overall statistics rmse wrf-ua: 159 w m -2 wrf-ucsd: 171 w m -2 stderr ‘bias-corrected rmse’
DESCRIPTION
Overall Statistics RMSE WRF-UA: 159 W m -2 WRF-UCSD: 171 W m -2 STDERR ‘Bias-corrected RMSE’ WRF-UA: 165 W m -2 WRF-UCSD: 164 W m -2 Ratios compared to NAM WRF-UA more accurate for 4/6 days Forecast horizon WRF-UCSD is more accurate for the first 6-7 hours of simulation time. - PowerPoint PPT PresentationTRANSCRIPT
RMSE Ratio to NAM STDERR Ratio to NAM
Date WRF-UA WRF WRF-UA WRF
8/10/2011 0.90 1.34 0.82 1.09
8/11/2011 0.92 0.61 0.92 0.498/12/2011 0.82 0.59 0.87 0.638/13/2011 0.36 0.51 0.34 0.538/14/2011 0.40 0.54 0.37 0.57
8/15/2011 0.70 0.79 0.66 0.76
Overall Statistics
RMSE- WRF-UA: 159 W m-2
- WRF-UCSD: 171 W m-
2
STDERR- ‘Bias-corrected RMSE’- WRF-UA: 165 W m-2
- WRF-UCSD: 164 W m-
2
Ratios compared to NAM- WRF-UA more
accurate for 4/6 days
Forecast horizon- WRF-UCSD is more
accurate for the first 6-7 hours of simulation time
8/10/2011- WRF-UCSD performed
much worse than WRF-UA- Not enough cloud
cover (afternoon)- Scale of clouds
- WRF-UA cloud scales are much smaller, despite WRF-UA having slightly coarser resolution (1.8 km vs. 1.3 km)
WRF-UA WRF-UCSD
1500 UTC
1900 UTC
2200 UTC
1700 UTC
2000 UTC
2200 UTC
WRF-UA WRF-UCSD8/12/2011- Largest improvement
over WRF-UA- Cloud field burnoff
- Timing captured by WRF-UCSD
- Too many afternoon clouds in WRF-UA
- WRF-UA has two distinct types of clouds from the model
Clouds generated by PBL scheme? Shallow convection?