1. Climate Forecasting Unit On the North Pacific reduced skill in near-term climate predictions of sea surface temperatures Virginie Guemas With the collaboration of : Francisco Doblas-Reyes, Fabian Lienert, Hui Du, Yves Soufflet

2. Climate Forecasting Unit Francisco J Doblas-Reyes : The Head Hui Du : Initial perturbations, sea ice Javier García-Serrano : AMO, African monsoon Virginie Guemas : Sea ice, North Pacific skill Fabian Lienert : regionalisation, PDO Melanie Davis : climate services Danila Volpi : initialisation techniques Luis Ricardo Rodrigues : ENSO, statistical models Aida Pintó : extremes Muhammad Asif : EC-Earth Oriol Mula-Valls : system administrator Domingo Manubens : autosubmit developer Our focus : Seasonal to decadal prediction We share, on request : 1)Autosubmit 2)Our decadal hindcasts 3)Monthly sea ice restarts 4)R diagnostic functions We run on : 1)Marenostrum ( Spain ) 2)ECMWF 3)HECTOR ( Scotland ) 4)Lindgren ( Sweden ) 5)Our local cluster

3. Climate Forecasting Unit On the North Pacific reduced skill in near-term climate predictions of sea surface temperatures Virginie Guemas With the collaboration of : Francisco Doblas-Reyes, Fabian Lienert, Hui Du, Yves Soufflet

4. Climate Forecasting Unit The climate prediction exercise Fig. 2 of Meehl et al. (2009, BAMS)

5. Climate Forecasting Unit Nov 60 Nov 65 Nov 70 Nov 75 Nov 80... Multi-model ensemble system with coupled initialized GCMs Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Leadtime = 10 years Obs The climate prediction exercise

6. Climate Forecasting Unit Multi-model ensemble system with coupled initialized GCMs Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Nov 60 Nov 65 Nov 70 Nov 75 Nov 80... Obs Leadtime = 5 years The climate prediction exercise

7. Climate Forecasting Unit Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Leadtime = 2-5 years Obs Multi-model ensemble system with coupled initialized GCMs Nov 60 Nov 65 Nov 70 Nov 75 Nov 80... The climate prediction exercise

8. Climate Forecasting Unit Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Leadtime = 2-5 years Obs Multi-model ensemble system with coupled initialized GCMs Nov 60 Nov 65 Nov 70 Nov 75 Nov 80... The climate prediction exercise

9. Climate Forecasting Unit Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Leadtime = 2-5 years Obs Multi-model ensemble system with coupled initialized GCMs Nov 60 Nov 65 Nov 70 Nov 75 Nov 80... The climate prediction exercise

10. Climate Forecasting Unit CERFACSUKMOIFM-GEOMAR ECMWFDePreSys EC-Earth 2m atmospheric temperature skill Correlation modelled and ERA40 T2M. Leadtimes : 2-5 years

11. Climate Forecasting Unit CERFACSUKMOIFM-GEOMAR ECMWFDePreSys EC-Earth Correlation modelled and ERSST SST. Leadtimes : 2-5 years Sea Surface Temperature skill

12. Climate Forecasting Unit Correlation between multi-model ensemble-mean and ERSST SST Sea Surface Temperature skill

13. Climate Forecasting Unit CERFACSUKMOIFM-GEOMAR ECMWFDePreSysEC-Earth Which major events are missed ? Sea Surface Temperature anomalies (155°E-235°E, 10°N-45°N)

14. Climate Forecasting Unit Outline Links with well-known variability modes ? Mechanism explaining the 1963 event Mechanism explaining the 1968 event Possible cause for the failure of forecasting systems

15. Climate Forecasting Unit Outline Links with well-known variability modes ? Mechanism explaining the 1963 event Mechanism explaining the 1968 event Possible cause for the failure of forecasting systems

16. Climate Forecasting Unit Two major warmings in the North Pacific Ocean in 1963 and 1968 Da Silva et al. (1994) turbulent Da Silva et al. (1994) total OAFluxes turbulent ERSST Sea Surface Temperature anomalies (155-235°E 10-45°N ) Upward surface heat fluxes anomalies (155-235°E 10-45°N ) 155-235°E 10-45°N

17. Climate Forecasting Unit No links with the main modes of variability http://www.esrl.noaa. gov/psd/data/climatei ndices/ Monthly values 12-month running mean

18. Climate Forecasting Unit Outline Links with well-known variability modes ? Mechanism explaining the 1963 event Mechanism explaining the 1968 event Possible cause for the failure of forecasting systems

19. Climate Forecasting Unit 1963: 3-dimensional structure of the anomaly Pattern of ERSST anomalies (155-235°E 10-45°N ) NEMOVAR temperature anomaly profile (180-205°E – 35-45°N) °C

20. Climate Forecasting Unit 1963: advection of the anomaly Hoevmuller ERSST anomalies ( 35°N-45°N latitude band) NEMOVAR backward trajectories computed with ARIANE ARIANE ( http://stockage.univ-brest.fr/~grima/Ariane/) April 1963 November 1960 °C

21. Climate Forecasting Unit 1963: modulation of the amplitude by mixing Hoevmuller ERSST anomalies ( 35°N-45°N latitude band) °C NEMOVAR climatological mixed layer depth ( 35°N-45°N ) m

22. Climate Forecasting Unit Outline Links with well-known variability modes ? Mechanism explaining the 1963 event Mechanism explaining the 1968 event Possible cause for the failure of forecasting systems

23. Climate Forecasting Unit 1968: 3-dimensional structure of the anomaly Pattern of ERSST anomalies (155-235°E 10-45°N ) °C NEMOVAR temperature anomaly profile (170-235°E – 10-35°N)

24. Climate Forecasting Unit 1968: upward transfer of the anomaly NEMOVAR temperature anomaly profile (170-235°E – 10-45°N)

25. Climate Forecasting Unit 1968: amplification by the atmosphere forcing Wind speed anomalies (170-235°E, 10-35°N) DFS4.3 NCEP NEMOVAR temperature anomaly profile (170-235°E – 10-45°N)

26. Climate Forecasting Unit Outline Links with well-known variability modes ? Mechanism explaining the 1963 event Mechanism explaining the 1968 event Possible cause for the failure of forecasting systems

27. Climate Forecasting Unit Stratification in the Pacific Ocean NEMOVAR-COMBINE CERFACS IFM-GEOMAR EC-Earth v2 FMA mixed layer depth (density criteria) in m

28. Climate Forecasting Unit Conclusion North Pacific region = region of lowest skill in near-term climate predictions ( 2- 5 years ) Major warm events around 1963 and 1968 missed by every forecasting systems = main cause for the low skill 1963 : heat anomaly advected along the Kuroshio-Oyashio and confined to an increasingly thinnner mixed layer 1968 : deep heat anomaly transferred toward the surface and then amplified by the atmospheric noise Biases in the Pacific ocean stratification could be responsible for the failure of the forecasting systems

29. Climate Forecasting Unit Thanks for your attention vguemas@ic3.cat