Air Quality Forecasting Gregory R. Carmichael University of Iowa USA
Getting better ? ….
Chemical
Figure 4
Courtesy John Reilly, MIT
Current CTMs Do Have Appreciable Skills In Predicting A Wide Variety Of Parameters INTEX B – STEM Forecasts
Courtesy John Reilly, MIT
CO
Figure 2
Figure 3
Courtesy John Reilly, MIT
O3
Figure 5
Courtesy John Reilly, MIT
NOy
Figure 6
Courtesy John Reilly, MIT
Figure 8
Aerosols
Figure 9
Tracking Air Masses During INTEX-B
Figure 15
Experiments such as TRACE-P and ACE-Asia employ mobile “Super-Sites” and study pollution outflow from source regions 50 N
40 N
China
Latitude
30 N
20 N
10 N DC-8 Flights P-3B Flights 0N 110 E
Spring 2001 120 E
130 E
140 E
Longitude
150 E
160 E
Post-mission analysis has shown that the inventory seems good for most species, except for high CO and BC observations in the Yellow Sea Model under-prediction
(Carmichael et al., JGR, 2003)
Comparison of New CO Inventory with Trace-P
Million tons
180
New estimate
160
Others Biomass Burning Transport Domestic Fossil Fuels
140
Domestic Biofuels Industry
120
Power
100
80
60
40
20
0 1999
2001
Trace-P
Ensemble Forecasting of Air Quality OZONE
* Persistence
* Single Forward Model w/o assimilation * Ensemble forecast (8 models) w/o assimilation (further improvements with bias corrections based on obs) McKeen et al., JGR, 2005
Ensemble Forecast Evaluation During Major Field Experiments PM2.5 Remains a Challenge O3
NEAQS2004
PM2.5 TexAQS -2006
McKeen et al., JGR, 2005 & 2009
Regional-Scale Chemical Analysis for Air Quality Modeling: A Closer Integration Of Observations And Models Transport Meteorology
Optimal analysis state
Chemical kinetics
CTM
Data Assimilation
Observations
Aerosols
Emissions
Improved: • forecasts • science • field experiment design • models • emission estimates • S/R relationships
Assimilation of ICARTT Ozone Observations -- Assessing Information Content
Assimilation Produces An Optimal State Space the importance of measurements above the surface!
w/o assimilation
with assimilation
Ozone predictions Example July 20, 2004
with
w/o assimilation
Chai et al., JGR 2007
Information below 4 km most important
Region-mean profile
Rapid Updates of Emissions Are Needed
Scaling factors
MILAGRO • Experimental Setup • Model performance and improvements. • Regional Effect on Ozone Production Regimes. • Effect of aerosol loading in ozone formation. • Conclusions Photo Credit: Frank Flocke.
Operational forecast products Emissions inventories •Mexico City anthropogenic emissions (UNAM, 3km, 1998) •NARSTO (36km) and EDGAR (100km) for rest of domain. •BEIS 3 Biogenic emissions. Forward modeling •60km (100x64) •12km (110x95) •60km tracer model •WRF V2.1.2 offline meteorology •18Z GFS from NCEP. •Boundary Conditions •RAQMS Global model.
60km tracer
60km Chemistry
12km chemistry
Adjoint analysis.
The high-resolution simulation yield different results not only due to the improved resolution of emissions, but also due to the terrain/landuse and wind field.
