This comparison will be examining 2 cases both of which occurred in June 2018. The first case occurred on evening of June 2nd when a squall line of thunderstorms moved through central Nebraska. Both the GFS/FV3 did a good job depicting this potential complex of thunderstorms on the 00Z run which verified well on MRMS radar loop. However, the FV3 did a much better job in handling the convective cold pool as this complex of storms moved through. In fact, the FV3 showed temperatures dropping into the 60s after the passage of the squall line which was verified on the Metar site ORD. The GFS seemed to miss this cold pool and kept 2 meter temperatures in the 70s and 80s.   

The second case  occurred on June 19th. In this case, both the FV3/GFS initialized a closed upper level over the Pacific Northwest on the 00Z run. However, there were some significant differences in each initialization. The GFS seemed to initialize the 00Z 500mb upper air analysis very well. On the other hand, the FV3 initialized the closed upper level low a bit further west and stronger than what was depicted on the 00Z upper air analysis. The GFS also did a better job handling the upper level low 12hr forecast based on the 12z upper air analysis, while the FV3 tended to keep the upper level low a bit further west and stronger. This caused some significant differences in the QPF over the mountains of western Wyoming. The FV3 projected 2 to 4 inches of precipitation over the mountains by June 19th at 12z, while the GFS predicted very light amounts. The final results from the MRMS 24hr QPE displayed little if any precipitation at 12z.