anders-kiaer
commented
10 months ago After video meeting between @sigurdp, @jorgenherje and myself:
We see two different feasible solutions to this:
1) We continue treating the historical vector as an "add-on", but instead of assuming they are all equal and showing only the first realization, we let the visualization of historical vectors use the same logic as the "standard vectors". I.e. if "Show individual realizations" is selected, we show all the different historical vectors on top of the standard vector realizations. Similarly, if "Statistical fan chart" we show fan chart for both historical vector and standard vector. Etc. A useful implementation optimization is if backend can verify the historical vectors are equal (which still will continue to be true for the majority of cases in SUMO probably), it can provide this information to the frontend such that it does not need to plot 100+ unnecessary traces on top of each other.
2) Move historical vectors from being an "add-on" to being shown in the same list as standard vectors. This might simplify the code, but still also might want the optimization with showing only one trace in case they are all equal? The downside with is that it would be more effort for the user to keep historical vector in "sync" with the corresponding simulated vector. It would also require that they are grouped on ensemble (i.e. show all different vectors in same plot) and not per vector which is the current default.
We decided to pursue 1).
A given historical vector can vary across the ensemble, e.g. if there are allocation uncertainty. Currently we only show for one realization.
What should we do instead? Should we treat them equally as as the "proper" vector (e.g. show statistical fan chart, if that is selected etc).?