Usingsatellite-derivedPM2.5 datasettoassistairpollutionmanagementinCalifornia

H-AQASTMember:MinghuiDiao(PI),FrankFreedman,SenChiao,AnaRivera*DepartmentofMeteorologyandClimateSciences;*DepartmentofGeography,SanJoseStateUniversity

Co-I:MohammadAl-Hamdan,UniversitiesSpaceResearchAssociation,NASAMarshallSpaceFlightCenter

Co-I:Akula Venkatram,DepartmentofMechanicalEngineering,UniversityofCaliforniaRiverside

StakeholderContact:Saffet Tanrikulu (BayAreaAirQualityManagementDistrict)CynthiaGarcia(CaliforniaAirResourceBoard)

Sang-Mi Lee(SouthCoastAirQualityManagementDistrict)

Collaborators:Meytar Sorek-Hamer&RobertChatfield(NASAARC)

SomeofthehighlightsAreviewpaperinpreparationforPM2.5

exposureestimatesUrban Transportation and Air Pollution1st Editionby Akula Venkatram (Author), Nico Schulte (Author)

ISBN-13: 978-0128115060ISBN-10: 0128115068

https://arcg.is/1XbzCy

AvisualizationwebsiteofsatelliteAODusingArcGISinterface

Averageof14daysin2016winter

SanJose

Acompositeanalysesof14daysin2016winterSanFranciscoBayAreaAverageofAOD:https://arcg.is/1XbzCy• AverageofMAIACAquaAODfields• NumberofDays:14(Jan27,Feb13,Feb22,Feb24,Feb

25,Feb29,Mar16,Nov2,Nov4,Nov8,Nov9,Nov13,Dec20,Dec29).

• NoHRRRwinds,onlyaverageAODimage.

SanFranciscoBayAreaaverageofAODArcGISvisualizationTestofindividualdaysinthewinterof2016November22016AQUA2:30PM:http://arcg.is/zGbTm

Selectioncriteria:(1) clearskies,(2) goodAODcoverage(3)500mb geopotentialheight≥576decameters

SFO

SanLeandro

SanFrancisco

UnionCity/Hayward

Oakland

Atwo-pageproposalsubmittedtoCaliforniaAirResourceBoardforAB617legislation

PredictedPM2.5 Quantile0.9over23days(preparedbyRobertChatfieldandMeytar Sorek-Hamer)

Figures prepared by Frank Freedman

PreparedBy:Dr.MohammadAl-HamdanUSRAatNASA/MSFCApril30,2018*DetailsaboutthesurfacingmethodsusedcanbefoundatAl-Hamdan etal.(2009,JAWMA;2014,Geocarto)

Sensitivityteststosurfacingalgorithms(1)TwomethodsforPM2.5 3-kmSurfacesAnnualMeanCompositein2016

PM2.5(ug/m3)

Inversedistanceweighted(IDW)

B-Splinemethod (2)Comparisonsoftwomethods

Validationstatisticsbasedon44non-FRMmonitorsindifferentlocationsthroughoutCalifornia

CorrelationCoefficient(R),RootMeanSquareError(RMSE),MeanError(ME)(Bias),andMeanAbsoluteError(MAE).

Comparisonsofthreecommonly-usedpubliclyavailablePM2.5 datasetsinthecontiguousU.S.

A B

DC

ArcGIS-generated county-level maps of PM2.5 in 2011

DiaoM.,T.Holloway,S.Choi,S.M.O’Neill,M.Z.Al-Hamdan,A.van Donkelaar,R.V.Martin,X.Jin,A.M.Fiore,D.K.Henze,F.Lacey,P.L.Kinney,F.Freedman,N.K.Larkin,Y.Zou,A.Vaidyanathan Methods,availability,andapplicationsofPM2.5 exposureestimatesderivedfromgroundmeasurements,models,andsatellitedatasets,inpreparation.

(1) CDCWONDERexhibitshigherPM2.5 andalargeregionalmaximumoverthecentralU.S.

(2) ForSouthernCalifornia,EPHTNshowsthehighestPM2.5 (over14μg/m3)

(3) DalhousieexhibitslowerPM2.5 overall,andismorespatiallyhomogeneousoverthewesternU.S.

CDC WONDER Tracking network

Dalhousie EPA monitor AirData

(Figure prepared by Grace Choi and Tracey Holloway)

StatisticaldistributionsofthreePM2.5 datasetsinthecontiguousUSin2011

DiaoM.,T.Holloway,S.Choi,S.M.O’Neill,M.Z.Al-Hamdan,A.van Donkelaar,R.V.Martin,X.Jin,A.M.Fiore,D.K.Henze,F.Lacey,P.L.Kinney,F.Freedman,N.K.Larkin,Y.Zou,A.Vaidyanathan Methods,availability,andapplicationsofPM2.5 exposureestimatesderivedfromgroundmeasurements,models,andsatellitedatasets,inpreparation.

(1) CDCWONDER:overallhighervalues

(2) Dalhousie:thelowestmeanvaluesofPM2.5 overall,andthelargeststandarddeviation

Figure prepared by Minghui Diao, Grace Choi and Tracey Holloway

Integrating Satellites into Health and Air Quality Management• 1. Satellite-derived PM2.5 grids

• Develop, apply and evaluate regression model for California;

• Construct AOD-PM2.5 surfaces for California for 2016 and 2017;

• Preparation of a review paper on PM2.5data availability, method and analyses, led by Minghui Diao

• 2. Visualization of satellite-derived PM2.5grids (mostly finished)• Develop visualization of MAIAC AOD

and derived PM2.5 on selected days (LA, Bay Area, Imperial Valley);

• Analysis and incorporation of HRRR wind fields.

• 3. Dispersion model simulation• Development and evaluation of the

dispersion model simulations

Tiger Team Participation• TT#1 led by Patrick Kinney

• Developed in a GIS at 1-km modeling grid that overlaps the MATES-IV modeling grid; Processed the remotely-sensed data of 2012, integrated into 1-km modeling grid;

• Deploy low-cost sensors in three Bay Area sites;

• MAIAC AOD and dispersion modeling analysis of PM10 and PM2.5 fields across Imperial Valley.

• TT#2 led by Susan O’Neill. Our group will contribute to the PM2.5 data derived from satellite data and use downscaling model to provide higher resolution data.

• 8+ academic talks; 5 stakeholders; research website on HAQAST project at SJSU:

• www.cloud-research.org

Year 2 Progress Update, PI Diao

ProjectOverview

Regression&SurfacingModelRegionalPM2.5=f(x,y)daily

Dailyaveragetimeseries

PM2.5(x,y,t)

FineScaleModelLocalPM2.5

=f(x,y)hourly

LocalPM2.5(x,y)(Daily)

2.DispersionModelingSystem

1.PM2.5Regression&Surfacing

DarkTarget(3-km)3.Blendfields

Dailyaveragehourlyresults

RegionalPM2.5(x,y)(Daily)

Meteorological&EmissionInputs

HRRRModel(NOAA– 3km)

CALTRANS/EMFAC/other

Meteorology(NLDAS– 12km)

PM2.5 Monitors(DailyEPAFRM)

MAIAC(1-km)4.Visualizationofthesatellite-basedPM2.5andAOD