plotting resids with PDC

[mnfienen]

In cases where ++ies_drop_conflicts is True, it would be nice to eliminate those dropped obs (especially if they are really really bad!) from plots like 1to1 plots. A simple flag could be supplied to pst.plot(kind='1to1', drop_pdc=True) for example, that reads the csv file with PDC-dropped obs and just removes them before plotting.

I can implement, but wanted to see if anyone has objections or other ideas of how to deal.

[jwhite-usgs]

I think we do need to get better at handling conflicts in general. But instead of dropping, what about having them washed out or otherwise colored differently or at least have that option? There is a more advanced PDC detection routine in helpers now so we could potentially just call that under the hood in the plotting so that we dont have an external file dependency (that may change in the future!). Just my thoughts.

[jtwhite79]

@mnfienen - any interest in this option still?

[mnfienen]

Yeah - I do think this might be useful (although I've been going rogue with 1to1 plots lately). Maybe could have two options for pdc_plot like 'drop' or 'gray'? If that's cool, maybe I can implement.

Also, for some recent work, found it valuable to plot 1to1 with error bars in both obs and modeled like the attached. Maybe can fold that into pyemu?

[jtwhite79]

Those plots look awesome! We could also just have variants on the 1-to-1 plots since all of us GW modelers are so obsessed with it - like a Pst.plot(kind="1to1-fancysauce") ? I like showing the meas error bars too!

[jtwhite79]

@mnfienen - any movement on this?

[mnfienen]

@jtwhite79 nope lol. I can try and revisit though!

[briochh]

@mnfienen, happy for me to move this to discussions? Might be good to revisit when whoever has the time?