METviewer: Time Series of Categorical Statistics Plot Output
If you successfully followed the instructions on the previous page, you should see a plot appear in the in the plot tab of the METviewer window. Your plot should look like this PNG IMAGE. And here is the corresponding plot XML File.
Studying the plot shows that the Critical Success Index (CSI) decreases for higher thresholds, as if often the case. This suggests that the forecast models have more skill at lower precipitation thresholds. This contrasts with the story told by the accuracy statistic, which increases as the threshold increases. This can be explained by the simplicity of the accuracy statistic, and the effect that always predicting no event can have for extreme events.
The following section discusses the output that METviewer generates for each plot. Click through the tabs above the METviewer plot to examine each:
- XML Tab: When the Generate Plot button is clicked, the web application serializes the information in the web page controls into an XML document and sends it across the internet to METviewer. This XML document can be used to generate a plot directly from METviewer using the command line interface, also called the batch engine. The batch engine can create many plots from a single XML document, using only small changes. Thus, a good way to use the METviewer web application is to "design" plots and then take the XML for the plot and extend it to create many different plots.
- Log Tab: The METviewer plot engine generates status output as it runs, and this information can be useful for finding problems.
- R script Tab: METviewer uses the R statistics package to generate plots by constructing an R script and running it in R. The R script that is constructed can be saved and modified if the users would like to make changes to the plot that are not supported in METviewer.
- R data Tab: Along with the R script generated by METviewer for the plot, the plot data is stored in a file that is read when the plot is generated in R. This file can be useful for problem solving, since it shows all of the data used in the plot and can be loaded directly into R for analysis.
- SQL Tab: All of the MET verification statistics output is stored in a database, which METviewer searches when it generates a plot. The SQL generated by METviewer to gather the data for a plot can be viewed, in case the user wants to explore or debug using the SQL directly.