Generate simulated time series from Hydromad model objects

Generate simulated time series from Hydromad model objects.

# S3 method for hydromad
predict(
  object,
  newdata = NULL,
  which = c("both", "sma", "routing"),
  ...,
  all = TRUE,
  feasible.set = FALSE,
  glue.quantiles = NULL,
  groups = NULL,
  FUN = sum,
  return_state = FALSE,
  return_components = FALSE
)

Arguments

object

an object of class hydromad.

newdata

a ts-like object containing a new time series dataset (replacing the original DATA argument given to the hydromad function).

which

selects either the SMA or routing model, or both models (the default). Note that if which = "routing", then newdata is treated as the effective rainfall (U).

...

any unmatched arguments will generate an error.

all

if TRUE, return the entire time series for which data exists. Otherwise, the warmup period (specified as an argument to hydromad or update) is stripped off.

feasible.set, glue.quantiles

if feasible.set is TRUE, then many simulations will be generated, using all parameter sets in the feasible set. This must have been previously specified using defineFeasibleSet. If glue.quantiles is NULL then all the simulated time series are returned. If it is c(0,1) then the overall bounds (minimum and maximum at each time step) are returned. Otherwise the specified quantiles are estimated using GLUE-type weighting.

groups, FUN

groups is an optional grouping variable, of the same length as the observed data in object, used to aggregate the observed and fitted time series. The function FUN is applied to each group.

return_state

passed to the SMA simulation function, to return state variables.

return_components

passed to the routing simulation function, to return flow components.

Value

simulated time series.

See also

hydromad, update.hydromad

Author

Felix Andrews felix@nfrac.org