soilFlux: Physics-Informed Neural Networks for Soil Water Retention Curves

The soilFlux package implements a physics-informed 1-D convolutional neural network (CNN1D-PINN) for estimating the complete soil water retention curve (SWRC) as a continuous function of matric potential, from soil texture, organic carbon, bulk density, and depth.

Details

Main functions

Function	Purpose
prepare_swrc_data()	Standardise raw soil data
fit_swrc()	Train the CNN1D-PINN model
predict_swrc()	Predict theta at given pF values
predict_swrc_dense()	Predict full SWRC curves
swrc_metrics()	Evaluate model performance
plot_swrc()	Plot retention curves
plot_pred_obs()	Predicted vs. observed plot
save_swrc_model() / load_swrc_model()	Persist the model
classify_texture()	USDA texture classification

References

Norouzi, A. M., et al. (2025). "Physics-Informed Neural Networks for Estimating a Continuous Form of the Soil Water Retention Curve." Water Resources Research. doi:10.1029/2024WR038149

See also

Useful links:

• https://github.com/HugoMachadoRodrigues/soilFlux

• https://hugomachadorodrigues.github.io/soilFlux/

• Report bugs at https://github.com/HugoMachadoRodrigues/soilFlux/issues

Author

Maintainer: Hugo Rodrigues rodrigues.machado.hugo@gmail.com (ORCID)