<%BANNER%>

Tests for Orbital Influences on the Geomagnetic Field, and Quarternary Magnetic Records from North Atlantic and Arctic D...

Permanent Link: http://ufdc.ufl.edu/UFE0041271/00001

Material Information

Title: Tests for Orbital Influences on the Geomagnetic Field, and Quarternary Magnetic Records from North Atlantic and Arctic Deep-Sea Sediments
Physical Description: 1 online resource (248 p.)
Language: english
Creator: Xuan, Chuang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: arctic, atlantic, excursions, geodynamo, magnetostratigraphy, matlab, orbital, paleointensity, paleomagnetism, periods, reversals, sediments, selfreversal, software, titanomaghemite
Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This dissertation investigated the possible connection between orbital variations and the Earth's magnetic field, and the origin of orbital periods in sedimentary relative paleointensity (RPI) records, using previously published data. Circular statistic methods were utilized to test whether there is any consistent relationship between the phase of orbital parameters and the timing of geomagnetic reversals or excursions. The results indicate no discernable tendency, disagreeing with orbital forcing on the geodynamo. Numerical simulations further indicate that precision of the current polarity timescales need to be improved for any firm relationship to be established. Wavelet analyses methods were employed to investigate the origin of orbital periods in the RPI records. In some records, significant coherence at orbital periods occurs between RPI and a particular magnetic grain-size proxy. Therefore, orbital periods in some RPI records are attributed to lithologic ?contamination? resulted from incomplete normalization of the natural remanent magnetization (NRM) record. Comparison of RPI records from different regions of the world in both the time and time-frequency domains imply that the ?contamination? does not debilitate most RPI records as a global signal that is primarily of geomagnetic origin. Calibrated RPI and oxygen isotope stack records (PISO-1500) were developed by simultaneously matching and stacking both RPI and oxygen isotope data for 13 pairs of high-resolution global records. Wavelet analyses on the PISO-1500 RPI stack record failed to show significant orbital periods, and no tendencies were found for RPI minima in the stack to occur at particular phases of orbital variations. The generation of high-resolution paleomagnetic data is often associated with processing large volumes of measurement data. MATLAB? software with graphical user interfaces was developed in this dissertation work to improve the efficiency of processing large volumes of paleomagnetic data and facilitates the calculation of paleomagnetic directions and RPI proxies. This new software incorporates new methods of analysis, particularly in the generation of RPI proxies. U-channel NRM measurements at Integrated Ocean Drilling Program (IODP) Site U1304 yield continuous high resolution paleomagnetic records for the last ~1.5 Ma. Sediments from IODP Site U1304 clearly recorded the Brunhes/Matuyama boundary, the Jaramillo subchron, and the Cobb Mountain subchron, as well as the Kamikatsura excursion and the Gardar excursion. Age model for the site is established by correlating IODP Site U1304 RPI record to the PISO-1500 RPI stack using automated dynamic programming method with limited number of tie points. No significant orbital periods were detected in RPI record from the site. Various evidences indicate that the episodic deposits of laminated diatom ooze throughout the IODP Site U1304 sediments, appear to dilute the magnetic concentrations of the sediments with elevated sedimentation rates, but do not debilitate the reliability of the acquired paleomagnetic direction and intensity data. Rock magnetic experiments carried out under various temperature ranges, along with scanning electron microscopy (SEM) and X-ray energy-dispersive spectroscopy (EDS) observations as well as X-ray diffraction (XRD) analyses, on bulk Arctic deep-sea sediments and magnetic extracts from seven cores collected by the Healy-Oden Trans-Arctic Expedition 2005 (HOTRAX05), indicate that (titano)magnetite and titanomaghemite are the magnetic remanence carriers. It appears that the titanomaghemite carries a chemical remanent magnetization (CRM) that is partially self-reversed relative to the detrital remanent magnetization (DRM) carried by the host titanomagnetite, causing the apparent magnetic ?excursions? in the Arctic deep-sea sediment records. The partial self-reversal could have been accomplished by ionic ordering during oxidation, thereby changing the balance of the magnetic moments in the ferrimagnetic sublattices that characterize titanomagnetite and titanomaghemite. The partial self-reversal process appears to have affected all the studied cores from different regions of the Arctic Ocean, with less alteration in core from the Yermak Plateau located at the edge of the Arctic Ocean.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Chuang Xuan.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Channell, James E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041271:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041271/00001

