1. Considering Climate Variability & Change TOGETHER Lisa Goddard International Research Institute for Climate & Society The Earth Institute of Columbia University goddard@iri.columbia.edu

2. Main Points The tropical Pacific climate is controlled by coupled ocean- atmosphere dynamics. Getting both these aspects right is necessary for getting a realistic estimate of climate change. Variability and change may have similarities, but they have many differences also. Change in the mean state does not necessarily imply a change in variability, and vice versa. We are really into the realm of ‘active research’ here. OUTLINE Possible ENSO-like changes in tropical Pacific mean state Possible changes in ENSO variability under climate change

3. ENSO-like Changes in Mean Climate Will it be La Niña-like? Will it be El Niño-like? - What does theory say? - What do the models say? - What do the observations say?

4. Tropical Pacific Trend Pattern vs ENSO Variability Source: IPCC AR4 WG1, Chp 10

5. Example of Tropical Pacific SST Change versus Composite El Nino SST Anomalies El Nino Composite (20 th C) Temperature Change (2071-2100)-(1971-2000) --------------- GFDL 2.0 --------------- --------------- GFDL 2.1 ---------------

6. ENSO-like Changes in Mean Climate Will it be La Niña-like? Will it be El Niño-like? - What does theory say? 2 Main theories: - Ocean thermostat mechanism - Weaker Walker Circulation mechanism

7. Cane, M.A., A.C. Clement, A. Kaplan, Y. Kushnir, D. Pozdnyakov, R. Seager, S.E. Zebiak, and R. Murtugudde, Twentieth-Century Sea Surface Temperature Trends, Science, 275, 957-960, 1997. 1. Theoretical mechanism for radiative forcing of ENSO-like states in the tropical Pacific Ocean Thermostat mechanism (Courtesy Mike Mann)

8. Mean modern conditions Ocean ‘thermostat’ mechanism (Clement et al. 1996) (Courtesy Mike Mann)

9. Uniform heating (Courtesy Mike Mann)

10. Larger temperature response in the West Cooling by upwelling opposes forcing in the East, reducing temperature response (Courtesy Mike Mann)

11. Larger temperature response in the West Cooling by upwelling opposes forcing in the East, reducing temperature response Coupled interactions (i.e. the Bjerknes feedback) amplify the East/west temperature difference (Courtesy Mike Mann)

12. RESULT: A new Pacific mean state with a STRONGER EAST/WEST GRADIENT (Courtesy Mike Mann)

13. 2. Theoretical mechanism for radiative forcing of ENSO-like state in the tropical Pacific Weaker Walker mechanism Water vapor in lower troposphere increases as expected by Clausius-Clapeyron (in the models), but precipitation does not, due to atmospheric energy constraints. This implies that circulation must slow down. (Held and Soden, 2006, J. Climate; Vecchi and Soden, 2007, J. Climate)

14. 2. Theoretical mechanism for radiative forcing of ENSO-like state in the tropical Pacific Weaker Walker mechanism Some suggestion that such a slow down in the Walker Circulation is being observed… Linear Trend in SLP in Observations & Model (Vecchi et al, 2006, Nature)

15. Ocean Thermostat or Weaker Walker Mechanism ? What do the models say? Model complexity seems to make a difference in what mechanism dominates the response to radiative forcing. Contains active ocean dynamics but very simplified atmosphere Contains atmospheric energy constraints but ocean dynamics are fixed Contains both fully dynamical and interactive ocean and atmosphere models. (Vecchi, Clement & Soden, 2008, EOS)

16. Ocean Thermostat or Weaker Walker Mechanism ? What do the observations say? The observations don’t answer the question either because there is so much uncertainty in the earlier data. (Vecchi, Clement & Soden, 2008, EOS)

17. Changes in El Niño Variability More or less? Stronger or weaker? - Does oceanic component of ENSO change? - Do teleconnections (at least in the tropics) change?

18. Tropical Pacific Trend Pattern vs ENSO Variability Source: IPCC AR4 WG1, Chp 8

19. Coelho, C.A.S. and L. Goddard, 2009. El Niño-induced tropical droughts in climate change projections. J. Climate, DOI: 10.1175/2009JCLI3185.1 Motivating Questions 1) How well do coupled models simulate tropical droughts associated with El Niño during the 20th century? 2) How do ENSO-related drought patterns change in 21 st century? 3) To what extent does the risk of presently defined drought during El Niño conditions change under climate change?

