The The Australian Air Quality Australian Air Quality Forecasting System (AAQFS)Forecasting System (AAQFS)

P.C. ManinsP.C. Manins11,, M.E.CopeM.E.Cope1,21,2, , G.D HessG.D Hess33, K.J. Tory, K.J. Tory33, ,

Sunhee LeeSunhee Lee11, K.Puri, K.Puri33, M.Young, M.Young44

11CSIRO Atmospheric Research, CSIRO Atmospheric Research, 22CSIRO Energy Technology, CSIRO Energy Technology, 33Bureau of Meteorology Bureau of Meteorology Research Centre Research Centre 44Environment Protection Authority of NSWEnvironment Protection Authority of NSW

GENERAL PROJECT OBJECTIVES

A project supported by A project supported by Environment Australia Environment Australia through the Natural through the Natural Heritage Trust.Heritage Trust.

SYDNEYSYDNEYMELBOURNEMELBOURNE

AUSTRALIAAUSTRALIA

Develop and implement a numerical air Develop and implement a numerical air quality forecasting system in Melbourne quality forecasting system in Melbourne and Sydney and Sydney –– AustraliaAustraliaDemonstrate the System in Sydney during Demonstrate the System in Sydney during the Olympics and Parathe Olympics and Para--Olympics (2000)Olympics (2000)

PORT PHILLIP BAY

260 280 300 320 340 360EASTING (km)

DND

BRI

FTSPSY

PTC

MTC ALP

PTHGLS

GVD

PLP BXH

5740

5760

5780

5800

5820

5840

NO

RTH

I NG

(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

Are spatial and temporal information Are spatial and temporal information needed from the forecast? needed from the forecast? (e.g. hour(e.g. hour--byby--hour, suburbhour, suburb--byby--suburb)suburb)

Support air quality management & policy Support air quality management & policy development? development? (e.g. VOC controls)(e.g. VOC controls)

Are monitoring data limited?Are monitoring data limited?(no extensive network)(no extensive network)??

Is a prognostic air pollution forecasting system worth the considerable effort?

Why not a use a statistical forecasting system? More…

Levels of Complexity More…

1.1. Embedded in a operational National Weather Embedded in a operational National Weather Forecasting System Forecasting System –– AAQFSAAQFS

2.2. Extension of Numerical Weather Forecasting Extension of Numerical Weather Forecasting Capability Capability –– e.g.e.g., Beijing, China, Beijing, China

3.3. NMHS seeking to develop both a national NMHS seeking to develop both a national numerical weather and pollution forecast numerical weather and pollution forecast –– Malaysia?Malaysia?

4.4. NMHS focussed on forecasting air pollution for NMHS focussed on forecasting air pollution for a limited region a limited region –– others?others?

AAQFS DESIGN FEATURES

Generate air quality forecasts twice per day for a Generate air quality forecasts twice per day for a period of 24period of 24––36 hours: 36 hours: (3 pm and 9 am).(3 pm and 9 am).Consider a range of air pollutants: Consider a range of air pollutants: NOxNOx, ROC, SO, ROC, SO22, O, O33, aerosol, air toxics., aerosol, air toxics.Resolve air quality at regional and suburb level Resolve air quality at regional and suburb level (5 km, 1 km).(5 km, 1 km).Generate a ‘business as usual’ forecast and a Generate a ‘business as usual’ forecast and a ‘greener emissions’ forecast.‘greener emissions’ forecast.

The Need for High Resolution More…

Leads to improved weather forecastsLeads to improved weather forecastsChanges in space and time importantChanges in space and time important

Necessary to resolve regional flowsNecessary to resolve regional flowsFor air pollution, wind For air pollution, wind trajectorytrajectory vitalvital

Boundary layer must be resolvedBoundary layer must be resolvedFor air pollution levels, For air pollution levels, mixing heightmixing height vitalvital

SYSTEM- FEATURES

MET ANALYSIS

EMISSIONSDATABASE

AQ + METOBS.