1000 800
6000
600 4000
400
2000
200 0
0 16
18
200
20 22 T IM E (U T C )
(p p b v )
4000
40
2000
O
80
2000 0
0 18
20 22 T IM E (U T C )
24
18
20 22 T IM E (U T C )
24
26
10000
F lig h t A ltit u d e O b s e rv e d 6 0 k m P r e d ic tio n 1 2 k m P r e d ic tio n
80
8000
(p p b v )
4000
10
0 16
60
6000
y
6000
16
0
40
4000
20
2000
N O
20
8000 6000
100
8000
10000
F lig h t A ltitu d e O b s e rv e d 6 0 k m P r e d ic tio n 1 2 k m P r e d ic tio n
120
10000
F lig h t A lt it u d e O b s e rv e d 6 0 k m P r e d ic tio n 1 2 k m P r e d ic tio n
A ltitu d e (m )
P r o p a n e (p p b v )
30
26
3
160
24
A ltitu d e (m )
C O (p p b v )
DC-8 flight #5 on 03/11
8000
26 0
0 16
18
20 22 T IM E (U T C )
24
26
A ltitu d e (m )
Resolution effect on model predictions
10000
F lig h t A ltitu d e O b s e rv e d 6 0 k m P re d ic tio n 1 2 k m P re d ic tio n
A ltitu d e (m )
1200
Experiments Like INTEX-A/ICARTT, INTEX-B/MILAGRO and Ace-Asia/Trace-P Provide Observations of Megacity Interactions at Various Scales "
Legend Texas
Legend
NOy (ppt)
March 19,2006 #
"
""
Filled Contours
Coahuila
MTBE (ppb)
Texas
0 - 0.001
Coahuila
155 - 300
0.001 - 0.002
300 - 500
##Nuevo Leon
500 - 700
"" ""Nuevo Leon
"
700 - 900
0.002 - 0.004
"
0.004 - 0.008 0.008 - 0.016
900 - 1,200
0.016 - 0.02
1,200 - 1,600
Tamaulipas
Zacatecas
Durango
2,000 - 3,000
0.02 - 0.04
Tamaulipas
1,600 - 2,000
0.04 - 0.06 Zacatecas
0.06 - 0.1
3,000 - 4,000
0.1 - 0.15
4,000 - 5,000 Zacatecas Aguascalientes
##
San Luis Potosi
#
# #
Michoacan
7,000 - 11,000
#
Jalisco Guanajuato
NOy
Aguascalientes
Guanajuato
Jalisco
""
"
" Michoacan
#
"
""
Yucatan
"
Hidalgo
""" "
Guerrero
1 - 2.2
"
Queretaro
#
Mexico Tlaxcala Distrito FederalPuebla
0.4 - 1
Veracruz
Campeche
Puebla
0.3 - 0.4
MTBE
"
Hidalgo
#
""
San Luis Potosi
Yucatan
Veracruz Queretaro
0.15 - 0.3
5,000 - 7,000
Data sources: Flocke & Abel
"
Mexico Distrito Federal Guerrero
Guerrero
Morelos
"" Campeche
Tlaxcala
" Puebla
"
"
Tabasco
Oaxaca
"
""
" "
"
""
"
Mena et al., 2008
MC Influence: March 11, VOC limited conditions. 10000
10000
1000
D C 8 F lig h t 3 . M a r c h 1 1 A ltitu d e o b s e rv e d fo re c a s t p o s t - a n a ly s is
8000
100
D C 8 F lig h t 3 . M a r c h 1 1 A ltitu d e o b s e rv e d m o d e le d p o s t - a n a ly s is
8000
NOy
80
100
4000
6000
60
4000
40
2000
20
N O y(p p b v )
A ltitu d e ( m )
O 3 /N O y r a tio
A ltitu d e ( m )
6000
10
O3/NOy
2000
0
0
1 12
13
14 lo c a l tim e ( h )
15
16
0 12
13
14 lo c a l tim e ( h )
15
16
Adjoint sensitivity on point along DC-8 path Originated in Mexico City. When MC outflow influences region, VOC lim conditions can be encountered. Modeled O3/NOy
Impact of Aerosols on Mexico City Photochemistry – MILAGRO Period
J-NO2 % Difference (without aerosol – with)/with Mena et al., ACP, 2009
Source Attribution at Global (and All) Scales is Becoming More Important (We need better tools) Air Quality Standards - Trends
Pollutant A
Concentration
