Modeling 2

Models Play a Critical Role in Linking Emissions to Aerosol and Trace Gas Distributions and Subsequent Effects

Modeled

Observed

07/29/09

1

Models are an Integral Part of Air Quality Studies • Field experiment planning • Provide 4-Dimensional context of the observations • Facilitate the integration of the different measurement platforms • Evaluate processes (e.g., role of biomass burning, heterogeneous chemistry….) • Evaluate emission estimates (bottom-up as well as top-down) • Emission control strategies testing • Air quality forecasting 07/29/09 • Measurement site selection

2

07/29/09

3

Air Quality Modeling: Improving Predictions of Air Quality (analysis and forecasting perspectives) Met model

Chemical, Aerosol, Removal modules

Predicted Quantity: e.g., ozone AQ violation

CTM

How confident are we in the models & predictions?

Emissions

07/29/09

Observations

4

Chemical Transport Model • 3D atmospheric transport-chemistry model (STEM-III)

∂ci 1 = −u ⋅ ∇ci + ∇ ⋅ ( ρK∇ci ) + f i (c) + Ei ∂t ρ where chemical reactions are modeled by nonlinear stiff terms

f i (c) = Pi (c) − Di (c)ci • Use operator splitting to solve CTM

M[Δ] t t  07/29/09

t

t  TXΔ2Δ2Δ2ΔΔ2Δ2Δ2 TY t  TZ t  C t TZ t  TY t  TX t  5

EULERIAN MODELS PARTITION ATMOSPHERIC DOMAIN INTO GRIDBOXES This discretizes the continuity equation in space

Solve continuity equation  for individual gridboxes

• Detailed chemical/aerosol models can  presently afford ­106 gridboxes • In global models, this implies a  horizontal resolution of ~ 0.5­1o (~50 to  100 km) in horizontal and ~ 0.5­1 km in  vertical • Chemical Transport Models (CTMs) use external meteorological data as input  (or run on­line) 07/29/09 6 • General Circulation Models (GCMs) compute their own meteorological fields From D. Jacob

07/29/09

7

07/29/09

8

07/29/09

9

SPECIFIC ISSUES FOR AEROSOL CONCENTRATIONS • A given aerosol particle is characterized by its size, shape, phases, and  chemical composition – large number of variables! • Measures of aerosol concentrations must be given in some integral  form, by summing over all particles present in a given air volume that  have a certain property • If evolution of the size distribution is not resolved, continuity equation  for aerosol species can be applied in same way as for gases • Simulating the evolution of the aerosol size distribution requires  inclusion of nucleation/growth/coagulation terms in Pi and Li, and size  characterization either through size bins or moments. 

condensation

coagulation

Typical aerosol  size distributions by volume

nucleation 07/29/09

10

From D. Jacob

LAGRANGIAN APPROACH: TRACK TRANSPORT OF POINTS IN MODEL DOMAIN (NO GRID)

• Transport large number of points with trajectories  from input meteorological data base (U) + random  turbulent component (U’) over time steps ∆t • Points have mass but no volume

position to+∆t U’∆t position to

U∆t

07/29/09

• Determine local concentrations as the number of  points within a given volume • Nonlinear chemistry requires Eulerian mapping at  every time step (semi­Lagrangian) PROS over Eulerian models: • no Courant number restrictions • no numerical diffusion/dispersion • easily track air parcel histories • invertible with respect to time CONS: • need very large # points for statistics • inhomogeneous representation of domain • convection is poorly represented 11 From D. Jacob • nonlinear chemistry is problematic

LAGRANGIAN RECEPTOR-ORIENTED MODELING Run Lagrangian model backward from receptor location,  with points released at receptor location only

• Efficient cost­effective quantification of source  influence distribution on receptor (“footprint”) • Enables inversion of source influences by the  adjoint method (backward model is the adjoint of  the Lagrangian forward model) 07/29/09

12

From D. Jacob

Air Quality Prediction: A Challenge of Scales and Integration Megacity Impacts

Air quality Analysis Regional Prediction

Global Assimilation

Satellite Products

07/29/09

Modified after Pierce NASA/Langley

13

Integrated Science Studies: Impacts of Global Composition on Regional Air Quality Global-Regional-Urban nesting of CTMs Effects of Boundary Conditions are significant and improve predictions (Tang et al., JGR 2007). Alaskan BB Impacts Northern Boundary

07/29/09

Assessment of continental inflow/outflow requires unified modeling/measurement strategy to accurately characterize coupling between the continental boundary layer, free troposphere, and longrange transport. 14