NWPGASP/LAPS

INVENTORY

CTM

AIR QUALITYFORECAST

EVALUATIONDATA

PACKAGE

Australian operationalAustralian operationalweather weather forecast modelsforecast models

EMSEMS--95 derivative95 derivative

CustomCustom--integrateintegrateinto NWPinto NWP

STUDY REGIONSVictoria-Melbourne

141.0 142.0 143.0 144.0 145.0 146.0 147.0

Longitude (deg. E)

-41.0

-40.0

-39.0

-38.0

-37.0

-36.0

-35.0

Latit

ude

(deg

. N)

MELBOURNE

WANGARATTA

BENDIGO

WARRAGULGEELONG

WODONGA

CASTLEMAINE

SWANHILL

HORSHAM

WARRNAMBOOL

KING IS

TASMANIA

BASS STRAIT

130 130 ×× 130; 130; ∆∆xx ~5 km ~5 km VICTORIAVICTORIA

SYDNEYSYDNEYMELBOURNEMELBOURNE

AUSTRALIAAUSTRALIA

144.2 144.4 144.6 144.8 145.0 145.2 145.4LONGITUDE (deg. E)

-38.4

-38.2

-38.0

-37.8

-37.6

LATI

TUD

E (d

eg. N

)

ALPH

BXHL MTCL

PTCK

PAIS

BRTN

RMIT

GSTH GRVD

PORT PHILLIPBAY

BASS STRAIT

MELBOURNE

GEELONG

130 130 ×× 96; 96; ∆∆xx ~1 km ~1 km

149.0 150.0 151.0 152.0 153.0

Longitude (deg. E)

-36.0

-35.0

-34.0

-33.0

-32.0

Latit

ude

(deg

. N)

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

STUDY REGIONSNew South Wales-Sydney

98 98 ×× 98; 98; ∆∆xx ~5 km ~5 km NSWNSW

SYDNEYSYDNEYMELBOURNEMELBOURNE

AUSTRALIAAUSTRALIA

150.5 150.6 150.7 150.8 150.9 151.0 151.1 151.2 151.3 151.4LONGITUDE (deg. E)

-34.1

-34.0

-33.9

-33.8

-33.7

-33.6

LATI

TUD

E (d

eg. N

)

BLAC

BRIN

CAMD CAMP

EARL LIDC

LIND

LIVE RAND

RICH

ROZE

ST.M

VYNE

WSTM

WOOL TASMAN

SEA

SYDNEY

MASCOT

PENRITH

SYDNEYSYDNEY 98 98 ×× 56; 56; ∆∆xx ~1 km ~1 km

Data Flows for AAQFS

GASP

CTMNSW

LAPS05VIC

CTMVIC

LAPS375

LAPS125

LAPS05NSW

SUPERCOMPUTERS

POST-PROCESSING,ARCHIVING,GRAPHICS,VERIFICATION

EPAVIC

EPANSW

EMAIL

ARCHIVE

WEBSITE

FTPSITE

EPANSW

EPAEPA--Victoria AAQFS Web PageVictoria AAQFS Web Page

http://www.epa.vic.gov.au/air/AAQFS

Daily Validation

15 March 200115 March 2001

OO33

NONOyy NONO22

VOCVOC

PERFORMANCE REVIEWPERFORMANCE REVIEW

Consider forecasts for some of 2000/2001 and all of 2001/2002 photochemical smog seasons

5 km forecasting domains

Assess the limit of predictability for forecasts of peak daily 1-hour ozone concentration

Verification- Air Quality modelling

Example: Sydney 7Example: Sydney 7--day ozone episode 21day ozone episode 21--27 January 200127 January 2001

149.0 150.0 151.0 152.0 153.0

EASTING

-36.0

-35.0

-34.0

-33.0

-32.0

NO

RTH

ING

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

40

45

50

55

60

65

70

75

80

85

90

95

100

105

110

115

120

OZONE- HOUR 16, 21/1/2001

149.0 150.0 151.0 152.0 153.0

EASTING

-36.0

-35.0

-34.0

-33.0

-32.0

NO

RTH

ING

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

40

45

50

55

60

65

70

75

80

85

90

95

100

105

110

115

120

OZONE- HOUR 16, 23/1/2001

149.0 150.0 151.0 152.0 153.0

EASTING

-36.0

-35.0

-34.0

-33.0

-32.0

NO

RTH

ING

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

40

45

50

55

60

65

70

75

80

85

90

95

100

105

110

115

120

OZONE- HOUR 16, 25/1/2001

149.0 150.0 151.0 152.0 153.0

EASTING

-36.0

-35.0

-34.0

-33.0

-32.0

NO

RTH

ING

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

40

45

50

55

60

65

70

75

80

85

90

95

100

105

110

115

120

OZONE- HOUR 16, 27/1/2001

(ppb)

