A Surprise in the North Pacific: Results from Applying a New Nonlinear Method for Calibra?ng Ice Cores Eric Kelsey 123 , Cameron Wake 1 , Erich Osterberg 4 and Karl Kreutz 5 1 Ins&tute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, 2 Dept. of Atmo. Sci. and Chem., Plymouth State University, Plymouth, NH 3 Mount Washington Observatory, Gorham, NH, 4 Dept. of Earth Science, Dartmouth College, Hanover, NH, 5 Climate Change Ins&tute, University of Maine, Orono, ME 1. Introduc?on Mo&va&on •They offer a valuable record of North Pacific and Arc&c climate variability because they sample the same synop&c scale phenomena at different eleva&ons (Figs. 1 and 2) •To obtain the best reconstruc&ons from the Eclipse and Mt. Logan ice cores, we must first understand the types of seasonal atmospheric circula&on paTerns that produce their ice core records 2. Data •Eclipse 9 and Mt. Logan 1,4 ice core data •Winter (DJFM) mean SLP gridded fields (2 o x2 o ) for the Northern Hemisphere north of 20 o N for 18712008 from the 20th Century Reanalysis dataset 10 are used for ice core calibra&on 3. Methodology •We divided the ranked ice core values (18722001) into 10 groups of 13 years (Fig. 2) •For each group, we examined the frequency of +/ SLP anomalies at each grid point to iden&fy common SLP paTerns •Composite mean SLP anomalies were ploTed with the SLP anomaly frequencies to iden&fy regions where high magnitude SLP anomalies occur frequently – these are called “frequency plots” •Groups that have high frequency (>70%) SLP anomalies and sta&s&cally significant composite SLP anomalies overlapping for a minimal synop&c scale area (2000 km x 2000 km = 4,000,000 km 2 ) are considered to have a significant SLP anomaly signal •Eclipse accumula&on and Mt. Logan [Na + ] are the only variables associated with a significant SLP anomaly paTern •The groups associated with significant SLP anomaly paTerns were used to reconstruct these SLP anomaly paTerns 8. References Contact: [email protected] 4a. Results: Individual Ice Core Calibra?on 5. Conclusions 6. Future Work Background •Mt. Logan annual [Na + ] (19481998) varies linearly with the strength of the Aleu&an Low (r=0.50, p<0.05) 1 •Eclipse and Mt. Logan ice core records are not well correlated, because they sample different air masses 2,3 (Fig. 2) •Stable isotope records and models suggest the moisture source at Mt. Logan is more distant than Eclipse 4,5 •The ~10% highest and lowest ice core values for Eclipse coldseason isotopes and accumula&on, and Mt. Logan accumula&on are consistently associated with strong and weak Aleu&an Low strength, respec&vely 6,7,8 •Here, we develop a novel calibra&on technique to isolate the climate value of these highest and lowest ice core values and allow for spa&al nonlineari&es by calcula&ng the frequency of sealevel pressure anomaly paTerns for small groups of ranked ice core values Fig. 4. Mean DJFM sealevel pressure (SLP) for 18722008 based on the 20th Century Reanalysis modelderived data9. SLP anomalies calculated in Figures 47 are calculated using this mean. The solid black circle iden&fies the loca&on of the Eclipse and Mt. Logan ice core sites. Ice Core Sta?s?cs Eclipse Mt. Logan Eleva&on 3017 m 5340 m Core length 345 m 186 m Accumula&on rate 1.38 m w.e. a 1 0.41 m w.e. a 1 Time period 1004 AD – 2001 AD 14000 yr BP 1998 Sample resolu&on 1015 cm 15 cm Reference horizons since 1872 40 10 Da&ng Error back to 1600 AD ~1% ~12% Ice Core Records Analyzed Accumula?on, δD, Na + , Mg 2+ ,K + , Ca 2+ , Cl , NO3 , SO4 , NH4 + Accumula&on, δ 18 O, K + , Mg 2+ , Na + , Ca 2+ , Cl , NO3 , SO4 , NH4 + •The two highest and two lowest [Na + ] groups contain consistent spa&al SLP anomaly paTerns (90100% of years) of high average magnitude (Fig. 5) •Spa&al nonlineari&es are evident between high and low ice core values (Figs. 5 and 6), and between the two highest [Na + ] groups •Short atmospheric residence &me of Na + may explain more distant –SLP anomalies in for second highest [Na + ] group •Surprisingly, increased storminess (–SLP anomalies) occurred in the western Pacific, far from Eclipse, for high accumula&on winters •Because the highest [Na + ] and accumula&on groups have overlapping –SLP anomalies, and the lowest [Na + ] and accumula&on groups have overlapping +SLP anomalies, frequency plots were made for shared and unshared years (Figs. 6 and 7) to help develop conceptual models that explain how these ice core records are built Fig. 6. Frequency plots for Eclipse winter accumula&on. Only the highest and lowest groups exhibit significant climate anomalies. The other groups (27th39th, 40th52nd and 53rd65th highest and lowest values; not shown) do not have significant climate anomalies. 