This function implements the reliability categorisation for forecasts of binary events as documented in Weisheimer et al. (2014). It has only been implemented for category forecasts with categories defined relative to the forecast and observed climatological distribution (i.e. without systematic bias).
weisheimer(
ens,
obs,
pthresh = 2/3,
nboot = 100,
brier.thresholds = seq(0, 1, 0.2),
...
)n x k matrix of n forecasts from k ensemble members
n verifying observations
probability threshold to convert to category forecasts. If negative, event falling below threshold is used, else, event above threshold is used.
number of bootstrap replicates to estimate 75 percent confidence interval
Thresholds used to bin the forecasts (see
brier)
additional arguments for compatibility with other scores