260 270 280 290 300 310 320 330 340 350 360

EASTING (km)

6200

6210

6220

6230

6240

6250

6260

6270

6280

6290

6300

NO

RTH

ING

(km

)

ROZELLE

WOOLOOWARE

LIVERPOOL

LIDCOMBE

BLACKTOWN

RICHMOND

WESTMEAD

VINEYARD

BARGO

CAMPBELLTOWN

APPIN

LINDFIELD

RANDWICK

Example: Sydney 7Example: Sydney 7--day ozone episodeday ozone episode2121--27 January 2001.27 January 2001.

VINEYARD- O3

0

20

40

60

80

100

120

140

0 24 48 72 96 120 144 168

TIME (hours)

CONC

ENTR

ATIO

N (p

pb) OBS

CTM

LIDCOMBE- O3

0

20

40

60

80

100

120

140

0 24 48 72 96 120 144 168

TIME (hours)

CON

CENT

RATI

ON

(ppb

)

OBSCTM

LIVERPOOL- O3

0

20

40

60

80

100

120

140

0 24 48 72 96 120 144 168

TIME (hours)

CO

NCE

NTR

ATI

ON

(ppb

)

OBSCTM

WOOLOOWARE- O3

0

20

40

60

80

100

120

140

0 24 48 72 96 120 144 168

TIME (hours)

CONC

ENTR

ATI

ON

(ppb

)

OBSCTM

PERFORMANCE REVIEWPERFORMANCE REVIEW

149.0 150.0 151.0 152.0 153.0

Longitude (deg. E)

-36.0

-35.0

-34.0

-33.0

-32.0

Latit

ude

(deg

. N)

NEWCASTLE

WOLLONGONG

MUSWELLBROOK

LITHGOW

SYDNEY

CANBERRA

ORANGE

PACIFIC OCEAN

REGIONALREGIONALFORECASTINGFORECASTING

PEAK 1-HOUR OZONE (Regional)

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

SYDNEY SOUTH WEST

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

SYDNEY NORTH WEST

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

150.5 150.6 150.7 150.8 150.9 151.0 151.1 151.2 151.3 151.4LONGITUDE (deg. E)

-34.1

-34.0

-33.9

-33.8

-33.7

-33.6

LATI

TUD

E (d

eg. N

)

BLAC

BRIN

CAMD CAMP

EARL LIDC

LIND

LIVE RAND

RICH

ROZE

ST.M

VYNE

WSTM

WOOL TASMAN

SEA

SYDNEY

MASCOT

PENRITH

NW

SW

EAST

SYDNEYSYDNEY-- Daily 1Daily 1--hour Ohour O33 maxmaxSUBSUB--REGIONAL FORECASTINGREGIONAL FORECASTING

SYDNEY EAST

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

SYDNEYSYDNEY-- Daily 1Daily 1--hour Ohour O33 maxmax

SUBURBSUBURB--LEVEL LEVEL FORECASTINGFORECASTING

150.5 150.6 150.7 150.8 150.9 151.0 151.1 151.2 151.3 151.4LONGITUDE (deg. E)

-34.1

-34.0

-33.9

-33.8

-33.7

-33.6

LATI

TUD

E (d

eg. N

)

BLAC

BRIN

CAMD CAMP

EARL LIDC

LIND

LIVE RAND

RICH

ROZE

ST.M

VYNE

WSTM

WOOL TASMAN

SEA

SYDNEY

MASCOT

PENRITH

NW

SW

EAST

PEAK 1-HOUR OZONE (Suburb)

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

SYDNEYSYDNEY-- Daily 1Daily 1--hour Ohour O33 maxmax

STATIONSTATION--LEVEL LEVEL FORECASTINGFORECASTING

150.5 150.6 150.7 150.8 150.9 151.0 151.1 151.2 151.3 151.4LONGITUDE (deg. E)

-34.1

-34.0

-33.9

-33.8

-33.7

-33.6

LATI

TUD

E (d

eg. N

)

