Graphical comparison of model and observed data

[rdstern]

There seems to be a book that explains the analyses using the openair package. Chapter 19 considers conditional quantiles and they seem to be useful?

Let me put this into context. We are usually comparing daily values and the comparisons may be for one station or for a number of stations.

a) One type of plot keeps the time-series structure. In these cases we must be careful when there is seasonal variation. We may ignore it, or get annual summaries for the comparisons. Or perhaps have separate comparisons for each month. This might be through getting monthly summaries. b) Another set of plots could concentrate on the seasonal structure c) Cutting across there is the distributions - how do they compare. That could be boxplots or density plots. I think q-q plots fit well here. The problem with those is that they compare the shapes - which is good, but they don't compare individual values, i.e. when one is high in the distribution, then what about the other? d) And where do the Taylor plots fit?

In c) I think that's where the conditional quantile plots come in?. ThisThey look pretty useful. And they seem pretty comprehensive in the openair package. So I suggest they be added to R-Instat. Perhaps it is the same dialogue as Taylor plots (using a radio button, or its own separate dialogue?

[Comfort123977]

@Cal-Insta and @Comfort123977 have gone through some graphs using hydroGOF and openair. Please find attached a copy of the related document. Graphical Comparison.pdf

[Cal-Insta]

@Comfort123977 and I have looked at some graphs in the hydroGOF package and openair package. In the hydroGOF package, we found the ggof plot (graphical goodness of fit) was interesting. There is also a plot2 function that gives a basic plot of the two-time series. There is the conditional quantile plot in the verification package but the conditional quantile plot in the openair package is more flexible.