mm5 assessment nelson seaman 10 october 2002
DESCRIPTION
MM5 Assessment Nelson Seaman 10 October 2002. The BRAVO Study: August-October 1999. MM5 Configuration:. Domains: Triply nested grids: Continental:Dx = 36 km Regional:Dx = 12 km High-resolution:Dx = 4 km Vertical resolution: 35 layers Lowest layer at 18 m AGL. MM5 domain locations. - PowerPoint PPT PresentationTRANSCRIPT
MM5 AssessmentMM5 Assessment
Nelson SeamanNelson Seaman10 October 200210 October 2002
The BRAVO Study:The BRAVO Study:
August-October 1999August-October 1999
MM5 Configuration:
Domains:Domains:• Triply nested grids:Triply nested grids:
• Continental:Continental: Dx = 36 kmDx = 36 km• Regional:Regional: Dx = 12 kmDx = 12 km• High-resolution:High-resolution: Dx = 4 kmDx = 4 km
• Vertical resolution:Vertical resolution:
• 35 layers35 layers• Lowest layer at 18 m AGLLowest layer at 18 m AGL
MM5 domain locations
12-km
4 km
MM5 Configuration:MM5 Configuration:
MM5v3.4 Physics for BRAVO:MM5v3.4 Physics for BRAVO:
• Kain-Fritsch deep convectionKain-Fritsch deep convection• Dudhia “simple-ice” microphysicsDudhia “simple-ice” microphysics• RRTM long-wave radiationRRTM long-wave radiation• Gayno-Seaman 1.5-order TKE PBLGayno-Seaman 1.5-order TKE PBL
MM5 Configuration:
Four-Dimensional Data AssimilationFour-Dimensional Data Assimilation
• Analysis Nudging: Assimilate analysesAnalysis Nudging: Assimilate analyses on 36-km and 12-km domains.on 36-km and 12-km domains.
• Observation Nudging: Assimilate BRAVO and Observation Nudging: Assimilate BRAVO and AIRS obs on 12-km and 4-km domains.AIRS obs on 12-km and 4-km domains.
• Surface Flux Nudging: For sensible and Surface Flux Nudging: For sensible and latent heat fluxes (Alapaty et al. 2001).latent heat fluxes (Alapaty et al. 2001).
MM5 Run Periods:
36-km and 12-km domains run for 36-km and 12-km domains run for the entire BRAVO period from the entire BRAVO period from August-October 1999August-October 1999
4-km domain run for 2 intensive 4-km domain run for 2 intensive modeling periods:modeling periods:
• 7-17 October 19997-17 October 1999• 15-25 August 199915-25 August 1999
Meteorological InputsWind Flow on October 7 at 3 p.m.
Surface Layer:18 m AGL
NCEP Surface Analysis
Emissions of Sea Salt Sodium
CMAQ
Surface-layer wind speed averaged on 4-km domain:Green: MM5 Asterisks: Observed
Boundary-layer wind speed averaged on 4-km domain:Red: MM5 Asterisks: Observed
Surface-layer wind direction averaged on 4-km domain:Green: MM5 Asterisks: Observed
Boundary-layer wind direction averaged on 4-km domain:Red: MM5 Asterisks: Observed
Surface-layer temperature averaged on 4-km domain:Green: MM5 Asterisks: Observed
Surface-layer mixing ratio averaged on 4-km domain:Green: MM5 Asterisks: Observed
Surface-layer wind speed averaged on 12-km domain:Green: MM5 Asterisks: Observed
Boundary-layer wind speed averaged on 12-km domain:Red: MM5 Asterisks: Observed
Surface-layer wind direction averaged on 12-km domain:Green: MM5 Asterisks: Observed
Boundary-layer wind direction averaged on 12-km domain:Red: MM5 Asterisks: Observed
Surface-layer temperature averaged on 12-km domain:Green: MM5 Asterisks: Observed
Surface-layer mixing ratio averaged on 12-km domain:Green: MM5 Asterisks: Observed
Statistical Summary of MM5 4-km Statistical Summary of MM5 4-km BRAVO Simulations: SurfaceBRAVO Simulations: Surface
Dates Wind Spd.(m/s)
Wind Dir.(degrees)
Temp.(C)
Mix. Ratio(g/kg)
7-10 Oct Rmse=1.61Bias= -0.25
MAE= 37.0Bias= +3.4
MAE= 2.28Bias=-1.62
MAE= 1.75Bias=+1.51
11-16 Oct Rmse=1.34Bias= -0.26
MAE= 54.6Bias= +3.1
MAE= 1.97Bias=-1.52
MAE= 1.18Bias=+0.52
15-20 Aug Rmse=1.37Bias= -0.33
MAE= 34.1Bias= -6.8
MAE= 2.51Bias=-2.29
MAE= 2.03Bias=+1.73
21-25 Aug Rmse=1.82Bias=+0.25
MAE= 41.1Bias= +4.0
MAE= 2.09Bias=-1.11
MAE= 1.61Bias=+0.99
Average: Rmse=1.54Bias= -0.15
MAE= 41.7Bias= +0.9
MAE= 2.21Bias=-1.64
MAE= 1.64Bias=+1.19
AVG
Benchmark: RMSE<2.0 m/s MAE<30.0 deg MAE<2.0 deg MAE<2.0 g/kg(by Environ) Bias < 0.5 m/s Bias <10.0 deg Bias<0.5 deg Bias<1.0 g/kg