BLAC

BRIN

CAMD CAMP

EARL LIDC

LIND

LIVE RAND

RICH

ROZE

ST.M

VYNE

WSTM

WOOL TASMAN

SEA

SYDNEY

MASCOT

PENRITH

NW

SW

EAST

PEAK 1-HOUR OZONE (Monitoring station)

0

50

100

150

200

0 50 100 150 200OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

144.2 144.4 144.6 144.8 145.0 145.2 145.4LONGITUDE (deg. E)

-38.4

-38.2

-38.0

-37.8

-37.6

LATI

TUD

E (d

eg. N

)

ALPH

BXHL MTCL

PTCK

PAIS

BRTN

RMIT

GSTH GRVD

PORT PHILLIPBAY

BASS STRAIT

MELBOURNE

GEELONG

MELBOURNEMELBOURNE-- Daily 1Daily 1--hour Ohour O33 maxmax

REGIONAL FORECASTINGREGIONAL FORECASTING

REGIONAL

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

MELBOURNEMELBOURNE-- Daily 1Daily 1--hour Ohour O33 maxmax

SUBSUB--REGIONAL FORECASTINGREGIONAL FORECASTING

144.2 144.4 144.6 144.8 145.0 145.2 145.4LONGITUDE (deg. E)

-38.4

-38.2

-38.0

-37.8

-37.6

LATI

TUD

E (d

eg. N

)

ALPH

BXHL MTCL

PTCK

PAIS

BRTN

RMIT

GSTH GRVD

PORT PHILLIPBAY

BASS STRAIT

MELBOURNE

GEELONG

EASTWEST

GEELONG

EAST

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

WEST

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

GEELONG

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

MELBOURNEMELBOURNE-- Daily 1Daily 1--hour Ohour O33 maxmax

144.2 144.4 144.6 144.8 145.0 145.2 145.4LONGITUDE (deg. E)

-38.4

-38.2

-38.0

-37.8

-37.6

LATI

TUD

E (d

eg. N

)

ALPH

BXHL MTCL

PTCK

PAIS

BRTN

RMIT

GSTH GRVD

PORT PHILLIPBAY

BASS STRAIT

MELBOURNE

GEELONG

SUBURBSUBURB--LEVEL FORECASTINGLEVEL FORECASTING

PEAK 1-HOUR OZONE (Suburb)

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

PEAK 1-HOUR OZONE (Monitoring station)

0

20

40

60

80

100

0 20 40 60 80 100OBSERVED CONCENTRATION (ppb)

FOR

ECA

ST C

ON

CEN

TRA

TIO

N (p

pb)

MELBOURNEMELBOURNE-- Daily 1Daily 1--hour Ohour O33 maxmax

144.2 144.4 144.6 144.8 145.0 145.2 145.4LONGITUDE (deg. E)

-38.4

-38.2

-38.0

-37.8

-37.6

LATI

TUD

E (d

eg. N

)

ALPH

BXHL MTCL

PTCK

PAIS

BRTN

RMIT

GSTH GRVD

PORT PHILLIPBAY

BASS STRAIT

MELBOURNE

GEELONG

STATIONSTATION--LEVEL FORECASTINGLEVEL FORECASTING

OO33>50 ppb>50 ppb

PERFORMANCE INDICESPERFORMANCE INDICES ((OO33 >> 50 ppb50 ppb))SYDNEY- NORMALISED BIAS

-30

-15

0

15

Regional Sub-regional

Suburb Station

Perc

enta

ge

SYDNEY- NORMALISED GROSS ERROR

0

10

20

30

Regional Sub-regional

Suburb Station

Perc

enta

ge

MELBOURNE- NORMALISED BIAS

-30

-15

0

15

Regional Sub-regional

Suburb Station

Perc

enta

ge

MELBOURNE- NORMALISED GROSS ERROR

0

10

20

30

Regional Sub-regional

Suburb Station

Perc

enta

ge

Detected = Detected = d/(d+md/(d+m) ) —— correct forecastscorrect forecastsFalse alarm = False alarm = f/(f+df/(f+d) ) —— missed events missed events

M ≥ O Observed

Model Yes No Total

Yes d f f+d

No m b m+b

Total d+m f+b

AAQFSAAQFS-- PERFORMANCEPERFORMANCE

Contingency TableContingency Table

SYDNEYSYDNEYOZONE DETECTION RATE- SYDNEY

0

20

40

60

80

100

40 60 80 100Exceedance concentration (ppb)

