ChemistryClimate Model Validation Activity for SPARC CCMVal Status

Chemistry-Climate Model Validation Activity for SPARC (CCMVal) Status and Workshop Goals Veronika Eyring (DLR), Ted Shepherd (Univ. Toronto), Darryn Waugh (JHU), Andrew Gettelman (NCAR), and Steven Pawson (NASA) CCMVal Workshop 2009; Toronto, 1 -5 June 2009

Motivation CCMVal-1 simulations

CCMVal approach to CCM Evaluation and analysis Process-oriented CCM evaluation CCMVal Evaluation Table Eyring et al. , BAMS, 2005

Example: CCMVal Diagnostic Table For Transport Process Diagnostic Variables Subtropical and polar mixing barriers PDFs of long-lived tracers N 2 O, CH 4, CFC-11, etc. ; Satellite and in situ Potential Vorticity (PV) (aircraft, balloons) chemical measurements and Meteorological Analyses Latitudinal gradients of long-lived tracers Correlations of long-lived tracers Meridional circulation Data Phase and amplitude of tropical CO 2 or H 2 O annual cycle in lower stratosphere (tape recorder) CO 2, H 2 O or idealized annually repeating tracer Satellite and in situ measurements Mean age Conserved tracer with linearly increasing concentration, SF 6 or CO 2; in situ measurements Correlation of interannual Total ozone and heat anomalies of total ozone flux at 100 h. Pa, zonal and Planetary wave flux and monthly means Satellite measurements, Meteorological Analyses Entire CCMVal Evaluation table: http: //www. pa. op. dlr. de/CCMVal_Evaluation. Table. html Eyring will be updated as part of the SPARC CCMVal Report et al. , BAMS, 2005

CCMVal-1: Model Assessment (13 CCMs) mainly focused on transport and polar dynamics diagnostics Temperature Bias Water Vapor Eyring et al. , JGR, 2006 Inorganic Chlorine Mean Age

CCMVal-1: Ozone Projections There are substantial differences in the date at which Cly returns to 1980 values varying from before 2030 to after 2050. (Primarily transport related. ) There is a similar large variation in the timing of recovery of Antarctic spring-time column ozone back to 1980 values. Value of Process-oriented evaluation of CCMs Eyring et al. , JGR, 2007

Unsatisfying from an assessment point of view. . . WMO, 2007: TWENTY QUESTIONS AND ANSWERS ABOUT THE OZONE LAYER: 2006 UPDATE Lead Author: D. W. Fahey

Quantitative Performance Metrics: Inorganic Chlorine Cly Grades: Waugh & Eyring, ACP, 2008 bad good

CCMVal-1 developed performance metrics 1 0 Waugh and Eyring, ACP, 2008 Potential benefits: 4 Allow visualization of performance for multiple aspects of the simulations. 4 Allow identification of incompletely modeled processes. 4 Enable quantitative assessment of model improvements for different versions of individual CCMs and for different generations (e. g. CCMVal-1 vs CCMVal-2). 4 Make it possible to explore the value of weighting the projections. Potential improvements: 4 Explore other metrics. 4 Consider other diagnostics (radiation & chemistry). 4 Better consider uncertainty in observations by using observations from multiple platforms and instruments. 4 Metrics for seasonal and interannual variability and trends. 4 Statistically more robust grading.

Weighting Ozone Projections For diagnostics and ozone projections considered there is, generally, only small differences between weighted and unweighted multi-model mean projections.

IPCC, WGNE and WGCM • The missing set of metrics that might be used to narrow the range of plausible climate projections through a comparison of model simulations with observations is one of the current principal source of uncertainty in model evaluation [IPCC, AR 4, WG 1, 2007]. • WGNE/WGCM metrics panel established (Members: Karl Taylor, Beth Ebert, Veronika Eyring, Peter Gleckler, Robert Pincus, Richard Wood) • IPCC Expert meeting on "Metrics" and "Assessing and Combining Multi-model Climate Projections", January 2010, NCAR, Boulder, USA

