Going to put this here since it is relevant to severe weather. I'm in Albany, NY this week for the NY State mesonet symposium. Several talks have focused on HRRR verification against the state mesonet sites and ASOS sites. Couple of conclusions that have come up over and over again regarding the HRRR:
1.) The HRRR systematically overpredicts surface winds.
2.) The HRRR performs poorly with initiating boundary layer clouds (cumulus) in days following air mass changes during June, July, August.
3.) The HRRR clearly overmixes the boundary layer, especially in SW flow events, and drops the surface dewpoint too quickly. This causes Surface Based CAPE (SBCAPE) to be between modeled 250-750 j/kg SBCAPE too low.
4.) The HRRR is too quick to mix out low level temperature inversions (The Wedge™ ) during spring convective events.