![Page 1: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/1.jpg)
Computational Aspects of Chemical Data Assimilation in Air Quality Models
Greg Carmichael (Dept. of Chem. Eng., U. Iowa)
Adrian Sandu (Dept. of Comp. Sci., Mich. Inst. Tech.)
John Seinfeld (Dept. Chem. Eng., Cal. Tech.)
Tad Anderson (Dept. Atmos. Sci., U. Washington)
Peter Hess (Atmos. Chem., NCAR)
Dacian Daescu (Inst. of Appl. Math., U. Minn.)
Tianfeng Chai, Center for Global & Regional Environmental Research, UIowa
Chemical Weather Forecasting
SatelliteProducts
Global Assimilation
RegionalPrediction
Public Impact
Requires Close Integration of Observations and Models
ICCS2003 Melbourne Australia, June 3, 2003
![Page 2: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/2.jpg)
The University of Iowa, USA
Characterization of Urban Signals
Science Support to Policy
UnderstandingUnderstandingUnderstanding
Field Experiments
Field Field ExperimentsExperiments
Long-termMonitoring
LongLong-- termtermMonitoringMonitoring
Satellites &Data Systems
Satellites &Satellites &Data Systems Data Systems
Regional and Global Simulations
Regional and Global Regional and Global SimulationsSimulations
PollutionPrediction
PollutionPollutionPredictionPrediction
PollutionDetection
PollutionPollutionDetectionDetection
Enhanced Enhanced Quality Quality of Lifeof Life
InformedInformedPolicyPolicy
DecisionsDecisions
ProcessProcessStudiesStudies UnderstandingUnderstandingUnderstanding
Field Experiments
Field Field ExperimentsExperiments
Long-termMonitoring
LongLong-- termtermMonitoringMonitoring
Satellites &Data Systems
Satellites &Satellites &Data Systems Data Systems
Regional and Global Simulations
Regional and Global Regional and Global SimulationsSimulations
PollutionPrediction
PollutionPollutionPredictionPrediction
PollutionDetection
PollutionPollutionDetectionDetection
Enhanced Enhanced Quality Quality of Lifeof Life
InformedInformedPolicyPolicy
DecisionsDecisions
ProcessProcessStudiesStudies DDDA
S
![Page 3: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/3.jpg)
TRACE-P/Ace-Asia EXECUTION
Emissions-Fossil fuel-Biomass burning-Biosphere, dust
Long-range transport fromEurope, N. America, Africa
ASIA PACIFIC
Satellite datain near-real time:MOPITTTOMSSEAWIFSAVHRRLIS
3D chemical model forecasts: - x - GEOS-CHyEM - CFORS - z
FLIGHTPLANNING
Boundary layerchemical/aerosolprocessing
ASIANOUTFLOW
Stratosphericintrusions
PACIFIC
![Page 4: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/4.jpg)
2 4 6 8 10T IM E (G M T )
0
200
400
600
800
1000
CO
(pp
bv)
0
2000
4000
6000
Alt
itu
de
(m)
ObservedModeledFlight Height
2 4 6 8 10T IM E (G M T )
4 0
6 0
8 0
1 0 0
O3 (
pp
bv)
0
2000
4000
6000
Alt
itu
de
(m)
ObservedModeledFlight Height
2 4 6 8 10T IM E (G M T )
0
0.5
1
1.5
2
2.5
NO
2 (pp
bv)
0
2000
4000
6000
Alt
itu
de
(m)
ObservedModeledFlight Height
2 4 6 8 10T IM E (G M T )
0
2
4
6
8
10
SO2 (
ppbv
)
0
2000
4000
6000
Alt
itu
de
(m)
ObservedModeledFlight Height
2 4 6 8 10T IM E (G M T )
0
1
2
3
4
5
SO4 (
ppbv
)
0
2000
4000
6000
Alt
itu
de
(m)
ObservedModeledFlight Height
P-3B
SO2
How Well Do Models Capture the Observed Features?
![Page 5: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/5.jpg)
P-3B
0.00
0.20
0.40
0.60
0.80
1.00
1.20
Te
mp
era
ture
H2
O
Win
d S
pe
ed
O3
SO
4
J[O
1D
]
SO
2
PA
N
Eth
en
e
Pro
pa
ne
CO
J[N
O2
]
Eth
an
e
No
y
Eth
yn
e
RN
O3
Be
nze
ne
+ T
olu
en
e
OH
AO
E
HN
O3
NO
2
NO
Co
rrela
tio
n C
oeff
icie
nt
R(<1KM)
R(1-3KM)
R(>3 KM)
Predictability – as Measured by Correlation Coefficient
Met Parameters are Best
![Page 6: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/6.jpg)
Overview of Research in Data Assimilation for Chemical Models. Solid lines represent current capabilities. Dotted lines represent new analysis capabilities that arise through the assimilation of chemical
data.