Material Information

Title: Tests for Orbital Influences on the Geomagnetic Field, and Quarternary Magnetic Records from North Atlantic and Arctic Deep-Sea Sediments
Physical Description: 1 online resource (248 p.)
Language: english
Creator: Xuan, Chuang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: arctic, atlantic, excursions, geodynamo, magnetostratigraphy, matlab, orbital, paleointensity, paleomagnetism, periods, reversals, sediments, selfreversal, software, titanomaghemite
Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This dissertation investigated the possible connection between orbital variations and the Earth's magnetic field, and the origin of orbital periods in sedimentary relative paleointensity (RPI) records, using previously published data. Circular statistic methods were utilized to test whether there is any consistent relationship between the phase of orbital parameters and the timing of geomagnetic reversals or excursions. The results indicate no discernable tendency, disagreeing with orbital forcing on the geodynamo. Numerical simulations further indicate that precision of the current polarity timescales need to be improved for any firm relationship to be established. Wavelet analyses methods were employed to investigate the origin of orbital periods in the RPI records. In some records, significant coherence at orbital periods occurs between RPI and a particular magnetic grain-size proxy. Therefore, orbital periods in some RPI records are attributed to lithologic ?contamination? resulted from incomplete normalization of the natural remanent magnetization (NRM) record. Comparison of RPI records from different regions of the world in both the time and time-frequency domains imply that the ?contamination? does not debilitate most RPI records as a global signal that is primarily of geomagnetic origin. Calibrated RPI and oxygen isotope stack records (PISO-1500) were developed by simultaneously matching and stacking both RPI and oxygen isotope data for 13 pairs of high-resolution global records. Wavelet analyses on the PISO-1500 RPI stack record failed to show significant orbital periods, and no tendencies were found for RPI minima in the stack to occur at particular phases of orbital variations. The generation of high-resolution paleomagnetic data is often associated with processing large volumes of measurement data. MATLAB? software with graphical user interfaces was developed in this dissertation work to improve the efficiency of processing large volumes of paleomagnetic data and facilitates the calculation of paleomagnetic directions and RPI proxies. This new software incorporates new methods of analysis, particularly in the generation of RPI proxies. U-channel NRM measurements at Integrated Ocean Drilling Program (IODP) Site U1304 yield continuous high resolution paleomagnetic records for the last ~1.5 Ma. Sediments from IODP Site U1304 clearly recorded the Brunhes/Matuyama boundary, the Jaramillo subchron, and the Cobb Mountain subchron, as well as the Kamikatsura excursion and the Gardar excursion. Age model for the site is established by correlating IODP Site U1304 RPI record to the PISO-1500 RPI stack using automated dynamic programming method with limited number of tie points. No significant orbital periods were detected in RPI record from the site. Various evidences indicate that the episodic deposits of laminated diatom ooze throughout the IODP Site U1304 sediments, appear to dilute the magnetic concentrations of the sediments with elevated sedimentation rates, but do not debilitate the reliability of the acquired paleomagnetic direction and intensity data. Rock magnetic experiments carried out under various temperature ranges, along with scanning electron microscopy (SEM) and X-ray energy-dispersive spectroscopy (EDS) observations as well as X-ray diffraction (XRD) analyses, on bulk Arctic deep-sea sediments and magnetic extracts from seven cores collected by the Healy-Oden Trans-Arctic Expedition 2005 (HOTRAX05), indicate that (titano)magnetite and titanomaghemite are the magnetic remanence carriers. It appears that the titanomaghemite carries a chemical remanent magnetization (CRM) that is partially self-reversed relative to the detrital remanent magnetization (DRM) carried by the host titanomagnetite, causing the apparent magnetic ?excursions? in the Arctic deep-sea sediment records. The partial self-reversal could have been accomplished by ionic ordering during oxidation, thereby changing the balance of the magnetic moments in the ferrimagnetic sublattices that characterize titanomagnetite and titanomaghemite. The partial self-reversal process appears to have affected all the studied cores from different regions of the Arctic Ocean, with less alteration in core from the Yermak Plateau located at the edge of the Arctic Ocean.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Chuang Xuan.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Channell, James E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041271:00001


This item has the following downloads:


Full Text

PAGE 1

1 TESTS FOR ORBITAL INFLUENCES ON THE GEOMAGNETIC FIELD, AND QUARTERNARY MAGNETIC RECORDS FROM NORTH ATLANTIC AND ARCTIC DEEPSEA SEDIMENTS By CHUANG XUAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Chuang Xuan

