addition of vpr results in model becoming unstable?

My model has three branches with distributed tributaries, one tributary, and one outlet structure. Everything was running fine in terms of water balance and maintained a stable water surface elevation. Upon the addition of a vpr for temperature, the model went unstable. If I assume isothermal conditions instead of a vpr it runs fine. The error message is: Unstable water surface elevation on day 182.000 negative surface layer thickness using minimum timestep at iteration 4 Segment, Surface layer thickness, m, Flow m3/s, U(KT,I) m/s, ELWS, m 6 0.50 -0.91 0.00 342.50 7 0.51 -141.32 -0.06 342.51 8 1.26 -21025.69 0.03 343.26 9 -0.48 -8861.90 -4.64 341.52 10 0.49 -8715.68 -6.37 342.49 11 0.50 -8715.00 -12.97 342.50 12 0.50 -8714.99 -37.78 342.50 My layers are 0.5 m thick and the water surface elevation never fluctuates more than this, so unsure why it thinks layers are going dry. The flow (m3/s) shown as negative seems rather large relative to any flows used in my model. Checked my inflow files and they are all in proper F8 format so I don't think it is misreading (plus lake surface elevation is stable so water balance is working fine). I tried reducing the time step but that did not solve the problem. If anyone can provide any insight here that would be appreciated!