It's often said that the ECMWF model is more accurate than the GFS. But what are the reasons why? Also, how does the NAM compare with these two and other forecasting models?
$\begingroup$ That's two different questions: (1) are there significant differences between different weather models' accuracy, and (2) if yes, why?. Who claims that ECMWF is more accurate than GFS? $\endgroup$– gerrit ♦May 13, 2014 at 19:32
$\begingroup$ @gerrit I don't think there's need for two questions, but I agree, this question should be revised to include some of many examples where ECMWF demonstrated higher skill than GFS. $\endgroup$– milancurcicMay 13, 2014 at 22:38
5$\begingroup$ "It's often said" are weasel words. I think we should avoid them here as they try to at wikipedia. $\endgroup$– naught101Jul 2, 2014 at 10:40
- Spatial (horizontal and vertical) and temporal resolution
- Dynamical core
- Handling of some complex non-linear physical processes (i.e. snow)
- Topography representation
- How the mathematical equations are numerically implemented
- The computer used
These are some of the main differences between 2 given numerical weather model.
With that being said, the same model will give different results when run on a different computer (Hong et al. 2013).
Hong S-Y, Koo M-S, Jang J, et al. (2013) An Evaluation of the Software System Dependency of a Global Atmospheric Model. Mon Weather Rev 141:4165–4172. doi: 10.1175/MWR-D-12-00352.1
A difference of accuracy between weather models can be due to different factors, one of these is certainly the availability of "good" initialization data. As pointed out in Bradley Ballish (NOAA/NCEP) in his AMS presentation the accuracy of GFS has improved drastically when ECMWf initaliziation data have been used. Furthermore, I think there is a big difference in computational power between the two centers, ECMWF center has bigger and faster supercomputers.
$\begingroup$ how does a faster supercomputer improve the forecast of a weather model? $\endgroup$– gansubOct 21, 2017 at 14:32
1$\begingroup$ a faster supercomputer can allow a more detailed (in spatial resolution, temporal resolution) and, in principle, more realistic modelisation. $\endgroup$ Oct 22, 2017 at 12:37