lasso modeling and measurements updatemay 2015 pilot project began june 2016 iniaal shcu simulaaons...
TRANSCRIPT
LASSOmodeling&measurementsupdateLESARMSymbio:cSimula:on&Observa:onWorkflow
WilliamIGustafsonJr.(PI)1,AndyVogelmann(Co-PI)2,ZhijinLi3,4,
XiaopingCheng3,SatoshiEndo2,TamiToto2,HengXiao1
JenniferComstock1
1PNNL,2BNL,3UCLA,4JPL
LASSOWebpage:h"p://www.arm.gov/science/themes/lasso
LASSOe-maillistsignup:h"p://eepurl.com/bCS8s5
TheLASSOVision
n BridgingthescalegapbetweenARMobservaAonsandmodels
n UseLEStoaddvaluetoARMobservaAonsn Self-consistentrepresentaAonoftheatmosphere
n ProvideadynamicalcontextfortheobservaAons
n Elucidateunobservableprocesses&properAes
n GenerateasimulaAonlibraryforresearchersn EnablestaAsAcalapproachesbeyondsingle-casementality
n ProvidetoolsnecessaryformodelerstoreproducetheLES
WhatcanyoudowithanObs+LESlibrary?
n AsanobservaAonalistn Informinstrumentremotesensingretrievalsn ConductObservaAonSystemSimulaAonExperiments(OSSEs)n TestimplicaAonsofradarscanstrategiesorflightpaths
n AsatheoreAciann GetesAmatesoffluxes&co-variabilityofvaluesn TestrelaAonshipsw/ohavingtorunthemodelyourself
n Asamodelern KnowaheadofAmewhichdayshavegoodforcingn Haveco-registeredobservaAonsathigh-resoluAonscalesn HaveinputsandcorrespondingoutputstotestparameterizaAons
Learningtowalkbeforewerun
n IniAalimplementaAonisaproof-of-conceptandtargetsshallowconvecAonatSGPn DeveloptheARMinfrastructurefordoingrouAneLES
n TestinterfacesandvisualizaAonmethodologiesforaccessingtheLASSOlibrary
n Long-rangevisionistolaterexpandtomulApleARMlocaAonsandaddiAonalcloudregimes
TheroadtoLESatSGP
GoalofrunningrouAneLESforshallow
convecAonin2017
The2-yrLASSOpilotproject
n Modularstructureforflexibilityandfuturegrowth
n Majorfocusonge[ngfuncAonalityinplaceandimprovingitoverAmen ImplemenAngmethods
thatalreadyworkn NoAmetoexperiment
withuntestedmethodologies
SGPMegasiteSupportsIntegratedMeasurements-ModelStrategy
n BoundaryLayerProfilingModules(4locaAons)n UpgradestoAOSandRamanLidarInstrumentsn Improvedsoilmoisturenetwork(17locaAons)n PlanstodeploymulAplescanningcloudradars
SGPMegasiteConcept:anetworkofspaAallydistributedfaciliAesdesignedtocharacterizetheheterogeneityinatmosphericandlandproperAesaroundtheSGPsite.ThisnetworkaddressesarangeofsciencethemesusinganobservaAonsandmodelingstrategy.
MeasurementEnhancements:
TheSouthernGreatPlainsMegasite
Service Layer Credits:
!(
!(
!(
!(
%%%
$+
$+
$+$+
E9 -
Ashton,
KS. 386m
E11 -
Byron,
OK. 360mE12 -
Pawhuska,
OK. 331m
E13 - Central
Facility. 318m
E15 -
Ringwood,
OK. 418m
E31 -
Anthony,
KS. 412m
E32 -
Medford,
OK. 328m
E33 -
Newkirk,
OK. 357m
E34 - Maple
City,
KS. 417m
E35 -
Tryon,
OK. 294m
E36 -
Marshall,
OK. 340m
E37 -
Waukomis,
OK. 379m
E38 -
Omega,
OK. 367m
E39 -
Morrison,
OK. 279m E40 -
Pawnee,
OK. 247m
E41 -
Peckham.
341m
IF4 -
Billings, OK
IF5 -
Garber, OK
IF6 - Deer
Creek, OK
IF8 -
Tonkawa, OK
IF9 -
Billings, OK
IF10 -
Lamont, OK
IF7 -
Nardin, OK
IF12
IF11
¹
$+ IF Locations (Precipitation Radar)
% IF11/12 (Scanning Cloud Radar)
!( EF/Boundary Layer Profiling Site
! EF Locations
! IF Locations
70 km Radius Around CF
0 4020
km
167 - 200
201 - 220
221 - 240
241 - 260
261 - 280
281 - 300
301 - 320
321 - 340
341 - 360
361 - 380
381 - 400+
2011 USGS 10m DEMElevation in meters.
25km
75km
Boundarylayerprofilingsites
9
ImprovedRetrievalswithUncertaintyEs:mates
MWR3C
AERI DopplerWindLidar
MWR3CCadedduetal.
LWP
PWV
AERIoeTurneretal.
Temp
WVMR
DLPROFWINDNewsometal.
WindSpeed&Direc:on
ECOR
Latent&SensibleHeat
SKYRAD
Radia:veFluxes
EnhanceknowledgeofspaAalheterogeneityofstatevariables,fluxes,andsoilproperAes.
