Comparison of EDGE_VELOCITY versus FLOW_ACROSS_FACE

[sjordan29]

While trying to clear up some memory for Clearwater-riverine, I found an interesting discrepancy between EDGE_VELOCITY and FLOW_ACROSS_FACE. There are instances when the EDGE_VELOCITY is equal to zero but the FLOW_ACROSS_FACE is nonzero. I suspect this is due to the coarser resolution of RAS output than the actual model timestep; i.e., maybe EDGE_VELOCITY pulls from the instantaneous velocity whereas the FLOW_ACROSS_FACE is the aggregate value of any sub-second flows across the face in RAS. I observe this in Plan 37.

@jrutyna- do you know if this is the case?

Implications:

• Advection Coefficient: this is calculated as follows, which was a remnant of some old logic that actually exposed this issue. I suspect we'll just want the advection coefficient to be equal the flow across the face, and not set this to zero if the edge velocity is 0.

mesh[FLOW_ACROSS_FACE] * np.sign(abs(mesh[EDGE_VELOCITY]))

• Flow Across Face: if we do change the advection coefficient as proposed above, then we could delete this variable (since it should be identical to the advection coefficient) in the matrix and free up some memory.