2013AMS Poster ECLP LOGP Calibration -...

A  Surprise  in  the  North  Pacific:  Results  from  Applying  a  New  Nonlinear  Method  for  Calibra?ng  Ice  Cores    

Eric  Kelsey123,  Cameron  Wake1,  Erich  Osterberg4  and  Karl  Kreutz5  1Ins&tute  for  the  Study  of  Earth,  Oceans,  and  Space,  University  of  New  Hampshire,  Durham,  NH,  2Dept.  of  Atmo.  Sci.  and  Chem.,  Plymouth  State  University,  Plymouth,  NH  

3Mount  Washington  Observatory,  Gorham,  NH,  4Dept.  of  Earth  Science,  Dartmouth  College,  Hanover,  NH,  5Climate  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  • Eclipse9  and  Mt.  Logan1,4  ice  core  data  • Winter  (DJFM)  mean  SLP  gridded  fields  (2o  x  2o)  for  the   Northern   Hemisphere   north   of   20oN   for  1871-­‐2008   from   the   20th   Century   Reanalysis  dataset10  are  used  for  ice  core  calibra&on  

3.  Methodology  • We   divided   the   ranked   ice   core   values   (1872-­‐2001)   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  km2)  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:  e.kelsey@plymouth.edu  

4a.  Results:  Individual  Ice  Core  Calibra?on  

5.  Conclusions  

6.  Future  Work  

Background  • Mt.   Logan   annual   [Na+]   (1948-­‐1998)   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  masses2,3  (Fig.  2)  • Stable   isotope   records   and   models   suggest   the  moisture   source   at   Mt.   Logan   is   more   distant   than  Eclipse4,5  

• The   ~10%   highest   and   lowest   ice   core   values   for  Eclipse   cold-­‐season   isotopes   and   accumula&on,   and  Mt.   Logan   accumula&on   are   consistently   associated  with   strong   and   weak   Aleu&an   Low   strength,  respec&vely6,7,8  

• 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   sea-­‐level   pressure  anomaly  paTerns   for   small   groups  of   ranked   ice   core  values  

Fig.  4.  Mean  DJFM  sea-­‐level  pressure  (SLP)  for  1872-­‐2008  based  on  the  20th  Century  Reanalysis  model-­‐derived  data9.  SLP  anomalies  calculated  in   Figures   4-­‐7   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   10-­‐15  cm   1-­‐5  cm  Reference  horizons  since  1872   40   10  

Da&ng  Error  back  to  1600  AD   ~1%   ~1-­‐2%  

Ice  Core  Records  Analyzed  

Accumula?on,  δD,  Na+,  Mg2+,  K+,  Ca2+,  Cl-­‐,  NO3

-­‐,  SO4

-­‐,  NH4+  

Accumula&on,  δ18O,  K+,  Mg2+,  Na+,  Ca2+,  Cl-­‐,  NO3

-­‐,  SO4-­‐,  NH4

+  

• The   two   highest   and   two   lowest   [Na+]   groups   contain  consistent   spa&al   SLP   anomaly   paTerns   (90-­‐100%   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  (27th-­‐39th,  40th-­‐52nd   and   53rd-­‐65th   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   1872-­‐2008.   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  cyclogenesis11,  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,  3493-­‐3507.  4Fisher,  D.,  C.  Wake,  K.  Kreutz,  K.  Yalcin,  E.  Steig,  P.  Mayewski,  and  others,  2004.  Geographie  physique  et  Quaternaire,  58,  337-­‐352.  5Fisher,  D.,  E.  Osterberg,  A.  Dyke,  and  others,  2008.  The  Holocene,  18,  667-­‐677.  6Kelsey,  E.P.,  C.P.  Wake,  K.  Yalcin  and  K.  Kreutz,  2012.  J.  Climate,  25,  6426-­‐6440.  7Moore,  G.W.K.,K.  Alverson,  and  G.  Holdsworth,  2002.  Ann.  of  Glaciol.,  35,  423-­‐429.  8Rupper,  S.,  E.J.  Steig,  and  G.  Roe,  2004.  J.  Climate,  17,  4724-­‐4739.  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,  1-­‐28.  DOI:  10.1002/qj.776.  11Hoskins,  B.J.,  and  K.I.  Hodges,  2002.  J.  Atmo.  Sci.,  59,  1041-­‐1061.  

• 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,  C a n a d a .  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 .   Adap ted   f r om  Holdsworth  20082.  

Fig.  5.  Winter  SLP  anomaly  frequency  plots  for  groups  of  ranked  annual  [Na+]  years  at  Mt.  Logan  for  1872-­‐1998.  

hPa  

Mt. Logan Annual [Na+]

Contours:  show  the  percentage  of  years   in  each  13-­‐year  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

both high

yr yr yr yr yr

high accum & high [Na+] high accum only high [Na+] only

2013  AMS  Annual  Conference