Unconfined calculation with perched water table

[martinveenvliet]

Hello,

For a complex problem where multiple confined (and some unconfined) aquifers are overlaying an unconfined aquifer I would like to calculate the correct flux from the overlaying confined layer to the unconfined layer. On the most complex locations we have a total of 3 perched water tables in the soil column.

From what I gather, there is an option in MODFLOW 6 for the correct calculation of fluxes from a confined layer to an underlying unconfined layer, namely the PERCHED option in the NPF package for calculation of fluxes (instead of taking the head difference between the two layers, taking difference between the head in the overlaying layer and the bottom of that layer).

However as these kind of complex unconfined models are often difficult to converge, it is almost always necessary to use the Newton Rhapson formulation. Unfortunately, the Newton Rhapson formulation does not support the PERCHED option.

Is there another way to correctly model the flux from a perched water table in MF6 using the Newton Rhapson formulation or am I simply missing something?

In the past (MF2005) we've modelled perched water tables by adjusting the resistance below the confined layer. As this is a MODFLOW 6 model and the vertical conductance between cells is calculated based on both the k33 of the confined and unconfined cell this does not seem possible (as changing the k33 value would also inherently change the incoming flux to the confined cell (from the overlaying layer)).

[christianlangevin]

Hi @martinveenvliet, I'm afraid that at the moment we don't have anything that will work for your particular case. There are two things that may help in the future.

1. There has been some work to get a vcont capability into MODFLOW 6 either through a vertical flow barrier or through specification of vcont for one or more cells. Not sure of the time frame, but it seems like this might work for you using the trick you mentioned.
2. MODFLOW-USG has a "flow to dry cell" option that that allows the Newton formulation to work with the perched conditions that you mention. We don't have this in MODFLOW 6, though we have thought about including it. Might be good for us to keep this on our radar.

Apologies for not having anything that can provide immediate relief.

[martinveenvliet]

Hi @christianlangevin

Thank you for your answer, would be curious to see what's possible in the future with either the vertical flow barrier or something similar to the flow to dry cell option. For now I will perform an iteration outside the entire model (let MODFLOW 6 finish simulation, adjust K33 values and rerun).

Just for reference (and maybe if someone else runs into this):
I also tried to use the TVK package in combination with the MODFLOW API, but as the NPF of that timestep (even with TVK) is initialized before the callback to the stress period and is not called again throughout the stress period/time step/iteration it was not possible to do any adjustments that result in a different K33 for that stress period.