Model Resolution, Transport and Removal also Contribute to Differences Emissions

Monthly mean concentrations

Temporal variation in concentrations

BC 07/29/09

15

Removal Processes Remain Poorly Characterized in Models Impact of Wet Removal on Predicted BC Progress limited by lack of understanding and observations

07/29/09

16

An Air Quality Modelling System Site data Topography

Land cover

Weather forecast and local data Model results

Observations

Site data processing

Emission inventories Industries Vehicular traffic

Chemical data

3D meteo modelling

Emissions modelling

Models interface

Area sources

3D chemical-transport modelling

Chemical 3D IC / BC processing

Monitoring data Model results

3D meteo with on-line chemistry Analysis of concentrations Comparison with regulatory norms and objectives

Concentrations: 3D & ground

Post-processing & graphic

07/29/09

17

Need to Estimate Emissions at Appropriate Scales

07/29/09

18

Detail: a matter of scale (3) SO2 emissions in the vicinity of Shanghai, China

1º x 1º

.5º x .5º

10’ x 10’

30’’ x 30’’

07/29/09

19

Emissions processing for AQM Inventories

Thematic data

“reference raw emission data” (point / line / area) SPACE DISAGGREGATION

TIME MODULATION

Modulation profiles

Residential

(hourly, daily, monthly) 2.5

Highways

Somma di FACTOR

NMVOC & PM SPECIATION & SIZE

2

DAY NOM_PROFIL

1.5

1 - Leggeri_Extraurbane 2 - Leggeri_Extraurbane

Deciduous

3 - Leggeri_Extraurbane 4 - Leggeri_Extraurbane 5 - Leggeri_Extraurbane 1

6 - Leggeri_Extraurbane 7 - Leggeri_Extraurbane

0.5

3

NonIndComb

Speciation & dimensional profiles

Formaldeide

gg_0_150 0 0

1

2

3

4

5

6

2.5 7

gg_150_300 8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

Acetaldeide

gg_300_500

23

gg_500_1000

HOUR

2

Aldeidi superiori

gg_1000_2000 P rodInd

1.5

Chetoni

SmalB usi DomA ctiv A irtraf

1

Colture VdA_auto_EA

0.5

VdA_comm_EA Decid

0

Conif

1

2

3

4

5

6

7

mese

8

9

10 11 12

ThP owP lants TotElP owReq Veicoli_VdA

Etilene Alcani (B.R.)

Latifoglie

Alcani (A.R.)

Conifere Auto

Aromatici (B.R.) Aromatici (A.R.) Olefine (B.R.) Olefine (A.R.) Isoprene Monoterpeni 0%

20%

40%

60%

80%

100%

Model-ready input 07/29/09

(hourly, gridded, speciated emissions)

20

Other types of emission data • Emissions estimated “on-line”, using current meteorology – desert dust – sea salt – forest fires

• Data from plants emissions monitoring systems • Real-time traffic

07/29/09

21

Other factors • Periodical update of inventory data (changes in activity levels, fuels, control equipment …)

• Special situations (traffic bans, events, …)

• Future scenarios …examples later

07/29/09

22

OUR MODELING SYSTEM FRAMEWORK

Assimilation Observations

Observations

07/29/09

23

STEM

Non reactive tracers: Anthropogenic CO, Biomass Burning CO, BC, OC, Sulfate aerosol, Dust, Anthropogenic Hg, Volcano SO2 Simple decaying rates Dry deposition Wet scavenging Horizontal Resolution: 60x60 km Vertical Resolution: WRF's half-sigma coordinate Time Resolution: 6hr Forecast length: +120hr where 00hr=00UTC Full Chemistry: SAPRC99 mechanism + on-line four-bin SCAPE II module on-line TUV for photolysis rates computation BC/IC: RAQMS global chemical transport model Horizontal Resolution: 60x60 km - North America Vertical Resolution: WRF's half-sigma coordinate Time Resolution: 6hr 07/29/09 24 Forecast length: +48hr where 00hr=00UTC 7th International Conference on Air Quality – Science and Application – 2009 Istanbu

07/29/09

25

Mission A strong Overview outflow event will appear from Saturday to Sunday

July 1 to 25 Model CO

Midwest Ohio etc NY-MA-MD TX-NM Southeast California Canada

2km wind field

07/29/09

26

BETTER UNDERSTANDING OF ARCTIC ATMOSPHERIC COMPOSITION AND CLIMATE Transport of mid-latitudes pollution to the Arctic Transport pathways for different pollutants - Contributions from different source regions - Source-receptor relationships. Boreal forest fires Compositions evolution of the fire plumes - aerosol optical properties and evolution - implications for regional and global atmospheric composition. Aerosol radiative forcing Regional radiative forcing from Arctic haze, fire plumes major sources of soot to the Arctic - soot deposition effect on ice albedo. Chemical processes HOx-NOx chemistry - regional extent of halogen radical chemistry - regional implications for ozone, aerosols, mercury - stratosphere-troposphere exchange effect on 07/29/09 27 tropospheric ozone in and theApplication Arctic. – 2009 Istanbul 7th International Conference on Air Quality – Science

