ATL: MATTHEW - Models

[tarheelprogrammer]

NNE movement it appears at hour 138. It could be going OTS from here.

See here:
http://www.tropicaltidbits.com/analysis/models/?model=gfs&region=atl&pkg=z500_mslp&runtime=2016092812&fh=138&xpos=0&ypos=200

[toad strangler]

Amazing consistency from the GFS the past couple days aside from the minor walks W earlier.

[MississippiWx]

I still believe the GFS is shooting this out of the Caribbean way too fast. GFS is obsessed with an upper level trough forming over the Gulf. Euro does not have this. GFS at 500mb just seems bogus.

[chris_fit]

A smidge east @ 144 compared to the 06z GFS - probably because it's stronger

[gatorcane]

Even the CMC has more ridging and less of a Gulf trough than the GFS. By the way the 12Z CMC has shifted west and is well west of the GFS through 90 hours the run is out too so far.

[OntarioEggplant]

How is it wrong if there's empirical evidence suggesting otherwise?

Note he's still saying the initial position is wrong. A wrong initial position most certainly affects the future path as we are seeing.

You're misunderstanding what he was saying. He said research has shown that by moving the model initialized vortex to wherever the NHC analyzes it may cause a degradation in forecast accuracy. So much goes into initializing the 0 hour conditions in a global model that just haphazardly shifting a low level vortex around my cause unintended consequences during the run. Maybe you're better off having the initial error and letting the model run with what it initialized with.

I understand that perfectly. It is still an incorrect initialization and an incorrect initialization has ramifications through the forecast. Maybe less than forcing it as he's talking about it, but it is still incorrect.

[RL3AO]

Both models have had a rough few days. There is very little certainty beyond day 5. Anyone saying model x will bust or model y is right is doing so with gut feeling and no scientific evidence. Be careful.

[KWT]

Upper ridge is coming down, could be a risk to the NE states on this run.

Any slower from the GFS and this will shift the threat further south.

[Blown Away]

12z GFS... 144 Hours... Cat 2/3 Moving N Through Central Bahamas... Slightly E of 06z

[Dean4Storms]

Wow, GEM has gone much further West!

[ronyan]

Ridge building down at hr 156.

[gatorcane]

Wow, GEM has gone much further West!

Yes trending toward the Euro, heading NW at 102 hours:

[cajungal]

I think timing will be the key.

[BensonTCwatcher]

major difference in pattern over Eastern Canada this run. Opens a weakness - will be out to sea.

- After Bahamas

What? There is literally very little difference from the past two runs on that. If anything it's between the 00Z and 06Z in strength.

That low pressure moving off consolidated and broke down the ridge much more this run. Also the 990mb deep low over the Dakotas, and the lack of movement of the trough digging down to TX are all differences. It may still get trapped, but it has an open door if it does not slow too much

[Dean4Storms]

GEM had a better initialization spot than the GFS did.

[gatorcane]

Despite the initial west shift, the 12Z CMC still ends up east of South Florida in the same spot as the 00Z:

[KWT]

Can't see Matthew escaping on the 12z GFS, its going to have to bend back westward with the strength of that upper high, its classic east coast set-up in a way.

[p1nheadlarry]

About to get stuck again?

[Alyono]

Of course, the UKMET shifts west

NEW TROPICAL CYCLONE FORECAST TO DEVELOP AFTER 6 HOURS
FORECAST POSITION AT T+ 6 : 13.8N 61.6W

LEAD CENTRAL MAXIMUM WIND
VERIFYING TIME TIME POSITION PRESSURE (MB) SPEED (KNOTS)
-------------- ---- -------- ------------- -------------
0000UTC 29.09.2016 12 14.1N 63.2W 1006 39
1200UTC 29.09.2016 24 14.3N 65.9W 1001 49
0000UTC 30.09.2016 36 14.2N 68.6W 993 58
1200UTC 30.09.2016 48 13.4N 70.5W 987 63
0000UTC 01.10.2016 60 13.1N 71.8W 983 65
1200UTC 01.10.2016 72 12.8N 72.9W 986 59
0000UTC 02.10.2016 84 12.8N 73.6W 989 52
1200UTC 02.10.2016 96 13.6N 74.5W 987 52
0000UTC 03.10.2016 108 15.2N 75.3W 979 62
1200UTC 03.10.2016 120 17.2N 75.3W 970 68
0000UTC 04.10.2016 132 19.0N 75.4W 970 63
1200UTC 04.10.2016 144 20.6N 75.3W 982 59

[JPmia]

Sounds like we should ignore this GFS run until we can verify the initialization and compare the model vs best track.. then 18z would bring us back to accuracy?