*Photo by N. Mellerski
See: weather maps stats by cycles (00z,06z,12z,18z)
Contact: Fanglin Yang ph: 301-7638000 x7296
Global Climate and Weather Modeling Branch
MODEL PERFORMANCE STATISTICS
Click here to see the new VSDB-based verification page
|
*Photo by N. Mellerski |
________________________________ |
See: weather maps stats by cycles (00z,06z,12z,18z)
Contact: Fanglin Yang ph: 301-7638000 x7296 |
EXTRATROPICS - lats 20-80 N and S, 500-hPa height(except where indicated).
Time series, anomaly correlations** in the extratropics
Time series of
rms height error for forecast days 5 and 6
in the extratropics - many models - updated daily.
Time series of
rms and mean height error by region for forecast day 5
in the extratropics - many models - updated biweekly.
Time series of
850-hPa rms vector wind error and mean windspeed error
by region for forecast day 3
in the extratropics - many models - updated biweekly.
Time series of
200-hPa rms vector wind error and mean windspeed error
by region for forecast day 2
in the extratropics - many models - updated biweekly.
Time series of
area-mean height error for forecast days 5 and 6
in the extratropics - many models - updated daily.
Decay curves out to forecast day 9
of time-averaged 500-hPa height anomaly correlation vs forecast length for
various wave numbers, many models - updated twice a month.
Decay curves out to forecast day 9
of time-averaged 1000-hPa height anomaly correlation vs forecast length for
various wave numbers, many models - updated twice a month.
Decay curves
2004,
2005
2006
2007
2008
out to forecast day 9 of annual average
500-hPa height anomaly correlation vs forecast length for
various wave numbers, many models - updated annually
Decay curves
2004,
2005
2006
2007
2008
out to forecast day 9 of annual average
1000-hPa height anomaly correlation vs forecast length for
various wave numbers, many models - updated annually
NEW! Animation of decay curves:
N Hem and
S Hem
annual averages of GFS 500-hPa anomaly correlations back to 1989
Growth curves
of time-averaged 500-mb RMS height error(above)r;
mean height error and 1000-500 mb thickness error(below) vs forecast
length, for many models - updated twice a month.
RMS
departure of forecast from climatology
vs forecast length for many models (for 500-hPa hgt) - updated twice a month.
Scatter diagrams
- daily anomaly correlation for various forecast lengths and wave numbers
model vs model - updated twice a month.
Anom correl as a function of verif date and fcst length
500-mb height, two models, zonal waves 1-20, days 1-9.
See also...
Intercomparison of stats by cycle
(00z,06z,12z,18z)
TROPICAL WINDS - rms and mean errors, anomaly correlations - 200 hPa and 850 hPa.
Rms vector wind error -
time series, forecast day 3, many models, updated daily.
archive to November, 2000
Rms error of u-component -
time series, forecast day 3, many models, updated daily.
Rms error of v-component -
time series, forecast day 3, many models, updated daily.
Mean windspeed error -
time series, forecast day 3, many models, updated daily.
Anomaly correlations of u-component
- time series, forecast day 3, many models, updated daily.
Anomaly correlations of v-component
- time series, forecast day 3, many models, updated daily.
Anomaly correlations of wind vector
- time series, forecast day 3, many models, updated daily.
Anomaly correlations of windspeed
- time series, forecast day 3, many models, updated daily.
Growth curves
of time-averaged rms vector wind error vs forecast length for
many models - updated twice a month.
Growth curves
of time-averaged mean windspeed error vs forecast length for
many models - updated twice a month.
Decay curves out to forecast day 9
of time-averaged anomaly correlation vs forecast length for the
v-component of the wind in the tropics at two levels -
various wave numbers, many models - updated twice a month.
-------------------------------------------------------------
* Acknowledgments: This site was first designed by Peter Caplan. Forecast scores are computed by Yuejian Zhu
** Each model verified against its own analysis.
The anomaly correlation score measures the ability of a numerical
forecast to represent general weather patterns of high pressure
ridges and low pressure troughs, and the locations and strengths of
fronts and major storms. It has a maximum value of 1.0 (a perfect
depiction of the location of major weather systems), while a score above
0.6 is still considered useful by forecasters. It is universally accepted
as a standard score by international weather modeling centers and the World
Meteorological Organization.