Issue with phreatic storage in semi-confined models

[Sietse-AG]

The docstring in timflow.transient.ModelMaq describes that when "topboundary" is "semi" and "phreatic top" is "True", "Sll" is interpreted as phreatic storage and not multiplied with the layer thickness. However, from lines 111 to 117 in transient/aquifer.py follows that this only happens when "topboundary" is set to "leaky" and not "semi".

[mbakker7]

Thanks for pointing this out. This is indeed incorrect. And it is something we are currently working on. The phreatictop keyword is removed entirely in PR #88 (currently in the dev branch). To specify a phreatic top aquifer, use the new topboundary="phreatic". A semi-confined aquifer with phreatic storage makes no sense (that was a typo in the docs), but a leaky layer top can potentially have a phreatic surface. This has to be developed, probably as part of computing the vertical head change in a leaky layer.

[Huite]

A semi-confined aquifer with phreatic storage makes no sense (that was a typo in the docs), but a leaky layer top can potentially have a phreatic surface.

I agree that "a semi-confined aquifer with phreatic storage" as a physical entity doesn't make sense stated as such, but if the first resistance is interpreted as a lumped drainage or infiltration resistance instead, I think its use is defensible.

E.g. in a "Hollandprofiel", I might be interested in some "average phreatic drawdown", with the well placed in the Pleistocene sand below. I would estimate a drainage resistance of e.g. 25 days for the ditches (using Ernst or something), and specificy a specific yield.
This is obviously a work-around, because the "semi-confining" layer doesn't have a thickness and I'd probably specificy a token thickness of 1 cm or something.

For concreteness, I'd specify:

• c=25
• kD = 10 (Phreatic aquifer)
• c=3000 (Holocence resistance)
• kD = 1000 (Pleistocene aquifer)
• etc.

Of course, having it available in a proper API / structure would be preferable to a work-around; I'm just highlighting that there is a use case for this concept.

[mbakker7]

Thanks for your suggestion, @Huite .
I think there may indeed be a use case for such a system.
If we want to support that, what would be a good name for a topboundary?
topboundary = "dutch" ?

[mbakker7]

topboundary="drained"?

[Huite]

Good question... "drained" seems to imply some non-linearity, maybe? Technical terms like "Robin" boundary aren't any good either.
Borrowing from MODFLOW, it would be most like a "general head boundary", which I don't think is the catchiest either. I've thought about it for quite some time, though, and all alternatives seem worse.

Another thought: is the setting consistent with the steady state model? Should it be?

topboundary : string, 'conf' or 'semi' (default is 'conf')
        Indicates whether the topboundary is confined ('conf') or
        semi-confined ('semi').

• Confined: no source/sink on top, specific storage coefficient, not specific yield
• Phreatic: no source/sink on top, specific yield. Is identical to "confined" for steady-state?
• Semi-confined: works for both, assumes there is a head boundary on top and assumes the aquitard is real and has thickness.
• Drained: works for both, head boundary on top, the top "aquitard"'s thickness doesn't have to be specified.

I think adding drained may be worthwhile for steady-state too. Working with a dummy thickness is never ideal, I believe.