Multi-model intercomparison of the impact of SORCE measurements in climate models TOSCA WG1 Workshop May 2012, Berlin K. Matthes (1), F. Hansen (1), J.D. Haigh (2), J.W. Harder (3), S. Ineson (4), K. Kodera (5,6), U. Langematz (7), D.R. Marsh (8), A.W. Merkel (3), P.A. Newman (9), S. Oberländer (7), A.A. Scaife (4), R.S. Stolarski (9,10), W.H. Swartz (11) (1) Helmholtz-Zentrum für Ozeanforschung Kiel (GEOMAR), Kiel, Germany; (2) Imperial College, London, UK; (3) LASP, CU, Boulder, USA; (4) Met Office Hadley Centre, Exeter, UK; (5) Meteorological Research Institute, Tsukuba, Japan; (6) STEL University of Nagoya, Nagoya, Japan; (7) Freie Universität Berlin, Institute für Meteorologie, Berlin, Germany; (8) NCAR, Boulder USA; (9) NASA GSFC, Greenbelt, USA; (10) John Hopkins University, Baltimore, USA; (11) JHU Applied Physics Laboratory, Laurel, USA
Introduction & Motivation Model Descriptions & Experimental Design Preliminary results from the multi-model comparison Summary Outlook Outline
„SIM‘s solar spectral irradiance measurements from April 2004 to December 2008 and inferences of their climatic implications are incompatible with the historical solar UV irradiance database […] but are consistent with known effects of instrument sensitivity drifts.“ „To prevent future research following a path of unrealistic solar-terrestrial behavior, the SORCE SIM observations should be used with extreme caution in studies of climate and atmospheric change until additional validation and uncertainty estimates are available.“ Introduction – Lean and DeLand (2012)
1.Do the SIM measurements provide real solar behavior or are they related to instrument drifts? 1.What are the effects of larger UV variability on the atmospheric response? Motivation – 2 Questions
„Top-Down Mechanism“ Gray et al. (2010)
„Top-Down“: Dynamical Interactions and Transfer to the Troposphere 10-day mean wave-mean flow interactions (Max-Min) u Stratospheric waves (direct solar effect) Tropospheric waves (response to stratospheric changes) EPF Matthes et al. (2006)
Significant tropospheric effects (AO-like pattern) result from changes in wave forcing in the stratosphere and troposphere which changes the meridional circulation and surface pressure Matthes et al. (2006) +2K ΔT Modeled Signal near Earth Surface Monthly mean Differences geop. Height (Max-Min) – 1000hPa
Uncertainty in Solar Irradiance Data Solar Max-Min NRLSSI vs. SATIRE Lean et al. (2005)Krivova et al. (2006) larger variation in Krivova data in and nm range SORCE measurements from 2004 through 2007 show very different spectral distribution (in-phase with solar cycle in UV, out-of-phase in VIS and NIR) => Implications for solar heating and ozone chemistry NRLSSI vs. SIM/SORCE
Participating Models Model Description & Experimental Design Caveat: all models used a slightly different experimental setup, so it won’t be possible to do an exact comparison! SOCOL, T42, L39, 0.01 hPa, nudged QBO, see talk by Eugene Rozanov this afternoon
Differences in Experimental Setup
Experimental Design Time series of F10.7cm solar flux SC23 „solar max“ 2004 „solar min“ : “solar max” (declining phase of SC23) 2007: “solar min” (close to minimum of SC23)
January Mean Differences (25N-25S) NRL SSI SORCE Shortwave Heating Rate (K/d)Temperature (K) larger shortwave heating rate and temperature differences for SORCE than NRL SSI data FUB-EMAC and HadGEM only include radiation, not ozone effects
January Mean Differences (25N-25S) Ozone (%)Temperature (K) larger ozone variations below 10hPa and smaller variations above for SORCE than NRL SSI data height for negative ozone signal in upper strat. differs between models NRL SSI SORCE
Large Multi Model Mean: all 5 models (FUB-EMAC, GEOS, HadGEM, IC2D, WACCM) Small Multi Model Mean: 3 models (GEOS, HadGEM, WACCM) Definition Ensemble Mean
Shortwave Heating Rate Differences January (K/d) Large multi-model mean (EMAC-FUB, GEOS, HadGEM, IC2D, WACCM) Small multi-model mean (GEOS, HadGEM, WACCM) NRL SSI SORCE NRL SSI shortwave heating rates: 0.2 K/d SORCE shortwave heating rates: 0.9 K/d (4x NRL SSI response)
Temperature Differences January (K) Large multi-model mean (EMAC-FUB, GEOS, HadGEM, IC2D, WACCM) Small multi-model mean (GEOS, HadGEM, WACCM) NRL SSI temperatures: 0.3 to 0.6 K (stratopause) SORCE temperatures: 1.5 to 1.8 K (5x NRL SSI response) colder polar stratosphere NRL SSI SORCE
Ozone Differences January (%) Large multi-model mean (GEOS, IC2D, WACCM) NRL SSI SORCE larger ozone variations below 10hPa and smaller variations above for SORCE than NRL SSI data height for negative ozone signal in upper strat. differs between models
Zonal Wind Differences January (m/s) Large multi-model mean (EMAC-FUB, GEOS, HadGEM, IC2D, WACCM) Small multi-model mean (GEOS, HadGEM, WACCM) consistently stronger zonal wind signals for SORCE than NRL SSI data wind signal in SORCE data characterized by strong westerly winds at polar latitudes, and significant and similar signals in NH troposphere NRL SSI SORCE
SORCE Differences NH Winter – small ensemble mean Zonal mean zonal wind (m/s) DecemberJanuaryFebruary downward extension of westerly zonal wind signals to the troposphere
