Observation error check against forecast

For each stratification the standard deviation of the forecast over all forecast cycles and models (mname), and in the case of an eps forecast, the ensemble members is computed. The difference between the observation and the forecast is expected to be smaller than a number of multiples of the standard deviation. The number of multiples of the standard deviation can be supplied or a default value used depending on the parameter.

Usage

check_obs_against_fcst(
  .fcst,
  parameter,
  num_sd_allowed = NULL,
  stratification = c("SID", "quarter_day")
)

Arguments

.fcst

A harp_df data frame, or a harp_list, with an observations column.

parameter

The observations column. Can be the column name, quoted, or unquoted. If a variable it should be embraced - i.e. wrapped in {{}}.

num_sd_allowed

The number of standard deviations of the forecast that the difference between the forecast and the observation must be smaller than.

stratification

The columns to stratify the data by when computing the allowed tolerance. In most cases the column must exist in the input data, but "quarter_day" can be passed to divide the observations into classes of [0, 6), [6, 12), [12, 18) and [18, 24) hour of day. The default behaviour is to stratify by station ("SID") and "quarter_day".

Value

An object of the sames class as .fcst with an attribute named removed_cases containing a data frame of the removed cases.