1. Factors Limiting the Current Skill of Forecasts: Flaws in Model and Initialization Center for Ocean-Land-Atmosphere studies (COLA) George Mason University (GMU) Emialia K. Jin Climate Test Bed Seminar Series 4 February 2009

2. Model Flaws  mean error, phase shift, different amplitude, and wrong seasonal cycle, etc Flaws in the way the data is used  data assimilation and initialization, chaos within non-linear dynamics of the coupled system Inherent limits to predictability  some times are more predictable than others, amplitude of SST anomalies with respect to ENSO phase Gaps in the observing system What is limiting the ENSO predictability? 2

3. Model Flaws  mean error, phase shift, different amplitude, and wrong seasonal cycle, etc Flaws in the way the data is used  data assimilation and initialization, chaos within non-linear dynamics of the coupled system Inherent limits to predictability  some times are more predictable than others, amplitude of SST anomalies with respect to ENSO phase Gaps in the observing system What is limiting the ENSO predictability? 3

4. Flaws in Model: Two Flavors of ENSO and Its Predictability Authors: Emilia K. Jin 1, J.-S. Kug 2, F.-F. Jin 2, J.-J. Luo 3, and T. Yamagata 3 1 George Mason Univ./COLA, 2 University of Hawaii, 1 George Mason Univ./COLA, 2 University of Hawaii, 3 FRCGC/JAMSTEC

5. Background and Objective  Conventional El Niño : “as a phenomenon in the equatorial Pacific Ocean characterized by a positive sea surface temperature departure form normal in the NINO 3.4 region greater than or equal in magnitude to 0.5C averaged over three consecutive months” (NOAA)  Different flavors of El Niño Trans- Niño (Trenberth and Stepaniak, 2001), Dateline El Niño (Larkin and Harrison 2005), El Niño Modoki (Ashok et al. 2007 ), Non-canonical ENSO (Guan and NIgam, 2008), Warm pool El Niño (Kug et al. 2008), etc. : Even though there are differences, the distinctive interannual SST variation over the central Pacific which becomes more active in recent year and significantly different global impact form conventional El Niño are common features.  The transition mechanisms and dynamical structure of two-types of El Nino are significantly different (Kug et al. 2008).  In this study, CGCM’s ability to predict the distinctive characteristics of two types of El Niño is investigated using two state-of-the-art CGCMs retrospective forecasts. 5

6. Observed Two Types of El Nino Kug et al., 2008 NINO4 NINO3 Composite of SST (Contour) and Rainfall (Shaded) (1982/83, 1986/87, 1997/98) (1990/91, 1994/95, 2002/03, 2004/05) Normalized NINO3 and NINO4 SST Warm-poolCold-tongue Either NINO3 SST or NINO4 SST is greater than their standard deviation Mixed 6

7. Observed DJF SST Anomalies Warm-pool Cold-tongue Mixed 7

8. RetrospectiveForecastRetrospectiveForecast Initial condition cases of 12 calendar months are analyzed. As observational counterparts, OISST, CMAP rainfall, and NCEP/NCAR reanalysis data are used. Model Lead month Ensemble Member PeriodAGCMOGCM FRCGC SINTEX 1291982-2006 ECHAM 4 T106 L19 OPA 8.2 2x2 L31 NCEP CFS 9151981-2006 GFS T62 L64 MOM 3 1/3x5/8 L27 Model and Dataset Courtesy of J.-J. Luo, T. Yamagata, and NCEP EMC In this study, forecast data is reconstructed with respect to lead time (monthly forecast composite). 8

9. Observed DJF SST Anomalies Warm-pool Cold-tongue Mixed 9

10. Simulated SODJFM SST Anomalies Forecast lead month 1 CFS SINTEX Shading is for model, and contour is for observation Warm-pool Cold-tongue Warm-pool Cold-tongue 10

11. Simulated DJF SST Anomalies Forecast lead month 6 Shading is for model, and contour is for observation Warm-pool Cold-tongue Warm-pool Cold-tongue CFS SINTEX Note: loss of predictability in the Warm Pool El Nino cases 11

12. Composite of SST Anomalies along the Equator Forecast lead month 7 CFS SINTEX Warm-pool Cold-tongue Mixed time Shading is for model bias, Contour is for observed composite Note: Positive anomaly and negative bias in the Warm Pool and Cold Tongue 12

