Evaluation of CMIP5 Simulated Clouds and TOA Radiation Budgets in the SMLs Using NASA Satellite Observations Erica K. Dolinar Xiquan Dong and Baike Xi University of North Dakota This talk is based on Dolinar et al. (2014, Clim. Dyn.) March 18, 2014 University of Washington, Seattle, WA Workshop on Clouds, Radiation, Aerosols, and the Air-Sea Interface in the S. Midlatitude Ocean
Motivation In many climate models, details in the representation of clouds can substantially affect the model estimates of cloud feedback and climate sensitivity. Moreover, the spread of climate sensitivity estimates among current models arises primarily from inter-model differences in cloud feedbacks. Therefore, cloud feedbacks remain the largest source of uncertainty in climate sensitivity estimates. IPCC Fourth Assessment Report (2007) Want to understand the impacts of simulated clouds on the TOA radiation budget and cloud radiative forcings in our current climate so that we may better predict the future climate 1
Satellite Products Clouds CERES MODIS SYN1degree Total Column Cloud Fraction CCCM (CloudSat, CALIPSO, CERES, MODIS) Vertically integrated Cloud Fraction Vertical Velocities (omega) MERRA Reanalysis Radiation CERES EBAF TOA radiation budgets TOA cloud radiative forcing (CRF) Products are Level-3 and have been either downloaded or provided by Science Team members *Caveat Observations have uncertainties (Dolinar et al. 2014) but are used as truth in this study 2
Study Groundwork 28 uncoupled - AMIP (atmosphere-only) models Climatologically prescribed SSTs 03/2000 02/2008 (8 years) SML: 70 30 South Ocean 3
Cloud Fraction (CF) Comparison Observations [81.5%] Multimodel Ensemble [69.3%] Bias [ 12.2%] Model simulated total cloud fraction is largely under estimated over the SMLs compared to CERES-MODIS observed CF 4
Cloud Water Path (Ice + Liquid) Observations [190.3 gm 2 ] A fair proxy for cloud optical depth Multimodel Ensemble [134.5 gm 2 ] Bias [ 55.8 gm 2 ] Model simulated cloud water path is largely under estimated in the SMLs compared to CERES-MODIS observation 5
CF Profile The under estimation of CF in the SML oceans is primarily a result of under estimated low- and midlevel (950 500 hpa) clouds. There does exist some over estimation of cloud fraction at higher levels (~250 hpa) At 850 hpa Multimodel Mean: 24.5% CCCM: 43.5% Bias: -19.0% 6 *Only 23 simulations available
Vertical Velocities Convective cloud types are commonly parameterized by the consideration of mass flux and vertical velocities while stratiform-type cloud schemes are based upon RH relationships At 850 hpa MERRA: 1.0 hpa day -1 (down) Multimodel Mean: -0.1 hpa day -1 Regime shift The dynamic forcing in this region is different (or slightly modified) than what is observed (reanalyzed) 7 Up Down *Only 26 simulations available
Vertical Velocities at 850 hpa Down Up Down Up The overall distribution of vertical velocities (convection/subsidence) at 850 hpa is correctly simulated by the multimodel ensemble in the Southern Mid-latitudes, but either the strength of the descending branch of the Hadley Cell is weaker or the ascending branch of the Ferrell Cell is stronger than reanalyzed ones 8
Cloud Fraction at ~850 hpa Observations [43.5%] Multimodel Ensemble [24.5%] Bias [ 19.0%] The largest biases at ~850 hpa coincide with the ascending/descending branches of the Hadley and Ferrell Cells 9
Summary I: CF Comparisons Total column cloud fraction is under estimated, on average, by the 28 model ensemble by 12.2% in the Southern mid-latitudes over the ocean Cloud water path is under estimated by 55.8 gm 2 Currently large uncertainties in observed CWC profiles Cloud fraction is under estimated by ~20% in the low-levels (~850 hpa) (23/28 models) Due to, but not limited to, a potential dynamical regime shift or lack of cloud water Would be interesting to analyze other simulated synoptic conditions What effect do these results have on the TOA radiation budget? 10
