Run ensemble data assimilation and/or produce forecasts

Run ensemble data assimilation and/or produce forecasts

Usage

run_da_forecast(
  states_init,
  pars_init = NULL,
  aux_states_init,
  obs,
  obs_sd,
  model_sd,
  working_directory,
  met_file_names,
  inflow_file_names = NULL,
  outflow_file_names = NULL,
  config,
  pars_config = NULL,
  states_config,
  obs_config,
  da_method = "enkf",
  par_fit_method = "perturb",
  obs_secchi = NULL,
  obs_depth = NULL
)

Arguments

states_init

array of the initial states. Required dimensions are [states, depths, ensemble]

pars_init

array of the initial states. Required dimensions are [pars, depths, ensemble]. (Default = NULL)

aux_states_init

list of initial conditions for auxillary states. These are states in the GLM that are require for restarting the model but are not included in data assimilation. These are states that are not associated with a value in model_sd.

obs

array; array of the observations. Required dimensions are [nobs, time, depth]

obs_sd

vector; vector of standard deviation for observation

model_sd

vector vector of standard deviations describing the model error for each state

working_directory

string; full path to directory where model executes

met_file_names

vector; vector of full path meteorology file names

inflow_file_names

vector or matrix;; vector of inflow file names

outflow_file_names

vector or matrix; vector of outflow file names

config

list; list of configurations

pars_config

list; list of parameter configurations (Default = NULL)

states_config

list; list of state configurations

obs_config

list; list of observation configurations

da_method

string; data assimilation method (enkf or pf; Default = enkf)

par_fit_method

string; method for adding noise to parameters during calibration

obs_secchi

list of secchi observations

obs_depth

list of depth observations

Value

a list

Details

Uses the ensemble data assimilation to predict water quality for a lake or reservoir. The function requires the initial conditions (states_init) for each state and ensemble member using an array with the following dimension order: states, depth, ensembles member. If you are fitting parameters, it also requires initial conditions for each parameter and ensemble member using an array (par_init) with the following dimension order: parameters, ensemble member. The arrays for states_init and pars_init can be created using the generate_initial_conditions() function, if starting from initial conditions in the states_config data frame or from observations in first time column of the obs array.