4b. Results: Combined Ice Core Calibra?on Fig. 6. Frequency plots for a) the four years shared by the highest accumula&on group and the two highest [Na+] groups, b) the nine unshared high accumula&on years, and c) the 10 unshared high [Na+] years for the second highest [Na+] group. Fig. 7. Frequency plots for a) the five years shared by the lowest accumula&on group and the two lowest [Na+] groups, b) the eight unshared high accumula&on years, and c) the 21 unshared high [Na+] years. Colored squares correspond to those in Fig. 8. Fig. 3. Mt. Logan annual mean [Na+] for 18722008. Black and blue numbers denote the highest and lowest 13 [Na+] years, respec&vely. We greatly appreciate the help from Dr. Jack Dibb and Dr. Joe Licciardi to improve this work. This work was funded by a Disserta&on Year Fellowship from the Univ. of New Hampshire Graduate School. •Mul&ple modes of winter SLP anomaly paTerns are revealed when separa&ng shared and unshared years •The –SLP anomalies (Fig. 6) are colocated with two regions of high Pacific cyclogenesis 11 , therefore, high accumula&on and high [Na + ] could be indicators of North Pacific cyclogenesis •Frequency plots of high and low accumula&on, and high and low [Na + ] exhibit spa&al nonlineari&es •Different dynamical characteris&cs of individual winter storms likely explain the different years of similar SLP anomaly paTerns between Fig. 6b and 6c •The calibrated ice core records agree with the winter North Pacific Index (NPI): 41 winters correct (21 high NPI, 20 low NPI), 12 incorrect, 49 winters without calibrated ice core values 1Osterberg, E.C., Mayewski, P.A., Kreutz, K.J., Fisher, D.A., Maasch, K.A., Sneed, S.B., in review. J. Geophys. Res.Atmospheres. 2Holdsworth, G.H., 2008. J. Geophys. Res., D08103, doi:10.1029/2007JD008639. 3Zdanowicz, C., G. Hall, J. Vaive, Y. Amelin, J. Percival, and others, 2006. Geochimica et Cosmochimica Acta, 70, 34933507. 4Fisher, D., C. Wake, K. Kreutz, K. Yalcin, E. Steig, P. Mayewski, and others, 2004. Geographie physique et Quaternaire, 58, 337352. 5Fisher, D., E. Osterberg, A. Dyke, and others, 2008. The Holocene, 18, 667677. 6Kelsey, E.P., C.P. Wake, K. Yalcin and K. Kreutz, 2012. J. Climate, 25, 64266440. 7Moore, G.W.K.,K. Alverson, and G. Holdsworth, 2002. Ann. of Glaciol., 35, 423429. 8Rupper, S., E.J. Steig, and G. Roe, 2004. J. Climate, 17, 47244739. 9Yalcin, K., C.P. Wake, K.J. Kreutz, M.S. Gemani, and S.I. Whitlow, 2007. J. Geophys. Res., 112, D08102, doi:10.1029/2006JD007497. 10Compo, G.P., J.S. Whitaker, P.D. Sardeshmukh, and others. Quarterly J. Roy. Meteorol. Soc., 137, 128. DOI: 10.1002/qj.776. 11Hoskins, B.J., and K.I. Hodges, 2002. J. Atmo. Sci., 59, 10411061. •Our new calibra&on technique iden&fied mul&ple SLP modes that produce these ice core signals •Increased winter storminess nearby and in the western Pacific produce high accumula&on at Eclipse •The rela&onship between ice core values and SLP is nonlinear; spa&al nonlineari&es were found between high and low groups of ice core values, and between the first and second highest [Na + ] groups •These ice core records may indicate the strength of cyclogenesis in the two dominant cyclogenic regions of the North Pacific •Significant spa&al nonlineari&es between the ice core records and SLP raise the possibility that other ice core records have similar nonlineari&es with climate variables •Perform intraseasonal analysis of meteorological features (e.g., storm tracks, moisture flux) to understand the types of cyclones and an&cyclones that are responsible for these ice core signals, especially high accumula&on at Eclipse •Use these results to constrain error bars on the Osterberg et al. (in review) 1 reconstruc&on of Aleu&an Low strength •Perform this new calibra&on on North Atlan&c ice cores 7. Acknowledgements Fig. 1. Map of the loca&ons of the Eclipse and Mt. Logan PR Col ice cores in the coastal St. Elias Mountain Range, Yukon, Canada. Adapted from Fisher et al. 20044. 3017 m 5400 m Fig. 2. Map of Eclipse and Mt. Logan PR Col eleva&ons with respect to isotopic layers in an extratropical cyclone as proposed by Holdsworth (2008) 2. Adapted from Holdsworth 20082. Fig. 5. Winter SLP anomaly frequency plots for groups of ranked annual [Na+] years at Mt. Logan for 18721998. hPa Mt. Logan Annual [Na + ] Contours: show the percentage of years in each 13year group that exhibited posi&ve (dark red) or nega&ve (blue) SLP anomalies at each grid point. Shading: indicates posi&ve (orange) and nega&ve (green) average SLP anomalies at 0.5 hPa intervals star&ng at +/0.5 hPa. Gray dashed line encloses sta&s&cally significant SLP anomalies (p<0.05). yr yr yr low accum & low [Na + ] low accum only low [Na + ] only Eclipse October-March Accumulation yr yr yr yr yr high accum & high [Na + ] high accum only high [Na + ] only 2013 AMS Annual Conference