s2dv (version 2.1.0)

UltimateBrier: Compute Brier scores

Description

Interface to compute probabilistic scores (Brier Score, Brier Skill Score) from the forecast and observational data anomalies. It provides six types to choose.

Usage

UltimateBrier(
  exp,
  obs,
  dat_dim = NULL,
  memb_dim = "member",
  time_dim = "sdate",
  quantile = TRUE,
  thr = c(5/100, 95/100),
  type = "BS",
  decomposition = TRUE,
  ncores = NULL
)

Value

If 'type' is 'BS' or 'FairStartDatesBS' and 'decomposition' is TRUE, the output is a list of 4 arrays (see details below.) In other cases, the output is an array of Brier scores or Brier skill scores. All the arrays have the same dimensions: c(nexp, nobs, no. of bins, the rest dimensions of 'exp' except 'time_dim' and 'memb_dim'). 'nexp' and 'nobs' is the length of dataset dimension in 'exp' and 'obs' respectively. If dat_dim is NULL, nexp and nobs are omitted.

The list of 4 includes:

• $bs: Brier Score

• $rel: Reliability component

• $res: Resolution component

• $unc: Uncertainty component

Arguments

exp

A numeric array of forecast anomalies with named dimensions that at least include 'memb_dim', and 'time_dim'. It can be provided by Ano().

obs

A numeric array of observational reference anomalies with named dimensions that at least include 'time_dim'. If it has 'memb_dim', the length must be 1. The dimensions should be consistent with 'exp' except 'dat_dim' and 'memb_dim'. It can be provided by Ano().

dat_dim

A character string indicating the name of the dataset dimension in 'exp' and 'obs'. The default value is NULL (no dataset). dimension, set NULL.

memb_dim

A character string indicating the name of the member dimension in 'exp' (and 'obs') for ensemble mean calculation. The default value is 'member'.

time_dim

A character string indicating the dimension along which to compute the probabilistic scores. The default value is 'sdate'.

quantile

A logical value to decide whether a quantile (TRUE) or a threshold (FALSE) is used to estimate the forecast and observed probabilities. If 'type' is 'FairEnsembleBS' or 'FairEnsembleBSS', it must be TRUE. The default value is TRUE.

thr

A numeric vector to be used in probability calculation (for 'BS', 'FairStartDatesBS', 'BSS', and 'FairStartDatesBSS') and binary event judgement (for 'FairEnsembleBS' and 'FairEnsembleBSS'). It is as quantiles if 'quantile' is TRUE or as thresholds if 'quantile' is FALSE. The default value is c(0.05, 0.95) for 'quantile = TRUE'.

type

A character string of the desired score type. It can be the following values:

• 'BS': Simple Brier Score. Use SpecsVerification::BrierDecomp inside.

• 'FairEnsembleBS': Corrected Brier Score computed across ensemble members. Use SpecsVerification::FairBrier inside.

• 'FairStartDatesBS': Corrected Brier Score computed across starting dates. Use s2dv:::.BrierScore inside.

• 'BSS': Simple Brier Skill Score. Use s2dv:::.BrierScore inside.

• 'FairEnsembleBSS': Corrected Brier Skill Score computed across ensemble members. Use SpecsVerification::FairBrierSs inside.

• 'FairStartDatesBSS': Corrected Brier Skill Score computed across starting dates. Use s2dv:::.BrierScore inside.

The default value is 'BS'.

decomposition

A logical value to determine whether the decomposition of the Brier Score should be provided (TRUE) or not (FALSE). It is only used when 'type' is 'BS' or 'FairStartDatesBS'. The default value is TRUE.

ncores

An integer indicating the number of cores to use for parallel computation. The default value is NULL.

Examples

 # \dontshow{
startDates <- c('19851101', '19901101', '19951101', '20001101', '20051101')
sampleData <- s2dv:::.LoadSampleData('tos', c('experiment'),
                                    c('observation'), startDates,
                                    leadtimemin = 1,
                                    leadtimemax = 4,
                                    output = 'lonlat',
                                    latmin = 27, latmax = 48,
                                    lonmin = -12, lonmax = 40)
 # }
sampleData$mod <- Season(sampleData$mod, monini = 11, moninf = 12, monsup = 2)
sampleData$obs <- Season(sampleData$obs, monini = 11, moninf = 12, monsup = 2)
clim <- Clim(sampleData$mod, sampleData$obs)
exp <- Ano(sampleData$mod, clim$clim_exp)
obs <- Ano(sampleData$obs, clim$clim_obs)
bs <- UltimateBrier(exp, obs, dat_dim = 'dataset')
bss <- UltimateBrier(exp, obs, type = 'BSS', dat_dim = 'dataset')