NATIONAL Local
Regional
International Natural Local Regional/Domestic Distant
Background
Distance/Time From Sources
Source Attribution is an Important and Challenging Problem Source/Receptor Relationships
Pollution Distributions
How Important Are Distant Sources To Local Air Quality? Model estimates for surface O3 pollution Annual mean surface O3 change from 20% Perturbation in NOx+CO+NMVOC regional anthrop. emissions
Source region:
NA NAEU EUEA EASA SAsum sum3 of 3 foreign regions sum3 Similar impact from 3 foreign regions
ppb
ppb
1.4 1.4 1.2 1.2 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 region: NA Receptor NA NA 0.45 Import sensitivity:
EU
EEA Asia
EU EAREGION SA 0.75 0.65 RECEPTOR
SSA Asia 0.45
EU Task Force on Hemispheric Transport of Pollutants. Interim Assessment Report 2008 http://www.unece.org/env/eb/Air.Pollution%20Studies.No.16.Hemispheric%20Transport.pdf
RECEPTOR REGION
NASA ARCTAS MISSION 2008 Tagged CO Along Flight Paths
S/R relations Arctic
Photosynthetic control of atmospheric carbonyl sulfide during the growing season Campbell, J.E.,1, Carmichael, G.R.1, Chai, T.1, Mena-Carrasco, M.1, Tang, Y.1, Blake, D.R.2, Blake, N.J.2, Vay, S.A.3, Collatz, G.J.4, Baker, I.5, Berry, J.A.6 Montzka, S.A.7, Sweeney, C.8, Schnoor, J.L.1, Stanier, C.O1 1-University of Iowa, 2-University of California, Irvine, 3-NASA Langley Research Center, 4-NASA-Goddard Space Flight Center, 5-Colorado State University, 6-Carnegie Institution of Washington, 7-NOAA Earth System Research Laboratory,8- University of Colorado.
A)
B)
12 Observa tion
Observa tion 10
Model GPP
Model
Model NPP
8 6
) m e(k d ltiu A
Climate models incorporate photosynthesisclimate feedbacks, yet we lack robust tools for large-scale assessments of these processes. DC8 airborne observations of COS and CO2 concentrations from NASA INTEX A experiment were analyzed during the growing season over North America with a 3-D atmospheric transport model. The persistent vertical drawdown of atmospheric COS was successfully modeled using the quantitative relationship between COS and photosynthesis that has been measured in plant chamber experiments. These results provide quantitative evidence that COS gradients in the continental growing season may have broad use as a measurement-based photosynthesis tracer.
4 2 0 365
370 CO2 (ppm)
375
380
405
430
455
480
COS (ppt)
COS is best modeled using photosynthesis uptake model (GPP), while CO2 reflects both uptake and respiration (NPP). The use of COS along with CO2 measurements allows the estimation of the individual fluxes due to photosynthesis and respiration, which can not be done using CO2 measurements by themselves. Data obtained from the DC 8 INTEX-A flights.
PACDEX Hiaper Air Craft Flight Path
Flight Tracks – Back Trajectories Animation
Flight 4: Japan to Alaska D . C . B. A .
D C A B
Flight 4: Japan to Alaska
A .
B .
D .
C .
D A
C B
Flight 4: Japan to Alaska
A .
B .
D .
C .
