NOTE: 12-h forecasts from RUC initialized at 0000 or 1200 UTC are forced to use 24-h old boundary conditions because they are run before new runs of the Eta. Therefore, the skill shown in these 12-h forecasts for RUC is somewhat poorer than that which would be obtained by delaying the 0000 or 1200 UTC RUC run until after the new Eta, so that they had the same boundary conditions. The 12-h RUC forecasts run at non-0000/1200 times have access to the new boundary conditions, and have higher skill. Thus, in a sense, these 12-h statistics show RUC at its worst.
Thick lines are for the 40-km MAPS. Thin lines are for the 60-km MAPS.
28 May - 15 Sep 1997 // 21 March - 16 April 1997 // Temperature (standard deviation) and wind (RMS vector difference) forecasts for difference from rawinsonde obs
28 May - 15 Sep 1997 // 22 March - 16 April 1997 // Height and relative humidity forecasts - standard deviation of difference from rawinsonde obs
28 May - 15 Sep 1997 // 22 March - 16 April 1997 // Temp / RH mean forecast errors
The 60-km MAPS runs at FSL using the same software as the 60-km RUC running at NCEP, but with the FSL data sets being ingested. This gives a closer control for comparison with the 40-km MAPS, which also uses the FSL data set. There are many missing dates for the 60-km MAPS runs, so these statistics do not match those for the Eta model, even though they cover the same time period.
28 May - 15 Sep 1997 // 28 March - 22 April 1997 // Temperature (standard deviation) and wind (RMS vector difference) forecasts for difference from rawinsonde obs
28 May - 15 Sep 1997 // 28 March - 22 April 1997 // Height and relative humidity forecasts - standard deviation of difference from rawinsonde obs
28 May - 15 Sep 1997 // 28 March - 22 April 1997 // Temp / RH mean forecast errors
NOTE: 12-h forecasts from MAPS initialized at 0000 or 1200 UTC are forced to use 24-h old boundary conditions because they are run before new runs of the Eta. Therefore, the skill shown in these 12-h forecasts for MAPS is somewhat poorer than that which would be obtained by delaying the 0000 or 1200 UTC MAPS/RUC run until after the new Eta, so that they had the same boundary conditions. The 12-h MAPS/RUC forecasts run at non-0000/1200 times have access to the new boundary conditions, and have higher skill. Thus, in a sense, these 12-h statistics show MAPS/RUC at its worst. Also, skill level should increase when the 40-km RUC is run over at NCEP, since its data sets and reliability are improved over those at FSL.
Temperature (standard deviation) and wind (RMS vector difference) analysis fit to rawinsonde obs
Height and relative humidity analyses - standard deviation of difference from rawinsonde obs
All METARs in 60km RUC domain - 97097-98. No quality control on
surface obs, so difference values are exaggerated for all analyses/models.
0h40 3h40 6h40 0h60 1h60
temp s.d. 1.6 2.2 2.3 3.0 3.6
temp mean -0.7 -1.3 -1.6 1.5 2.0
wind rmsve 4.4 4.7 4.6 5.3 5.5
wind spd bias 0.2 -0.1 -0.1 0.4 0.6
(97100)
3hdp s.d. 0.7 1.0 1.1 0.7 0.8
3hdp bias 0.0 -0.3 0.2 -0.1 0.2
METARs west of 105 W
temp s.d. 2.1 2.7 2.8 3.4 3.6
temp mean -0.9 -2.0 -2.6 4.8 4.6
wind rmsve 3.2 3.6 3.8 3.0 3.5
wind spd bias 0.2 -0.2 -0.2 0.6 0.6
Last updated: 16 Apr 97