Temporal scale analysis of AIRPACT5 Performance for O3 and PM2.5 in the Pacific Northwest

Tsengel Nergui, Serena Chung*, Yunha Lee, Joseph Vaughan, and Brian Lamb

Laboratory for Atmospheric Research, Washington State University*U.S. Environmental Protection Agency

NW-AIRQUEST Annual Meeting, June 14-16, 2017, Richland, WA

2

OperationalEvaluationfor2016:Summer(49sites):Goodagreement(MFBs>± 30%at11sites)Winter(28sites):Overestimation(MFBs>± 30%at21sites)

Ozone: Mean Fractional Error (MFE) and Mean Fractional Bias (MFB)

BenchmarkforO3:NME<±15%andNME<35%(aboutsameasMFB<±30%andMFE<50%)

3

PM2.5 : Mean Fractional Error (MFE) and Mean Fractional Bias (MFB)

OperationalEvaluationfor2016:Summer(103sites):Underestimationatmostsites(~60%MFBs)Winter(129sites):Overestimatedinurban(~70%MFBs),underestimatedinruralareas(~60%MFBs)

BenchmarkforPM2.5:MFB<±30%andMFE<50%

Objective

To assess the AIRPACT-5 ability for reproducing the important temporal scale components embedded in observed O3 and PM2.5 concentrations

4

DataandMethod

• HourlyO3 concentrations,~50AQSsites

• HourlyPM2.5 concentrations,~140AQSsites

• TheAIRPACT-5outputsfor2016

• Spectralanalysis(temporalscaleseparation)usingtheKolmogorov–Zurbenko(KZ)filtering

5

6

Analyzingatimeseriesindistincttemporalscales

Kolmogorov–Zurbenko(KZ)filter(Zurbenko,1986)

Atimeseries=Mean+Varioustemporalscalefluctuations+Trend

7

Example:O3 timeseriesdecomposition(Site#530090013,RURAL/Forest,WA)

Baselin

e(>21

days)

— Seasonalvariationofsolarradiation

— Depositionduetochangesinsurfaceproperties

— Slowchangingprecursors’emissions

Syno

ptic

(3-21days) — Changesinweatherconditions

(stagnanthighpressuresystem,frontalpassage)

— Associatedchangesinmixingheights,cloudcover,andcirculationpatterns

Diurna

l(11-36

hrs) — Diurnalpatternofsolarradiation

— Differencebetweendaytimeproduction&nighttimeremovalfor O3

Intrad

ay(<11

hrs) —Convectivemixing

— Localemissionschanges— Photolysisratesassociatedwith

theactinicflux

8

Variance Contributions Component Correlations(7sites)

(25sites)

Summertime O3:

9

Findings for summertime O3:

• VariancecontributionsforobservedO3:― Diurnal(~67%),Synoptic(~15%),Baseline(~12%),and

Intraday(~5%)components.― Diurnalcomponentishigherinurbanvs.ruralareas(72%vs.

64%).― Thebaseline/synopticcomponentstendtobehigherinruralvs.

urbanareas(18%vs.12%).• Modelunderestimatedvariancecontributionfordiurnal(~16%less)

andoverestimatedforsynoptic(~15%more)andbaseline(~6%more)components.

• DiurnalcomponentwasthebestcorrelatedbecauseofinherentcyclicalnatureofthediurnalprocessforO3production/destruction.

• Correlationsatthebaselinescalearebetterthanthoseforsynopticandintradaytimescales,butvarybysite(medianr =0.6,MAD= 0.3).

HOURLYPM2.5CONCENTRATIONS(88101:FRM/FEM~40SITES)(88502:NON-FRM/FEM~100SITES)

10

11

Variance Contributions Component Correlations(45sites)

(39sites)

Wintertime PM2.5:

12

• VariancecontributionsforobservedPM2.5:― Diurnal(~30%),baseline(~26%),synoptic(~25%),and

Intraday(~15%)components.― Contributionofintradaycomponentincreasesfromurbanto

ruralsites(13%to18%).• Modeltendstogiveslightlyhighermeanforurban/suburbansites

andlowerforruralsites.• Themodelunderestimatedvariancesoftheintraday,diurnal,and

baselinecomponents(~10%lessforeach)andoverestimatedforsynopticcomponent(~7%more).

• Correlationsbetweenmodeledvs.observedcomponentsincreasefromdiurnaltobaseline,butdisplayalargevariability(medianR=0.2-0.8,MAD= 0.3).

Findings for wintertime PM2.5:

Spectral analysis shows…• The diurnal and baseline components

are the most important temporal scales for O3 and PM2.5 concentrations.

• Improving the baseline component:

13

Baseline (>21 days)— Seasonal variation of solar radiation— Deposition due to changes in surface

properties— Seasonal allocation of O3 precursors’

emissions

Variancecontributionsfromdifferenttemporalscalesbasedonentiretimeseries

What next?

THANKYOU.

FEEDBACKSANDQUESTIONS?

14