PAGE 3

3 To my parent s and my wife

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my advisor, Jim Channell, for giving me the great opportunity to study in this Ph.D program, and for all his help and constant support over the last few years. I also thank my dissertation committee members Neil Opdyke, Dave Hodell, Mark Yang, Ray Russo, and Joe Meert for supervis ing my Ph.D. research and for their sugg estions that have improved my work This dissertation also benefited from help, contribution and support from many other colle a gues : Kainian Huang, Ray Thomas Ann Heatherington, Leonid Polyak, Jason Curtis, and John Jaeger. My research was made possible by financial support of sources including the University of Florida Alumni Fellowship the University of Florida Gibson Dissertation Fellowship, and the National Science Foundation. Support was also provided by the College of Liberal Arts and Sciences, the Graduate School, and the department of Geological Sciences at the University of Florida, and by the Institute for Rock Magnetism at the University of Minnesota.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 ABSTRACT ................................................................................................................... 13 CHAPTER 1 INTRODUCTION .................................................................................................... 16 2 TESTING THE RELATIONSHIP BETWEEN TIMING OF GEOMAGNETIC REVERSALS/EXCURSIONS AND PHASE OF ORBITAL CYCLES USING CIRCULAR STATISTICS AND MONTE CARLO SIMULATIONS ........................... 19 Introduction ............................................................................................................. 19 Data ........................................................................................................................ 20 Methods .................................................................................................................. 23 Phase Calculation ............................................................................................ 23 Circular Statistics .............................................................................................. 23 Monte Carlo Simulation .................................................................................... 24 Results .................................................................................................................... 25 Conclusions ............................................................................................................ 30 3 ORIGIN OF ORBITAL PERIODS IN THE SEDIMENTARY RELATIVE PALEOINTENSITY RECORDS .............................................................................. 43 Introduction ............................................................................................................. 43 The Recognition and Interpretation of Orbital Cycles in RPI Records .................... 46 Data and Methods .................................................................................................. 48 Results and Discussion ........................................................................................... 53 Conclusions ............................................................................................................ 62 4 STACKING PALEOINTENSITY AND OXYGEN ISOTOPE DATA FOR THE LAST 1.5 MYR (PISO 1500) ................................................................................... 74 Introduction ............................................................................................................. 74 The Match and the Stack ........................................................................................ 76 Comparison with other Stacks ................................................................................ 80 Discussion .............................................................................................................. 81 Conclusions ............................................................................................................ 86

PAGE 6

6 5 UPMAG: MATLAB SOFTWARE FOR VIEWING AND PROCESSING UCHANNEL OR OTHER PASSTHROUGH PALEOMAGNETIC DATA ................... 99 Introduction ............................................................................................................. 99 Description of UPmag ........................................................................................... 101 UVIEW ............................................................................................................ 102 UDIR .............................................................................................................. 104 UINT ............................................................................................................... 106 Conclusions .......................................................................................................... 107 6 QUATERNARY PALEOMAGNETIC RECORD FROM DIATOM RICH SEDIMENTS AT IODP SITE U1304 (SOUTHERN GARDAR DRIFT, NORTH ATLANTIC) ........................................................................................................... 117 Introduction ........................................................................................................... 117 Sampling and Methods ......................................................................................... 118 Rock Magnetic Properties of the Sediments ......................................................... 120 Natural Remanent Magnetization and RPI Proxies ............................................... 121 Paleointensity Based Age Model and Detection of Orbital Periods in the RPI Record ............................................................................................................... 125 Conclusions .......................................................................................................... 128 7 SELF REVERSAL AND APPARENT MAGNETIC EXCURSIONS IN ARCTIC SEDIMENTS ......................................................................................................... 142 Introduction ........................................................................................................... 142 Magnetic Properties of Core HLY05036JPC ....................................................... 145 XRD and SEM Observations ................................................................................ 149 Self reversal in Titanomaghemite ......................................................................... 151 Conclusions .......................................................................................................... 153 8 ORIGIN OF APPARENT MAGNETIC EXCURSIONS IN DEEPSEA SEDIMENTS FROM MENDELEEVALPHA RIDGE (ARCTIC OCEAN) ............... 164 Introduction ........................................................................................................... 164 NRM Measurements ............................................................................................. 166 Rock Magnetic Studies ......................................................................................... 169 SEM and XRD Analyses ....................................................................................... 174 Discussion ............................................................................................................ 176 Conclusions .......................................................................................................... 179 9 PALEOMAGNETIC AND ROCK MAGNETIC STUDIES ON DEEPSEA SEDIMENTS FROM LOMONOSOV RIDGE AND YERMAK PLATEAU ............... 192 Introduction ........................................................................................................... 192 Materials and Methods .......................................................................................... 193 Results and Discussions ....................................................................................... 196 Conclusions .......................................................................................................... 204

PAGE 7

7 10 CONCLUSIONS AND F UTURE WORK ............................................................... 216 LIST OF REFERENCES ............................................................................................. 225 BIOGRAPHICAL SKETCH .......................................................................................... 247

PAGE 8

8 LIST OF TABLES Table page 2-1 Ages of the 8 best established excursions in the Brunhes and Matuyama Chrons ................................................................................................................ 33 2-2 pvalue of Rayleigh test for phase data in Figure 23, Figure 24, and Figure 2-5 ...................................................................................................................... 34 2-3 Monte Carlo simulation of influence of age uncertainties and number of data points on Rayleigh test p value ........................................................................... 35 2 -4 pvalue of Rayleigh tests for phases of eccentricity and maximum obliquity envelope ............................................................................................................. 36 3-1 Location, water depth, length, sedimentation rate, normalizer, and estimated age for relative paleointensity records used ....................................................... 65 4-1 Records used in the construction of the PISO 1500 stac ks. ............................... 88 4-2 Correlation coefficient and percentage of significant area .................................. 89 5-1 The default UF measurement data file format .................................................. 108 5-2 The default UF *.dir file format .......................................................................... 109 7-1 ............... 155 8-1 Location, length, water depth, and age model information for cores studied in this chapter ....................................................................................................... 181 8-2 titanomaghemite. .............................................................................................. 182

