This repository contains the data needed to create the plots of Producing accurate hydrological distributional predictions using Bayesian long-short term memory networks in order to enable reviewers and readers to reproduce the plots.
One can load the data into a Python-dictionary named data using the following code:
file_name = 'fig05.pickle'
with (open(file_name, 'rb')) as f:
data = pickle.load(f)
This dictionary contains descriptive key names for various data contained in the plots. We present the different dictionary keys per plot:
- date_range: x-axis (dates in daily resolution)
- obs: Observations
- preds_means: Model predictions
- lower_bound: Lower bounds of prediction interval
- upper_bound: Upper bounds of prediction interval
- steps: x-axis (training steps)
- p_factor: average P-factor-values for training steps
- date_range: x-axis (dates in hourly resolution)
- obs: Observations
- preds_means: Model predictions
- lower_bound: Lower bounds of prediction interval
- upper_bound: Upper bounds of prediction interval
- larsim: LARSIM-simulations
- date_range: x-axis (dates in daily resolution)
- obs: Observations
- preds_means: Model predictions
- lower_bound: Lower bounds of prediction interval
- upper_bound: Upper bounds of prediction interval
- larsim: LARSIM-simulations