doi of this repo:
This repository provides example model code according to:
Wunsch, A., Liesch, T., Cinkus, G., Ravbar, N., Chen, Z., Mazzilli, N., Jourde, H., and Goldscheider, N.: Karst spring discharge modeling based on deep learning using spatially distributed input data, Hydrol. Earth Syst. Sci., 26, 2405–2430, 2022, https://doi.org/10.5194/hess-26-2405-2022.
Contact: wunsch.andreas.edu@gmail.com
ORCIDs of first author:
A. Wunsch: 0000-0002-0585-9549
For a detailed description please refer to the publication. Please adapt all absolute loading/saving and software paths within the scripts to make them running. Our models are implemented in Python 3.8 (van Rossum, 1995) and we use the following libraries and frameworks: Numpy (van der Walt et al., 2011), Pandas (McKinney, 2010; Reback et al., 2020), Scikit-Learn (Pedregosa et al., 2011), Unumpy (Lebigot, 2010), Matplotlib (Hunter, 2007), BayesOpt (Nogueira, 2014), TensorFlow 2.3 and its Keras API (Abadi et al., 2015; Chollet, 2015). Large parts of the code comes from Sam Anderson - please check the according publication: Anderson, Sam and Valentina Radic. "Evaluation and interpretation of convolutional-recurrent neural networks for regional hydrological modelling" HERE. The model code runs best on a GPU, unfortunately we cannot provide original data due to republication restrictions of some parties. However, all data is accessible, please refer to the publication for references.