PAGE 9

9 LIST OF FIGURES Figure page 2-1 Comparing differences in different reversal time scales .................................... 37 2-2 Definition of phase of orbital cycles corresponding to a reversal/excursion. ....... 38 2-3 Circular plot of phase of actual obliquity ............................................................ 39 2-4 Circular plot of phase of actual eccentricity ....................................................... 40 2-5 Circular plot of phase of maximum obliquity envelope. ....................................... 41 2-6 Comparing relative paleointensity records with orbital obliquity ........................ 42 3-1 Location of sites discussed in this study. ............................................................ 66 3-2 Wavelet analyses for ETP curve and LR04. ....................................................... 67 3-3 Percentage of significant area at different significance levels ........................... 68 3-4 Local wavelet power spectra of RPI, NRM, and ARM/IRM records ................... 69 3-5 Cross wavelet power spectra between RPI record and other parameters .......... 70 3-6 Wavelet analyses for synthetic RPI and ARM records. ...................................... 71 3-7 Squared wavelet coherence between physical parameters, RPI records, and oxygen isotope records ..................................................................................... 72 3-8 Comparing RPI records in the time domain and timefrequency domain. ........... 73 4-1 Location of the 13 coupled isotope and relative paleointensity records ............. 90 4-2 Oxygen isotope and RPI data from the 13 sites after matching. ......................... 91 4-3 Squared wavelet coherence between RPI records and bertween oxygen isotope records before and after matching. ........................................................ 92 4-4 The PISO 1500 oxygen isotope and RPI stacks ................................................ 93 4-5 PISO 1500 compared with other RPI stack records .......................................... 94 4-6 Wavelet analyses for PISO 1500 oxygen isotope stack and RPI stack ............. 95 4-7 Virtual axial dipole moment calibration of the PISO 1500 RPI stack ................. 96

PAGE 10

10 4-8 Circular plot of phases of orbital eccentricity, obliquity and precession corresponding to RPI lows in PISO 1500 RPI stack .......................................... 97 4-9 Comparison of the PISO 1500 RPI stack with oxygen isotope stack ................. 98 5-1 A flow chart summarizing the logic and major steps in the UPmag software. ... 110 5-2 The UVIEW graphical user interface ............................................................... 111 5-3 The UDIR graphical user interface. ................................................................. 113 5-4 The UINT graphical user interface. ................................................................... 116 6-1 Location of IODP Site U1304 ........................................................................... 130 6-2 Hysteresis ratio s of IODP Site U1304 sediments on D AY plot ......................... 131 6-3 ARM versus plot for IODP Site U1304 sediments ......................................... 132 6-4 Cores splice image, diatom distribution, and magnetic concentrations at IODP Site U1304 .............................................................................................. 133 6-5 Component directions from IODP Site U1304 compared to diatom distribution at the site.. ........................................................................................................ 134 6-6 Orthogonal projections of NRM demagnetization data for representative samples of IODP Site U1304 ........................................................................... 135 6-7 Histograms of component inclinations for samples from diatom rich and nondiatom intervals ............................................................................................... 136 6-8 RPI proxies from IODP Site U1304 with R values compared to diatom distribution. ....................................................................................................... 137 6-9 IODP Site U1304 susceptibility, inclination, and RPI records on age, compared to PISO 1500. .................................................................................. 138 610 Depthage curve and sedimentation rates for IODP Site U1304, compared to diatom distribution at the site and the PISO 1500 oxygen isotope stack. ......... 139 611 Power spectrum of the IODP Site U1304 RPI record ...................................... 140 612 Wavelet analyses on IODP Site U1304 RPI and PISO 1500 RPI stack .......... 141 7-1 Location of Core HLY0503-0 6JPC. .................................................................. 156 7-2 Core HLY05036JPC component directions and NRM int ensity ...................... 157

