NAM

[Joel Wright]

Does anyone know what's going on with the NAM? It seems for awhile now to be performing quite poorly at anything beyond about 42hrs.

Models this year seem to be having a difficult time. Which could partly be from the unusual pattern we've all been in.

I'm mainly picking on the NAM here because it used to perform very well, quite consistantly. Sure it had it's down times, but for the most part it seemed rock solid.

Anyone have any inside scoop? :wink:

 

[Jason McKittrick]

Since they installed some new physics in the model last spring the NAM has been having trouble with the precip and dewpoint forecasts. My observation this year is that the NAM can no longer accurately forecast precip beyond 12 hours, it seems to be horribly dry. The GFS has been far superior which I cannot believe I am saying because before this spring it was no where close to as skilled as the NAM.

 

[samuel stone]

Yeah, the nam has sucked lately, it appears to be screwing up just as badly on tommorows possible severe outbreak also, it has been all over the map on this system!

 

[Mike Gauldin]

Agreed, it pushed the low about 350 miles south over 6 hours, then back 450 miles to the northeast another 6 hours later.

This is why drugs are bad for forecast models.

 

[Joel Wright]

Originally posted by Jason McKittrick
The GFS has been far superior which I cannot believe I am saying because before this spring it was no where close to as skilled as the NAM.

It's nice to see the GFS become a little more accurate. Although it would have been nicer if the NAM hadn't have see-sawed down the other way.

Hopefully over time they'll resolve the physics problem within the model. I would assume the problem's already known by the model R&D. I'm hoping they can get it back on track soon. No pun intended. 8)

 

[mikegeukes]

NAM model changes 3 May 2005: What changed, what is forecast impact?
http://meted.ucar.edu/nwp/pcu2/NAMMay2005.htm

More information on this change bundle is in the NCEP/EMC Mesoscale Modeling Branch's 106-slide briefing.
http://www.emc.ncep.noaa.gov/mmb/mmbpll/Sp...v3_document.htm

NOAA NWS NCEP HPC INTERACTIVE MODEL BIAS PAGE
http://www.hpc.ncep.noaa.gov/mdlbias/

Mike

 

[Bill Tabor]

Personally I'm not impressed with what I've seen from the GFS last couple of years particularly more than several days out. It's burned me too many times.

As for NAM - I used to feel that the ETA was pretty accurate and dependable. They screwed it up by adding some stuff than increased sbcape a few years ago. Not sure what was changed between the ETA and NAM other than a name change so far - not convinced it is a better model yet.

Will be exciting to see what the WRF framework and model changes will bring to our model quality.

 

[Justin Turcotte]

The NAM has had a horrendous dry bias this winter. I don't even look at the QPF output much anymore. The NAM still does a fair job with wave progression, UA temps and wind, surface boundaries, and other goodies. When using the NAM for forecasts I often find myself using its level RH fields for POPs. The GFS and GEM have been very good for operational forecasts. Of course, the GUT model (instinct and pattern recognition) is still a viable option. NAM seemed to do a good job during the convective season but even then it often didn't capture the southern extent of MCSs like the GFS.

 

[Scott Olson]

"The model is much moister — dewpoints, precipitable water, and CAPE are higher and heavy precipitation is more widespread, reducing or eliminating a dry bias."

The NAM seemed to overforecast dewpoints consistently all the way up until mid May. IMO it has performed quite bad in the 84-48hr timeframe, even the GFS seems to sometimes have better contunity for a medium range forecast. Seems there has been several times where the NAM was not amplified enough on a trough coming in. Otherwise it's still been a useful model once an event draws near for good percipitation forecasts (it nailed the snowfall on our first blizzard pretty well.)

-Scott Olson.