Model Diagnostic Discussion NWS Weather Prediction Center College Park MD 1217 PM EDT Sat Sep 14 2019 Valid Sep 14/1200 UTC thru Sep 18/0000 UTC ...See NOUS42 KWNO (ADMNFD) for the status of the upper air ingest... 12Z Model Evaluation with Preferences and Forecast Confidence ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ...Great Lakes into Northeast CONUS... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Preference: General Model Blend Confidence: Above average A very flat but tightly packed height pattern exists across the CONUS, particularly the northern CONUS at the start of the forecast period, with a subtle shortwave crossing ND into MN today. This wave will cross the Great Lakes and help to build some broad troughing across the Northeast. The 00z ECMWF is a bit deeper than other guidance but with little significance weather-wise to support the general model blend at above average confidence. ...Deep Latitude Trof Entering the West by Monday; Southwest shortwave... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Preference: 1/3rds 12z GFS & 2/3rds 00z ECMWF Confidence: Average GOES-WV suite denotes a nice outflow to the shortwave across the Desert Southwest, this will move northward with similar evolution in the guidance suite. However, timing is going to be an issue but this is mostly affected by the timing of the upstream digging large scale trof entering N CA Monday. There has been some growing consensus with the UKMET/GFS slowing through Monday, but typical of bias, by Tuesday the split in timing increases. The UKMET is well outpaced over the NAM/GFS followed by the lagging ECMWF/CMC. This appears to be a traditional timing issue, so a typical GFS/ECMWF blend will likely work best with average confidence (given spread is moderate to large by Day 3). So employing a 1/3rd GFS and 2/3rds ECWMF solution works out well here. ...Gulf of Mexico TUTT and Surface Wave... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Preference: 12z NAM/GFS and 00z ECWMF blend Confidence: Slightly above average The TUTT cell moving west across the Gulf continues to look impressive with a solid trailing surface wave. The 12z NAM and ECMWF are very similar in depth and timing of both features, while the GFS is in a semi-biased weaker phase relative to both but timed well. The 00z UKMET is typical bias in tropical systems with strong convective response leading to a compact wave that feeds back on itself...slowing and shifting north of the remaining guidance. The 00z CMC evolves the surface wave well, but the TUTT is a bit too weak and slow, so the surface wave lags. So while not favored in a blend, it does provide confidence to the evolution of the system. So a 12z NAM/GFS and 00z ECMWF blend is preferred in the Gulf at slightly above average confidence. ...Tropical Storm Humberto... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Preference: Official NHC Forecast Best Proxy: 00z CMC through 48hrs; 12z GFS/NAM thereafter Confidence: Average Humberto is starting to have a stronger/deeper appearance at this time. There remains a solid split in the guidance with the ECMWF/UKMET continuing to be east and eventually south of the NHC track after the turn at 36 to 48hrs. The 12z GFS/NAM both start initially slower and slightly east of the 00z guidance but still are west and northwest over the next two days before turning east. The 00z CMC is closest to the track through 48hrs before accelerating after the eastward turn...moving away from the other guidance. It is at this point the GFS/NAM both move along the eastward track after 48hrs with better timing, perhaps a bit slow. Overall, the best proxy in a general model blend, but to not blend out the features too much; a 00z CMC through 48hrs shifting to GFS/NAM thereafter works closest to the NHC forecast. There is growing consensus as the guidance envelope tightens, as such confidence is average in utilizing this proxy for the 15z NHC forecast. Model trends at www.wpc.noaa.gov/html/model2.html 500 mb forecasts at www.wpc.noaa.gov/h5pref/h5pref.shtml Gallina