As a user, I want the WRES to allow for custom bins for summary statistic histograms (rather than a simple bin count)

[HankHerr-NOAA]

I was going to write an email, because I think we might want to reject this ticket, but decided on a ticket so that the conversation is not lost.

I was attempting to generate a plot similar to Slide 26 in the Tech Demo (not going to share link here) where the NWM team summarized a difference in KGE between NWM v3.0 and v2.2 across features in a histogram. My attempt to produce something similar used retrospective simulations for NWM v3.0 and v2.1 and features across MARFC. The plot is attached; link at bottom. You can see that outliers are making the plot difficult to read. In fact, I had to use 200 (!) bins to get the detail I got. The NWM team got around the outlier issue by having upper and lower bins that stretch to infinity. The result was a plot that was "ready for primetime", versus a WRES output that is not (in my opinion).

To support that, the WRES would likely need to allow for custom bins to be defined, instead of just a bin count, with the option for bins that stretch to infinity.

The reason we might want to reject this ticket is because I don't think we advertise the WRES as being able to generate plots that are "ready for primetime". Its a calculating engine, producing numbers, and we always anticipated users (machine or human) would then take those numbers and build displays. Having said that, our users are generating images, and some may want to share them directly with their stakeholders.

Let me know if you would like to reject this ticket or keep it and put it in the backlog. Thanks,

Hank

[james-d-brown]

I don't mind you retaining the ticket but, as you said, any solution would need to be commensurate with the idea that graphics are a nice-to-have only with WRES, and we are not going to tailor them too much. But it may be reasonable to define an upper and lower bucket that catches everything outside.