Statistical Summary of MM5 4-km Statistical Summary of MM5 4-km BRAVO Simulations: 80-1500 mBRAVO Simulations: 80-1500 m
Dates Wind Spd.(m/s)
Wind Dir.(degrees)
Temp.(C)
Mix. Ratio(g/kg)
7-10 Oct Rmse=1.72Bias= -0.83
MAE= 15.1Bias= -2.1
11-16 Oct Rmse=1.56Bias= -0.93
MAE= 22.5Bias= -0.5
15-20 Aug Rmse=1.42Bias= -0.34
MAE= 21.7Bias= -7.0
21-25 Aug Rmse=1.50Bias=-0.29
MAE= 20.6Bias= -3.2
Average: Rmse=1.55Bias= -0.60
MAE= 20.0Bias= -3.2
AVG
Statistical Summary MM5 12-km Statistical Summary MM5 12-km BRAVO Simulations: 40 mBRAVO Simulations: 40 m
Dates Wind Spd.(m/s)
Wind Dir.(degrees)
Temp.(C)
Mix. Ratio(g/kg)
7-10 Oct Rmse=1.52Bias=+0.42
MAE= 38.2Bias= +0.4
MAE= 1.26Bias= -0.48
MAE= 1.08Bias=+0.24
11-16 Oct Rmse=1.25Bias=+0.26
MAE= 30.7Bias= -1.1
MAE= 1.36Bias= -0.64
MAE= 1.01Bias=+0.14
15-20 Aug Rmse=1.30Bias=+0.11
MAE= 33.0Bias= -7.0
MAE= 1.37Bias= -0.59
MAE= 1.69Bias=+0.65
21-25 Aug Rmse=1.47Bias=+0.21
MAE= 24.5Bias= +0.3
MAE= 1.43Bias=-0.65
MAE= 1.66Bias=+0.31
Average: Rmse=1.39Bias=+0.25
MAE= 31.6Bias= -1.9
MAE= 1.36Bias=-0.59
MAE= 1.36Bias=+0.34
AVG
Benchmark: RMSE<2.0 m/s MAE<30.0 deg MAE<2.0 deg MAE<2.0 g/kg(by Environ) Bias < 0.5 m/s Bias <10.0 deg Bias<0.5 deg Bias<1.0 g/kg
Statistical Summary MM5 12-km Statistical Summary MM5 12-km BRAVO Simulations: 80-1500 mBRAVO Simulations: 80-1500 mDates Wind Spd.
(m/s)Wind Dir.(degrees)
Temp.(C)
Mix. Ratio(g/kg)
7-10 Oct Rmse=1.60Bias= -0.16
MAE= 13.3Bias= -1.5
11-16 Oct Rmse=1.37Bias= -0.33
MAE= 14.8Bias= -0.9
15-20 Aug Rmse=1.36Bias= -0.27
MAE= 17.2Bias= -0.9
21-25 Aug Rmse=1.48Bias= -0.22
MAE= 17.3Bias= -1.3
Average: Rmse=1.45Bias= -0.25
MAE= 15.7Bias= -1.2
AVG
Conclusions:
MM5 has reproduced the observed wind MM5 has reproduced the observed wind speeds quite well, compared to speeds quite well, compared to benchmarks. RMSEs and biases are benchmarks. RMSEs and biases are low, both in the surface layer and low, both in the surface layer and planetary boundary layer.planetary boundary layer.
• BRAVO BRAVO transport ratestransport rates should be should be represented well.represented well.
Conclusions:
MM5’s wind directions have fairly large MAE in the MM5’s wind directions have fairly large MAE in the surface layer of the 4-km domain, but are quite surface layer of the 4-km domain, but are quite good MAE in the PBL.good MAE in the PBL.
Wind direction MAEs are significantly more Wind direction MAEs are significantly more accurate on the 12-km domain.accurate on the 12-km domain.
Wind biases are very low on both domains.Wind biases are very low on both domains.
• BRAVO BRAVO transport directionstransport directions should be should be represented well.represented well.
Conclusions:
MM5’s surface temperatures have very low MM5’s surface temperatures have very low MAE and average bias on the 12-km domain, MAE and average bias on the 12-km domain, where the Alapaty et al. surface-flux nudging is where the Alapaty et al. surface-flux nudging is used.used.
Surface temperatures on the MM5 4-km Surface temperatures on the MM5 4-km domain have MAEs and biases that are larger domain have MAEs and biases that are larger than desirable.than desirable.
• BRAVO BRAVO thermal processes and PBL depthsthermal processes and PBL depths should should be represented well, especially on the 12-km domain.be represented well, especially on the 12-km domain.
Conclusions:
MM5’s surface mixing ratios have very low MM5’s surface mixing ratios have very low MAEs and biases on the 12-km domain, where MAEs and biases on the 12-km domain, where the Alapaty et al. surface-flux nudging is used.the Alapaty et al. surface-flux nudging is used.
Surface mixing ratios on the MM5 4-km Surface mixing ratios on the MM5 4-km domain have low MAEs, but biases are larger domain have low MAEs, but biases are larger than desirable.than desirable.
• BRAVO BRAVO boundary-layer clouds and convectionboundary-layer clouds and convection should be represented well, especially on the 12-km should be represented well, especially on the 12-km domain. However, the default soil moisture is too domain. However, the default soil moisture is too high.high.