SpecsVerification (version 0.5-3)

PlotRankhist: Plotting function for rank histograms

Description

Plots a rank histogram in different modes.

Usage

PlotRankhist(rank.hist, mode = "raw")

Arguments

rank.hist

A vector or rank counts.

mode

Either "raw" (default) or "prob.paper". Whether to draw the raw rank histogram, or the rank histogram on probability paper. See Details.

Details

The plotting modes currently implemented are:

raw (the default): A simple bar plot of the counts provided by the `rank.hist` argument.

prob.paper: The individual counts given by `rank.hist` are transformed to their cumulative probabilities under the binomial distribution with parameters `N` and `1/K`, where `N=sum(rank.hist)` and `K=length(rank.hist)`. This transformation makes possible an assessment of the observed rank counts under the hypothesis of equally likely ranks. The y-axis on the left indicates the cumulative probabilities. The intervals on the right of the plot indicate central 90, 95, and 99 percent _simultaneous_ confidence intervals. That is, if all ranks were equally likely on average, approximately 90 percent of all rank histograms would be _completely_ contained in the 90 percent interval and approximately 10 percent of all rank histograms would have _at least_ one bar that falls outside this interval.

References

Anderson J.L. (1996). A Method for Producing and Evaluating Probabilistic Forecasts from Ensemble Model Integrations. J. Climate, 9, 1518--1530. Broecker J. (2008). On reliability analysis of multi-categorical forecasts. Nonlin. Processes Geophys., 15, 661-673.

See Also

Rankhist, TestRankhist

Examples

# NOT RUN {
data(eurotempforecast)
rank.hist <- Rankhist(ens, obs)
PlotRankhist(rank.hist, mode="prob.paper")
# }