The AGM has two options for estimating sub-grid scale variance. The first method uses
rawinsonde uncertainties (300 km X 300 km area) as a function of altitude as sampling scale
variance in the gridded area. The second method attempts to better represent the sampling
scale turbulence on the ECMWF grid scale and to provide "realistic" variability in the winds
based upon the mean wind field. The model represents the uncertainties by scaling the mean
ECMWF wind speed by 20%. Using one day of the ECMWF data, SWA computed the global
distribution of variance for 12 atmospheric pressure levels. Comparisons of the total
global averages with the NMC OI rawinsonde profiles suggest that the simulated
variances are not unreasonable.
Comparison of the NMC rawinsonde variances and the
global average variances obtained by the LSM using
a 24-hour sample of ECMWF data.
| Pressure (mb)
| NMC Old
| NMC New
| AGM Global Average
|
| 1000
| 1.8
| 1.4
| 1.44
|
| 850
| 1.8
| 2.2
| 1.65
|
| 700
| 2.4
| 2.4
| 1.81
|
| 500
| 3.8
| 2.8
| 2.42
|
| 400
| 4.7
| 3.4
| 3.00
|
| 300
| 5.9
| 3.4
| 3.61
|
| 250
| 5.9
| 3.2
| 3.87
|
| 200
| 5.9
| 3.0
| 4.00
|
| 150
| 5.5
| 2.7
| 3.83
|
| 100
| 4.9
| 2.5
| 3.12
|
| 70
| 4.9
| 2.5
| 2.80
|
| 50
| 3.9
| 2.7
| 2.80
|