Model code: Comparison of artificial neural networks and reservoir models for simulating karst spring discharge on five test sites in the Alpine and Mediterranean regions

Guillaume Cinkus, Andreas Wunsch, Naomi Mazzilli, Tanja Liesch, Zhao Chen, Nataša Ravbar, Joanna Doummar, Jaime Fernández-Ortega, Juan Antonio Barberá, Bartolomé Andreo, Nico Goldscheider and Hervé Jourde

Preprint:

Cinkus, G., Wunsch, A., Mazzilli, N., Liesch, T., Chen, Z., Ravbar, N., Doummar, J., Fernández-Ortega, J., Barberá, J. A., Andreo, B., Goldscheider, N., and Jourde, H.: Comparison of artificial neural networks and reservoir models for simulating karst spring discharge on five test sites in the Alpine and Mediterranean regions, Hydrol. Earth Syst. Sci., 1–41, https://doi.org/10.5194/hess-2022-365, 2022.

Author ORCIDs:

• Guillaume Cinkus 0000-0002-2877-6551
• Andreas Wunsch 0000-0002-0585-9549
• Naomi Mazzilli 0000-0002-9145-5160
• Tanja Liesch 0000-0001-8648-5333
• Zhao Chen 0000-0003-0076-7079
• Nataša Ravbar 0000-0002-0160-1460
• Joanna Doummar 0000-0001-6146-1917
• Jaime Fernández-Ortega 0000-0003-0183-3015
• Juan Antonio Barberá 0000-0003-3379-0953
• Bartolomé Andreo 0000-0002-3769-7329
• Nico Goldscheider 0000-0002-8428-5001
• Hervé Jourde 0000-0001-7124-4879

Description

This repository contains the following elements:

• Reservoir model code
  • KarstMod files for each studied system
  • R script for performing the snow routine
• ANN model code
  • Code for each individual site (only slight differences between the files)
  • Dummy data files to illustrate the input data structure

Workflows

Snow routine

The snow routine is detailed in the appendix D of the manuscript. The routine is inspired from the work of Chen et al. (2018), which successfully simulated spring discharge of a mountainous karst system heavily influenced by snow accumulation and melt. The workflow is:

1. Get time series of (i) precipitation, (ii) temperature, and (iii) potential clear-sky solar radiation (if needed)
2. Define subcatchment (if needed) then calculate their areas and their relative proportion to the whole catchment
3. Apply the snow routine function for each subcatchment. We recommend to shift the temperature time series according to an appropriate temperature gradient scaling with altitude. The inputs for the snow routine function are:
   • temperature vector (T,1)
   • precipitation vector (T,1)
   • potential clear-sky solar radiation vector (T,1)
   • model parameters vector: temperature threshold, melt factor, refreezing factor, water holding capacity of snow and radiation coefficient (T,5)
4. Apply the relative proportion of each subcatchment to their corresponding P time series (output of the snow routine)
5. Sum up the P time series of each subcatchment

If working without solar radiation, radiation coefficient parameter needs to be 0 and potential clear-sky solar radiation must be a vector of 0 of the same length as temperature and precipitation time series.

KarstMod

Information on the KarstMod platform can be found in the section 3.2 of the manuscript. The main workflow is:

1. Prepare the input data
2. Open the appropriate KarstMod file (if needed)
3. Import the input data
4. Define warm-up/calibration/validation periods
5. Define Output directory
6. Run calibration

It is possible to modify the model parameters, the objective function, the number of iterations, the maximum time, and other options. The Save button allows to save the new modifications and to get a new KarstMod file.

ANNs

1D-Convolutional Neural Networks for karst spring discharge modeling. For details please see the according publication.

Dependencies: Python 3.8, Tensorflow 2.7, BayesianOptimization 1.2, Numpy 1.21, Pandas 1.4, Scipy 1.7, Scikit-learn 1.0, Matplotlib 3.5

Resources

For more details about the KarstMod platform, please refer to the User manual provided below.

For more details about the hydrological models, please refer to the section 3 of the manuscript.

Download KarstMod: https://sokarst.org/en/softwares-en/karstmod-en/

Download KarstMod User manual: https://hal.archives-ouvertes.fr/hal-01832693

References

Chen, Z., Hartmann, A., Wagener, T., Goldscheider, N., 2018. Dynamics of water fluxes and storages in an Alpine karst catchment under current and potential future climate conditions. Hydrology and Earth System Sciences 22, 3807–3823. https://doi.org/10.5194/hess-22-3807-2018