D A
C B w/o China
PACDEX Flight 3 May 2nd 2007 STEM Aerosol Japan to Japan At low altitudes you have higher concentration of marine air (Sea Salt aerosol) At high altitudes the aerosol contribution is mainly from dust At mid-altitudes the aerosol contribution is mainly from continental pollution/dust 2.0
10
2.0
1.5
50
20
20
1.5
15
0
15
1.0
3
5
10
3
3
Altitlude (x10 m)
3
3
10
Total Dust (ug/m )
0.5
Sulfate (ug/m )
0.0
3
-20
0.5
Total BC (ug/m )
0
-10
Total OC (ug/m )
Sea Salt (ug/m )
1.0
0.0 -0.5
-50 -30
-1.0
-100
-1.5
-40 122.0
122.1
122.2
122.3 122.4 Jullian Day
122.5
5 -0.5
-1.0 122.6
0
-5
0 122.7
PACDEX Flight 3 May 2nd 2007 Measured CCN comparison to STEM aerosol data 2000
2.0
A
B
2.0
1.5
50 0
1.5
15
5
3
3
3
3
Sea Salt (ug/m )
10
-0.5 -4000
5
-6000 122.0
-1.5 122.1
122.2
122.3
122.4 Jullian Day
122.5
-0.5
-1.0 122.6
Total Dust (ug/m )
10
0.0
-1.0
-100
0.5
3
0.0
Total BC (ug/m )
0.5
Sulfate (ug/m )
CCN Concentration (#/cm )
1.0
3
-50
-2000
20
15
1.0
Total OC (ug/m )
0
20
0
-5
0 122.7
PACDEX Flight 3 May 2nd 2007 Measured CCN comparison to STEM aerosol data 2000
C
Cloud 2.0
D
2.0
1.5
50 0
1.5
15
5
3
3
3
3
Sea Salt (ug/m )
10
-0.5 -4000
5
-6000 122.0
-1.5 122.1
122.2
122.3
122.4 Jullian Day
122.5
-0.5
-1.0 122.6
Total Dust (ug/m )
10
0.0
-1.0
-100
0.5
3
0.0
Total BC (ug/m )
0.5
Sulfate (ug/m )
CCN Concentration (#/cm )
1.0
3
-50
-2000
20
15
1.0
Total OC (ug/m )
0
20
0
-5
0 122.7
PACDEX Flight 3 May 2nd 2007
CCN aerosol from continental pollution/dust sources
Measured CCN comparison to STEM aerosol data
E
2000
2.0
2.0
1.5
50 0
1.5
15
5
3
3
3
3
Sea Salt (ug/m )
10
-0.5 -4000
5
-6000 122.0
-1.5 122.1
122.2
122.3
122.4 Jullian Day
122.5
-0.5
-1.0 122.6
Total Dust (ug/m )
10
0.0
-1.0
-100
0.5
3
0.0
Total BC (ug/m )
0.5
Sulfate (ug/m )
CCN Concentration (#/cm )
1.0
3
-50
-2000
20
15
1.0
Total OC (ug/m )
0
20
0
-5
0 122.7
Our Analysis Approach
Prediction/ Analysis State
Data Assimilation
Documenting improvement (ICART) 0.14
probability
0.12
Forecast
0.1
NEI2001- Frost LPS NEI2001FrostLPS*
0.08 0.06 0.04 0.02 0 -100 -80 -60 -40 -20
0
20
40
60
80 100
% O3 bias
Left: Quantile-quantile plot of modeled ozone with observed ozone for DC-8 platform, data points collected at altitude less than 4000m, STEM-2K3, Forecast: NEI 1999, Post Analysis: NEI2001-Frost LPS*. MOZART-NCAR boundary conditions Right: Probability distribution of % ozone bias for Forecast (NEI 1999) and post analysis runs (NEI2001-FrostLPS and NEI2001FrostLPS*) for DC-8 measurements under 4000m. Mena et al., JGR, 2007
NAS 2008 www.nap.edu/catalog.php?record_id=12540
What’s wrong with these pictures?
Soil Moisture Networks
PM2.5
O3
Common Measurement Needs Threads
X important gaps may exist;
so inadequate that no
Observations Priorities Stemming from Common Threads MOST NEEDED:
Height of the planetary boundary layer Soil moisture and temperature profiles High resolution vertical profiles of humidity Measurements of air quality and atmospheric composition above the surface layer NEEDED: Direct and diffuse radiation Vertical profiles of wind Sub-surface temperature profiles (e.g., under pavement) Icing near the surface Vertical profiles of temperature Surface turbulence parameters
61