PAGE 11

11 7-3 Orthogonal projection of AF demagnetization and thermal demagnetization for Core 0 6JPC samples ................................................................................. 158 7-4 Thermal demagnetization of threeaxis isothermal remanent magnetizations of Core 0 6JPC bulk sediments ......................................................................... 159 7-5 Magnetization derived from hysteresis loops measured at increasing temperatures and magnetization measured during thermal cycling. ................ 160 7-6 Low temperature rock magnetic measurement data for Core 06JPC samples. .......................................................................................................... 161 7-7 EDS elemental mapping of micronsized grains from an unheated magnetic extract from Core 06JPC ................................................................................. 162 7-8 XRD results for unheated and heated magnetic extracts from Core 06JPC .... 163 8-1 Location of Cores 08JPC, 10JPC, 11JPC, and 13JPC retrieved by the HOTRAX05 ...................................................................................................... 183 8-2 Component direction data of Cores 08JPC, 10JPC, 11JPC, and 13JPC ........ 184 8-3 Orthogonal projection of thermal demagnetization and AF demagnetization for samples from Cores 08JPC, 10JPC, 11JPC, and 13JPC .......................... 185 8-4 Saturation magnetization and coercivity of remanence derived from hysteresis loops measured at increasing temperatures, and magnetizations monitored during heating and cooling. .............................................................. 186 8-5 Low temperatrue rock magnetic measurement data. ....................................... 187 8-6 IRM acquisition curves for magnetic extracts from Cores 08JPC and 10JPC 188 8-7 SEM and EDS analyses for micronsized grains of magnetic extracts from Cores 08JPC and 10JPC sediments ............................................................... 189 8-8 XRD results for Cores 08JPC and 10JPC samples ......................................... 190 8-9 Conceptual models for the coercivity spectra and the blocking temperature spectra. ............................................................................................................. 191 9-1 Location of HLY0503 Cores 20JPC and 22JPC. .............................................. 206 9-2 Component direction records of Cores 20JPC and 22JPC, and ARM record from Core 22JPC. ............................................................................................. 207 9-3 ARM of Core 22JPC on age model constructed by correlating ARM to PISO 1500. ................................................................. 208

PAGE 12

12 9-4 Orthogonal projection of thermal demagnetization and AF demagnetization with NRM intensity versus thermal steps for samples from Cores 20 JPC and 22JPC. .............................................................................................................. 209 9-5 Thermal demagnetization of threeaxis isothermal remanent magnetizations for bulk sediment samples from Cores 20JPC and 22JPC .............................. 210 9-6 IRM acquisition data for magnetic extracts from Cores 20JPC and 22JPC ...... 211 9-7 M odeling of the IRM gradient data for magnetic extracts from Cores 20JPC and 22JPC. ....................................................................................................... 212 9-8 Susceptibility of magnetic extracts form Cores 20JPC and 22JPC monitored on heating and cooling .................................................................................... 213 9-9 SEM and EDS analyses for micronsized grains of magnetic extracts from Cores 20JPC and 22JPC. ................................................................................ 214 910 High resolution XRD results for Cores 20JPC and 22JPC magnetic extracts 215

PAGE 13

13 Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy TESTS FOR ORBITAL INFLUENCES ON THE GEOMAGNETIC FIELD, AND QUARTERNARY MAGNETIC RECORDS FROM NORTH ATLANTIC AND ARCTIC DEEP-SEA SEDIMENTS By Chuang Xuan May 2010 Chair: James E.T. Channell Major: Geology This dissertation investigated the possibl e connection between orbital variations and the Earth's magnetic field, and the origin of orbital periods in sedimentary relative paleointensity (RPI) records, using previously published data Circular statistic methods were utilized to test whether there is any consistent relationship between the phase of orbital parameters and the ti ming of geomagnetic reversals or excursions. The results indicate no discernable tendency, disagreeing wit h orbital forcing on the geodynamo. Numerical simulations further indicate that precision of the current polarity timescales need to be improved for any firm relationshi p to be established. Wavelet analyses methods were employed to investigate the origin of orbital periods in the RPI records. In some records, significant coherence at orbital periods occurs between RPI and a particular magnetic grain-size pr oxy. Therefore, orbital peri ods in some RPI records are attributed to lithologic contamination resulted from incomplete normalization of the natural remanent magnetization (NRM) reco rd. Comparison of RPI records from different regions of the world in both the time and time-frequency domains imply that the contamination does not debilitate most RPI re cords as a global signal that is primarily

PAGE 14

14 of geomagnetic origin. Calibrated RPI and o xygen isotope stack records (PISO-1500) were developed by simultaneously matching and stacking both RPI and oxygen isotope data for 13 pairs of high-resolution global records. Wavelet analyses on the PISO-1500 RPI stack record failed to show signific ant orbital periods, and no tendencies were found for RPI minima in the stack to occur at particular phases of orbital variations. The generation of high-resolution paleoma gnetic data is often associated with processing large volumes of measurement data. MATL AB software with graphical user interfaces was developed in this disse rtation work to improve the efficiency of processing large volumes of paleomagnetic data and facilitates t he calculation of paleomagnetic directions and RPI proxies. This new software incorporates new methods of analysis, particularly in the generation of RPI proxies. U-channel NRM measurements at Integrated Ocean Drilli ng Program (IODP) Site U1304 yield continuous high resolution pale omagnetic records for the last ~1.5 Ma. Sediments from IODP Site U1304 clearly recorded the Brunhes/Matuyama boundary, the Jaramillo subchron, and t he Cobb Mountain subchron, as well as the Kamikatsura excursion and the Gardar excurs ion. Age model for the site is established by correlating IODP Site U1304 RPI record to the PISO -1500 RPI stack using automated dynamic programming method with limited number of tie points. No significant orbital periods were detected in RPI record from the site. Various evidences indica te that the episodic deposits of laminated diatom ooze throughout the IODP Site U1304 sediments, appear to dilute the magnetic concentrations of the sediments with el evated sedimentation rates, but do not debilitate the reliability of the acquired paleomagnetic direction and intensity data.