13. Interannual Variability of NINO3 and NINO4 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Observed Nov Dec CFS SINTEX NINO3NINO3 NINO4NINO4 13

14. Scatter Diagram of Normalized DJF NINO3 vs. NINO4 NINO3 Index NINO4 Index CFS SINTEX Lead month 1 Lead month 7 14

15. Relationship between NINO3 and NINO4 CFS SINTEX COR=0.69 15

16. Impact of Couplde Model Error on Predictability 1 st mode SEOF of SST (Low frequency mode) Obs. long run 1 st month 9 th month 5 th month NCEP CFS JJA SINTEX-F MAM Forecast lead month Correlation Temporal correlation of PC timeseries with observation Pattern correlation of eigenvector with free long run SINTEX-F NCEP CFS SINTEX-F NCEP CFS Correlation coefficients with respect to lead month Jin and Kinter, 2009 Climate Dynamics With increase of the lead month, the forecast ENSO mode progressively approaches to the model intrinsic mode in free coupled run and departs from the observed. 16

17. PRCGCSINTEXPRCGCSINTEX NCEPCFSNCEPCFS Free long run Free long run forecast forecast 202-year simulation Analyzing last 200 years (200-yr climatology) 52-year simulation Analyzing last 50 years (50-yr climatology) 1982-2004 period 9 members 12 calendar months ICs 12 months lead 1981-2003 period 15 members 12 calendar months ICs 9 months lead Luo et al. 2005 Saha et al. 2006 Model and Dataset Courtesy of J.-J. Luo, T. Yamagata, and K. Pegion 17

18. Scatter Diagram of Normalized DJF NINO 3 vs. NINO 4 From free long run of two CGCMs NINO3 Index NINO4 Index Obs. CFS SINTEX 1950-200550 years200 years 0.690.820.86  Model Flaw: One Flavor of El Nino COR=(NINO3, NINO4) Shading: Observed; models do not capture observed behavior 18

19. Observed Composite of Precipitation Anomalies Warm-pool Cold-tongue Forecast lead month 6 CFS SINTEX Obs. 19

20. 500 hPa GPH Anomalies CFS SINTEX Warm-pool Cold-tongue Forecast lead month 6 Obs. 20

21.  In two state-of-the-art CGCMs, the forecast skill of El Niño is investigated focusing on two flavors of El Niño: Warm-pool and cold- tongue.  As the lead month of forecast increases, the models fail to distinguish between two flavors of El Niño.  Both models have difficulties to reproduce the nonlinear relationship between NINO3 and NINO4 SST anomalies.  From the free long run, models tend to simulate the mixed mode of El Nino rather than warm-pool or cold-tongue El Niño.  Tropical precipitation and extratropical circulation anomalies associated with two flavors of El Niño are not captured by models. Summary 21

22. Model Flaws  mean error, phase shift, different amplitude, and wrong seasonal cycle, etc Flaws in the way the data is used  data assimilation and initialization, chaos within non-linear dynamics of the coupled system Inherent limits to predictability  some times are more predictable than others, amplitude of SST anomalies with respect to ENSO phase Gaps in the observing system What is limiting the ENSO predictability? 22

23. Flaws in the initialization: Impact of Ocean Initialization in CCSM3.0 Re-forecast Experiments Authors: Emilia K. Jin 12, B. Kirtman 23, D.-H. Min 3, K. Ashok 4 and H-I. Jeong 4 1 George Mason Univ., 2 COLA, 3 Univ. of Miami/RSMAS, 1 George Mason Univ., 2 COLA, 3 Univ. of Miami/RSMAS, 4 APEC Climate Center

24. APCCCOLA Initialization Atm NCEP/NCAR Reanalysis 2 (1, 3, 5 th of November) Random conditions from CCSM 3.0 long run via AMIP (No observation) Ocean SST-nudged scheme (Luo et al. 2005) GFDL MOM3 ODA (Rosati and Harrison, 2002) Member 36 Lead month 712 Period 1982-20031982-1998 Reference Ashok et al. (2009)Kirtman and Min (2009) Model and Dataset Courtesy of K. Ashok and H. Jeong (APEC Climate Center), and Ben Kirtman and Dug-Hong Min (Univ. of Miami) RetrospectiveForecastRetrospectiveForecast CCSM 3.0: CAM3 T85L26 + POP 1.4 gx1v3 L40 Initial condition case of November are analyzed. As observational counterparts, OISST and CMAP rainfall are used. 24

