Model Diagnostic Discussion
NWS Weather Prediction Center College Park MD
200 PM EST Mon Dec 31 2018

Valid Dec 31/1200 UTC thru Jan 04/0000 UTC

...See NOUS42 KWNO (ADMNFD) for the status of the upper air
ingest...

12Z Model Evaluation...with Preferences and Confidence
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

...Overall pattern across the CONUS...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Preference: 12z NAM/GFS/ECMWF blend
Confidence: Average

19z update: The 12z CMC and UKMET remain on the faster side of the
spread with respect to the southern stream moving across the
southern Plains on Thursday. However, it is slower than its
previous run and now the overall spread has lessened with the rest
of the 12z guidance now in. The ECMWF remained consistent compared
to its previous run and so while still a bit faster than the GFS,
again the spread has lowered. With this in mind, the preferred
blend for this cycle will remain in place (12z NAM/GFS/ECMWF
blend).

---previous discussion---

The two main weather systems over the CONUS through the forecast
period lie across the central US, with a northern stream shortwave
expected to race across the northern Plains and Upper Midwest
Tuesday Night into Wednesday, while a closed low over the
southwest US slowly ejects out into the southern Plains as it
opens up. The phasing (or lack of) and subsequent speed remains
the model diagnostic challenge this cycle.

The northern stream energy appears to be well handled by the
deterministic models (00z and early 12z look). The GFS is a bit
more amplified compared to the UKMET/CMC/ECMWF solutions but its
axis seems to match up well with the general consensus.

The southern stream closed low has more problems, though perhaps
better agreement compared to 24 hours ago. The 00z CMC remains too
fast and is on the edges of the ensemble spread from 02.00z
onward. It ejects the wave way too fast across Texas, though by
the end of the forecast period (04.00z) its displacement is not as
high as it slows down and the other models catch up. The 12z GFS
is on the slower side of the deterministic guidance, but within
the ensemble spread. The 12z run has trended faster compared to
the 06z version. For the ECMWF, it is actually faster than GFS,
but its dprog/dt shows a slowing trend and therefore better
agreement overall. The 12z NAM seems reasonable as a compromise
through 03.12z but then the upper low drifts further north than
the rest of the other models. Finally, the UKMET is on the faster
solution but not to the degree that the CMC is. Surface low track
variability increases substantially right at the end of the
period, with most of the deterministic and ensemble clusters
showing a range from TX to AL/TN.

So overall, a blended approach of the GFS/NAM/ECMWF (completely
discarding the CMC) probably will wash out some of the slow/fast
biases seen in the models and yield a reasonable solution for the
mass fields at this time. 


Model trends at www.wpc.ncep.noaa.gov/html/model2.shtml
500 mb forecasts at www.wpc.ncep.noaa.gov/h5pref/h5pref.shtml

Taylor