20. ENSO Characteristics in IPCC Models Observations IPCC models Other Coupled Models Differences in Magnitude of ENSO events… … and their frequency Source: AchutaRao and Sperber, 2006

21. DATA IPCC Coupled Models: (20C3m; A2) “WEAK” Models –CGCM3_T47, MIROC_mres, GISS-ER “MODERATE” Models –NCAR_CCSM3, UKMO_HadCM3, GFDL_CM2.0 “STRONG” Models –CNRM_CM3, MPI_ECHAM5, GFDL_CM2.1 Observations: UEA-CRU Precipitation

22. “Good” ENSO Models “MODERATE” Models –NCAR_CCSM3, UKMO_HadCM3, GFDL_CM2.0 “STRONG” Models –CNRM_CM3, MPI_ECHAM5, GFDL_CM2.1 Guilyardi (2006) tropical Pacific climatology e.g. strength of seasonal cycle (winds & SST), seasonal cycle locking, Interannual coupling strength van Oldenborgh et al. (2005) coupled air-sea variability e.g. spatial structure and power spectrum of SSTa, wind stress & thermocline variability, air-sea coupling parameters

23. El Niño Events in IPCC Models Anomalies relative to evolving current climatology TOO WEAK MODERATE TOO STRONG Threshold based on 75 th %-ile of climatology (Coelho & Goddard, 2009, J. Climate)

24. El Niño Magnitude NOT Projected to Change Too much event-to-event variability to attribute change in any particular realization (Coelho & Goddard, 2009, J. Climate)

25. El Niño Frequency NOT Projected to Change Conclusions based on any single model is likely to contain substantial sampling bias (Coelho & Goddard, 2009, J. Climate)

26. ENSO Frequency Sensitivity to Record Length (Wittenberg, 2009)

27. Motivating Questions 1) How well do coupled models simulate tropical droughts associated with El Niño during the 20th century? 2) How do ENSO-related drought patterns change in 21 st century? 3) To what extent does the risk of presently defined drought during El Niño conditions change under climate change?

28. Tropical Drought and El Niño Note: 5 month lag between max. Nino 3.4 SSTA and extent peaks (B. Lyon, 2004, GRL)

29. IPCC C20C Model Composites El Niño (DJF) 30-year moving climatology Models categorized by ENSO Variance IPCC models with reasonable ENSO variability reproduce reasonable El Nino teleconnections compared to seasonal forecast models. Strength of composite teleconnections scales with strength of ENSO events. (Coelho & Goddard, 2009, J. Climate)

30. IPCC Model Composites: 20C vs 21C El Niño (DJF) 30-year moving climatology Models categorized by ENSO Variance (Coelho & Goddard, 2009, J. Climate)

31. Precipitation Changes DJF (2071-2100) – (1971-2000) Models categorized by ENSO Variance “WEAK” “MODERATE” “STRONG” No statistically significant tendency in patterns of mean precipitation change related to strength of a model’s ENSO variability. (Coelho & Goddard, 2009, J. Climate)

32. How to Make a Multi-Model (MM) Ensemble How to indicate robustness or confidence in the MME? – MM Mean – % models agreeing in sign of change – MM Signal > Inter-model Noise – Choosing (or weighting) a subset of models according to ??

33. Precipitation Changes (%) (2090-2099) – (1980-1999) Mean + % of models agreement White areas where less than 2/3 models agree in sign of change. Stippled areas where more than 90% of models agree in sign of change.

34. Precipitation Changes (mm/day) (2090-2099) – (1980-1999) Mean + (S/N > 1) Stippling denotes areas where the magnitude of the multi-model ensemble mean exceeds the inter-model standard deviation.

35. Concept of Layering El Niño-related Drought Risk Based on 20 th Century Observations: Anthropogenic Precipitation Changes Based on IPCC Models for 21 st Century: +  El Niño-related Drought Risk Projected for end of 21 st Century

36. Concept of Layering

37. Drought Risk Now & In the Future DJF: End 20 th C vs End 21 st C (Coelho & Goddard, 2009, J. Climate)

38. Drought Risk Now & In the Future JJA: End 20 th C vs End 21 st C (Coelho & Goddard, 2009, J. Climate)

39. Using ENSO Fidelity to Vet Models DJF JJA Much of this encouraging result is due to sampling, (i.e. only using 4 models rather than 20+) (Coelho & Goddard, 2009, J. Climate)

40. CONCLUSIONS: El Niño Variability (from CMIP3) No discernible change in relative El Niño strength or frequency in IPCC models (according to individual models) IPCC models, that do contain ENSO, exhibit realistic ENSO teleconnection patterns, which do not appear to change due to global warming. Layering of information: trends from IPCC models + variability from observations (in the tropics) provides more meaningful guidance on drought risk than using the models alone. Local confidence and specificity in such projections must be communicated carefully. CONCLUSIONS: Tropical Pacific Mean State Changes in the tropical Pacific mean state appear to be controlled by both ocean dynamics, and energy constraints of the atmosphere