SORCE Geopot. Height Differences January (gpdm) 500 hPa100 hPa NAO/AO positive signal during solar max 10 hPa
Solar Cycle & NAO Solar Max: NAO positive (high index) Colder stratosphere => stronger NAO, i.e. stronger Iceland low, higher pressure over Azores amplified storm track mild conditions over northern Europe and eastern US => dry conditions in the mediterranean
Solar Cycle & NAO Solar Max: NAO positive (high index) Solar Min: NAO negative (low index) Matthes (2011)
Consistently larger amplitudes in 2004 to 2007 in solar signals for SORCE than for NRL SSI data in temperature, ozone, shortwave heating rates, zonal winds and geopotential heights Larger ozone variations below 10hPa and smaller variations above for SORCE than NRL SSI data; height for negative ozone signal in upper stratosphere differs between models Solar cycle effect on AO/NAO contributes to substantial fraction of typical year-to-year variations and therefore is a potentially useful source of improved decadal climate predictability (Ineson et al. (2011)) Results for the SORCE spectral irradiance data are provisional because of the need for continued degradation correction validation and because of the short length of the SORCE time series which does not cover a full solar cycle Summary
Paper on multi-model comparison to be submitted before 31 st July coordinated sensitivity experiments within the SPARC-SOLARIS Initiative for a typical solar max (2002) and solar min (2008) spectrum from the NRL SSI, SATIRE and the SORCE (and possibly other data or reconstructions? SCIA, COSI?) data to investigate the atmospheric and surface climate response between the models in a more consistent way SOLARIS/HEPPA workshop 9-12 October 2012 in Boulder Outlook
Estes Park/RMNP, Thank you very much!
Shortwave Heating Rate Differences January (K/d) EMAC-FUBGEOS IC2DHadGEMWACCM NRL SSI SORCE NRL SSI shortwave heating rates: 0.2 to 0.3 K/d SORCE shortwave heating rates: 0.7 to >1.0 K/d (3x NRL SSI response)
Temperature Differences January (K) EMAC-FUBGEOS IC2DHadGEMWACCM NRL SSI SORCE NRL SSI temperatures: 0.5 to 1.0 K (stratopause) SORCE temperatures: 2.5 to 4.0 K (4-5x NRL SSI response) colder polar stratosphere
Ozone Differences January (%) EMAC-FUBGEOS IC2DHadGEMWACCM NRL SSI SORCE larger ozone variations below 10hPa and smaller variations above for SORCE than NRL SSI data height for negative ozone signal in upper strat. differs between models
Ozone Differences January (%) Large multi-model mean (EMAC-FUB, GEOS, HadGEM, IC2D, WACCM) Small multi-model mean (GEOS, IC2D, WACCM) NRL SSI SORCE larger ozone variations below 10hPa and smaller variations above for SORCE than NRL SSI data height for negative ozone signal in upper strat. differs between models
Annual Mean Tropical Profiles Temperature (K)Ozone (%) SPARC CCMVal (2010)
SORCE Wind Differences NH Winter EMAC-FUBGEOS IC2DHadGEMWACCM Dec Jan Feb
Zonal Wind Differences January (m/s) EMAC-FUBGEOS IC2DHadGEMWACCM NRL SSI SORCE consistently stronger zonal wind signals for SORCE than NRL SSI data wind signal in SORCE data characterized by strong westerly winds at polar latitudes, and significant and similar signals in NH troposphere
SORCE Wind Differences NH Winter Large multi-model mean (EMAC-FUB, GEOS, HadGEM, IC2D, WACCM) Small multi-model mean (GEOS, HadGEM, WACCM) Dec Jan Feb
SORCE Geopot. Height Differences January (gpdm) EMAC-FUBGEOS HadGEMWACCM 500 hPa 100 hPa 10 hPa NAO/AO positive signal during solar max strongest for HadGEM and WACCM
SORCE Geopot. Height Differences January (gpdm) Large multi-model mean (EMAC-FUB, GEOS, HadGEM, WACCM) Small multi-model mean (GEOS, HadGEM, WACCM) 500 hPa 100 hPa 10 hPa NAO/AO positive signal during solar max strongest for HadGEM and WACCM
Solar Min Surface Pressure Signal Ineson et al. (2011) Model (HadGEM) Observations (Reanalyses) 90 (95%) significances 25 (50%) of interannual standard deviation
SORCE/SIM measurements from 2004 to 2007: increased solar spectral irradiance at UV and IR wavelengths even as solar and TSI decreased => SIM spectral data into climate models => „the effects of solar variability on temperature throughout the atmosphere may be contrary to current expectations“ (Haigh et al., 2010) => higher solar activity cools Earth But: SIM trends relative to TSI and solar activity during solar min => unlikely to be solar in origin „It is doubtful that simulations of climate and atmospheric change using SIM measurements are indicative of real behavior in the Earth‘s climate and atmosphere.“ (Lean and DeLand, 2012) „SIM‘s solar spectral irradiance measurements from April 2004 to December 2008 and inferences of their climatic implications are incompatible with the historical solar UV irradiance database […] but are consistent with known effects of instrument sensitivity drifts.“ (Lean and DeLand, 2012) „To prevent future research following a path of unrealistic solar-terrestrial behavior, the SORCE SIM observations should be used with extreme caution in studies of climate and atmospheric change until additional validation and uncertainty estimates are available.“(Lean and DeLand, 2012) Motivation