1. Synthesis of Modes of Ocean-Ice-Atmosphere Covariability in the Arctic System from Multivariate Century-Scale Observations Martin Miles Environmental Systems Analysis Research Center, Boulder, CO Mark Serreze National Snow and Ice Data Center, University of Colorado, Boulder, CO James Overland Pacific Marine Environmental Laboratory, Seattle, WA

2. Overall objective Quantitatively synthesize modes of (co)variability – and changes in these modes – in the Arctic and subpolar North Atlantic ocean–ice–atmosphere system in the past one to two centuries

3. Specific objectives (4): Assemble, update and systematize the longest continuous time series of oceanographic and meteorological measurements, sea ice observations and climate indices Integrate a subset of the multidecadal to century-scale Unaami Data Collection with a set of century-scale time series from the Arctic and subpolar North Atlantic, complemented by NCEP and other fields

4. ab (a) Approximate locations of multidecadal time series of climate ( ) atmosphere ( ), ocean ( ), sea ice ( ), terrestrial ( ), biology ( ) and fisheries ( ) variables in the Unaami Data Collection. (b) Locations of the mostly century-scale SAT measurement series >64ºN, as in Overland et al. (2004).

5. Approximate locations of century-scale time series of climate ( ) atmosphere ( ), ocean ( ) and sea ice ( ) variables from the Arctic and subpolar North Atlantic. Color coded same as in previous figure.

6. Specific objectives (cont.) Quantitatively characterize the ocean–ice– atmosphere system, through identifying recurrent modes of (co)variability – distinguishing modes other than the AO/NAO – in ocean, ice and climate parameters Seasonal to multidecadal time scales

7. Specific objectives (cont.) Quantitatively document changes in modes of variability and covariability – i.e., re-organization of the signals and linkages that may represent ‘regime shifts’ in the system

8. Specific objectives (cont.) Synthesize our results together with other observational and modelling analyses, to develop improved understanding of the arctic ocean–ice–atmosphere system and interactions with the subpolar North Atlantic 8

9. Overarching science questions What are the modes of variability other than the Arctic/North Atlantic Oscillation (AO/NAO) and what is their relative importance in different regions and seasons? How do modes of variability change through time and how unusual are recent changes compared to longer-term observations or model-based results?

10. Methodological approach Data synthesis: Integrated, consistent multi-method, multivariate statistical analysis Reconcile previous research: heterogeneous

11. Methodological approach Data synthesis: Integrated, consistent multi-method, multivariate statistical analysis Simple and advanced:  time series analysis techniques in time and frequency domains  spatial analysis – composites, correlations, EOF, SVD

12. Modes of variability concept Organized, recurring patterns (spatial and temporal) of variability and covariability  Statistical or data modes  Physical or dynamical modes

13. a)Arctic Oscillation b)Pacific Pattern (Pacific North American - PNA) AO amplitude small in recent years! 20 th Century Arctic Climate Patterns SLP-based

14. Surface Air Temperature (SAT) Anomalies Arctic Climate Patterns 1977–88 (PNA+) 1989–1995 (AO+) 2000–2005 (Arctic Warm) Pacific North American Arctic Oscillation Modified from Overland and Wang, Geophys. Res. Lett., 2006 New Recent Pattern

15. New Recent Arctic Climate Pattern (Arctic Warm) Cross-section, zonal annual T anomalies Temperature Anomalies 2000–2005 SAT anomalies (spring)

16. Source: Semenov and Bengtsson, Geophys. Res. Lett., 2004 Primary statistical modes of 20 th century SAT variability EOF1 AO/NAO EOF2 PNA

17. AO/NAO Pattern: Anomalies in SAT (shown), SST, SSS and sea ice in the Atlantic Arctic 1989-1995 AO/NAO+ 1969-1971 AO/NAO– EOF1 AO/NAO Source: Semenov and Bengtsson, Geophys. Res. Lett., 2004

18. Date unknown “So extensive and dangerous a work…” 12 primary stations 12+ auxilary United States Russia Netherlands Sweden Russia UK/Canada United States Denmark Germany Finland Austria Norway 1882-1883 Arctic Climate Patterns evident in First IPY data

19. Characteristic large-scale patterns evident Std. dev. from ref. mean 1968–1997 NAO Index data provided by the Climate Analysis Section, NCAR, Boulder, USA, Hurrell (1995). SAT anomalies related to fluctuations in atmospheric circulation Similar in scale to modern values DEC JANFEB SLP SAT WINTER C W L NAO +2.4 H NAO -1.4 C 1882-1883

20. Primary statistical modes of 20 th century SAT variability EOF1 AO/NAO EOF2 PNA EOF3 Not related to known physical mode of variability Early 20 th century warming period

21. EOF3 2006 winter (Arctic Warm, except Bering and Siberia) Recentmost Arctic Warmth vis-à-vis Early 20 th Century Warming

22. EOF3 2006 winter (re-scaled) Greenland–Barents Sea region unusually warm: Jan Mayen and Svalbard (+13 C) anomalies! Recentmost Arctic Warmth vis-à-vis Early 20 th Century Warming Feedback mechanism? Barents Sea SAT amplification – BSO pressure-gradient and sea ice? (Bengtsson et al., J. Climate (2004).

23. Recentmost Arctic Warmth (cont.) 2006 winter pressure- and wind-anomalies Winter 2006 sea ice Feedback mechanism? Barents Sea SAT amplification – BSO pressure-gradient and sea ice? (Bengtsson et al., J. Climate (2004).

24. Many science questions to be addressed:  Recent new arctic climate pattern: completely new or similar to early 20th century; seasonal aspects, feedbacks?  Decadal to interannual quasi-periodic modes of covariability: patterns, spatial and temporal; regime shifts?  Early 20th century warming: onset and underlying mechanisms?  AMO: manifestation in sea ice and other arctic records; mechanisms interactions; role in recent warming?  Atlantic Arctic: anomalous 19th-20th century vis-á-vis pan- Arctic temperature reconstructions?

25. Insight into many of these science questions will be gained in this ARCSS synthesis project Integrated multivariate statistical analysis of long time series of SAT, SST, sea ice and SLP

26. Examples of SST data and variability evident in Faroe- Iceland gateway

27. Example: Historical sea ice observations – Greenland Regional series from ice charts, e.g., Danish chart from 1910s. Source: DMI (Schmith and Hansen, 2003)

28. Examples of variability evident in multivariate observations in the 19 th and 20 th centuries

30. Results and deliverables  Data set: Time series, standardized and metrics  Synthesis papers general: Modes of variability synthesis focused:Multidecadal oscillation in sea ice records Early 20th-century warming onset Science, Nature, Geophys. Res. Lett., J. Clim., Clim. Change Timetable based around thematic papers

31. Summary Contribution to the ARCSS Synthesis Long-term baseline data and information from under-utilized multivariate observations that bridge a gap between contemporary measurements and paleoenvironmental data Self-consistent ‘metrics’ quantifying quasi- periodic variability, covariability and changes evident over the past ~50–200 years, thereby “setting the record(s) straight”

32. Summary Contribution to the ARCSS Synthesis Improved understanding of the temporal and spatial patterns, linkages and mechanisms of change in the arctic system Potential linkages to several SASS projects

33. Evidence for a multidecadal oscillation in arctic sea ice Evident in other 20 th -century sea ice analyses, e.g., Polyakov and others (2003), Johannessen and others (2004) Temperature and sea ice: Annual sea ice extent vs. air temperature in the Arctic (60-90°N) Zakharov ice extent (10 6 km 2 )

34. Observational data: Historical observations – Nordic Seas region Source: NPI

35. End of slideshow