Differences between GFS and NAM

[Michael Wurzburger]

Here's hoping for a couple of great days ahead! With that being said, I have a couple of questions regarding the GFS and NAM models. Is there ever truly a time in which these models agree with each other? I realize that they will never be equal, but lately it seems as though, after each run, the analysis of one doesn't come close the other. Is the same set of data used for each model run? If that is the case, what are the parameters that would cause such a difference? Thanks for the help.

Michael

 

[MClarkson]

Are you checking the time stamps to make sure the time is equal? Over short time frames the GFS and NAM will usually be pretty close to eachother.

 

[Michael Wurzburger]

Definately. I'm always making sure the the Z runs match.

 

[Chad Cowan]

Here's hoping for a couple of great days ahead! With that being said, I have a couple of questions regarding the GFS and NAM models. Is there ever truly a time in which these models agree with each other? I realize that they will never be equal, but lately it seems as though, after each run, the analysis of one doesn't come close the other. Is the same set of data used for each model run? If that is the case, what are the parameters that would cause such a difference? Thanks for the help.

Michael

The reason that the GFS and NAM charts you're looking at for tomorrow and Friday don't seem to match up well is because of their differences in handling the precipitation. I'm assuming the largest difference that you're seeing (and one that had been present for the past few days of model runs) is that the GFS has the warm front (~ > 70F sfc temps) displaced much further north than the NAM has been showing. The models' solutions of the early-mid day elevated convection has a HUGE impact on what happens at the surface throughout the day on the subsequent surface charts (temp, dewpoint, CAPE, CINH, etc) and the fact that this elevated convection is quite difficult to forecast with much certainty adds exponentially more uncertainty to forecasting for surface based convection later in the day.

In a nutshell, keep the model's various precipitation forecasts in mind and on the day-of, whichever one of those is most accurate at the time, you should put quite a bit of credence in that model's solution moving on.

EDIT: I think I should add that there can be pretty large timing differences between the NAM and the GFS also, and these differences cause more forecast variability at the surface. Typically the GFS is thought of as being too progressive (fast moving) with systems, as of the past 6 months or so, though, I think the GFS has been doing a remarkable job with timing and amplitude.

 

[Scott Weberpal]

The NAM has a resolution of 12km while the GFS (to 180 hours) has a resolution of (I believe) 35km - which makes a huge difference in how it handles convective precip.

I almost never use the GFS inside of 84 hours.

 

[rdale]

If you want to gain any value out of using the models, it's just about mandatory that you understand how a model works. You have to sit through the level 1 intro to modeling at http://www.meted.ucar.edu/topics_nwp.php - and if you really want to understand, go through level 2 as well.