Data for: A hybrid physics–ML framework for integrating groundwater dynamics into land surface modeling

This dataset supports the manuscript “A hybrid physics–ML framework for integrating groundwater dynamics into land surface modeling.” It includes data used for model training and evaluation; additional data for post-processing and result analysis, including physical-model outputs with free-drainage boundaries; trained model checkpoints; outputs from land-surface-model coupled tests; training log files; runnable model templates; and figure-generation scripts.

The accompanying README describes the archive structure, data usage, and steps for reproducing the analyses. The associated model code and detailed documentation are available through the linked Zenodo archive (doi.org/10.5281/zenodo.20791904) and the actively maintained GitHub repository (github.com/aureliayang/ParFlow-nn)Version update (26 September 2026): A new archive, revised_supplementary_material_20260926.tar.gz, has been added.