) methods. Our methodology compares different fine-tuned models and evaluates performance using accuracy and cosine similarity metrics across task types. Results show that the 8B-DeepSeek-Llama variant achieved the strongest overall performance, while the 32B model overfitted and the 1.5B model underperformed – demonstrating that larger size is not always advantageous and highlighting the need to match model capacity to dataset size. This work demonstrates that effective domain adaptation requires careful consideration of architecture, parameter count, and task complexity. By establishing performance and identifying the limits of current fine-tuning approaches, we took a concrete step toward building HydroLLM as a robust, domain-specific language model for hydrological analysis and decision support.

domain-specific AI

fine-tuning

HydroLLM

hydrology

large language models (LLMs)

natural language processing (NLP)

Details

Title: Subtitle: Towards HydroLLM: approaches for building a domain-specific language model for hydrology
Creators: Dilara Kizilkaya - University of Iowa
Yusuf Sermet - Tulane University
Ibrahim Demir - Tulane University
Resource Type: Journal article
Publication Details: Journal of hydroinformatics, Vol.27(10), pp.1652-1666
DOI: 10.2166/hydro.2025.100
ISSN: 1464-7141
eISSN: 1465-1734
Publisher: IWA PUBLISHING
Grant note: National Oceanic and Atmospheric Administration (NOAA)University of Alabama: NA22NWS4320003 NSF: NAIRR240072
This project was funded by the National Oceanic and Atmospheric Administration (NOAA) via a cooperative agreement with the University of Alabama (NA22NWS4320003) awarded to the Cooperative Institute for Research to Operations in Hydrology (CIROH). We also acknowledge NSF grant NAIRR240072 for research computing on multimodal language models in hydrology.
Language: English
Electronic publication date: 09/18/2025
Date published: 10/01/2025
Academic Unit: Electrical and Computer Engineering; Civil and Environmental Engineering; IIHR--Hydroscience and Engineering; Injury Prevention Research Center
Record Identifier: 9984966334602771

Metrics

15 Record Views

See more details