www.umbc.edu

DeterminationofMixingLayerHeightandASOS:

Testbed,AlgorithmsandNetwork

RubenDelgado,BelayDemozAtmosphericLidarGroup

JointCenterforEarthSystemsTechnologyUniversityofMaryland,BaltimoreCounty

InternationalCooperativeforAerosolPrediction(ICAP)8th WorkingGroupMeeting:

LidarDataandItsUseinModelVerificationandDataAssimilationJuly13,2016

www.umbc.edu

•Inversionalgorithms,optical,chemicalandphysicalpropertiesofatmosphericaerosols,gases,andclouds.

•BoundaryLayerDynamics(AirQualityandWindEnergy)

•ContinentalandintercontinentalplumetransporttoEasternUSandCaribbean.

•AOD-PM2.5EstimatorDevelopmentfromGround,SatelliteObservations,NWFandGlobalModels

•Newremotesensingtechnologiesforatmosphericobservations.

UMBCAtmosphericLidarGroupResearchAreas

www.umbc.edu

MixingLayerHeight(MLH)

• Diagnosticvariableatmospherictransportanddispersionforecastingmodels.

• WithoutrealisticMLHmodelshavelargeerrorsthatresultininadequatepublicprotectionagainstunhealthyairquality.

• NationalResearchCouncilhasrecommendeda“networkofnetworks”1• After60yearsofremotesensingresearch,itisastoundingthatthePBLisnotmeasuredregularlythroughoutitsdiurnalcycle

1- NRC.2009.ObservingWeatherandClimatefromtheGroundUp:ANationwideNetworkofNetworks.Washington,DC:NationalAcademyPress.

www.umbc.edu

June2015CanadianSmokeEvent

Elastic lidar backscatter image shows aerosolsaloft (1.5-3 km) on June 10th. The particles beganto mix causing increased near surface particlepollution.

The 11th shows a homogenous layer, smokemixed with the mixing layer which extends up to1.5 km.

UMBCSmogBlog:http://alg.umbc.edu/usaq

www.umbc.edu

MLHAlgorithms

• LidarandwindprofilersMLHcanprovidecontinuoustemporalresolutionatmosphericprofilesforverificationandvalidationofforecastsandmodels,onwhetherthephysicsanddynamicspackagesarecorrectinmodels.

*Compton etal.(2013),J.Atmos.Ocean.Tech.,doi:10.1175/JTECHD-12-00116.1

www.umbc.edu

JointNOAA/ARLNOAA/NCEPFieldStudy- September,2009

1-Develop an urban meteorological evaluation database to investigate theevolution and spatial variability of the urban atmospheric boundary layer mixingheight.

2-Evaluate various instrument platforms for detecting mixed layer height.

3- Accurate assessment of boundary layer information at finer scales shouldimprove the Nation's ability to assess the effects of a toxic release (in support toHomeland Security).

*Project supported demonstration of NOAA's Real-Time Mesoscale Analysis(RTMA) of PBL information for use by plume dispersion modelers.

www.umbc.edu

JointNOAA/ARLNOAA/NCEPFieldStudy

LidarmeasurementshelpedtoidentifyproblemswithautomaticPBLHcalculationfromaircraftprofiles(ACARS).

www.umbc.edu

DISCOVERAQSummer2011@Beltsville,MD

CL51Ceilometer MicroPulse Lidar(MPL) HowardRamanLidar

Laser:InGaAsPower:8.9mWWavelength:910nmAlgorithm:BL-View

Laser:Nd-YLFPower:25mWWavelength:532nmAlgorithm:Wavelet

Laser:Nd-YAGPower:8WWavelength:355nmAlgorithm:Wavelet

AlgorithmComparison

www.umbc.edu

RecommendationofCeilometerPBLHeightsforAssimilation/VerificationofForecastProducts

UMBC: Belay Demoz, Ruben Delgado, Kevin Veermesch; Howard University: RicardoSakai; NWS: Dennis Atkinson, Michael Hicks, Jason Chasse (Program Manager NextGenAviation Weather at NOAA/NWS/ OS&T)

• UMBC algorithm being used to retrieve MLH from the NWS Vaisala’s CL31 ceilometers, as partof a Proof of Concept CL31 Test bed.

