Need a generalized approach for dealing with diversion from confluence #9
Description
On 7/18/2018, an issue with the current integrated approach was brought to light by the Russian River model.
When a diversion occurs from a confluence, and the water available to the diversion should only reflect the water in one of the tributaries (not both), there is currently no means for accommodating this.
The reason for diverting from the confluence rather than the upstream end of the segment from which water is to be made available is because currently, GW-SW interaction terms are added/deducted from the downstream end of the segment (so to divert from the upstream end would be to divert water without first factoring GW-SW gains and losses)
The reason that the segment from which water is to be diverted can't simply be broken in two is because this would require new SFR segment numbering, which would mess up many other references in GSFLOW, including the subbasin numbering in PRMS, IRUNBND in UZF, and likely others. So in other words, once the construction of a GSFLOW model is far along it becomes too late to add in new demands into the SFR stream network because this would break a number of connections previously established during the model's construction.
During the phone call, Enrique offered a solution that will add nodes just upstream of confluences from which water can be diverted, and the link between these new nodes and the original confluence node will not have a counterpart in SFR, but won't need one.
(3) & (4) are depicted in the example networks below with further notation describing the problem.
The example .xy file used to create the figures also is attached in case other would like to modify and screen capture for posting to this issue. (To use the file, change the .txt file back to .xy after downloading and MODSIM will be able to open it. Github won't host .xy file types)