Reflected SW and OLR Flux differences (Model CERES) eled TOA reflectedswfluxishigherwhile OLR is lower CERES observations over the SMLs se results do not make physical sense compared to restimated CF and CWP in model
TOA SW and LW CRF differences (Model CERES) CRF = All - Clr The magnitude of TOA SW (LW) CRF cooling (warming) is underestimated in the SMLs Regions of positive (negative) biases are consistent with the SW (LW) radiation flux results 12
The simulated magnitude of the Net CRF cooling is under estimated in the SMLs but there does exist an area of stronger cooling due to clouds between S. America and Australia in the models Summary II: TOA Radiation Results All TOA radiation and the cloud radiative heating/cooling is under estimated in the SMLs Areas of over estimated SW/Net cooling due to clouds Results are consistent with each other but not with corresponding CF and CWP results Less clouds, more reflection/cooling and less outgoing/warming? How? A topic for further consideration and research 13
Acknowledgements Workshop organizers Drs Jonathan Jiang and Hui Su at JPL for their help and support over the past year Research group at UND All of you! 14
Questions erica.dolinar@my.und.edu 15
Backup 16
Modeling Center (or Group) Model Name Grid Spacing Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of ACCESS1.0 1.875 1.25, L38 Meteorology (BOM), Australia Beijing Climate Center, China Meteorological Administration BCC-CSM1.1 BCC-CSM1.1 (m) 1.25 1.25, L26 2.8125 2.8125, L26 College of Global Change and Earth System Science, Beijing Normal University, China BNU-ESM 2.8125 2.8125, L26 Canadian Centre for Climate Modeling and Analysis CanAM4 2.8125 2.8125, L35 National Center for Atmospheric Research (NCAR), USA CCSM4 1.25 0.9375, L26 Community Earth System Model Contributors (NSF- DOE-NCAR), USA CESM1 (CAM5) 1.25 0.9375, L30 Centro-Euro-Mediterraneo per I Cambiamenti Climatici, Italy CMCC-CM 0.75 0.75, L31 Centre National de Recherches Meteorologiques/ Centre Europeen de Recherche et Formation Avancees en Calcul CNRM-CM5 1.4 1.4, L31 Scientifique, France Commonwealth Scientific and Industrial Research Organization in collaboration with Queensland CSIRO-Mk3.6.0 1.875 1.875, L18 Climate Change Centre of Excellence, Australia LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences CESS, Tsinghua University FGOALS-s2 FGOALS-g2 2.8125 1.666, L26 2.8125 3.0, L26 NOAA Geophysical Fluid Dynamics Laboratory, USA GFDL-CM3 GFDL-HIRAM-C180 GFDL-HIRAM-C360 2.5 2.0, L48 0.625 0.5, L32 0.3125 0.25, L32 NASA Goddard Institute for Space Studies, USA GISS-E2-R 2.5 2.0, L29 Met Office Hadley Centre, United Kingdom HadGEM2-A 1.875 1.25, L38 Institute for Numerical Mathematics, Russia INM-CM4 2.0 1.5, L21 Institut Pierre-Simon Laplace, France IPSL-CM5A-LR IPSL-CM5A-MR IPSL-CM5B-LR 3.75 1.875, L39 2.5 1.25, L39 3.75 1.875, L39 Atmosphere and Ocean Research Institute (The University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine- MIROC5 1.4 1.4, L40 Earth Science and Technology Max Planck Institute for Meteorology, Germany MPI-ESM-MR MPI-ESM-LR 1.875 1.875, L95 1.875 1.875, L47 Meteorological Research Institute, Japan MRI-AGCM3.2H MRI-AGCM3.2S MRI-CGCM3 0.5625 0.5625 0.1875 0.1875 1.125 1.125, L35 Norwegian Climate Centre 17 NorESM1-M 2.5 1.875, L26
Relative Humidity 15-20% uncertainty in AIRS RH Stratiform type clouds are commonly parameterized with the consideration of relative humidity Relative humidity is over estimated at all levels (with the exception of one model below 900 hpa) BUT we do not know which models contain both liquid and ice RHs so we will not put any faith in these results 18 *Only 13 simulations available
Summary Variable Observed Mean* Ensemble Mean Mean Bias** Cloud Fraction 81.5 69.3 ± 8.0 12.2 Cloud Water Path 190.3 134.5 ± 47.0 55.8 TOA Reflected SW 105.3 103.6 ± 8.1 1.7 TOA Outgoing LW 223.8 222.5 ± 3.9 1.3 TOA SW CRF 63.1 60.8 ± 8.9 2.3 TOA LW CRF 28.9 27.0 ± 5.2 1.9 TOA Net CRF 34.2 33.8 ± 5.8 0.4 *Observed values are from CERES MODIS/EBAF ** Mean biases in CRFs correspond to the relative warming/cooling effects 19