• The algorithm development for MLH from CL31 ceilometers to be implemented at nationwideASOS sites, as support of scientific efforts of the NWS Sterling Field Support Center.

www.umbc.edu

DeterminationofPlanetaryBoundaryLayerHeightwithDopplerWindLidar

QinLiu1,BrianCarroll1,ThomasRieutord2,AlanBrewer3,AdityaChoukulkar3,RubenDelgado1

1UniversityofMaryland,BaltimoreCounty,Baltimore2Météo-France,Toulouse,France

3National Oceanic and Atmospheric Administration

• CollaborationwithNOAAESRL.• Thepurposeofthisstudyistoevaluatetheplanetaryboundarylayer

heightretrievalsfromDopplerwindlidars.• Analysiswasappliedtodatacollectedfromthetwolidar systems

duringtheJuly-August2014DiscoverAQandLUMEXcampaigns.• Thiscomparisonaidsapplicationsinairqualityandwindenergy

forecasting.

www.umbc.edu

PeakDetectionMethodUsingHaarWaveletTransform

BowtieandVerticalScanrange-correctedintensityprofilesandhorizontalwindspeedanddirection.

www.umbc.edu

K-meansAlgorithm• Initializetheseeds(clusters)• Calculatethedistancefromeachpointtoeachcluster• Assigneachpointtotheclosestcluster• Redefinetheclustersasthecentroidofpointsassigned• Repeattheprocessuntiltheintra-clustervariancenolonger

decreases

InitialConditions• Twoclustersused,assigntophalfoftheprofiletooneclusterand

lowerhalftotheothercluster

ClusterAnalysis

www.umbc.edu

ConvergentTest

• CalculateEuclideandistancebetweeneachpointtotheclusterandintra-clustervariance

• Thealgorithmstopswhentheintra-clustervariancearenolongerdecreasing• TheMLHisdefinedastheheightwheretheclustertransitions

www.umbc.edu

BadGood

• Pros:fasterthanrandomseeding,resultsareconsistenteverytime• Cons:resultisnotaccurateifthere’smissingdatainonesingleprofile• ValidationandSensitivityofAlgorithmiscurrentlyevaluatedwithPECAN

Elastic,Raman,DopplerLidar,MicrowaveRadiometerandSoundingsdatasets

ClusterAnalysis

www.umbc.edu

Update:SavingCeilometerDatafromthe

AutomatedSurfaceObservingSystem(ASOS)

www.umbc.edu

Ceilometers!!!

ThermodynamicProfilingTechnologiesWorkshop12-14April,2011

NRCstudy:ObservingWeatherandClimatefromtheGroundUp:ANationwideNetworkofNetworks(2009)

Regional Testbed

ASOSData:Motivation

16

www.umbc.edu17

• GOAL:DescribehowASOSceilometerbackscatterdatawouldbeusedifNWScouldprovideit.

• WhatvaluewouldthedataprovidetotheNation?Neededtoanswerthefollowingquestions:

– WhatdataareavailablefromtheCL31?– Whatisthequalityofthedata?– Howoftenisthedataavailable?– Howwouldthedatabesavedwithoutoperationalinterference.

• Listtheavailableapplicationsforbackscatterdata• Describetheresearchthatisunderwayorrequired• Listchallengesforresearch-to-operations(RTO)• Chartacourseofactiontoachievegoals

ASOSCeilometerWorkshop:NWS/Sterling,VA;March22,2012

www.umbc.edu

Chet Schmitt (OPS 22): ASOS Ceilometer Workshop, NWS/Sterling, VA. March 22, 2012

ASOSCeilometerSites

18

www.umbc.edu

ASOS(Instrumentation/Issues)

19

Problem-2:Limited bandwidth transmission to main frame ASOS computer CT12K CL31

Problem-1:Quality of the lidars Problem-3:

• Inertia• “Operational”!

ß Each ASOS

www.umbc.edu

StepsRequiredBeforeWeCanStart…ASOSCL31DataPollingatNWS- Sterling,VAStep1:CollectandevaluateCOTSceilometer’sprofiledatainalocalnetwork[Completed].