Rat

e (%

)

RegionSub-regionSuburb

OZONE FALSE ALARM RATE- SYDNEY

0

20

40

60

80

100

40 60 80 100Exceedance concentration (ppb)

Rat

e (%

)

RegionSub-regionSuburb

OZONE DETECTION RATE- MELBOURNE

0

20

40

60

80

100

40 60 80 100Exceedance concentration (ppb)

Rat

e (%

)

RegionSub-regionSuburb

OZONE FALSE ALARM RATE- MELBOURNE

0

20

40

60

80

100

40 60 80 100Exceedance concentration (ppb)

Rat

e (%

)

RegionSub-regionSuburb

MELBOURNEMELBOURNE

AAQFS vs. PERSISTENCEAAQFS vs. PERSISTENCE(Sub(Sub--regional)regional)

DETECTION RATE

0

1

2

3

4

40 60 80 100Exceedance concentration (ppb)

AAQ

FS /

PER

SIST

ENC

E SydneyMelbourne

FALSE ALARM RATE

0.0

0.5

1.0

1.5

2.0

40 60 80 100Exceedance concentration (ppb)

AAQ

FS /

PER

SIST

ENC

E SydneyMelbourne

TIME

DO

SAG

E

TWOTWO--SCENARIO FORECASTSCENARIO FORECAST

Why is AAQFS special?

Provides twiceProvides twice--daily forecasts of AQI and daily forecasts of AQI and 18 pollutants for 18 pollutants for EPAsEPAsShows the daily development of pollution Shows the daily development of pollution (highly instructive/other applications)(highly instructive/other applications)Because prognostic, unusual events handledBecause prognostic, unusual events handledCan explore results with offCan explore results with off--line toolsline toolsAlternative scenarios, special locationsAlternative scenarios, special locations

Why is AAQFS unique?

Resolution: 5 km Met, 1 km AQResolution: 5 km Met, 1 km AQIntegral verification of yesterday’s forecastIntegral verification of yesterday’s forecastProven operational within a NMHSProven operational within a NMHSResponsive to daily changes in Responsive to daily changes in EPAEPA--supplied emissions datasupplied emissions dataMinimal demand on resources by Minimal demand on resources by EPAsEPAs‘Green’ scenarios can be run on special days‘Green’ scenarios can be run on special days

Experience with Generating Emissions Inventories More…

PopulationPopulation--based emissions can produce based emissions can produce quite acceptable results (no industry)quite acceptable results (no industry)Industry must be treated explicitlyIndustry must be treated explicitlySimple Simple biogenicsbiogenics scheme works wellscheme works wellPollution inflows may be much more Pollution inflows may be much more important in other national settingsimportant in other national settingsSeasonal, diurnal and weatherSeasonal, diurnal and weather--related related emissions changes should be done emissions changes should be done inin--line for simplicity and speed.line for simplicity and speed.

Prognostic Model + Emissions Inventory → Applications More…

StandStand--alone applications useful (TAPM)alone applications useful (TAPM)Air quality management planning/scenariosAir quality management planning/scenariosMonitoring network designingMonitoring network designingIndustrial complex emissions managementIndustrial complex emissions managementSurveillance of urban pollution emissionsSurveillance of urban pollution emissionsUrban design applicationsUrban design applicationsAssessments of transport options/technologiesAssessments of transport options/technologiesAirshed emissions taxes/tradingAirshed emissions taxes/tradingWindWind--Power prospecting!Power prospecting!

AAQFS Experiences More…

Emissions Inventories Emissions Inventories –– our biggest problemour biggest problemWindWind--blown dust, other particle sources, blown dust, other particle sources, difficult difficult –– effort by Met. Service of Canada effort by Met. Service of Canada commendablecommendableTimeliness and quality of air pollution Timeliness and quality of air pollution monitoring data is vital for warm starts monitoring data is vital for warm starts (assimilation)(assimilation)GRS chemistry adequacyGRS chemistry adequacyRoutine verification of forecastsRoutine verification of forecastsBackgrounds/domainBackgrounds/domain--size issuesize issueCooperation between Agencies is importantCooperation between Agencies is importantUptake by others is slow Uptake by others is slow –– patience!patience!