config.toml

# HOOPLApy config file

[operations]
calibration = true  # Run calibration
simulation  = true  # Run simulation
forecast    = true  # Run forecast

[dates]
calibration.begin = 1997-01-01T03:00:00
#calibration.begin = 2000-01-01T03:00:00
calibration.end   = 2007-01-10T00:00:00
simulation.begin  = 2010-01-01T03:00:00
simulation.end    = 2015-01-01T00:00:00
forecast.begin    = 2015-04-15T03:00:00
forecast.end      = 2016-07-01T00:00:00

[general]
verbose          = true  # Verbose. Display information about computing
time_step        = '3h'  # Computation time step. Choose between '3h' and '24h'
compute_pet      = true  # Compute PET
compute_snowmelt = true  # Compute snowmelt
compute_warm_up  = true  # Add warm up before modelling
export_light     = true  # Export fewer data(/results) to save space
overwrite        = true  # Overwrite existing files created by HOOPLA
seed             = 42    # Seed for random number generation. If seed is 'None', then use random seed each run.
parallelism      = false # Run models in parallel

[calibration]
export         = true  # Export calibrated parameters to ./Data for future Simulat/Forecastce calibration is performed
calibrate_snow = true # Calibrate snow module (if 0, default values are used)
method         = 'SCE' # Choose between 'DDS' and 'SCE'
remove_winter  = true  # Remove the Quebec "ice months" (dec, jan, fev, mar)
score          = 'RMSE' # Performance criteron (RMSE, MSE, NSE, etc.)
maxiter        = 500   # Maximum number of iteration during calibration
SCE.ngs        = 25     # Number of Complexes for the SCE optimization

[forecast]
issue_time       = 6     # Hour of the day for which a forecast is issued (can be several per day ex: [6 12 18 24])
perfect_forecast = true # Use meteorological observations as meteorological forecast
horizon          = 80    # Horizon of the forecast (in time steps)
meteo_ens        = false # Use meteorological ensemble forecast

[data]
do_data_assimilation = false      # Perform data assimilation
tech                 = 'EnKF' # Choose either 'EnKF' (Ensemble Kalman Filter), 'PF' (particle filter), or 'PertOnly' (perturbation of inputs only)
Uc_Q                 = 0.1    # Discharge (standard deviation=10# * Qobs)
Uc_Pt                = 0.5    # Rainfal (standard deviation=50# * Pt mm, reference:Liu et al 2012,Reichl 2002)
Uc_T_pet             = 2      # temperature for PET (std deg cel)
Uc_T_snow_melt       = 2      # temperature for snow melt(std in deg cel)
Uc_T_min             = 2      # min temperature for snow melt(std in degree cel)
Uc_T_max             = 2      # max temperature for snow melt(std in degree cel)
Uc_E                 = 0.1    # PET (standard deviation=10# * E). Value used only if petCompute=0
dt                   = 8      # delta t between two correction steps
N                    = 50     # Ensemble size

PF.resample_tech   = 'systematic_resampling' # resampling technique. Either 'multinomial_resampling' or 'systematic_resampling'
PF.resample_thresh = inf                     # Particle effective ensemble size before resampling. Included in [0 N]. 0 = never resample, ..., N = resample at each time step

[models]
hydro_models = ['HydroMod1']
pet_models   = ['Oudin']
sar_models   = ['CemaNeige']
da_models    = ['EnsembleKalmanFilter']