PAGE 15

15 Rock magnetic experiments carried out under various temperature ranges, along with scanning electron microscopy (SEM) an d X-ray energy-dispersive spectroscopy (EDS) observations as well as X-ray diffrac tion (XRD) analyses, on bulk Arctic deep-sea sediments and magnetic extracts from seven cores collected by the Healy-Oden TransArctic Expedition 2005 (HOTRAX05), indicate that (titano)magnetite and titanomaghemite are the magnetic rem anence carriers. It appears that the titanomaghemite carries a chemical remanent m agnetization (CRM) that is partially selfreversed relative to the detrital remanent magnetization (DRM) carried by the host titanomagnetite, causing t he apparent magnetic excursions in the Arctic deep-sea sediment records. The partial self-reve rsal could have been accomplished by ionic ordering during oxidation, t hereby changing the balance of the magnetic moments in the ferrimagnetic sublattices that characte rize titanomagnetite and titanomaghemite. The partial self-reversal process appears to have affected all the studied cores from different regions of the Arctic Ocean, with less alterati on in core from the Yermak Plateau located at the edge of t he Arctic Ocean.

PAGE 16

16 CHAPTER 1 INTRODUCTION Recognition of the Earths magnetic field can be traced back to the 2nd century AD when the ancient Chinese discovered that magnets tend to align themselves in the north-south direction. Since then, the a ccumulation of knowledge on the Earths magnetic has revolutionized means of navigation, and led to further interest on the direction and intensity of the geomagnetic field, and the application of the paleomagnetic record in the geosciences. Be yond the limited dire ct measurement or documentation of the Earths magnetic field, which extend back only a few centuries, information on the ancient geomagnetic field relies on the natural remanent magnetization (NRM) carried by rocks and sediments. Paleomagnetic direction and intensity data acquired from sediments and volcanics have been critical for providing continuous high-quality magnetic records that are important for understanding the geodynamo as well as for stratigraphic applications. Magnetic studies on deep-sea sediments have highlighted topical questions that need further investigation. For instance, or bital periods have been reported in a number of globally-distributed sedimentary relative paleointensity (RPI) records. Do they imply orbital forcing on the geodynamo or environmental/lithologic contamination of the RPI records? Brunhes-aged deep-sea sediments from the Arctic Ocean often record decimeter-scale negative inclination interval s. Are those negative in clination intervals due to magnetic excursions and special behavior of the geom agnetic field in the Arctic area, or do they represent post-depositional distortions of the paleomagnetic records? Continuous sedimentary magnetic records that span the last ~1.5 m illion years are fairly rare, especially in sediments that have mean sedimentation ra te of >10 cm/kyr.

PAGE 17

17 Instrumental advances that allow rapid dat a accumulation from deep-sea sediments has generated the need for processing data in real time through suitable data visualization and analytical software. This dissertation comprises 8 main c hapters that attempt to address the above mentioned issues, along with this introducti on chapter (Chapter 1) and a conclusion chapter (Chapter 10). In chapter 2, statistical methods were utilized to test whether there is any consistent relationship betw een the phase of orbita l parameters and the timing of geomagnetic reversals or excursions. Numerical simulations were employed to understand the possible influenc e of age uncertainties on t he tests. Chapter 2 was published in Earth and Planetary Science Letters (Volume 268, Pages 245-254, 2008). Using published deep-sea sedimentary record s and wavelet analyses methods, Chapter 3 investigates the origin of the orbital periods reported in sedimentary RPI records. Chapter 3 was published in Physics of the Earth and Planetary Interiors (Volume 169, Pages 140-451, 2008). In Chapter 4, calibrat ed RPI and oxygen isotope stack records were developed by simultaneously matching and stacking both RPI and oxygen isotope data for 13 pairs of high-resolution gl obal records. Chapter 4 was published in Earth and Planetary Science Letters (Volume 283, Pages 14-23, 2009). A MATLAB software (UPmag) with graphical user interfaces designed for easy and rapid analysis of NRM and laboratory-induced remanent magnetization data for uchannel samples or core sections is introduc ed in Chapter 5. Chapter 5 was published in Geophysics, Geochemistry, Geosystems (doi:10.1029/2009G C002584, 2009). Chapter 6 presents high resolution paleomagnet ic records spanning the last ~1.5 Ma, acquired from Integrated Ocean Drilling Program (IODP) Site U1304 sediments that