Step2:EvaluatemethodsofPollingASOSceilometersforprofiledatawithoutinterferingwithASOSfunctions

Ceilometerprofilesat1minresolutionwerecollectedformonthsusingadatalogger

20

Nointerferenceobservedthatcouldbetracedtotheinstallationofthedataloggerontheceilometer!

www.umbc.edu

TwoExamples:• PM-studies:ScalingSatellite-measuredAODandPM-Correlations• Nighttimeconvection: PECANexperiment[Elevatedstorm]

Morecasesstudies:•LimitednetworkofCeilometer: Baltimore-Washington-area-Network•CL31vsCT12VsCL51:Anexampleofcomparativedata•CL31datastatistics:Cloudbaseabove12000ft needstobereported•PBLstudy: PBLfromCL31:Multi-algorithmcomparison

MoreonAirQualityApplications•FireandAirquality:Thecaseof9-10June2015•Volcanicashmonitoring: HowcouldASOShelp?

21

CL31:CaseStudiesList

www.umbc.edu

UMBCMPLCL31

CHM 15k

HUCHM 15kCL31 &51

CT12

NWSCL31MPL

NCWCPCL31 ES

CT12

HUCT12

SERCCT12

GSFCMPL

SigmaMPL

22

CurrentCL31PBLProjectSitesMap• MLHprocessingassessmentandevaluationinprogress.CollaborationwithNASAMPLNeT (E.Welton andJ.Lewis)

• UMBCSmogBlog:MonitoringofMLandotherAQissues

• Ceilometerinter-comparisoninprogress• NWSengagementinASOS

www.umbc.edu

• First1-2signalreturnsintheCL-31profilecanbeusedtoinferPMloading.Clearsky,RH<62%used[Munkel etal.(2007):Vermeesch etal.(2011),others]

• “Correctandscale”satelliteopticaldepthmeasurementsforAQstudies[Chuetal.,2013- DISCOVER-AQsite;]– Lietal.2016usedCT25atBeltsvilleBeltsville,Maryland,Vermeeschetal.,2011

23

PossibleASOSuseinAQ

www.umbc.edu

PECAN:CL31NetworkDemonstrationforSevereStormResearch

24

CL31RS-232

DatacollectionRaspberryPI(Linux

board)Cellnetwork

DataProcessing,storageandvisualization

A 24/7 real-time access/op

HU - CL31

Other - CL31

• Plains Elevated Convection At Night http://catalog.eol.ucar.edu/pecanA multi-agency, multi-university field observation over Kansas to investigate the sources of nightime summer elevated convection.

• CL31 (4) were used for realtime network demonstration (see figure below)

• NWS SOO and field sites from Dodge City and Wichita, Kansas were collaborators and allowed siting of two of the CL31s.

• Data collection algorithm and electronics developed and tested.

www.umbc.edu

PECAN:3June2015BoreCase

Pictures taken at FP2

SPOL

DDCFP2 ICT

• Undular Bores are one of the suspected event thattransport moisture upward priming the nightimeatmosphere for destabilization and severe storms.

• An accurate statistics of occurrence and observation islacking, hence PECAN. This is bore case observed earlyon 3 June 2015, during PECAN.

• The CL31 network reveals the spatial evolution andduration of this bore.

www.umbc.edu

PECAN: 3June2015BoreCase

26

CL31 network data from PECAN – no operational instrument is capable of capturing this event in such detail. Equivalent ASOS data is plotted, showing data lost. Analysis of these data sets is ongoing in PECAN.

What we measure

What is saved by ASOS

www.umbc.edu

ASOSprojectMilestonesandFuture• CL31PBLProofofConceptcompleted• Managementapprovaltoproceed• DatacollectionfromASOSdemonstrated• CaseStudiesCompleted

– PBL,PECAN,Fireetc,(severestorm)– demonstrationnetworkcompleted

• Morecasestudy/dataanalysis• WorkingonWMOVolcanicAshexpertteam• BAMSpaperindraft

• AlgorithmAssessment/TestinginASOSOperationalEnvironmentComplete(plannedDecember2017)

• AlgorithmIncorporatedintoASOS*(plannedJune2018)*dependentuponASOSACU/DCPupgradecompletion

Com

plet

ed

InProgress

Futu

re

www.umbc.edu

EndofPresentation

Thankyou!

www.umbc.edu

Extra

www.umbc.edu

Instrumenttypes:CL31,CT12K,CL51

CL31

CT12

CL51

BACK

Currentinstrumentintest• Lufft:CHM15K(UMBC,HU)• Vaisala:CL51,CL31,CT12K

OtherLidars:• HU-RamanLidar• ALVICE(NASARaman)

PlantoworkwithMPLnet (J.Welton onPBLandsuch)

www.umbc.edu

• CL31datastatistics:• Cloudheightstatisticsabove12Kft needstobereported

~30%- Not reported

31

CloudStatistics@12Kft+

BACK

www.umbc.edu

GoodComparisonDay:SuccessinmultiplelayerswithQIof3.

