EUROCS DEEP CONVECTI VE CLOUDS DI URNAL CYCLE of DEEP CONVECTI ON OVER LAND CNRM, ECMWF, LMD, Met Of f ice, NCAR & SMHI (Europe, Fr ance, Sweden, UK & US) Guichard & P et ch Beau, Belj aars, Chaboureau, Cheinet , Grabowski, Grandpeix, Grenier, J akob, J ones, Koehler, Laf ore, Piriou, Redelsperger, Royer, St irling, Tailleux, Tomasini EUROCS, 16 December 2002, Madrid
OBJECTI VES � t o document perf ormances of GCMs � t o design a f ramework t o address t he problem (a simple case study f or CRMs & SCMs) � t o analyse t he case, f indings, conclusions � t o improve paramet erisat ions & GCMs
Colin Jones Regional Climate Modelling most frequently occuring time of max precipitation in a diurnal cycle (June 10-July 31 1993, from hourly accumulations) 1 2 0 2 4 6 8 10 12 14 16 18 20 22 24 hour (local time) � t he model capt ures t he broad early-lat e evening max of rainf all � errors is in t he SE, could be relat ed t o t he proximit y of model boundaries
COMMON CRMs/ SCMs CASE STUDY 1 : an « observed case » to assess our models over land Sout hern Great Plains (US) GCSS WG4 Case3a � 4-day runs wit h deep convect ion � f orcings prescribed f rom observat ions (large scale advect ion, surf ace f luxes) 2 : building an « idealized case » to address the diurnal cycle of deep convection over land and its representation in models � because most event s of t he « observed case » not linked t o our aims (not designed f or t his purpose) � mot ivat ed by Bet t s & J akob (2002) error in t he diurnal cycle of deep convect ion: shared by short & long- t erm GCM runs reproduced in SCM runs
THE SI MULATI ONS : 6 SCMs & 4 CRMs model t ype lab (model name) part icipant s SCM CNRM (ARPEGE, Clim & WF) Beau, Grenier, Piriou Chaboureau, J akob, SCM ECMWF (I FS) Koehler, Becht old SCM LMD (LMDz) Tailleux SCM Met Of f ice (UM) Pet ch SCM SMHI (close t o HI RLAM) J ones CRM CNRM (mésoNH) Chaboureau & Tomasini CRM CNRM (comeNH) Guichard CRM Met Of f ice (UM) Pet ch CRM NCAR (UM) Grabowski
I n practice � Preparat ion of t he case � Def init ion of a «base list s» of relevant diagnost ics t imes series & t ime-height series of select ed f ields mean prof iles, convect ive f luxes, subgrid-scale moment s, Q1, Q2, radiat ive heat ing rat es, cloud f ract ion, cloud wat er… � CRMs : Lx ~ 500 km Dx ~ 250m t o 2km Dz ~ st ret ched 70-700m or less � SCMs : operat ional version � Closer lab-lab CRM-SCM direct collaborat ions, e.g. CNRM-ECMWF � Sharing t he work, e.g. f or CRMs, sensit ivit y st udies: UK Met Of f ice: spat ial resolut ion CNRM: sub-grid scale represent at ion
SI MULATI ON OF DEEP CONVECTI ON OVER LAND THE OBSERVED CASE � broad conclusions consist ent wit h Xu et al. (2002) & Xie et al. (2002) ( new t est f or more t han 50% of models) an example : comparison wit h obs, min-max envelope f or CRMs & SCMs min max � bet t er agreement & less scat t er among CRM result s t hat SCM ones
comparison CRMs & SCMs (paramet ers f or which direct observat ions are not available) � obviously room f or CRMs improvement s (« cold » microphysics) � however much more consist ency among CRMs t han SCMs � very weak convect ive downdraught s in several SCMs
comments Pet ch et al. f or CRMs (2002) import ance of t he resolut ion and of t he represent at ion of subgrid-scale processes � because the good representation of boundary layer processes is essential f or this issue hour SCMs & GCMs issues complex int eract ions among paramet erisat ions involved � why convection occurs or not ? and how ? � identif ying the major weaknesses to correct them in priority
