the mm5 prognostic meteorological model
DESCRIPTION
The MM5 Prognostic Meteorological Model. Define the physics of the domain properly and the meteorology fields will be defined properly. Current CCOS Episodes. July 09-13, 1999 July 31, -August 02, 2000. Meteorology Field Evaluations. Objective Approaches: - PowerPoint PPT PresentationTRANSCRIPT
The MM5 PrognosticMeteorological Model
Define the physics of the domain properlyand the meteorology fields will be defined properly.
Current CCOS Episodes
July 09-13, 1999
July 31, -August 02, 2000
Meteorology Field Evaluations
Objective Approaches:-- statistical evaluation simulated and observed
meteorological parameter values-- statistical evaluation of observed and simulated
air quality parameter values
Subjective Approaches:-- spatial comparison-- conceptual review
1
3
33i
sitesi
obs
i
obs
i
sim
OOOMNB
Air Quality Model Performance
Ozone Model : Ozone Concentration = 85 ppb
Mean Normalized Bias +/- 15 %
Alternative Wind Fields
CALMET (objective/prognostic hybrid)
MM5 Prognostic without Obs. FDDA1/
MM5 Prognostic with Obs. FDDA1/
1/ numerious itternations
Jul 31 Aug 01 Aug 02 UPkR NB UPkR NB UPKR NB ppb % ppb % ppb %---------------------------------------------------CMHb ModelSF Bay Area 0.97 +06 1.14 -04 1.16 -41Sacramento 1.35 +09 0.99 00 0.99 -10Southern SJV 1.10 -02 1.03 -10 0.88 -09
MM5_N1 Model (w/ FDDA)SF Bay Area 0.88 +01 1.05 -11 1.04 -37Sacramento 1.22 +02 1.00 -10 0.90 -18Southern SJV 0.95 -07 0.88 -17 0.73 -19
MM5_N2 Model (wo/ FDDA)SF Bay Area 0.98 +03 1.11 -23 1.16 -14Sacramento 1.32 +08 0.93 -18 0.96 -03Southern SJV 1.06 -03 1.03 -11 0.78 -19---------------------------------------------------
Ozone Model Performance (USEPA, 1991)* for the CCOS July/August, 2000 Episode Using CAMx/SAPRC99f
UPkR -- Unpaired Peak Ratio NB -- Paired Mean Normalized Bias
* simulations meeting USEPA model performance guidelines are highlightedCalifornia Air Resources Board/PTSD April, 2005
“There are 3 kinds of lies:lies,damn lies,and statistics”
(attrib: Benjamin Disraeli)
Statistical Analysis
500
500
500
500
500
500 500
500
500
500
500500
500
500
10001000
1000
2000
2000
2000
3000
Observed surface winds for July 31, 2000 at 0200 PDT. Windvectors are shown as 1-hour wind run.
500
500
500
500
500
500 500
500
500
500
500500
500
500
10001000
1000
2000
2000
2000
3000
Observed surface winds for August 01, 2000 at 0200 PDT. Windvectors are shown as 1-hour wind run.
500
500
500
500
500
500 500
500
500
500
500500
500
500
10001000
1000
2000
2000
2000
3000
Observed surface winds for August 02, 2000 at 0200 PDT. Windvectors are shown as 1-hour wind run.
500
500
500
500
500
500 500
500
500
500
500500
500
500
10001000
1000
2000
2000
2000
3000
Simulated surface winds for August 02, 2000 at 0200 PDT using the MM5 model without observational FDDA (MM5_N2)
500
500
500
500
500
500 500
500
500
500
500500
500
500
10001000
1000
2000
2000
2000
3000
Simulated surface winds for August 02, 2000 at 0200 PDT using the MM5 model with observational FDDA (MM5_N1)
USEPA. 2005. “Guidance on the Use of Models and Other Analyses in Attainment Demonstrations for the 8-hour OzoneNAAQS” Draft Final. USEPA. February, 2005.
“…if used improperly, FDDA can significantly degrade overallmodel performance and introduce computational artifacts.Inappropriately strong nudging coefficients can distort the magnitude of the physical terms in the underlying … equationsand result in ‘patchwork’ meteorological fields with stronggradients between near-site grid cells and the remainder of thegrid.”