BadComparisonDay:NotePBLHmissingafter1200UTC.

32

BL-view/UMBC)

Selectioncriteriaforcomparison?

www.umbc.edu

PBL(LCLasfiler)

33

GoodPBLComparisonDay:• ExcellentmorningPBLevolution• LCLfilter removesday-nighttransitionsissues

BaddayComparisonDay:• PBL– definitionissue• LCLfilterdoesn’thelp

AmongthePBLMethodscompared:Liu&Liang,Hicksetal2015;Waveletbased;RiB – Richardsonnumber,HeffterBL-View(Vaisala)

“Bad Day” “Good Day”

www.umbc.edu

Summary:CL31comparisontableRoutine Pros Cons CommentsHicks •GoodformorningPBL

•LCLfilterhelpfulinpruning• NWSorigin

•Day-Nighttransachallenge•LCLfilterremoveselevatedNBL

PublishedinBLMHicksetal.,2015;Combinessomeoftheerror-function,Meteorology,andcanrunonarchive.

UMBC •PerformanceasHicksetal.;•Comparedtoradar-SNR•Comparedtoothers

• Day-Nighttransachallenge PublishedinComptonetal.,2013.localsourceandsimilartoHicksetal.AlsousedbyMDEetc.

BL-View • Runsinrealtimenow• welltested/robust(NWSSeattle,Vancouver,EU,etc)• designedfortheceilometer

•1-softwareto1-instrument• Notnetworkcapable• limitstheprofileto4.5km

Severalpapers.Commercialbacking.Costly,inrelativeterms,unlessnegotiated.

Recommendation:• AcombinationofUMBC/Hicksmethodsbeusedoncurrentdata.• Alow-cost,networkcapable,commercialsoftwareisdesirable.• Abilitytoprocessesreal-timeaswellasarchivedataisdesired.

34BACK

http://so2.gsfc.nasa.gov/pix/special/2009/redoubt/redoubt_all.html

HowcouldASOShavehelpedinVolcanicashstudies?

CouldtheCL31haveseentheash?

Stepsused:• OMI/CALIPSOforplumeboundary

• Estimateaerosol“loading”abovebackground.

• LocateifwithinCL31range

• Speculateifitwouldhavebeendetectedandmeasured.

OMI:ColumnSO2

Redoubt– PlumeofMarch23,2009

www.umbc.edu

CL31:FireandSmokepollutionhttp://alg.umbc.edu/usaq/

Black Carbon at the Beltsville site over the week leading up to and through the 6/11 event. Joel Dreessen -MDE-<joel.dreessen@maryland.gov

Redoubt:Lostopportunity

• Plumeof03/23/2009wouldhavebeendetectedbyASOSlidars

• WouldhaveassistedNWS-Alaskaregioninmonitoring.

BACK

DISCOVERAQSummer2011@Beltsville,MD

CL51Ceilometer MicroPulseLidar(MPL) HowardRamanLidar

Laser:InGaAsPower:8.9mWWavelength:910nmAlgorithm:BL-View

Laser:Nd-YLFPower:25mWWavelength:532nmAlgorithm:Wavelet

Laser:Nd-YAGPower:8WWavelength:355nmAlgorithm:Wavelet

AlgorithmComparison

CL31PBLReview

Summary:BLView avgdiff(m)

Hicksavgdiff(m)

UMBCavgdiff(m)

Method

STAB

LE

LiuLiang 710.327 12.066 1221.472RiB 770.776 173.336 1158.482Heffter 761.597 54.004 1243.862AVERAGE 768.476 123.15 1236.075

CONVE

CTIVE LiuLiang 196.37 -604.388 ****

RiB 527.195 -175.566 ****Heffter -299.849 -1151.683 ****AVERAGE 227.076 -732.185 ****

Lidar-sonde differences:

• BL-View data close to the sonde is chosen from the 3-choices.

• UMBC/Hicks lowest reported is chosen

• Explore smarter method for comparing algorithms

NB: PBLH comparisons under ideal conditions – revealed +/- 200m (IHOP2002)

39

Peak-basedThreshold

TheMLHisdefinedasthehighestpointconnectedtothegroundintheprofile