Proposal regarding DEV_MINIMUM_SATURATED_THICKNESS

[tandreasr]

Hi Chris,
we've been very happy to see, that you seem to bring in the above mentioned NPF option in the next future.
Which release do you plan to make that option permanent - already in 6.4.2 or later on?

While testing, it actually helped us to solve a lot of problems we had with models featuring very thin aquifers thicknesses ,
which used to oscillate due to drying and rewetting issues.

But we've got one suggestion regarding that option.
Currently it is a faktor between 0..1 which scales the aquifer thickness of every cell.
In our opinion it would be much better to make it an absolute value in [LENGTH_UNITS] (as specified in DIS* packages).
This is due to the fact that in very inhomogenous aquifers (regarding their cell thicknesses) the relative approach leads to problems.

As an example:
We've got a lot of models where aquifers differs thickness between 0.1m at the borders and more than 30m in the central model regions.
The relative approach
EITHER leads for a small DEV_MINIMUM_SATURATED_THICKNESS (of for example 0.01) to the fact, that we get very small minimum thicknesses of 1E-3m at the border (leading to failing convergence due to massive solver oscillations)
OR for a bigger value (e.g. 0.5) the convergence at the border is fine, but in the central region we get a minium saturated thickness of 15m...
The better way would actually be to specify an overall absolute minimum saturated thickness of for example 0.1m for that example.
We've already tested out that absolute approach and are quite happy with the results.
If you support our proposal, I could send you the source code changes (just 3 minor changes from our point of view).

Thanks and best regards
Andreas

[langevin-usgs]

We will take a deeper look at this. Our dev options are typically for development and testing, and are not available in official releases. If you continue to have good success with the minimum saturated thickness dev option, we may consider it in the future, but we'd need to think through all the ramifications (for transport, particle tracking, etc.). Keep us posted.

[tandreasr]

Hi Chris,
sounds good for me.
And yes: as I've written before, we've got quite a couple of models we didn't get to converge otherwise due to very small saturated thicknesses being calculated in the range of 1e-3 and below, thus leading to instabilities/oscillations.
When using that option, they instantly behaved very well :-)
So: in our opinion that option does make a lot of sense.
Have a nice holiday!
Best regards
Andreas