In order to be able to better analyze FlowNet results in the Brugge example, the reference simulation should output some additional summary vectors. A few ones that I noticed to be missing are: FOPT, FWPT, FWIT, FOIP, FPR
The report dates are monthly during the first years of the reference simulation and every 6 months towards the last years. This causes the frequency of observations extracted from the reference simulation for the FlowNet runs to vary too much, which ends up giving too much weight on matching the reference simulation at earlier dates. We should enable monthly report dates throughout the entire simulated period. This will also improve the use of the resampling option to vary the frequency of the observations used for training the FlowNet ensemble
FOPT, FWPT, FWIT, FOIP, FPR
resampling
option to vary the frequency of the observations used for training the FlowNet ensemble