Why is the 06z GFS performing so much worse than other runs?

[Edge Weather]

The 06z GFS is performing MUCH worse than any other model run over the past 31 days.  Usually it is close to the other runs in terms of accuracy.  Does anyone know why this is happening currently?

 

The difference is most notable in the 6 day 500 mb chart posted below.  Look at the much lower verification score than the other runs.

 

[dtk]

The 06z GFS is performing MUCH worse than any other model run over the past 31 days.  Usually it is close to the other runs in terms of accuracy.  Does anyone know why this is happening currently?

 

The difference is most notable in the 6 day 500 mb chart posted below.  Look at the much lower verification score than the other runs.

See thread here:

http://www.americanwx.com/bb/index.php/topic/32149-gfs-12z-and-0z-runs-compared-to-6z-and-18z-runs/

 

Or presentation here:

http://www.emc.ncep.noaa.gov/gmb/wx24fy/doc/GFS4cycle_GCWMB_briefing_13dec2012_fyang.pdf

 

In summary:

1.  The 6z cycle is found to be slightly "less skillful" for the same lead time as the other cycles, but the differences are likely not statistically significant.

2.  To answer your direct question, a 30-day sample is not nearly long enough to prescribe statistically significant differences for a 6 day lead time (predictability isn't large enough).

3.  For short-to-medium length lead times, the 6z cycle will be better than or equal in skill the 0z cycle valid at the same time (i.e. the 114h forecast from 6z compared to the 120h forecast from 0z).

 

Lastly, this particular sample seems to really suffer from the 18-28th of October.  There was a bunch of data-loss issues back in October (some of it documented, some of it creeping in earlier than advertised).  The 6z cycle is the one most likely to be impacted by such data loss and this probably warrants further investigation....