The Relative Operating Characteristic Skill Score (ROCSS; Kharin and Zwiers, 2003) is based on the ROC curve, which gives information about the hit rates against the false-alarm rates for a particular category or event. The ROC curve can be summarized with the area under the ROC curve, known as the ROC score, to provide a skill value for each category. The ROCSS ranges between minus infinite and 1. A positive ROCSS value indicates that the forecast has higher skill than the reference forecasts, meaning the contrary otherwise.
ROCSS(
exp,
obs,
ref = NULL,
time_dim = "sdate",
memb_dim = "member",
dat_dim = NULL,
prob_thresholds = c(1/3, 2/3),
indices_for_clim = NULL,
cross.val = FALSE,
ncores = NULL
)A numerical array of ROCSS with the same dimensions as 'exp' excluding 'time_dim' and 'memb_dim' dimensions and including 'cat' dimension, which is each category. The length if 'cat' dimension corresponds to the number of probabilistic categories, i.e., 1 + length(prob_thresholds). If there are multiple datasets, two additional dimensions 'nexp' and 'nobs' are added.
A named numerical array of the forecast with at least time and member dimension.
A named numerical array of the observation with at least time dimension. The dimensions must be the same as 'exp' except 'memb_dim' and 'dat_dim'.
A named numerical array of the reference forecast data with at least time and member dimension. The dimensions must be the same as 'exp' except 'memb_dim' and 'dat_dim'. If there is only one reference dataset, it should not have dataset dimension. If there is corresponding reference for each experiement, the dataset dimension must have the same length as in 'exp'. If 'ref' is NULL, the random forecast is used as reference forecast. The default value is NULL.
A character string indicating the name of the time dimension. The default value is 'sdate'.
A character string indicating the name of the member dimension to compute the probabilities of the forecast and the reference forecast. The default value is 'member'.
A character string indicating the name of dataset dimension. The length of this dimension can be different between 'exp' and 'obs'. The default value is NULL.
A numeric vector of the relative thresholds (from 0 to 1) between the categories. The default value is c(1/3, 2/3), which corresponds to tercile equiprobable categories.
A vector of the indices to be taken along 'time_dim' for computing the thresholds between the probabilistic categories. If NULL, the whole period is used. The default value is NULL.
A logical indicating whether to compute the thresholds between probabilistic categories in cross-validation. The default value is FALSE.
An integer indicating the number of cores to use for parallel computation. The default value is NULL.
Kharin, V. V. and Zwiers, F. W. (2003): https://doi.org/10.1175/1520-0442(2003)016
exp <- array(rnorm(1000), dim = c(lon = 3, lat = 2, sdate = 60, member = 10))
ref <- array(rnorm(1000), dim = c(lon = 3, lat = 2, sdate = 60, member = 10))
obs <- array(rnorm(1000), dim = c(lon = 3, lat = 2, sdate = 60))
ROCSS(exp = exp, obs = obs) ## random forecast as reference forecast
ROCSS(exp = exp, obs = obs, ref = ref) ## ref as reference forecast
Run the code above in your browser using DataLab