PAGE 18

18 contain episodic deposits of laminated diatom ooze. Reliability of paleomagnetic records acquired from diatom-rich intervals wa s also discussed in this chapter. In Chapter 7, rock magnet ic studies as well as X-ray energy-dispersive spectroscopy and X-ray diffraction analyses were performed on Core 06 recovered by the Healy-Oden Trans-Arctic Expedition 20 05 (HOTRAX05) to the Mendeleev Ridge (Arctic Ocean), to understand the magnetic mineralogy of the sediments and whether partial self-reversal is a possible explan ation for the apparent magnetic excursions recorded in the core. C hapter 7 was published in Earth and Planetary Science Letters (Volume 284, Pages 124-131, 2009). Studi es on HOTRAX05 Cores 08, 10, 11, 13 further along the Mendeleev-Alpha Ridge are reported in Chapter 8, to further understand the origin of low/ negative NRM inclinations th at are present in these sediments. Chapter 8 is now in press in Geophysics, Geochemistry, Geosystems (doi:10.1029/2009GC002879, in press). Chapter 9 extends the study areas to the Lomonosov Ridge (central Arctic Ocean) and the Yermak plateau located at the edge of the Arctic Ocean, to gauge the regional importance of the possible self-reversal mechanism. Chapter 10 concludes the major findings of this dissertation and looks into future work that could further our underst anding of these research topics.

PAGE 19

19 CHAPTER 2 TESTING THE RELATIONSHIP BETWEEN TIMING OF GEOMAGNETIC REVERSALS/EXCURSIONS AND PHASE OF ORBITAL CYCLES USING CIRCULAR STATISTICS AND MONTE CARLO SIMULATIONS Introduction There has been intermittent interest in the influence of orbital periods on the geomagnetic field that can be traced back to Blackett's experiments in the 1950s (Blackett, 1952). Geodynamos driven by precessional forces were advocated in the 1960s (e.g. Malkus, 1968) and are still thought to be viable (e.g. Vanyo and Dunn, 2001; Tilgner, 2005), although Rochester et al. (1975) and Loper (1975) have argued that the energy available fr om precession is insufficient to drive the geodynamo. Orbital periods in sedimentary relative paleointens ity records have been considered evidence for orbital influence on the geodynamo (Kent and Opdyke, 1977; Channell et al., 1998; Yamazaki, 1999; Yamazaki and Oda, 2002). Orbi tal periods in paleomagnetic data may, however, be attributed to lithologic/climatic c ontamination of the sedimentary relative paleointensity records (Kent, 1982; Guyodo et al., 2000; Roberts et al., 2003). Kent and Carlut (2001) found no discernable tendency for reversals or excursions to occur at a consistent amplitude or phase of obliquity or eccentricity by comparing the histogram of obliquity and eccentricity va lues corresponding to ages of the last 21 reversals and 6 excursions in the Brunhes Chron with the histogram of orbital parameters over the same age ranges. These authors used the Lourens et al. (1996) polarity timescale for the last 5.5 Myrs the Brunhes excursion chronologies of Langereis et al. (1997), and astronomical solu tions from Laskar (1990). Fuller (2006) has recently revived the debate by compar ing the timing of polarity reversals with current orbital solutions for obliquity (Laskar et al., 2004), utilizing the ATNTS2004

PAGE 20

20 timescale (Lourens et al., 2004). Fuller (200 6) determined the phase of the obliquity signal at time of reversal, and, for the last nine reversals covering the last 3 Myr, demonstrated that reversals preferentially o ccurred during decrease from maxima within the 41 kyr obliquity cycle (Figure 7 of Fulle r, 2006). After comparing the occurrence of the last 17 reversals with t he smoothed maximum obliquity envelope, he also suggested that reversals preferentially occur when the average amplitude of the obliquity signal is lower than the mean. In addition, he noted a co incidence of paleointensity minima in the Sint-800 relative paleointensity stack (Guyodo and Valet, 1999) with minima in the orbital solution for obliquity, and a preferred duration of 30-40 kyr (corresponding to an obliquity cycle) for polarity subchrons in t he last 13 Myr of the ATNTS2004 timescale (Lourens et al., 2004). Here, we expand on Fuller's analysis by assessing the relationship of reversal/excursion age to the phase of or bital obliquity, the phase of the obliquity envelope, and the phase of eccentricity. Th rough Monte Carlo simulations, we provide estimates of the sensitivity of these results to reversal/e xcursion age uncertainties. Data The orbital solutions used here are those fo r obliquity and eccent ricity from Laskar et al. (2004). This recent La2003 integrati on (Laskar et al., 2004) has been improved with respect to La93 (Laskar et al., 1993) by us ing direct integration of the gravitational equations for orbital motions, and by improving the dissipative contributions. For eccentricity, the solution is considered to be precise over the last 40 Myr because eccentricity depends on the orbital part of the so lution (Laskar et al., 2004; Plike et al., 2004). The solution for precession and obliquity is, however, less accurate due to the uncertainties that remain from tidal dissipation in the Ea rth-Moon system, which

