Interesting note from NWS Norman regarding the 12z model initializations. FWIW, the 21z RAP edged west with the main precip shield, while the 18z Euro maintained.
The complicating factor with this forecast is that for precipitation
amounts and types with the cyclone, the models widely diverge in
solutions and the impacts that result from the solutions. In
discussions with WPC, the GFS/NAM 12z initialization had a 30 meter
initialization error in the 500mb height fields, which should have a
sharper upper-level trough. For this forecast more weight was put
towards the ECMWF solution for precipitation location and type,
however the forecast for precipitation type/impacts could
drastically change with any shifts in path of the cyclone.