CCMVal Diagnostic Tool (requires NCL & Python installed) Andrew Gettelman (NCAR), Veronika Eyring (DLR), Greg Bodeker (NIWA), Irene Cionni (DLR), Chris Fischer (NCAR), Mike Neish (Univ. Toronto), Hamish Struthers (NIWA), Ted Shepherd (Univ. Toronto), Hisako Shiona (NIWA) & Charlotte Pascoe (BADC) Facilitate the model evaluation for CCMVal, e. g. · Allow quick looks at standard diagnostic plots & output diagnostic variables · Produce climatology files from CCMVal CF-compliant model output. Include the diagnostics of the previous round of CCMVal evaluation so that we don't have to start from scratch each time · ensures progress · allows to assess quickly where we stand with the new CCMVal simulations · helps to increase the standard for CCM evaluation Expandable and extensible · Useful for multiple model groups & those analyzing models · SPARC CCMVal report can extend tool (i. e. provide diagnostics once report is finalized) · Extend with tropospheric diagnostics (in collaboration with Jean-Francois Lamarque)

E 06 FIG 01. ncl: Figure 1 of Eyring et al. , JGR, 2006. /var_att/ta_att. ncl info@fig 01_lat_max = (/90. , -60. /) info@fig 01_lat_min = (/60. , -90. /) info@fig 01_season = (/"DJF", "MAM", "JJA", "SON"/) info@fig 01_ref. Model = (/"ERA-40"/) info@fig 01_clim. Obs =(/"NCEP", "UKMO"/) ; Climatological Observation file info@fig 01_clim. Obs_file =(/"input_data/OBS/CCMVal 1_1980_2000_NCEP_Obs _T 2 Mz_ta. nc", "input_data/OBS/CCMVal 2_19922001_UKMO_Obs_C 2 Mz_ta. nc"/) ; C 2 Mz file Climatological mean temperature biases for latitudes between info@fig 01_lat_max and info@fig 01_lat_min for the info@fig 01_seasons. Biases are calculated relative to info@fig 01_ref. Model. The grey area shows info@fig 01_ref. Model plus and minus 1 standard deviation about the climatological mean. The climatological means info@fig 01_clim. Obs are included.

CCMVal-2 (in support of WMO/UNEP 2010 and AR 5) § 18 CCMs participate in the 2 nd round of CCMVal (CCMVal-2). § CCMVal-2 reference and sensitivity simulations defined (Eyring et al. , 2008) § CCMVal-2 simulations of the future begin in 1960, and most continue to 2100. § Earlier starting date allows a more accurate determination of the milestone when total ozone returns to pre-1980 levels § Extended simulations allow multi-model ozone projections and an analysis of the causes of these projected changes throughout the 21 st century. § CCMVal-2 data request output is collected in Climate and Forecast (CF) standard compliant Net. CDF format at BADC. § Allows automatic software to work on the output (e. g. CCMVal diagnostic tool and net. CDF operators). § Same format as for CMIP 5 simulations. § Base output for core diagnostics 3 D (lat, lon, p) monthly mean fields. In addition, instantaneous output for a subset of diagnostics is collected (e. g. UTLS). § CCMVal Data Policy: § To release the model data to the SPARC CCMVal author teams at an early stage, a PHASE 0 has been added to the existing Phase 1 and 2 of the CCMVal data policy. § In PHASE 0 CCMVal-2 data are only accessible for the authors and models PIs.