similar t ype of f ramework as t he 1st one: THE I DEALI ZED � 27 Mai 1997 of GCSS case 3a repeat ed t wice CASE (prescribed large-scale adv. & surf ace heat f luxes) � st art ing in t he morning inst ead of t he evening night time rainf all event s t end t o occurs earlier in SCMs t han in CRMs local solar time (hour)
What are the main reasons accounting f or these dif f erences ? (beginning) precipitation development of large CAPE CAPE during dayt ime associat ed wit h boundary layer evolut ion by design, st rong link wit h CAPE in many paramet erizat ions (CAPE closure) julian day (1997) Not much account of CI N 20 days in exist ing f ormulat ions precipitation atmospheric prof ile ~ 10 km thermal equilibrium level pseudo- adiabat CIN CAPE heigth dry ~ 1 km adiabat f ree convection level f ree convection level values derived f rom observat ions CI N condensation level (Xie et al. 2001) température Particule P adapt ed f r om Roux (1991)
Convective I nhibition (CI N) large f luct uat ions linked t o t he surf ace but in SCMs in CRMs I n CRMs, signif icant increase of CI N associat ed wit h deep convect ion in SCMs, more largely cont rolled by t he diurnal cycle of t he surf ace t he impact of convect ion on CI N is t oo weak
What are the main reasons accounting f or these dif f erences ? (cont inue)
What are the main reasons accounting f or these dif f erences ? (cont inue) snapshots of cloud + rain water content in CRM run transition regime in CRMs, (UK UM CRM, Petch) corresponding t o t he build up of convect ion: a f eat ure a « shallow » non-precipitating transition period which « broadly coherent » wit h last one to a few hours in CRMs … several previous observed st udies total …before the development of deep convection clouds 15 km cold clouds Lx: 300 km which f actors control the lenght of this phase? Wylie & Woolf (2002) role of buoyancy, wind shear, moist ure…
in CRMs saturation def icit cloud f raction rain mixing ratio
SCMs: saturation def icit SCMs: upward convective mass f lux
Chaboureau, Becht old, ECMWF GCM Köhler, Belj aars, et al. f ormulat ion of t he trigger f unction adding a represent at ion of local subgrid-scale f orcing (correct represent at ion of shallow convect ion required) 3D mode (i.e. ; f ull GCM) 1D column mode Rhodonia, Amazonia idealized EUROCS case (1-mont h composit e) (4-day composit e)
Chaboureau, diurnal cycle of rainf all in the Tropics Becht old, Köhler, Belj aars et al. diurnal composit e, Hovmüller diagram (20S-20N avg) OBS Feb 1999 (TRMM data ) Amazonia Africa ECMWF GCM Standard Feb 1999 ECMWF GCM Modif ied Feb 1999
CONCLUSI ON, PERSPECTI VES � document at ion of GCMs & RCMs weaknesses/ diurnal cycle of deep convect ion � assess CRMs/ SCMs over land wit h GCSS/ ARM case � design an idealized case t o address t he problem and it worked! i.e., deep convect ion occurs ear lier t han observed in many SCMs runs (consist ent wit h result s of GCMs) � bet t er result s/ consist ency among CRMs t han SCMs (T & q, cloud paramet ers: agreement wit h previous st udies) � CRM runs : t he t reat ment of t he BL is crucial increased horizont al resolut ion &/ or subgr id-scale processes raising issues concerning t he gap bet ween shallow and deep convect ion (numer ical invest igat ion) � SCM runs: ident if ying t he maj or weknesses & t est ing modif icat ions � no succession of dry-shallow-deep regimes in SCMs, dry t o deep direct ly � sensit ivit y t o t he t riggering cr it eria � dist inct diurnal cycle of st abilit y in CRMs & SCMs (weak par am. downdr af t s) � improvement of GCMs � f irst result s quit e successf ull, ongoing act ivit y & more need t o be done
Recommend
More recommend