Guidance on Use of Data Assimilation
1500
1500
1500 1500
1500
1500
1500
1500
1500
1500
1500
15001500
1500
1500
1500
1500
1500
2000
2000 2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000 2000
2000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
Simulated Mixing Heights (m) for August 01, 2000 at 1700 PDTusing the MM5 model without FDDA (MM5_N2)
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
15001500
1500
1500
1500
1500
1500 1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000 2000
2000
2000
2000
2000
2000
20002000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
3000
3000
3000
3000
3000
3000
3000
3000
3000
30003000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
Simulated Mixing Heights (m) for August 01, 2000 at 1700 PDT using the MM5 model with observational FDDA (MM5_N1)
1500
1500
1500
15001500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
20002000
2000
2000
2000
2000
2000
2000
2000
2000
2000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
Simulated Mixing Heights (m) for August 02, 2000 at 1700 PDT using the MM5 model with observational FDDA (MM5_N1)
1500
1500
1500
1500
1500 15
00
1500
1500
1500
1500
1500
1500
15001500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000 20
00
2000
2000
2000
2000
2000
2000
20002000
20002000
2000
2000
2000
2000
2000
2000
2000
2000
20002000
2000
2000
2000
3000
3000
3000
3000
Simulated Mixing Heights (m) for July 11, 1999 at 1700 PDT using the MM5 model with observational FDDA (MM5_N1)
1500
1500
1500
1500
1500
1500
1500
1500
1500 1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
20002000
2000
2000
2000
2000
2000
3000
3000
3000
30003000
Simulated Mixing Heights (m) for July 12, 1999 at 1700 PDT using the MM5 model with observational FDDA (F02)
ABL Height Comparisons
(Colored contours are TKE, and dots indicate the observed ABL height)
Simulated 500-m winds for August 01, 2000 at 0600 PDT using the MM5 model with without FDDA (MM5_N2)
Simulated 500-m winds for August 01, 2000 at 0600 PDT using the MM5 model with observational FDDA (MM5_N1)
500
500
500
500
500
500
500
500500
500
1000
1000
1000
2000
2000
2000
2000
2000
Simulated surface winds for August 01, 2000 at 1400 PDT using the MM5 model with without FDDA (MM5_N2)
500
500
500
500
500
500
500
500500
500
1000
1000
1000
2000
2000
2000
2000
2000
Simulated surface winds for August 01, 2000 at 1400 PDT using the MM5 model with observational FDDA (MM5_N1)
500
500
500
500
500
500
500
1000
Simulated surface winds for July 31, 2000 at 1700 PDT using the MM5 model with without FDDA (MM5_N2)
500
500
500
500
500
500
500
1000
Simulated surface winds for July 31, 2000 at 1700 PDT using the MM5 model with observational FDDA (MM5_N1)
CAMx/MM5 (w/FDDA)/SAPRC99 July 31, 2000
Simulated Surface winds for and ozone concentrations for July 31, at 1300 PDTusing the MM5 model with observational FDDA (B01) and CAMx/SAPRC99
Simulated Surface winds for and ozone concentrations for July 31, at 1400 PDTusing the MM5 model with observational FDDA (B01) and CAMx/SAPRC99
Simulated Surface winds for and ozone concentrations for July 31, at 1500 PDTusing the MM5 model with observational FDDA (B01) and CAMx/SAPRC99
Simulated Surface winds for and ozone concentrations for July 31, at 1600 PDTusing the MM5 model with observational FDDA (B01) and CAMx/SAPRC99
Simulated Surface winds for and ozone concentrations for July 31, at 1700 PDTusing the MM5 model with observational FDDA (B01) and CAMx/SAPRC99
July, 1990 Episode
500
500
500
500
500
500
500
1000
1000
2000
Simulated surface winds for July 9, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
500
500
500
500
500
500
500
1000
1000
2000
Simulated surface winds for July 10, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
500
500
500
500
500
500
500
1000
1000
2000
Simulated surface winds for July 11, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
500
500
500
500
500
500
500
1000
1000
2000
Simulated 500-m winds for July 10, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
500
500
500
500
500
500
500
1000
1000
2000
Simulated 500-m winds for July 11, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
500
500
500
500
500
500
500
1000
1000
2000
Simulated 500-m winds for July 12, 1999 at 1100 PDT using the MM5 model with observational FDDA (F02)
Jul 10 Jul 11 Jul 12 Jul 13 UPkR NB UPkR NB UPKR NB UPKR NB -na- % -na- % -na- % -na- %--------------------------------------------------------------
CMHb ModelSF Bay Area 1.11 +07 1.09 -10 0.92 -08 1.19 -24Sacramento 1.04 -03 1.07 -12 1.20 -04 1.08 -07Central SJV 1.09 -14 0.91 -10 1.05 +01 1.03 +09Southern SJV 0.92 -17 0.88 -25 1.39 -02 1.29 +10
F02 Model (w/ FDDA)SF Bay Area 0.91 -14 1.04 -07 0.99 -02 -- --Sacramento 0.74 -25 0.93 -18 1.13 -09 -- --Central SJV 0.81 -20 0.78 -26 0.85 -21 -- --Southern SJV 0.72 -33 0.78 -29 1.28 -13 -- --
--------------------------------------------------------------
Ozone Model Performance (USEPA, 1991)* for the CCOS July, 1999 Episode Using CAMx/SAPRC99f
UPkR -- Unpaired Peak Ratio NB -- Paired Mean Normalized Bias
* simulations meeting USEPA model performance guidelines are highlightedCalifornia Air Resources Board/PTSD April, 2005
Hypothesis
Meteorological fields generated using MM5 will tend to be more diffusive with lower pollutant concentrations spread over largerareas.
Inert Tracer Analysis
Arbitrary Grid Cell in the Delta: ~ Pittsburg ~ San Francisco
Daily Inert Surface-Level Emissions: 0600-0800 PDT
Concentration Intervals: ~ * 3.1
Color Tags: CAMx/MM5 CAMx/CALMET
Concluding Remarks
Aside from the uncertainties inherent in the MM5 Prognostic model, the use of observational FDDA distorts the simulated wind fields leading to inconsistent flow patterns, incoherent mixing heights, and increased mass divergence,. These effects may misrepresent ozone formation in complex modeling domains, and overestimate the dilution of air pollutants transported over any significant distance.
Concluding Remarks (cont.)
Using almost any standard of objective or subjective comparison, based on either meteorological or air quality simulation results, the meteorological fields generated using the MM5 prognostic model are not as satisfactory those generated using the CALMET hybrid model.
Concluding Remarks (cont.)
The successful application of the MM5 model for the generation of meteorological inputs required for air quality modeling in California will not happen until a number of deficiencies are addressed. Among them:
-- the model is too sensitive to changes in terrain elevation. -- relatively large air temperature errors suggest poor representation of the surface energy balance. -- observational FDDA can not be relied upon to improve wind field performance in a fine-scale domain with complex topography