CCMVal-2 Models Model Group and Location Horiz. resolution Vert. Layers / Upper Boundary 1 AMTRAC 3 GFDL, USA 2˚ (lat) x 2. 5˚(lon) 48 L / 0. 002 h. Pa 2 CAM 3. 5 NCAR, USA 1. 9(lat)x 2. 5(lon) 26 L / 2. 2 h. Pa 3 CCSRNIES, University of Tokyo, Japan T 42 34 L / 0. 012 h. Pa 4 CMAM MSC, Univ. of Toronto, York Univ. , Canada T 31 71 L / 0. 000637 h. Pa 5 CNRM-ACM Meteo-France, France T 42(GCM), T 21(CHEM) 60 L / 0. 07 h. Pa 6 E 39 CA DLR, Germany T 30 39 L / 10 h. Pa 7 EMAC MPI Mainz, Germany T 42 90 L/ 0. 01 h. Pa 8 GEOSCCM NASA/GSFC, USA 2˚ x 2. 5˚ 72 L / 0. 01 h. Pa 9 LMDZrepro IPSL, France 2. 5˚ x 3. 75˚ 50 L / 0. 07 h. Pa 10 MRI, Japan T 42 68 L / 0. 01 h. Pa 11 NIWA-SOCOL NIWA, NZ T 30 39 L / 0. 01 h. Pa 12 SOCOL PMOD/WRC and IAC ETHZ, Switzerland T 30 39 L / 0. 01 h. Pa 13 ULAQ University of L'Aquila, Italy 10˚ x 22. 5˚ (CHEM) 26 L / 0. 04 h. Pa 14 UMETRAC NIWA, NZ 2. 5˚ x 3. 75˚ 64 L / 0. 01 h. Pa 15 UMSLIMCAT University of Leeds, UK 2. 5˚ x 3. 75˚ 64 L / 0. 01 h. Pa 16 UMUKCA-METO Met. Office, UK 2. 5˚ x 3. 75˚ 64 L / 84 km 17 UMUKCA-UCAM University of Cambridge, UK 2. 5˚ x 3. 75˚ 64 L / 84 km 18 WACCM (v. 3) NCAR, USA 4˚ x 5˚ 66 L / 4. 5 x 10 -6 h. Pa　

Documentation of Model Improvements Inorganic Chlorine CCMVal-1 CCMVal-2

SPARC CCMVal Report on Evaluation of CCMs 18 CCMs in support of WMO/UNEP 2010 and IPCC AR 5 Ø Ø In the past there has been insufficient time to evaluate CCM performance thoroughly while preparing the Ozone Assessments. The goal of the SPARC CCMVal report is to provide useful and timely information for the WMO/UNEP 2010 & IPCC AR 5 and an up-to-date evaluation of CCMs, a reassessment of the projections of ozone and UV radiation through the 21 st century, and the impact of stratospheric changes on climate. Structure and Authors (around 100 authors are analyzing the CCMVal-2 data): · Executive Summary {Eyring, Shepherd, Waugh plus chapter Lead Authors} · Chapter 1: Introduction {Eyring, Shepherd, Waugh} · Part A: Chapter 2: Chemistry Climate Models and Scenarios {Morgenstern, Giorgetta, Shibata} · Part B: Process evaluation * Chapter 3: Radiation {Fomichev, Forster} * Chapter 4: Dynamics {Butchart, Charlton} * Chapter 5: Transport {Neu, Strahan} * Chapter 6: Chemistry and microphysics {Chipperfield, Kinnison} * Chapter 7: UTLS {Gettelman, Hegglin} · Part C: Chemistry-Climate Coupling * Chapter 8: Natural Variability {Manzini, Matthes} * Chapter 9: Long-term Projection of Stratospheric Ozone {Austin, Scinocca} * Chapter 10: Effect of the Stratosphere on Climate {Baldwin, Gillett} Timelines: Currently under final external review, published Jan-March 2010; JGR Special Issue

Chapter 7: The UTLS in chemistryclimate models Michaela I. Hegglin & Andrew Gettelman Seok-Woo Son, Masatomo Fujiwara, Simone Tilmes, Laura Pan, Peter Hoor, Huikyo Lee, Gloria Manney, Thomas Birner, Gabriele Stiller, Markus Rex, Stefanie Kremser, Don Wuebbles • CCMVal-2 included for the first time the comprehensive validation of CCM performance in the tropical and extra-tropical UTLS. • Major efforts were needed in order to ü find and process suitable observations for comparison ü establish robust diagnostics for the CCMs (which operate in a free-running mode and are multi-dimensional) ü turn the diagnostic into a quantitative grade • Trend analyses for tropopause measures and chemical species were also included.

The global UTLS

Seasonal cycle in H 2 O The right representation of the water vapour is critical for both the radiative and chemical properties of the UTLS (and the stratosphere in general), and should be mainly determined by the seasonal cycle in the tropical cold point temperatures. • The models show major difficulties in reproducing the H 2 O seasonal cycle. • Model means, phase, and amplitude of the seasonal cycle exhibit a wide range. • At lower altitudes, the models indicate too strong mixing across the tropopause.