![Page 7: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/7.jpg)
![Page 8: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/8.jpg)
Advantages of Adjoint Sensitivities
![Page 9: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/9.jpg)
![Page 10: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/10.jpg)
We are Developing General Software Tools to Facilitate the Close Integration of Measurements and Models
The framework will provide tools for: 1) construction of the adjoint model; 2) handling large datasets; 3) checkpointing support; 4) optimization; 5) analysis
of results; 6) remote access to data and computational resources. (www.cs.mtu.edu/~asandu/Software/Kpp)
![Page 11: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/11.jpg)
![Page 12: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/12.jpg)
![Page 13: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/13.jpg)
![Page 14: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/14.jpg)
![Page 15: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/15.jpg)
![Page 16: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/16.jpg)
![Page 17: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/17.jpg)
Advantage Of Adjoints Is That We Get Sensitivities; e.g., Influence Functions (top view)
(top:Mar.1-3, bottom:Mar.22-24; from left to right: O3, NO2, HCHO)
![Page 18: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/18.jpg)
Summary• Nonlinear, stiff chemistry important part of
chemical transport models;• One-step methods (RK, Ros) seem preferable
w/ adjoint sensitivity;• KPP: efficient code for simulation and
direct/adjoint sensitivities;• Preliminary 3D assimilation results with parallel
adjoint STEM-III;• Developing/exploring new optimization
techniques (using 2nd order sensitivities), adaptive grids…..
![Page 19: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/19.jpg)
Target Applications Air quality forecasting in
urban environments; Integration of measurements
and models to produce a consistent/optimal analysis data set for atmospheric chemistry field experiments (e.g., Ace-Asia);
Inverse analysis to produce a better estimate of emissions;
Design of observation strategies to improve chemical forecasting capabilities.
PORT PHILLIP BAY
260 280 300 320 340 360
EASTING (km)
DND
BRI
FTSPSY
PTC
MTC ALP
PTHGLS
GVD
PLP BXH
5740
5760
5780
5800
5820
5840
NORTHIN
G(km)
LIGHT
MODERATE
HEAVY
AIR QUALITY FORECAST-MELBOURNE
AIR QUALITY FORECASTAIR QUALITY FORECAST--MELBOURNEMELBOURNE
NORTH EAST
HOUR
IND
EX
NORTH EAST
HOUR
IND
EX
Tomorrow will be fine and sunnyTomorrow will be fine and sunny--with moderate to heavy air pollutionwith moderate to heavy air pollution
PORT PHILLIP BAY
260 280 300 320 340 360
EASTING (km)
DND
BRI
FTSPSY
PTC
MTC ALP
PTHGLS
GVD
PLP BXH
5740
5760
5780
5800
5820
5840
NORTHIN
G(km)
LIGHT
MODERATE
HEAVY
AIR QUALITY FORECAST-MELBOURNE
AIR QUALITY FORECASTAIR QUALITY FORECAST--MELBOURNEMELBOURNE
PORT PHILLIP BAY
260 280 300 320 340 360
EASTING (km)
DND
BRI
FTSPSY
PORT PHILLIP BAY
260 280 300 320 340 360
EASTING (km)
DND
BRI
FTSPSY
PTC
MTC ALP
PTHGLS
GVD
PLP BXH
5740
5760
5780
5800
5820
5840
NORTHIN
G(km)
LIGHT
MODERATE
HEAVY
AIR QUALITY FORECAST-MELBOURNE
AIR QUALITY FORECASTAIR QUALITY FORECAST--MELBOURNEMELBOURNE
NORTH EAST
HOUR
IND
EX
NORTH EAST
HOUR
IND
EX NORTH EAST
HOUR
IND
EX
NORTH EAST
HOUR
IND
EX
Tomorrow will be fine and sunnyTomorrow will be fine and sunny--with moderate to heavy air pollutionwith moderate to heavy air pollution
TWO-SCENARIO TWO-SCENARIO FORECASTFORECAST
![Page 20: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/20.jpg)
• Ability of forecast models to represent the individual processes controlling air pollution formation and transport.
& intercontinental
regional
Application: The Design of Better Observation Strategies to Improve Chemical Forecasting Capabilities.
Data Assimilation Will help us Better Determine Where and When to Fly and How to More Effectively Deploy our Resources (People, Platforms, $$s)
We Plan to Test These Tools Including Adaptive Measurements in
the Summer of 2004
![Page 21: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/21.jpg)
![Page 22: Computational Aspects of Chemical Data Assimilation in Air Quality Models Greg Carmichael (Dept. of Chem. Eng., U. Iowa) Adrian Sandu (Dept. of Comp. Sci.,](https://reader035.vdocument.in/reader035/viewer/2022062518/5697bf731a28abf838c7f09c/html5/thumbnails/22.jpg)