Dynamical responses to volcanic forcings in climate model simulations

1. Dynamical responses to volcanic forcings in climate model simulations DynVar workshop 22.04.13 Matthew Toohey with Kirstin Krüger, Claudia Timmreck, Hauke Schmidt

2. Motivation • What would happen if a large volcanic eruption occurred tomorrow? • Every seasonal to decadal climate forecast made prior to the eruption would become obsolete. Thompson et al. (2009) Thompson et al. (2012)

3. Motivation

4. “Winter Warming” Robock and Mao (1992)

5. Post-volcanic dynamical anomalies Schmidt et al., 2013 13 eruptions Christiansen, 2008 Baldwin and Dunkerton. 2001

6. Stratospheric mechanism Stenchikov et al. (2002)

7. Model results • A number of studies have reported realistic simulation of post-volcanic NH dynamical anomalies (Graf et al., 1993, 1994; Mao and Robock, 1998; Kirchner et al., 1999; Shindell et al., 2001; Rozanovet al., 2002; Stenchikov et al., 2002; Collins, 2004; Shindellet al., 2003, Shindell et al. 2004) • But multi-model studies (e.g. CMIP, CCMVal-2) have not produced a convincing picture of model behavior.

8. CCMVal-2 post-eruption T anomalies Ch. 8 in SPARC, CCMVal Report, 2010

9. CMIP5 9 eruptions 13 models 72 members 9 eruptions n=18 Sea level Pressure 9 eruptions 13 models 72 members 4 eruptions n=8 50 hPa Geopotential height Driscoll et al. 2012

10. CMIP5 CMIP5 Low-top High-top ERA-interim Charlton-Perez et al., 2013

11. Stratospheric mechanism ? ? Stenchikov et al. (2002)

12. The question • Why don’t CMIP5 models show strong NH winter vortices (i.e., negative polar cap z50 anomalies) after volcanic eruptions? • Either • Response is not real (just chance?) • Models are flawed • Implementation of volcanic aerosol forcing is flawed • Volcanic aerosol forcing is flawed

13. CMIP volcanic forcings Ammann (2003)/(2007) Sato et al. (1990)/GISS/Stenchikov 0.4 0.3 0.2 0.1 0 Jul 91 Jul 91 Jan 91 Jan 92 Jan 92 Jan 91 Jan 92 Jan 92 • Pinatubo and El Chichon based on SAGE observations • Recently updated with OSIRIS observations Oct 2001 - present • Best estimate sulfur mass injection, distributed via parameterized stratospheric transport model

14. CMIP Volcanic forcings • Notes: zonal mean, monthly mean, for pre-satellite era eruptions, spatial distribution of aerosols poorly constrained Sato et al. (1990)/GISS/Stenchikov

15. Experiment • Part 1: • Use MAECHAM5-HAM, a coupled aerosol-climate model, to simulate the evolution of stratospheric sulfate aerosol after a Pinatubo-like eruption. • Part 2: • Use MPI-ESM, a high-top CMIP5 model, and replace the prescribed Pinatubo volcanic forcing from historical simulations with forcing sets built from Part 1.

16. MPI-ESM • MPI-ESM: full Earth System model, with atmosphere, ocean, carbon cycle, vegetation components. • Atmospheric component ECHAM6. • “low resolution” (LR, T63/L47), configuration used here (no QBO). • Volcanic aerosols are prescribed • CMIP5 historical simulations use Stenchikov et al. (1998) forcing data set -> monthly mean, zonal mean aerosol extinction, single scattering albedo, and asymmetry factor

17. MAECHAM5-HAM • ECHAM: GCM developed at MPI-M, Hamburg • Middle atmosphere version: 39 vertical levels up to 0.01 hPa (~80 km) • T42 horizontal resolution • Climatological sea surface temperatures, no QBO, no chemistry • HAM: Aerosol microphysical module • Modified for simulation of stratospheric volcanic aerosols • Models aerosol growth, radiative effects, eventual removal HAM Aerosol growth Radiative effects Inject SO2 at 24 km Transport to troposphere, rainout! SO2→ H2SO4 ECHAM5 Aerosol transport via atmospheric circulation

18. MAECHAM5-HAM Pinatubo simulations Tooheyet al (2011, ACP) • Simulations of 17 Tg eruption, June 15, 15.3°N • Excellent agreement with ERBE TOA SW flux anomalies observed after Pinatubo eruption. • Little to no dependence on eruption longitude.

19. Modeled aerosol transport months after eruption months after eruption Toohey et al. (2011)

20. HAM July eruption simulations: DJF1 Temperature Geopotential height Zonal wind n=12

21. DJF1 z50 anomalies July eruptions April, July and October eruptions n=36 n=12

22. AOD: July eruption ensemble variability

23. Weak and Strong vortex composite AOD July eruptions n=12

24. Vortex strength ~ AOD gradient? Polar cap gph anomaly calculated as area mean over 70-90N. AOD gradient at 60N as AOD(60-90N) – AOD(50-60N)

25. Vortex strength ~ AOD gradient? Strong Vortex AOD gradient across vortex Aerosol heating gradient? If we want our prescribed aerosols to force a strong vortex, the forcing had better take the form of a strong vortex.

26. MPI-ESM Pinatubo forcing experiment r1,r2,r3 r4,r5,r6 r7,r8,r9 Stenchikov (CMIP5) HAM weak HAM strong

27. Aerosol extinction at 550 nm Stenchikov HAM weak HAM strong

28. MPI-ESM: tropical 50 hPa T

29. MPI-ESM: DJF1 T and u anomalies Stenchikov HAM weak HAM strong Temperature (K) u wind (m/s)

30. MPI-ESM: DJF1 z50 anomalies Low-top High-top ERA-interim

31. MPI-ESM: DJF1 z50 anomalies Low-top High-top ERA-interim

32. MPI-ESM: DJF1&2 z50 anomalies CMIP5 Low-top High-top ERA-interim

33. Aerosol extinction at 550 nm Stenchikov HAM weak HAM strong

34. Extinction at 550 nm August Arfeuille et al. ACPD 2013

35. Volcanic forcing, the next generation • CCMI: Surface Area Densities (SADs), stratospheric heating rates, and radiative properties, based on SAGE_4λ retrievals (Tom Peter and BeipingLuo, ETHZ) • Model-basedaerosol reconstructions becoming available for pre-satellite era eruptions. Tambora: Arfeuille et al. (2013) vs. Crowley (2008)

36. Conclusions • For a CMIP5 historical-style simulation of Pinatubo, we can control the strength of the (ensemble mean) post-eruption NH winter vortex with the aerosol forcing set • Vortex strength ~ AOD gradient across vortex edge • Likely that dynamical response to volcanic eruptions can be „improved“ by using different forcing data sets. • Future work will show whether new volcanic forcing sets lead to better dynamical responses in climate models.

37. Volcanic vs. Anthropogenic forcing