\name{taylor.diagram} \alias{taylor.diagram} \title{ Taylor diagram } \description{ Display a Taylor diagram} \usage{ taylor.diagram(ref,model,add=FALSE,col="red",pch=19,pos.cor=TRUE, xlab="",ylab="",main="Taylor Diagram",show.gamma=TRUE,ngamma=3, gamma.col=8,sd.arcs=0,ref.sd=FALSE,sd.method="sample", grad.corr.lines=c(0.2,0.4,0.6,0.8,0.9), pcex=1,cex.axis=1,normalize=FALSE,mar=c(5,4,6,6),...) } \arguments{ \item{ref}{numeric vector - the reference values.} \item{model}{numeric vector - the predicted model values.} \item{add}{whether to draw the diagram or just add a point.} \item{col}{the color for the points displayed.} \item{pch}{the type of point to display.} \item{pos.cor}{whether to display only positive (\samp{TRUE}) or all values of correlation (\samp{FALSE}).} \item{xlab,ylab}{plot axis labels.} \item{main}{title for the plot.} \item{show.gamma}{whether to display standard deviation arcs around the reference point (only for \samp{pos.cor=TRUE}).} \item{ngamma}{the number of gammas to display (default=3).} \item{gamma.col}{color to use for the gamma arcs (only with pos.cor=TRUE).} \item{sd.arcs}{whether to display arcs along the standard deviation axes (see Details).} \item{ref.sd}{whether to display the arc representing the reference standard deviation.} \item{sd.method}{Whether to use the sample or estimated population SD.} \item{grad.corr.lines}{the values for the radial lines for correlation values (see Details).} \item{pcex}{character expansion for the plotted points.} \item{cex.axis}{character expansion for the axis text.} \item{normalize}{whether to normalize the models so that the reference has a standard deviation of 1.} \item{mar}{margins - only applies to the \samp{pos.cor=TRUE} plot.} \item{...}{Additional arguments passed to \samp{plot}.} } \details{ The Taylor diagram is used to display the quality of model predictions against the reference values, typically direct observations. A diagram is built by plotting one model against the reference, then adding alternative model points. If \samp{normalize=TRUE} when plotting the first model, remember to set it to \samp{TRUE} when plotting additional models. Two displays are available. One displays the entire range of correlations from -1 to 1. Setting \samp{pos.cor} to \samp{FALSE} will produce this display. The -1 to 1 display includes a radial grid for the correlation values. When \samp{pos.cor} is set to \samp{TRUE}, only the range from 0 to 1 will be displayed. The \samp{gamma} lines and the arc at the reference standard deviation are optional in this display. Both the standard deviation arcs and the gamma lines are optional in the \samp{pos.cor=TRUE} version. Setting \samp{sd.arcs} or \samp{grad.corr.lines} to zero or FALSE will cause them not to be displayed. If more than one value is passed for \samp{sd.arcs}, the function will try to use the values passed, otherwise it will call \samp{pretty} to calculate the values. } \value{ The values of \samp{par} that preceded the function. This allows the user to add points to the diagram, then restore the original values. This is only necessary when using the 0 to 1 correlation range. } \references{ Taylor, K.E. (2001) Summarizing multiple aspects of model performance in a single diagram. Journal of Geophysical Research, 106: 7183-7192. } \author{Olivier Eterradossi with modifications by Jim Lemon} \examples{ # fake some reference data ref<-rnorm(30,sd=2) # add a little noise model1<-ref+rnorm(30)/2 # add more noise model2<-ref+rnorm(30) # display the diagram with the better model oldpar<-taylor.diagram(ref,model1) # now add the worse model taylor.diagram(ref,model2,add=TRUE,col="blue") # get approximate legend position lpos<-1.5*sd(ref) # add a legend legend(lpos,lpos,legend=c("Better","Worse"),pch=19,col=c("red","blue")) # now restore par values par(oldpar) # show the "all correlation" display taylor.diagram(ref,model1,pos.cor=FALSE) taylor.diagram(ref,model2,add=TRUE,col="blue") } \keyword{misc}