ARCTAS Arctic Research of the Composition of the Troposphere from Aircraft and Satellites Part of the international IPY/POLARCAT Arctic field program for atmospheric composition NASA TROPOSPHERIC CHEMISTRY PROGRAM ARCTAS Partners: NASA, NOAA, DOE, NSF, Canada, CNRS, DLR MEASURE

SATELLITE

07/29/09

VALIDATION ASSIMILATION ANALYSIS

MODEL 28

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

ARCTAS AIRCRAFT FLEET DC-8 in situ chemistry and aerosol Ceiling 37 kft, range 4000 nmi, endurance 9 h Payload: O3, H2O, CO, CO2, CH4, NOx and Hox hemistry, BrO, mercury, NMVOCs, halocarbons, SO2. HCN/CH3CN, actinic fluxes, aerosol composition, aerosol mass and number concentrations, aerosol physical and optical properties, remote ozone and aerosol.

P-3B radiation and in situ aerosol Ceiling 30 kft, range 3800 nmi, endurance 8 h Payload: optical depth, radiative flux, radiance spectra, aerosol composition, black carbon 29

07/29/09 7th International Conference on Air Quality – Science and Application – 2009 Istanbul

SATELLITE TEAMS Satellite: CALIPSO (CALIOP); MetOp(GOME-2); AQUA (MODIS, AIRS); TERRA (MODIS, MISR, MOPITT); AURA (MLS, TES, HIRDLS, OMI); ENVISAT (SCIAMACHY) Capabilities: Aerosol Optical Depth properties, CO, Ozone, BrO, NO2, HCHO, Methan

MODEL

Forecasts/analyses: GEOS-5, GOCART, GEOS-Chem, STEM, MOZART, RAQMS

GROUND TEAMS

Ground Teams: UAF, NATIVE, ARC-IONS 07/29/09

30

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

MISSION DEPLOIMENTS Spring deployment: April 2008 Fairbanks and Barrow, Alaska Thule, Greenland Iqaluit, Canada

Summer deployment: June – July 2008 California, Cold Lake and Yellowknife, Canada Thule Greenland 07/29/09

31

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

FLIGHT PLANNING AIMS TO DEFINE THE BEST FLIGHT MENU ACCORDING TO: weather conditions sampling opportunities progress in meeting mission objectives UNIVERSITY OF IOWA Forecasting Team Purpose: Provide Gridded Data Emissions for all chemical transport models. Daily-bases High Resolution 3-D Meteorological and Chemical Forecast

07/29/09

32

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

07/29/09

33

07/29/09

34

07/29/09

35

07/29/09

36

07/29/09

37

07/29/09

38

07/29/09

39

07/29/09

40

07/29/09

41

07/29/09

42

07/29/09

43

07/29/09

44

07/29/09

45

07/29/09

46

07/29/09

47

07/29/09

48

07/29/09

49

07/29/09

50

07/29/09

51

07/29/09

52

07/29/09

53

How did we do?

07/29/09

54

DC-8 SPRING FLIGHT 17th of April European Plume

Anthropogenic CO 5.5 km by S. Kulkarni, CGRER

Asian Plume Biomass CO 5.5km

07/29/09

Biomass CO curtain plot 55

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

DC-8 SPRING FLIGHT 17th of April Black Carbon

Speed

Temp

Missed Polar Pick European Asian Plume Plume

Biomass CO tracer vs. measured Hydrogen Cyanide (HCN )

RH

07/29/09

56

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

th DC-8 SUMMER FLIGHT 09 of July Siberian-Asian

Plume

AOD

by S. Kulkarni, CGRER

BC 07/29/09

5.5km 57

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

DC-8 SUMMER FLIGHT 09th of July Speed

Direction

Composite satellite image http://amrc.ssec.wisc.edu/

Temp

RH 5.5km 07/29/09

58

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

DC-8 SUMMER FLIGHT 09th of July Biomass CO tracer vs. measured Hydrogen Cyanide (HCN ) Siberian Asian Plum

