- Python 62.6%
- Jupyter Notebook 35.1%
- Shell 2.3%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| 1_model_training.ipynb | ||
| 2_high_res_WTD_map_generation.py | ||
| 2_high_res_WTD_map_generation.sh | ||
| 3_convert_to_TIFF.py | ||
| environment_high_res_wtd.yml | ||
| LICENSE | ||
| random_forest_sklearn.py | ||
| README.md | ||
High Resolution Static Water Table Depth Estimation over the Contiguous United States
This repo includes scripts supporting the analysis of the 2025 high resolution water table depth (WTD) static paper (Ma et al., 2025). The scripts were developed to train a random forest model for estimating the long-term mean WTD across the contiguous United States (CONUS) using available observations, and to generate a long-term mean WTD map at a 1 arc-second (~30 m) resolution for the CONUS.
The final dataset is accessible via hf_hydrodata. It can also be visualized on the HydroGEN platform.
For questions, please contact Yueling Ma (y.ma@fz-juelich.de).
Model Training & Deployment
The following notebooks and scripts used for model training and deployment can be found in the main folder:
random_forest_sklearn.py: The function of a random forest model developed with scikit-learn.1_model_training.ipynb: The notebook used to train a random forest model for estimating 1 arc-second resolution long-term mean WTD over the CONUS. Links to some input datasets are provided in the notebook.2_high_res_WTD_map_generation.py: The script used for generating the high resolution WTD map along with its associated uncertainty maps. Input datasets are too large to share. One can adapt this script to their own datasets and domains.2_high_res_WTD_map_generation.sh: The Slurm job script used to run2_high_res_WTD_map_generation.py.3_convert_to_TIFF.py: The script used for converting the resulting npy files from2_high_res_WTD_map_generation.pyto TIFFs.
Environment
The python modules used in this repo can be installed using environment_high_res_wtd.yml.
conda env create -f environment_high_res_wtd.yml
conda activate high-res-wtd
About
For more information, please refer to the paper: https://doi.org/10.1038/s43247-025-03094-3
Funding
This research has been supported by the U.S. National Science Foundation Convergence Accelerator Program (grant no. CA-2040542). This work was conducted using Princeton Research Computing resources at Princeton University which is consortium of groups led by the Princeton Institute for Computational Science and Engineering (PICSciE) and Office of Information Technology's Research Computing.