It seems that using hord_tr=10 is related to strange radar reflectivity drops in the supercell simulation when using NSSL microphysics. Changing to hord_tr=8 alleviates the issue.
This may suggest some kind of problem with hord_tr=10, but would need more investigation. The issue occurs near the top of a strong updraft (~40m/s) and strong horizontal divergence (U and V magnitudes on the order of 20-30m/s)