Embed Size (px)
Citation preview
Slide 1
October 2011
Verification for polar regions Scores computed for polewards of 65°
NB proposed for CBS is polewards of 60°
Verification at ECMWF using available fields from other centres
Done for Z500 only
All verification against analysis (each centre against own analysis)
ERA-Interim scores shown as reference (ERA is fixed model and assimilation system)
Slide 1
Slide 2
ECMWF operational and ERA-Interim (1990-2011)
Z500 ACC=80%, 12-month moving average
N Pole: clear improvement in system around 2000, and consistently better than ERA beyond 2002. But the apparent change 2001-2002 and 2008-09 are matched in ERA
S Pole: clear sustained improvement in 1990s; still positive trend
ERA changes: either atmospheric variability or changes to observing system
October 2011 Slide 2
N Pole
S Pole
Slide 3
Comparison with other centres (2000-2011) N pole
Day 3 forecasts (T+72)
Z500, 12-month moving average
Each centre verified against own analysis
ERA-I shown for reference
October 2011 Slide 3
rms error
ACC
Slide 4
Comparison with other centres (2000-2011) S pole
Day 3 forecasts (T+72)
Z500, 12-month moving average
Each centre verified against own analysis
ERA-I shown for reference
October 2011 Slide 4
rms error
ACC
Slide 5
Comparison with other centres (2000-2011) N pole
Day 5 forecasts (T+120)
Z500, 12-month moving average
Each centre verified against own analysis
ERA-I shown for reference
October 2011 Slide 5
rms error
ACC
Slide 6
Comparison with other centres (2000-2011) S pole
Day 5 forecasts (T+120)
Z500, 12-month moving average
Each centre verified against own analysis
ERA-I shown for reference
NB some dates missing for CMC in 2009 – affects these scores for 2009 (other years OK)
October 2011 Slide 6
rms error
ACC
Slide 7
Comparison with other centres (2000-2011) N pole
Variability (activity) of forecast and analysis fields: standard deviation of anomalies
Day 5 forecasts (T+120)
Z500, 12-month moving average
ERA-I shown for reference
Compared to the analysis, Met Office forecast now rather underactive; CMC overactive (this can affect the rms errors)
NB some dates missing for CMC in 2009 – affects these scores for 2009 (other years OK)
October 2011 Slide 7
analysis
forecast
Slide 8
Comparison between analyses (N Pole)
Differences between the analyses of different centres
Z500 30 day moving average
Decrease over last decade in the difference between the analyses of different centres
October 2011 Slide 8
rms difference
mean difference
Slide 9
Comparison between analyses (S Pole)
Differences between the analyses of different centres
Z500 30 day moving average
Decrease over last decade in the difference between the analyses of different centres
October 2011 Slide 9
rms difference
mean difference
Slide 10
Comparison with other centres (JJA 2011) arctic
Mean scores over 3 months (June, July, August 2011)
Includes JMA
This for polewards of 60° (the new proposed CBS area)
All centres have slightly lower rmse and higher ACC for this larger area
In winter (DJF) it is opposite (error larger in the larger area, consistent with higher average activity towards mid-latitudes)
The definition of the area does not affect the relative performance of the different models
October 2011 Slide 10
Slide 11
Comparison with other centres (JJA 2011) N pole
Mean scores over 3 months (June, July, August 2011)
Includes JMA
This for polewards of 65° (for comparison with previous slide)
October 2011 Slide 11
Slide 12
Comparison with other centres (JJA 2011) antarctic
Mean scores over 3 months (June, July, August 2011)
Includes JMA
This for polewards of 60° (the new proposed CBS area)
October 2011 Slide 12
