LLMs for Hydrological Model Parameter Optimization
Abstract
The curated code and processed data used to generate the main figures and summary statistics in this study are provided in the public release package associated with this manuscript. The release package is organized into `code/` and `data/` folders and includes (1) cleaned HBV model scripts and processed HBV forcing / streamflow inputs, (2) processed comparison tables for the standalone-optimizer and LLM-initialized evolutionary experiments, and (3) the text, rule-extraction, and SHAP tables used for the interpretability analyses. All hard-coded credentials and local proxy settings were removed from the released scripts before archiving. The distributed VIC source code and routing environment used in the original experiments depend on third-party model components and large external forcing / geospatial inputs; therefore, the release package preserves the custom VIC calibration wrappers and processed result tables, while the full external VIC inputs should be obtained from their original providers. The underlying external datasets referenced in the manuscript include SRTM DEM, MODIS MCD12Q1 land cover, FAO HWSD soil data, CN05.1 meteorological forcing, and Tangnaihai station discharge records from the Yellow River Conservancy Commission.
Fields of Study & Subjects
Hydrology and Watershed Management Studies
Hydrological Forecasting Using AI
Environmental Monitoring and Data Management
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 16 March 2026
- Language
- EN