GEWEX Process Evaluation Study on Upper Tropospheric Clouds & Convection

Transcription

1 GEWEX Process Evaluation Study on Upper Tropospheric Clouds & Convection GEWEX UTCC PROES advance understanding on feedback of UT clouds large-scale modelling necessary to identify most influential feedback mechanisms -> models should be in agreement with observations understand relation between convection, cirrus anvils & radiative heating provide observational metrics to probe process understanding Coordination: C. Stubenrauch & G. Stephens Claudia Stubenrauch Laboratoire de Météorologie Dynamique / IPSL, France GEWEX Data and Analysis Panel meeting Nov 2018, Lisbon, Portugal

2 GEWEX UTCC PROES highlights 2018 UTCC PROES website: goals, meeting presentations, references, links 3 rd GEWEX UTCC PROES meeting hosted by C. Stubenrauch, Sorbonne University, Paris, Oct participants Observational analyses of mesoscale convective systems, Water vapor & convective transport, Process studies, Climate variation & feedbacks, Parameterizations & model diagnostic studies very positive feedback from participants to continue this kind of focused meetings link to SPARC? 2019: write review article on this topic & promote process-oriented cloud system diagnostics as additional constraint in evaluation of model parameterizations 2

3 3 rd GEWEX UTCC PROES meeting contributions NASA CloudSat-lidar R05, A-Train -> radiative properties of UT clouds T. L Ecuyer tropical MSC data base ( ), IR geostationary, 30min, 4km T. Fiolleau, D. Bouniol convective systems (midlat Jul 2015) Cb tracking, Ci from ANN, geostationary L. Bugliaro monitoring deep convection, AMSU-B,MHS ( ), 0.2 J.F. Rysman RH predicted from cloud classes of diff. thermodyn. phase G. Carella rad. heating rates predicted from cloud & atmospheric properties C. Stubenrauch IAGOS commercial flight in situ observations UT RH ice (>1995) & N ice (> 2011) A. Petzold Straeole 2 balloon campaigns in tropical UTLS (2019, 2024) R. Plougonven assessment of heating rates from reanalyses (used for transport) B. Legras probe convective transport pathways by chemical tracers of diff. life times J. Luo simulating microphysical & dynamical processes of deep convection S. van den Heever NAO cloud anomalies > radiative heating anomalies -> dampening G. Papavasileiou amplification of hydrological cycle under ENSO G. Stephens tropical margin maintenance H. Masunaga middle trop. water vapor dominance in super-greenhouse effect R. Ramaswamy Ice crystal formation by convective detrainment in ECHAM-HAM S. Münch Using Cloud System Concept to assess bulk ice schemes in LMDZ GCM M. Bonazzola Parameterization of cold pool population in LMDZ GCM J.-Y. Grandpeix E 3 SM Earth System Model parameterization improvements & impacts P.-L. Ma 3

4 Motivation & Approach critical to climate feedback of UT clouds : cirrus radiative heating in upper troposphere Heating will be affected by: areal coverage emissivity distribution vertical structure (layering) Climate warming : change in convective intensity & coverage, height of convective systems & emissivity structure of the anvils? d[cov cold conv /cov conv ]/dt s = +18 ± 5 % / C (T<210K) d(thci/anv)/dt s = ± / C This then affects the heating gradients -> large-scale circulation preliminary Goal: understand relation between convection & radiative heating induced by cirrus anvils Method: 1) IR Sounders provide cloud height & emissivity; sensitive to cirrus 2) Cloud System Concept relates the anvil properties to processes shaping them 3) expand radar-lidar nadir track vertical structure laterally across UT cloud systems 4

5 UTCC PROES Strategy meetings: Nov 2015, Apr 2016, Mar 2017, Oct 2018 working group links communities from observations, radiative transfer, transport, process & climate modelling focus on tropical convective systems & cirrus originating from large-scale forcing Cloud System Concept, anchored on IR sounder data (horizontal extent & convective cores/cirrus anvil/thin cirrus based on p cld, ε cld ) -> relationships between anvil properties & convective strength build synergetic data (vert. dimension, atmosph. environment, temporal res.) determine heating rates of different parts of UT cloud systems follow snapshots by Lagrangian transfer -> evolution & feedbacks investigate how cloud systems behave in CRM studies & in GCM simulations (under different parameterizations of convection/detrainment/microphysics) 5

6 link anvil structure to convective depth Protopapadaki et al. ACP years AIRS; tropical UT cloud systems (p cld -p tropopause < 250 hpa or p cld < 440 hpa); convective core (Cb): ε cld >0.98; mature systems: Cb fraction within system AIRS AMSR-E synergy Deeper convective cores -> stronger max rain rate -> T cb min good proxy for convective strength Deeper convection leads to relatively more thin cirrus within larger anvils (similar land / ocean) relation robust using different proxies : T min Cb / LNB(max mass) increasing convective depth Why? H1: UT environmental predisposition (at higher altitude larger RH, T stratification) H2: UT humidification from cirrus outflow Does the relationship change in a warmer climate? CRM GCM 6

7 link heating rates to convective depth via a complete 3-D description of UT cloud systems & their environment (from ERA5) 1) along nadir tracks: categorize CloudSat FLXHR-lidar vertical structure & heating rates wrt cloud type (p cld & ε cld ), for different atmospheric situations 2) expand nadir track info across UT cloud systems & environment: develop optimized non-linear regression models : deep neural network learning techniques relate most suitable cloud & atmospheric properties from IR sounders & meteorological reanalyses to vertical structure & heating rates 7