STEM FULL CHEMISTRY NORTH AMERICA DOMAIN OZONE

07/29/09

59

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

CONCLUSIONS •For the selected flights, the major targets highlighted during the pre-flight meetings have been reasonably met. •The WRF model was able to capture the important features of synoptic conditions (fronts positioning, clouds cover, general circulation) and Boundary Layer flow. •Computed-measured comparisons showed the System’s capability to describe the trends seen for the observations •The preliminary results coming out from the present study represent an encouraging starting point for further analysis based on the feedback of 07/29/09 60 measurements into the models in order to better 7th International Conference on Air Quality – Science and Application – 2009 Istanbul understand the Arctic Environment and to improve

ACKNOWLEDGEMENTS I would like to thank the ARCTAS Measurement Team for permission in using the measurements, CGRER and the University of Iowa and the Space Science and Engineering Center, University of WisconsinMadison, for the satellite imagine.

07/29/09

61

7th International Conference on Air Quality – Science and Application – 2009 Istanbul

07/29/09

62

OPERATOR SPLITTING IN EULERIAN MODELS Reduces dimensionality of problem

• Split the continuity equation into contributions from transport and local terms:

Ci  Ci   t  t 

TRANSPORT

 dCi    dt  

LOCAL

 dCi  Transport  advection, convection:    U  Ci   dt  TRANSPORT Local  chemistry, emission, deposition, aerosol processes:  dCi   dt   

 Pi (C)  Li (C) LOCAL

 … and integrate each process separately over discrete time steps:

Ci (to  t )  (Local)•(Transport)  Ci (to ) These operators can be split further: • split transport into 1­D advective and turbulent transport for x, y, z    (usually necessary)  split local into chemistry, emissions, deposition (usually not necessary)63 •07/29/09

SPLITTING THE TRANSPORT OPERATOR

• Wind velocity U  has turbulent fluctuations over time step ∆ t:

U(t )  U  U '(t )

Time­averaged component (resolved)

Fluctuating component (stochastic)

• Split transport into advection (mean wind) and turbulent components:

Ci 1  U  Ci    KCi t 

  air density K  turbulent diffusion matrix

advection             turbulence (1st­order closure)

• Further split transport in x, y, and z to reduce dimensionality.  In x direction:

Ci Ci 1  Ci  u  ( K xx ) t x  x x 07/29/09

advection operator

turbulent operator

U  (u , v, w)

64

SOLVING THE EULERIAN ADVECTION EQUATION Ci Ci  u t x

•   Equation is conservative: need to avoid  diffusion or dispersion of features. Also need  mass conservation, stability, positivity… • All schemes involve finite difference  approximation of derivatives : order of  approximation → accuracy of solution • Classic schemes: leapfrog, Lax­Wendroff,  Crank­Nicholson, upwind, moments… • Stability requires Courant number u∆ t/∆ x < 1  … limits size of time step • Addressing other requirements (e.g., positivity)  introduces non­linearity in advection scheme 07/29/09

65

VERTICAL TURBULENT TRANSPORT (BUOYANCY) • generally dominates over mean vertical advection

• K­diffusion OK for dry convection in boundary layer (small eddies) • Deeper (wet) convection requires non­local convective parameterization Convective cloud (0.1­100 km)

detrainment Model vertical levels

updraft

downdraft

entrainment

07/29/09

Model grid scale

Wet convection is  subgrid scale in global  models and must be  treated as a vertical  mass exchange  separate from transport  by grid­scale winds. Need info on convective  mass fluxes from the  model meteorological  driver. 

66

LOCAL (CHEMISTRY) OPERATOR: solves ODE system for n interacting species For each species i   1, n  dC i  Pi (C)  Li (C) dt

C  (C1 ,...Cn )

System is typically “stiff” (lifetimes range over many orders of magnitude) → implicit solution method is necessary. • Simplest method: backward Euler.  Transform into system of n algebraic  equations with n unknowns  C(to  t )

Ci (to  t )  Ci (to )  Pi (C(to  t ))  Li (C(to  t )) t

i   1, n 

Solve e.g., by Newton’s method.  Backward Euler is stable, mass­conserving,  flexible (can use other constraints such as steady­state, chemical family  closure, etc… in lieu of ∆ C/∆t ).  But it is expensive.  Most 3­D models use  higher­order implicit schemes such as the Gear method. 07/29/09

67

COMPOSITION? Solve continuity equation for chemical mixing ratios Ci(x, t) Lightning

Eulerian form:

U = wind vector

Ci  U  Ci  Pi  Li t

Pi = local source

Transport

 Lagrangian form:

Chemistry Aerosol microphysics

dCi  Pi  Li dt

Volcanoes  Fires

Human Land biosphere activity

       of chemical i

Li = local sink

Ocean 

Courtesy John Reilly, MIT 07/29/09

69