LAK packages

[Milos Novotny]

I am updating a model grid with a lake that will be filled in a predictive simulation. I am flexible and can use NWT/USG/MF6 but with the upstream-weighting approach. I am trying to understand the differences in LAK packages and whether any of these will be effective for the simulation or whether using a highK/S approach will be better. I have used the Council (1998) LAK2 package with SURFACT on other projects successfully.

My understanding is the LAK3/LAK7 package requires the volume of the lake to be represented by noflow cells (or possibly combination of noflow and stage-area-volume defined) AND that the model grid needs to be horizontal in the lake (Assumptions and Limitations section, Merritt and Konikow, 2000). This would need to be configured in the model layering which is less than ideal.

Alternately,  the MF6 LAK package is characterized as being similar to the RIV package (as is LAK2) in Documentation for the MODFLOW 6 Groundwater Flow Model (Langevin et al, 2017) and does not indicate the requirements for noflow cells and horizontal layers mentioned above. 

If this is the case, I would like to just assign the LAK boundary conditions in the top model layer (with the top equal to sloping land surface) and define stage-area-volume separately and not use noflow cells to represent the volume.

Additionally, it would be useful if I could define some LAK6 connections to GWF cells independent of their layering (i.e. connect cells that are in lower layers but similar elevation as the physical lake).

Any explanation on LAK package practical limitations and assumption and/or guidance is appreciated!!