rcrv {verification}

Reduced centered random variable

Description

The RCRV provides information on the reliability of an ensemble system in terms of the bias and the dispersion. A perfectly reliable system as no bias and a dispersion equal to 1. The observational error is taken into account

Usage

      rcrv(obs,epsMean,epsVariance,obsError)
       

Arguments

obs

A vector of observations

epsMean

A vector of the means of the ensemble

epsVariance

A vector of the variances of the ensemble

obsError

Observational error

Value

bias

The weighted bias between the ensemble and the observation. A value equal to 0 indicates no bias. A positive (negative) value indicates a positive (negative) bias

disp

The dispersion of the ensemble. A value equal to 1 indicates no dispersion. A value greater (smaller) then 1 indicates underdispersion (overdispersion)

y

Vector of y. Mean of y equals bias and standard deviation of y equals dispersion

obsError

Observational error (passed to function)

Author(s)

Ronald Frenette <[e-mail]>

References

G. Candille, C. P. L. Houtekamer, and G. Pellerin: Verification of an Ensemble Prediction System against Observations, Monthly Weather Review,135, pp. 2688-2699

Examples

data(precip.ensemble) 
#Observations are in the column
obs<-precip.ensemble[,3] 

#Forecast values of ensemble are in the column 4 to 54
eps<-precip.ensemble[,4:54]   

#Means and variances of the ensemble
mean<-apply(eps,1,mean)
var<-apply(eps,1,var)

#observation error of 0.5mm 
sig0 <- 0.5 

rcrv(obs,mean,var,sig0)