TSM produces extreme values for small volume / large surface area combo

[sjordan29]

When I was coupling riverine and modules to run the most recent Plan 38 from @jrutyna, I found that TSM produced extreme negative results that caused riverine's solver fail.

Cell 113 starts dry and wets between timestep 3,874 to 3,875, when it then has a very small volume through the end of the simulation. This is what flagged a few questions for me -- not sure if these should be addressed in modules or just in the linking:

Observations

• Zero volume cells: TSM produces an inf result. That seems fine to me, you don't really need to pass in zero-volume cells anyways, so I've just been handling this on the BMI / model-linking side and I think that's appropriate.
• Cells with low volumes and high surface area: TSM produces a extremely low negative value (e.g., -1.94059847e+10 degrees C).
  • I'm pulling the surface area from the RAS output HDF file in this case (Geometry/2D Flow Areas/TestArea/Cells Surface Area), which doesn't have a time dimension. Therefore, the surface area is just the surface area of the cell itself, not necessarily the wet area. @jrutyna - is that tracked in RAS or its output?
  • Feeding this value causes riverine to produce null results, but riverine runs just fine on its own, with that cell tracking just above about 12 degC. The extreme value likely makes the matrix unsolvable.
  • I think this is something we should address in TSM, since liquid water won't be lower than 0 C. I'd think we'd want to prevent it from producing unrealistic results... but this is also a 'garbage in, garbage out' scenario. When I drop the surface area to a more reasonable value based on the volume, we get more realistic output from tsm.