GLOBAL UTLS METRICS TABLE

KEY FINDINGS • Despite their relatively low horizontal resolution, the CCMs perform reasonably well in resolving the complex processes and structures in the UTLS. Models with semi. Lagrangian transport schemes tend to be overly diffusive and perform worst. • The models perform least in representing the amplitude and phase in water vapour, both in the tropics and the extra-tropics. This seems particularly worrisome in the light of the expected feedback role of water vapour. KEY ISSUES AND RECOMMENDATIONS • The UTLS is still relatively sparsely sampled by observations. This limits our confidence in the quantitative evaluation of model performance in the UTLS. New observations are needed especially for O 3 and H 2 O with a vertical resolution better than 1 km and a horizontal resolution better than 100 km, especially in the SH and the tropics. • In this round of CCMVal, our main focus was on evaluating the representation of dynamics, and transport and mixing in the UTLS. However, testing chemistry should be part of future model validation efforts, including tropospheric CCMs used for the IPCC. Future model development is needed that brings together tropospheric and stratospheric chemistry climate models.

SPARC CCMVal Report Chapter 8: Natural Variability of Stratospheric Ozone Lead Authors: Elisa Manzini and Katja Matthes Co-Authors: Christian Blume, Greg Bodeker, Chiara Cagnazzo, Natalia Calvo, Andrew Charlton-Perrez, Anne Douglass, Pier Giuseppe Fogli, Lesley Gray, Junsu Kim, Kuni Kodera, Markus Kunze, Cristina Pena Ortiz, Bill Randel, Thomas Reichler, Gera Stenchikov, Claudia Timmreck, Matt Toohey, and Shigeo Yoden

Motivation Stratospheric ozone is known to vary in response to a number of natural factors, such as the seasonal and the 11 -year cycles in solar irradiance, the QBO, ENSO, variations in transport associated with large-scale circulations (i. e. , Brewer Dobson circulation) and dynamical variability associated with the annular modes. Aerosols from volcanic eruptions can also affect stratospheric ozone, although their effects depend on the background atmospheric composition. Ozone observations have demonstrated variations on a large number of spatial and temporal scales. Objective: The goal of this Chapter is to evaluate how well CCMs simulate natural stratospheric ozone variability, based on our current knowledge about links between ozone variations and natural forcings.

Tropical variability measure: Detrended, deseasonalized and filtered -48 months) time series. Monthly zonal mean standard deviation (5 o. S-5 o. N). Zonal wind (m/s) (9 Ozone density (DU/km) TOP PANELS: The QBO (via nudging of the zonal winds or vorticity) is assimilated in the chemistry climate models (8 models out of 17 models) CCMVal-2 simulations: in color, ERA 40 (left) and SAGE (right) black.

QBO, key results: • The modeling of the QBO in the CCMVal-2 models is judged to be still at a primitive stage. A very few models are capable to spontaneously simulate the salient features of QBO ozone variations. • Some AGCM in recent years have been able to simulate a quite realistic QBO in zonal winds and related dynamical quantities, but it does not seem that this expertise has passed to the CCMs, possibly also because of the computational and/or developmental constrain of the additional chemical modelling. • The nudging of the QBO induces substantial errors in QBO ozone variations, notably in the amplitude of the column ozone. RECOMMENDATIONS: Advance the development of the modeling of the QBO is called for. This modeling problem is outstanding. It is an open issue that cannot be tackled by assimilating selected properties of the observed QBO in the CCMs.

SPARC CCMVal Report Chapter 9: Long-term projections of stratospheric ozone Lead Authors: John Austin and John Scinocca Co-Authors: Trevor Bailey, Luke Oman, David Plummer, David Stephenson, and Hamish Struthers

Ozone Changes in CCMVal-2 Vertical profile results of the MLR analysis for the CCMVal-2 models in the latitude band 10°S -10°N. (a) Ozone in the year 2000 (b) Ozone change from 2000 to 2100 (c) EESC change from 2000 to 2100, (d) Contribution of the EESC change to the ozone change. (e) and (f) are the same as (c) and (d), except for temperature. From Chapter 9, Figure 9. 4 of SPARC CCMVal (2010).