25. Ocean Initialization SST-nudgedschemeSST-nudgedscheme MOM3 ODA APCC re-forecasts: 2-dimensional ocean initialization COLS re-forecasts: 3-dimensional ocean initialization GFDL MOM3 Ocean Data Assimilation Grid interpolation to POP 1.4 gx1v3 L40 6 atm. ICs 25

26. Root-mean-square error of NINO Indices Forecast lead month Nov Dec Jan Feb Mar Apr May 26

27. Anomaly Correlation Coefficients of NINO Indices Forecast lead month Nov Dec Jan Feb Mar Apr May 27

28. Temporal Correlation Coefficients of SST Anomalies DJF (1 st season)MAM (2 nd season) 28

29. Temporal Correlation Coefficients of SST Anomalies 29 Differences of Correlation (APCC minus COLA)

30. Pattern Correlation of SST Anomalies Pattern Correlation Coefficients Year 1 st season (lead month 2-4) (160-280E, 30S-30N) (40-160E, 30S-30N) APCC member COLA member 30

31. 1984 SST anomalies along the Equator Forecast lead month 31

32. ACC of NINO 3.4 Index (November IC) Colored dots denote14 CGCMs re-forecasts from DEMETER and APCC/CliPAS (Jin et al. 2008). Tier-1 MME: 10 CGCM multi-model ensemble except NASA, UH, APCC and COLA. Dynamic-Statistical Model: MME of modified CZ model, two statistical models and persistence Black line denotes persistence.

33. RMSE of NINO 3.4 Index (Nov IC) Colored dots denote14 CGCMs re-forecasts from DEMETER and APCC/CliPAS (Jin et al. 2008). Tier-1 MME: 10 CGCM multi-model ensemble except NASA, UH, APCC and COLA. Dynamic-Statistical Model: MME of modified CZ model, two statistical models and persistence Black line denotes persistence.

34. NINO3: Warm minus Cold composite SST anomalies Influence of Systematic Error on CFS Forecast Skill  Warm composite (82/83, 86/87, 91/92, 97/98) - Cold composite (84/85, 88/89, 98/99, 99/00)  Dashed lines denote composite for Hindcasts at different lead times Observation CFS long run (Hindcast composite) Forecast lead month Correlation CORR. with respect to lead month based on 1 st SEOF mode of SST Correlation between 1 st PCs based on observation and hindcasts at different lead times Correlation between 1 st PCs based on long run and hindcasts at different lead times Jin and Kinter, Climate Dynamics 2009  Model Flaw: Slow coupled dynamics of CGCM 34

35. DJF SST Climatology A 138-year long run of CCSM3.0 Retrospective forecasts 1 st season of Nov IC

36. Year Simulated Nino34 Index (CCSM3.0) 36

37. Composite Analysis of Nino34 Index Warm minus Cold composite El Nino composite La Nina composite For CCSM3.0 free long run, events more than one standard deviation of DJF NINO 3 index is selected and 32 El Niño and 24 La Niña is picked up. Warm composite (82/83, 86/87, 91/92, 97/98) - Cold composite (84/85, 88/89, 98/99, 99/00)

38. DJF Correlation of SST with Precipitation Local SSTNINO3.4 Re-Forecasts (Nov IC) Free long run 38

39. DJF Precipitation Climatology A 138-year long run of CCSM3.0 Retrospective forecasts 1 st season of Nov IC 39

40. Temporal Correlation Coefficients of Precipitation Anomalies DJFMAM 40

41.  In this study, the intercomparison of long-lead coupled prediction experiments has conducted focusing on the multiple sets of retrospective forecasts with two types of initial conditions using same CCSM3.0 CGCM: 2- dimensional ocean initialization (SST nudged scheme) and 3-dimensional ocean initialization (MOM3 Ocean Data Assimilation).  Focusing on ENSO forecast, the ocean initialization of the COLA re-forecasts causes the remarkable improvement of forecast skill in spite of the large systematic errors of model. On the other hand, the APCC re-forecasts has little advantages of ocean initialization and the influence of model’s systematic errors are quite large.  These results emphasize the importance of initialization of forecast model, in particular ocean component. Summary 41

42. Emilia K. Jin kjin@cola.iges.org