Pure Python implementation of the GR4J hydrologic rainfall-runoff model.
Requires Python 2.6 or greater (mostly tested with Python 2.7 on Linux).
If Cython is available the Python code will be compiled by Cython during installation. Running the unit tests requires numpy and pandas. These dependencies are all optional as the base code is pure Python.
Install from Github:
pip install git+https://github.com/amacd31/gr4j.git
A single method that takes input rainfall and potential evapotranspiration data, a parameters dictionary and optional starting state.
See Perrin et al. 2003 paper (referenced below) on the parameters for the model.
from gr4j import gr4j
params = { 'X1': 303.627616, 'X2': 0.32238919, 'X3': 6.49759466, 'X4': 0.294803885 }
states = { 'production_store': 0.60 * params['X1'], 'routing_store': 0.70 * params['X3'] }
rainfall = [ 14.1, 3.7, 7.1, 9.3 ]
potential_evap = [ 0.46, 0.46, 0.47, 0.47 ]
simulated_flow = gr4j(rainfall, potential_evap, params, states)
Implemented as a learning exercise. Code based on the paper 2003 paper by Perrin et al. and the website summary of the model. Results compared to the output of the Excel version of GR4J: https://webgr.irstea.fr/wp-content/uploads/2013/04/GR4J_en.xlsx
The design of a single functional method was chosen as part of this learning exercise with a mind to later implement an OpenCL version in C for parallel usage. Later work may update the design of this package, but that is not currently planned for the near future.
Cloned in order to learn how to use the code and to implement automatic parameter estimation using python optimization packages. Learning exercises are in https://github.com/dvictori/learn_gr4j
Perrin, Charles, Claude Michel, and Vazken Andréassian. "Improvement of a parsimonious model for streamflow simulation." Journal of Hydrology 279, no. 1 (2003): 275-289.
Operation of GR4J: https://webgr.irstea.fr/modeles/journalier-gr4j-2/fonctionnement_gr4j/?lang=en