Situa:onally-dependentLES–not24/7
n SimulaAonsdonewhencondiAonsaremetn “RouAne”insteadof“operaAonal”n 1–2monthlagAmetoaccountforneededVAPsandcomputeAme
n ConAngentonavailableobservaAons&presenceofShCun NewcloudclassificaAonschemebySunnyLim
Low clouds Congestus Deep Conv. Altocumulus Altostratus Cirrostratus Cirrus
Ensembleofforcingdatasets
n Employforcingensemblestoaddressforcinguncertainty
n TesAng3forcingmethodsduringpilotphase
n ARMconstrainedvariaAonalanalysis(VARANAL)
n ECMWF-analysis–basedforcing
n MulA-scaledataassimilaAon(MS-DA)n BasedonWRFwithGSIusingtwodifferenterrorcovariancescales
n IniAallyusing3D-VarDAandplantotesthybridEnKFDA
Mul:-scaledataassimila:on
n Startwithagriddedbackgroundfield,e.g.FNLn IncorporateARMT,q,andwindprofilesfromboundary/intermediate
faciliAesn Providesa3-Dvolumeevery3hoursforgeneraAngLESforcing
ARSCLCloudFrac:onCloudFrac:onfromΔx=2kmMS-DAAveragedover25km Averagedover75km
Averagedover150km Averagedover300km
Modelselec:on&configura:on
n EvaluaAngbothSAM&WRFduringpilotphase
n TesAngdoubly-periodic&nestedLESdomains
n IniAallytargeAng25-kmdomain,Δx=100m
n Currenttestswithmodeltopneartropopause
Databundles
n DatabundleswillprovidetheprimaryinterfaceforLASSOdatan ContainobservaAon&modeloutputincomparableforms
n AppropriateAmesampling
n Instrumentsimulators
n Includepre-computedquick-looks,metrics,anddiagnosAcstoassistdatadiscoveryandquickenuseranalysis
n Include3-Dmodelfields,profilestaAsAcs,model-basedbudgetterms,andGCMparameterizaAonterms.
n ForcingsandiniAalcondiAonsprovidedforuserstoconducttheirownsensiAvitystudies.
OObbss
.. && CC
aassee
CCllaass
ssiifificcaa
ttiioonn LLEESS IInnppuuttss
&& OOuuttppuuttss
MMeettrriiccss && DDiiaaggnnoossttiiccss
Databundlemetrics
n LESsimulaAonsassessedusingdiagnosAcsbasedonARMobservaAonsofcloudandenvironmentalproperAes
n E.g.,Timeseries,Regressionanalyses,Taylordiagrams,Heatmaps,andPhasespacerelaAonships
n SkillsscoresbasedonthediagnosAcswillquanAfythequalityofthesimulaAonsn e.g.,LWP,cloudfracAon,temperature,watervaporconcentraAon,
relaAvecloudforcing,verAcalvelocity,etc.
n Searchablemetricswillassistusersinfindingcasesofinterest.
Datadiscovery
n SearchonmetricvaluesandselectedcondiAonsn EvaluaAngwaystosearchbasedonmodel-observaAonagreementandcase
descriptors
n TesAngmethodstopermitrobustsearchcapabiliAesandon-the-flyeventcomparisonn E.g.,CassandraNoSQLdatabasen TieredlevelsofcomplexitywithuserneedsandsophisAcaAon
Dataaccess&sojwaretools
n Reproducibilityisparamountn Willprovideallworkflowsomware,modelcode,andconfiguraAonstoenableotherstoreproducetheruns
n DesigningworkflowtorunonbothARMandotherDOEcompuAngfaciliAes
n Goalofenablingeasierdataaccess/transferfromtheARMArchiveviaGlobus
LASSOTimeline
May2015 Pilotprojectbegan
June2016 IniAalShCusimulaAonsfromspring-summer2015madeavailable• Ensembleofforcings• LESsimulaAonsfromSAMandWRF(bulkmicrophysics)• ObservaAonsincomparableform• FirstcutatmetricsanddiagnosAcs
January2017 2ndbatchofShCusimulaAonsfromspring-summer2016• Willincludeinfluenceofboundaryfacilityprofiles• Bothbulkandspectral-binmicrophysicsversions
April2017 AddiAonaltestcasesforyear-roundshallowcloudcondiAonsBetasomwaresuiteRecommendedconfiguraAonsforongoingsimulaAons
May2017 PilotprojectoverandtransiAontorouAnesimulaAonmode
LearnmoreaboutLASSO
n Breakoutsession,Wed.7:30–9:30p.m.n Posters
n 145:Gustafson,TheLASSOWorkflowPilotProject
n 147:Endo,LASSOWorkflow:EnsembleforcingsandLESsensiAvity
n 146:Vogelmann,LASSOWorkflow:model-observaAon“datacubes”
n 148:Comstock,Boundarylayerprofilingmodules…
n 137:Lim,Developmentofcloud-typeclassificaAonalgorithms…
n 139:Kollias,RadarnetworkapproachtocharacterizeShCuatSGP
n 138:Krishna,Large-scaledataanalysisandvis.forARMusingNoSQL
n Website:h"p://www.arm.gov/science/themes/lasso
n E-maillist:h"p://eepurl.com/bCS8s5