MMET cases are established in-part to address situations where operational models do not appripriately predict high impact and/or routine weather events. In addition to the cases provided via MMET, over the last several years, EMC's Model Evaluation Group (MEG) has identified persistent or common deficiencies in the NCEP modeling suite. Established in 2012, the weekly MEG discussions have helped increase communication and feedback among model developers, NWS forecast offices, and the research community. The MEG meets weekly to examine model performance, including subjective and objective verification, of recent meteorological events as well as investigating cumulative performance of operational and experimental models.

Below is subset of cases of interest and/or persistent operational model issues (source: weekly MEG meetings).

Cases Identified in MEG

Case	Case Details
26 Apr 2016	Models struggled with storms developing on the dryline
9-11 Mar 2016	Medium-range forecast of heavy precipitation over Ark-La-Tex
16 Feb 2016	Cold air damming eroded too quickly; GFS was too warm, especially across NC; Fire Wx NAM nest was best (large differences in low-level winds in parent and nest); HRRR was better in west VA than NAM
14 Dec 2015	NAM had moist absolutely unstable layer (MAUL)
16 Nov 2015	Tornado outbreak over western Plains
30 Oct 2015	Central TX flooding
3-4 Oct 2015	South Carolina flooding
29 Aug 2015	Pacific NW wind storm; GFS and NAM failed to give advance warning
12-13 Jul 2015	Derecho event, spanning from MN to KY; Parallel NAM and HRRR performed better than operational models with propagation and placement
8 May 2015	Moderate risk bust; NAM was more unstable than other models; NAM had CAPE recover much too quickly
26-27 Mar 2015	Eastern TX bow echo; NAM had too strong of cold pools compared to ARW, had too strong of LLJ, and CAPE recovered too quickly after early morning convection
27 Jan 2015	NY blizzard "bust"
25 Jan and 16-17 Feb 2015	GFS too cold 2-m temperatures over snowpack
22 Jan 2015	Philadelphia ice event
14 Dec 2015	NAM had moist absolutely unstable layer (MAUL)

Reviewed Top GFS Shortcomings

While MMET focuses on mesoscale modeling, understanding the performance of the global model is also informative. Below are common issues with the GFS:

• Late afternoon 2-m cold, wet bias in E. US
▪ 2-m temperature can plummet between 21 and 00 UTC, partially due to collapse of boundary layer
▪ 2-m temperature can drop rapidly over snow cover with light winds
• GFS inversions not strong enough
• Biases in low-level winds over CONUS
• US precipitation skill and biases
▪ Skill lower than other centers
▪ Areas of widespread over-prediction at small thresholds
▪ Too dry medium amounts
▪ Popcorn pattern in West in Summer
• Southern Plains too warm (12Z) 2-m temperature
▪ Great Plains still too warm and dry in summer
▪ CAPE underestimated
• Biases compared to radiosondes
• GFS too fast, too weak with mid-latitude systems
• Inconsistent forecasts, even at short ranges
▪ One out of every 3-4 events
▪ Run-to-run jumpiness