Ma et al. 2025 random forest water table depth and uncertainty at the 1 arcsec resolution

A high-resolution water table depth (WTD) map for the contiguous United States using machine learning methods trained on over one million well observations compiled from multiple groundwater databases spanning 1914-2023. A random forest model with 300 decision trees was trained on 80% of these data using input variables including climatology (precipitation, temperature, PME), subsurface properties (hydraulic conductivity, soil texture), and topographic features (elevation, slope, distances to streams), achieving test performance of r = 0.79, RMSE = 14.94 m, and NSE = 0.62.

You can download the full WTD tiff file using Zenodo or using hf_hydrodata.get_raw_file(). You can get subsets of the dataset as a numpy array using hf_hydrodata.get_gridded_data(), or download a file containing a subset of the data as a file with projection information as a TIFF or NetCDF file using hf_hydrodata.get_gridded_file(). You can select a subset using the filter options: grid_bounds, latlon_bounds or huc_id.

Dataset Name: ma_2025

Data Source: ma_2025

Data Collection or Processing Notes:

Long-term mean water table depth estimates were obtained using the median of tree outputs from the trained random forest model. The uncertainty was assessed based on the interquantile range of the tree outputs from the random forest model.

Citations:

Please refer to the following citations for more information on this dataset and cite them if you use the data

Extent and Resolution:

  • Available Date Range:

  • Grid: conus2_wtd.30

    • Spacial Resolution: 24.14076631 meters

    • XY Grid Spacial Extent: 246056 x 144287

    • LatLon Spacial Exent: -126.88755692881833, 21.8170599154073, -64.7677149695924, 53.20274381640737

    • Origin (meters): -2848561.29, -1724573.11

    • Projection: +proj=lcc +lat_1=30 +lat_2=60 +lon_0=-97.0 +lat_0=40.0000076294444 +a=6370000.0 +b=6370000

Variables

This describes the available variables of the dataset. Use the dataset, variables and temporal_resolution in python access functions as described in the Working with Gridded Data, and Working with Point Observations.

Subsurface Variables

variable

description

temporal_resolution

units

grid

3D

water_table_depth

Water table depth

static

m

conus2_wtd.30

no

wtd_uncertainty

Uncertainty in water table depth estimation from a random forest model

static

m

conus2_wtd.30

no