How the Models Did with with Dennis.......preliminary...

[Derecho]

Caveats...this is done with Stomtrakker; may have major differences with NHC verification due to a variety of reasons. None of this has been "quality controlled."

It wasn't a really long-lasting storm so i'll just post 72 hour performance by the major well known models....lowest to highest error:


1) CONU Consensus Model: 54 mile error at 72 hours

2) NHC Official forecast: 68 miles

3) GUNA Consensus Model: 80 miles

4) NOGAPS: 98 miles

5) UKMET: 108 miles

6) GFS: 111 miles

7) GFDL: 116 miles

8) BAMD: 139 miles

9) NAM: 189 miles

10) Canadian (CMC): 263 miles


Som really egregiously bad forecasts and flopping around with the bottom two up there; Canadian only barely edged out "Extrapolated Motion" and did WORSE than CLIPER...

CLIPER is a purely statistically based history model, something like 40 years old, and is used as a baseline comparison by NHC to designate "skill".

[LCfromFL]

Interesting post, Derecho, thanks!

[birdwomn]

down thru the GDFL, it looks like the did pretty well to me. Unfortunately, there will be much complaining because of the flooding east of the storm and the fact that people who didn't get hit very badly evacuated.

Basically, I think people want an exact answer on where and how bad the damage will be. That information is almost impossible to give because there are so many variables in hurricanes: wind, rain, storm surge, tides, coastal topography, etc.

Ok, I will stop...I am "preaching to the choir here"