Evaluating the Influence of Temporal Data Resolution on the Performance of Deep Neural Networks for Karst Groundwater Forecasting

By: Xiao Xia Liang, Dany Lauzon, Erwan Gloaguen, Julien Straubhaar, Maxime Claprood, and Philippe Renard

The models used are LSTM, GRU auto-encoder, and Graph WaveNet

The Graph WaveNet codes adapted for this work is the improved version from Shleifer et al., (2019)

Data

Data can be found here

Package Requirements

• python
• pandas
• numpy
• scikit-learn
• scipy
• torch
• tensorflow