HydroModels.jl is a modern hydrological modeling framework that extends and enhances SUPERFLEX's design philosophy. Built on Julia language and the SciML (Scientific Machine Learning) ecosystem, it combines flexible model construction with computational efficiency, particularly supporting deep learning integration in hydrological modeling.
- Flexible Model Construction: Supports development of lumped, semi-distributed, and distributed hydrological models
- Deep Learning Integration: Enables neural network integration for enhanced flux calculations and dynamic parameter estimation
- Computational Efficiency: Leverages Julia's high-performance capabilities and the SciML ecosystem
- Gradient-Based Optimization: Supports advanced parameter optimization techniques by Optimization.jl
- Comprehensive Framework: Provides tools for both traditional hydrological modeling and modern machine learning approaches
HydroModels.jl offers:
- Easy implementation and customization of hydrological models
- Integration with Julia's scientific computing ecosystem
- Support for various modeling approaches:
- Traditional conceptual models
- Neural network enhanced models
- Distributed hydrological systems
using Pkg
Pkg.add("HydroModels")- I am not a professional full-time software developer, so the code may have imperfections. If you find any issues, please post them in the issues section, and I will respond promptly.
- The software has some issues. When solving ODE files, using EnzymeVJP from SciMLSensitivity.jl generally provides more efficient gradient calculations,
but there are problems with multi-node ODE solving. I will fix this as soon as possible and begin trying to calculate gradients using Enzyme.jl and compile code using Reactant.jl. - To be responsible, if you only want to build a hydrological model or a neural network-embedded hydrological model, I would recommend building it manually. My main purpose in writing this package is to reduce the difficulty of model development and to serve future large model generation tasks.
HydroModels.jl is an open-source project available on Github. We welcome contributions from the community to help advance deep learning applications in hydrology.