8 1) heating rates of UT cloud systems, sampled along track AIRS UT cloud systems collocated to Lidar-CloudSat FLXHR heating rates wrt to ε cld, p cld, cloud LW heating altitude (km) km T Cb < 225K T Cb > 225K preliminary LW heating cloud clear sky convective core Ci anvil thin Ci anvil cloud clear sky heating rate (K/day) clear distinction of heating associated with each category cold convective systems have a larger thin Ci heating 8

9 2) Cloud radiative LW heating rates via deep learning results improve when initial cloud info extended by atmospheric info no overfitting, mean absolute error 0.5 K/day! ε cld critical variable=> models per cloud type very promising, but needs deeper assessment preliminary 17.5 km thin Cirrus Cirrus Cb 0 km data predicted K/day models reproduce very well the LW heating rates 9

10 UT Cloud System Concept to assess GCM parameterizations analyze GCM clouds as seen from AIRS/IASI, via simulator & construct UT cloud systems -> evaluation of GCM convection schemes / detrainment / microphysics spatial res. 2.5 x 1.25 Goal: build coherent v m - De parameterization nominal fall speed v m = 0.3 x f(iwc) horizontal cloud system emissivity structure sensitive to v m, De M. Bonazzola, LMD De = f(t), ε = f(de, IWC) scaled v m too small compared to observations v m = 0.9 x f(iwc, T) v m increases with IWC & T, v m closely related to De Deng & Mace 2008 v m increase with IWC weaker towards warm T Field et al. 2007, Furtado et al PSD moment parameterization De = f(v m ) Heymsfield et al Rad. balance -> precip. efficiency, UT hum variability AIRS 7 Jan

11 process-oriented UT cloud system behaviour Data control v m =0.3 x f(iwc) De = f(t) empirical v m & De = f(v m ) PSDM v m & De = f(v m ) PSDM v m, D m & D e = f(d m ) preliminary increasing age of system increasing convective depth including T dependency of v m -> larger spread in T of UT cloud systems (in better agreement with obs.) New process-oriented diagnostics based on Cloud System Concept powerful constraint more realistic v m De very promising: leads to more realistic anvil size & ε horizontal structure (increasing thin Ci) development 11

12 Discussion UTCC PROES workshops seem to be very inspiring, as they focus on specific thematic & they link different scientific communities (observation & modelling at diff. scales) Is it the role of GEWEX to foster this kind of workshops or should they continue in a different framework (WCRP, GEWEX SPARC)? Different synergetic data analyses -> complementary information on processes Do results present a coherent picture, and if not why (differences in proxies: convection, cloud systems, etc)? write review article on this topic UT Cloud System Data base (0.5 x 0.5 ) (AIRS IASI) AIRS-IASI Cloud Observation Simulator & UT Cloud System Data at GCM spat. resolution Cloud System analysis can be automatized for GCM evaluation Cloud System analysis may be used to investigate mechanisms in CRM studies promote process-oriented cloud system diagnostics as additional constraint in evaluation of climate model parameterizations synergies : meteorological reanalyses -> thermodynamics & dynamics A-Train active lidar radar -> vertical structure within cloud systems A-Train MODIS & CERES -> smaller scale visible opt. depth & evaluation of fluxes MCS from geostationary imagers & precipitation (opaque anvils) -> life cycle? (test proxy of Cb fraction within system to indicate life cycle stage) 12

13 Update of GEWEX Cloud Assessment Data base Claudia Stubenrauch Laboratoire de Météorologie Dynamique, IPSL, France GEWEX Data and Analysis Panel meeting Nov 2018, Lisbon, Portugal

14 to facilitate assessments, climate studies & model evaluation monthly statistics per observation time at 1 x 1 (netcdf): averages variability histograms properties: Cloud Assessment L3 Database cloud amount, height, radiative & bulk microphysical properties height stratified (H/M/L) & liquid / ice statistics 2D ISCCP histograms: COD-CP (CEM-CP) -> cloud types IR-NIR-VIS Radiometers, IR Sounders, VIS-SWIR Radiometers exploit different parts of EM spectrum: How does this affect climatic averages & distributions? Stubenrauch et al., WCRP report 2012, BAMS

15 Cloud Assessment updated versions PATMOS-x (AVHRR) NOAA MODIS-CERES NASA updated datasets AIRS-CIRS AERIS, (Stubenrauch et al. 2017) MISR CALIPSO-ScienceTeam NASA v3, 3D, COD filter (0.3), opaque (unfortunately not without COD filter) new datasets CLARA(AVHRR) EUMETSAT CM SAF ( IASI-CIRS AERIS, (Stubenrauch et al. 2017) HECTOR beta (HIRS) CM SAF, CIRS method AVHRR-Cloud_cci (v2.0) ESA Cloud-cci, (Stengel et al. 2017) in preparation: ISCCP (HGG, NOAA), HIRS-NOAA 15

16 Conclusions & thoughts for discussion GEWEX Cloud Assessment database: 12 global state of the art datasets (in 2008) joint effort to build consistent database Utility of database: assessment of new datasets climate studies (CFMIP-OBS database used for model evaluation) -> demand from users to update the database Update (same data format, similar website structure in cooperation with AERIS): 5-7 datasets of first round + 4 new datasets 1 analysis & note to BAMS with summary website will be constructed in coop. with AERIS in Jan / Feb 2019 & can be updated every year Improvements due to retrieval problems, but it looks like main conclusion stays the same: sensitivity to cirrus instrument/ retrieval method dependent Ancillary data affect low-level cloud amount Next step should be really multi-instrument retrievals or intelligent combination of different data sets / variables 16

17 even if instantaneous cloud properties are not very accurate, the synergy of different variables provide invaluable potential for improving our understanding of clouds synergy also important for model evaluation: compare correlations of physical variables or statistics organized by weather states or cloud systems -> GEWEX PROES just be aware of assumptions made in retrieval 17