This repository contains the data for the paper
'Covariance resampling for particle filter -- state and parameter estimation for soil hydrology'. All plots contained in the paper can be reproduced using this data.

The zip-archive 'results.zip' contains eight HDF5 files.

Files ending with '..._data.h5' contain the data of the synthetic truth:
'measurement_positions': position [cm] of the measurements
Following data have the time [s] in the first row:
'measurement_stddevs': standard deviation of the measurements
'measurements': actual measurements
'true_state': true water content state
'true_state_parameter': true values of the Mualem-van Genuchten parameters,
where the first pair is the decadic logarithm of the saturated conductivity function,
the second pair is the parameter n-1 and the third pair is alpha.

Files ending with '..._da_run.h5' contain the particle filter run,
where the data has the time [s] in the first row:
'weights_prior': initial weights (equally weighted)
'weights_forecast': weights after the forecast
'weights_weighted': weights after updating with the measurements
'weights_analysis': weights after resampling

In the folder 'state' and 'state_parameter' is the temporal development
of each ensemble member saved, including their mean.

The main run of the paper, used for the plots in section 4 & 5 is saved in:
'resampling_main_data.h5'
'resampling_main_da_run.h5'

The run without data assimilation (resampling is switched off) from the start is
saved in:
'free_run_da.h5'

The particle filter runs with a model error used for the plots in section 6.2
are saved for the 10% bias in:
'10_percent_bias_data.h5'

'10_percent_bias_da_run.h5'

and for the 20% bias in:
'20_percent_bias_data.h5'

'20_percent_bias_da_run.h5'

Since the runs in section 6.1 and the supplementary are too large, the analyzed data is saved in:
'seed_dependency_analysis.h5'

Section 6.1:
The folders 'adaptivity_1_0', 'adaptivity_1_1', 'adaptivity_1_2' and 'adaptivity_1_3'
contain the ensemble members, mean and variance of the water content used for figure 10 in the paper.

Supplementary: The rest of the data in 'seed_dependency_analysis.h5 ' is part of
the supplementary. The mean parameter value at the last assimilation time step is saved for each parameter with the different realizations in x-direction and the
ensemble sizes in y-direction.
The ensemble sizes are different for each factor of gamma:
for gamma=1.0: from: 20 to 180 with stepsize 20
from: 200 to 600 with stepsize 50
from: 700 to 1000 with stepsize 50
for gamma=1.1: from 20 to 300 with stepsize 20
for gamma=1.2: from 10 to 180 with stepsize 10
for gamma=1.3: from 10 to 160 with stepsize 10