PLSTM-Reg v1.0: a regional physics-encoded LSTM model for simulating reservoir operations without site-specific training

Yu, Bin; Chen, Yanan; Zheng, Yi

doi:10.5194/gmd-19-10699-2026

Articles | Volume 19, issue 21

https://doi.org/10.5194/gmd-19-10699-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/gmd-19-10699-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 19, issue 21

Model description paper

|

11 Nov 2026

Model description paper |

| 11 Nov 2026

PLSTM-Reg v1.0: a regional physics-encoded LSTM model for simulating reservoir operations without site-specific training

Bin Yu, Yanan Chen, and Yi Zheng

Supplement

https://doi.org/10.5194/gmd-19-10699-2026-supplement

Data sets

Meteorological Forcings and Reservoir Operations Data for PLSTM-Reg v1.0 (CONUS) Bin Yu, Yanan Chen, Yi Zheng https://doi.org/10.5281/zenodo.19334272

MERIT and GRSAD Subsets (CONUS) for PLSTM-Reg v1.0 Reservoir Modeling Bin Yu, Yanan Chen, Yi Zheng https://doi.org/10.5281/zenodo.19284811

Historical Operation Data of 256 Reservoirs in Contiguous United States Yanan Chen et al. https://doi.org/10.4211/hs.092720588e2e4524bf2674235ff69d81

Daymet: Daily Surface Weather Data on a 1 km Grid for North America (4.1) M. M. Thornton et al. https://doi.org/10.3334/ORNLDAAC/2129

Global Dam Watch database version 1.0 Bernhard Lehner et al. https://doi.org/10.6084/m9.figshare.25988293.v1

Global Aridity Index and Potential Evapotranspiration (ET0) Database: Version 3.1 Robert Zomer and Antonio Trabucco https://doi.org/10.6084/m9.figshare.7504448.v7

MERIT and GRSAD Subsets (CONUS) for PLSTM-Reg v1.0 Reservoir Modeling Bin Yu et al. https://doi.org/10.5281/zenodo.19284811

Model code and software

PLSTM-Reg v1.0: A regional physics-encoded LSTM model for simulating reservoir operations under data scarcity Bin Yu https://doi.org/10.5281/zenodo.18265198

NeuralHydrology - A Python library for Deep Learning research in hydrology (Version v.1.3.0) [Computer software] Frederik Kratzert et al. https://doi.org/10.5281/zenodo.7063259

delineator.py: Fast, accurate watershed delineation using hybrid vector- and raster-based methods and data from MERIT-Hydro (1.3) Matthew Heberger https://doi.org/10.5281/zenodo.7314286

Short summary

We introduce a deep learning approach that extracts shared operating patterns from diverse, data-rich reservoirs and transfers this knowledge to limited-data reservoirs, yielding skillful simulations of storage and release. Using satellite observations, the model also reconstructs historical behavior in no-data reservoirs. Together, this work provide a scalable foundation for addressing pressing data gaps in reservoir operations and better supporting large-scale and long-term water management.