Ozone Changes in CCMVal-2 1980 baseline-adjusted multi-model trend (MMT) estimates of annually averaged total ozone for the latitude range 60 -90°S (heavy dark grey line) (upper panels). The baseline-adjusted IMT estimates, and unadjusted fit to the observations are additionally plotted. The lower panel shows the same analysis of CCMVal-2 data but for a baseline adjustment employing a 1960 reference date From Chapter 9, Figure 9. 11 of SPARC CCMVal (2010).

Ozone Changes in CCMVal-2 Date of return to 1960 (left) and 1980 (right) values for the annual average (tropical and midlatitude) and spring (polar) total ozone column derived from the IMT (colored symbols) and MMT (large black triangles) estimates for CCMVal-2 (left and right respectively in each latitude band). From Chapter 9, Figure 9. 21 of SPARC CCMVal (2010).

SPARC CCMVal Report Chapter 10: Effects of the stratosphere on the troposphere Lead Authors: Mark Baldwin & Nathan Gillett Co-Authors: Piers Forster, Ed Gerber, Michaela Hegglin, Alexey Karpechko, Junsu Kim, Paul Kushner, Olaf Morgenstern, Thomas Reichler, Seok-Woo Son, Kleareti Tourpali

Effect of ozone hole depletion/recovery on climate y-axis is SH: polar cap 50 h. Pa ozone x-axis is: (a) SH polar cap 100 h. Pa temperature (b) TP pressure poleward of 50 S (c) latitude of 850 h. Pa zonal wind maximum (d) SH Hadley cell boundary at 500 h. Pa

Effect of climate change on clear-sky UV radiation High northern latitudes will see a long-term reduction in UV Tropics will see a long-term increase

Effect of ozone depletion/recovery and climate change on stratosphere-to-troposphere ozone flux Effect of ozone depletion/recovery evident in SH Strengthening Brewer-Dobson circulation will increase ozone flux CCMVal models on the high side of AR 4 range, also of the observational estimates

AR 5 WG 1 draft outline • Ch. 1: Introduction • Ch. 2: Observations: Atmosphere and Surface • Ch. 3: Observations: Ocean • Ch. 4: Cryosphere • Ch. 5: Information from climate archives • Ch. 6: Carbon and other biogeochemical cycles • Ch. 7: Clouds and aerosols • Ch. 8: Anthropogenic and natural radiative forcing • Ch. 9: Evaluation of climate models • Ch. 10: Detection and attribution of climate change: from global to regional • Ch. 11: Near-term climate change: projections and predictability • Ch. 12: Long-term climate change: projections, commitments, reversibility • Ch. 13: Sea-level change • Ch. 14: Climate phenomena and their relevant for future regional climate change

SPARC and IGAC possible points of entry Ch. 2: Observations: Atmosphere and Surface · Changes in atmospheric circulation · Spatial-temporal patterns of climate variability Ch. 7: Clouds and aerosols · Aerosol types, including black carbon · Geoengineering involving aerosols Ch. 8: Anthropogenic and natural radiative forcing · Radiative forcing changes: solar and volcanic · Effects of atmospheric chemistry and composition, incl. GHGs · Radiative forcing due to emissions from aviation and shipping Ch. 9: Evaluation of climate models · Performance metrics, ensembles and their use · New model components and coupling · Representation of processes and feedbacks in climate models · Simulation of recent and longer-term records · Simulation of regional patterns, variability and extremes

Ch. 10: Detection and attribution of climate change: from global to regional Atmospheric and surface changes Ch. 11: Near-term climate change: projections and predictability · Climate change projections for the next few decades · Predictability of decadal climate variations and change · Regional climate change, variability and extremes · Atmospheric composition and air quality · Possible effects of geoengineering Ch. 12: Long-term climate change: projections, commitments, reversibility · Projections for the 21 st century Ch. 14: Climate phenomena and their relevant for future regional climate change · Patterns of variability · Monsoons · Interconnections among phenomena

AR 5 WG 2 draft outline (30 chapters!) SPARC and IGAC point of entry Ch. 11: Human health · Air quality and human health [with WG 1 authors on air quality] I asked Chris Field (WG 2 co-chair) about climate/ozone/UV aspects, and he said that climate/ozone/health issues would be part of this section. He said there are considering a separately identified group of atmospheric chemistry authors.