lecture2-lauritzen - 图文 

AtmosphereModelingI:Intro&DynamicstheCAM(CommunityAtmosphereModel)FV(FiniteVolume)andSE(Spectralelement)dynamicalcoresPeterHjortLauritzen

AtmosphericModeling&PredictabilitySection(AMP)ClimateandGlobalDynamicsLaboratory(CGD)NationalCenterforAtmosphericResearch(NCAR)asi-uniformglobalsphericalgridsconsideredfornextgenerationmodereg.lat-loncubed-sphereVoronoiYin-Yang)noNCARglobalcommunicationissponsoredbytheNationalScienceFoundation(NSF)CESM1 F1850, ATM component, BGPPeterHjortLauritzen(NCAR)AtmosphereModelingI:Intro&DynamicsAugust14,20171/361

Atmosphereintro

Multi-scalenatureofatmospheredynamicsResolvedandun-resolvedscales

‘De?ne’dynamicalcoreandparameterizations

CAM-FVdynamicalcore(current‘workhorse’dynamicalcore)HorizontalandverticalgridEquationsofmotion

TheLinandRood(1996)advectionscheme

Finite-volumediscretizationoftheequationsofmotionThe‘CD’gridapproachVerticalremappingTracers

CESMsimplermodels(focusonCAM)

OtherdynamicalcoreoptionsinCAM

CAM-EUL(Eulerian):Basedonsphericalharmonicfunctions

CAM-SE(Spectral-Elements):DefaultdynamicalcoreinCAMforhigh(hydrostatic)horizontalresolutionapplications

2

3

4

PeterHjortLauritzen(NCAR)AtmosphereModelingI:Intro&DynamicsAugust14,20172/36

Domain

Source:NASAEarthObservatory

PeterHjortLauritzen(NCAR)AtmosphereModelingI:Intro&DynamicsAugust14,20173/36

HorizontalcomputationalspaceRedlines:regularlatitude-longitudegridGrid-cellsizede?nesthesmallestscalethatcanberesolved(=e?ectiveresolution!)Manyimportantprocessestakingplacesub-grid-scalethatmustbeparameterizedLooselyspeaking,theparameterizationscomputegrid-cellaveragetendenciesduetosub-grid-scaleprocessesintermsofthe(resolvedscale)atmosphericstateInmodelingjargonparameterizationsarealsoreferredtoasphysics(whatisunphysicalaboutresolvedscaledynamics?)PeterHjortLauritzen(NCAR)AtmosphereModelingI:Intro&DynamicsAugust14,20173/36E?ectiveresolution:smallestscale(highestwave-numberk=ke?)that

modelcanaccuratelyrepresent

ke?canbeassessedanalyticallyforlinearizedequations(VonNeumannanalysis)Inafullmodelonecanassesske?usingtotalkineticenergyspectra(TKE)of,e.g.,horizontalwind??v(seeFigurebelow)E?ectiveresolutionistypically4-10grid-lengthsdependingonnumericalmethod!Spectral (Eulerian) Coreeffective resolution?FV coreeffective resolution lo g en e r gy de n s it y kinetic energy (m2/s2) 2 Δ x wavelength correct spectrum kinetic energy (m2/s2) 2 Δx wavelength correct spectrumno divergencedampingmodel spectra model spectrum (short wavelengths aliased to longer wavelengths) log k spherical wavenumber (n)spherical wavenumber (n)k!FigurefromSkamarock(2011):(left)Schematicdepictingthepossiblebehaviorofspectraltailsderivedfrommodelforecasts.(right)TKE(solidlines)asafunctionofsphericalwavenumberfortheCCSM?nite-volumedynamicalcorederivedfromaquaplanetsimulations.ThetotalKEisbrokenintodivergent

androtationalcomponents(dashedlines)andthesolidblacklinesshowsthek?3slope.

PeterHjortLauritzen(NCAR)

AtmosphereModelingI:Intro&Dynamics

August14,2017

3/36