PAGE 21

21 manifests largely as a small change in precession frequency, and appears in the obliquity solution as a time offset. Lourens et al. (2004) provided an estimate for uncertainty in the astronomic solution due to ti dal dissipation by plotting the differences in age of correlative minimum values in th e obliquity and precession cycles between two La2003 solutions that include the present-day and half the present-day tidal dissipation value for the last 25 Myr. According to this analysis, errors in as tronomical ages over the last 10 Myr should be of t he order of 0.1-0.2% (10-20 kyr) and possibly even less. At ~23 Ma, the differences between the tw o solutions reach three cycles, which correspond to a maximum uncertain ty of ~68 kyr in precession, or ~123 kyr in obliquity (Figure 21.7 in Lourens et al., 2004). Laska r et al. (2004) expected the solution for obliquity to be valid over the last 20 Myr with a 5% error in tidal dissipation; however, the error may increase to 10% beyond 20Ma. An uncertainty of 10%in the tidal dissipation term corresponds to an uncertainty in the orbi tal solution of ~16 kyr after 20Ma, and ~63 kyr after 40 Ma(Laskar et al., 2004). This implies an uncertainty of b16 kyr due to tidal dissipation in the orbital solu tion for obliquity during the last 20 Myrs. For the last 15 years, the Cande and Kent (CK95) polarity timescale (Cande and Kent, 1992, 1995) has been the standard for stra tigraphic studies deal ing with the last 80 Myr. The 1995 version of this timescale utilized astrochronologically determined reversal ages for the last 5 Myrs (S hackleton et al., 1990; Hilgen, 1991) and radiometrically-calibrated marine magnetic anom aly (MMA) spacings prior to 5 Ma. The ATNTS2004 timescale (Lourens et al., 2004) fo r the Cenozoic incorporates many of the astrochronological timescale calibrations t hat have become available since 1995. Since the publication of ATNTS2004, the timescale of Billups et al. (2004) provides alternative

PAGE 22

22 age constraints in the 15-25 Ma interval from tuning of the oxygen isotope record at ODP Site 1090 (South Atlantic). Astronomically calibrated reversal ages from equatorial Pacific at ODP Site 1218 (Plike et al., 2006) also cover the 15-25 Ma interval. Recently assigned astronomical ages with very small estimated uncertainties for reversal boundaries between 8.5 Ma and 12.5 Ma are based on the Monte dei Corvi section in northern Italy (Hsing et al., 2007). Additional estimates of re versal ages for part of this interval (9.3-11.2 Ma) are available for ei ght polarity chron boundar ies recorded at ODP Site 1092 from the South Atlantic (Evans et al., 2007).We utilize these six polarity timescales spanning different time intervals, and a compilat ion of excursion ages (Table 2-1), to test the relationship between the ages of reversals and excursions and the phases of obliquity, of eccentricity and of the envelope of obliquity. The main uncertainties in astronomically tuned reversal ages depend on the accuracy of the astronomical solution from which the target was derived, the accuracy of the tuning, and any lag between orbital forcing and response. To gauge uncertainties in polarity timescales, we plot the age diffe rences between the six polarity timescales cited above, including the Shackleton et al. (1995) timescale from ODP Leg 138 (equatorial Pacific), and the AT NTS2004 timescale (Figure 2-1). For the last ~5 Myrs, differences among ATNTS2004, CK95, and the Shackleton et al. timescale are quite small (<50 kyr). Reversal ages in CK95 pr ior to 5 Ma were not astronomically determined, and differences between CK95 and ATNTS2004 exceed 800 kyr in the early Miocene. Differences among astronom ically calibrated reversal timescales (Lourens et al., 2004; Billups et al., 2004; Plike et al., 2006; Hsing et al., 2007; Evans et al., 2007) reach 200 kyr in the Miocene (Figure 2-1).

PAGE 23

23 Methods Phase Calculation To determine if the reversals and excursions occur at a preferr ed phase of orbital cycles (obliquity or eccentricity), we ca lculate the phase corresponding to reversals and excursions since 25 Ma with the definition of phase as follows. The local maximum of obliquity or eccentricity is defined as 0, the following local mini mum is defined as 180, and the following local maximum as 360. If a reversal or excursion occurs at time which is between a local maximum at time and a local minimum at time (Figure 22), the phase corresponding to that reversal or excursion can then be calculated using Equation 2-1. (2-1) Circular Statistics Mardia and Jupp (2000) describe the Raylei gh test as a simple and powerful way to test for uniformity in circ ular distributions. The null hypothesis of a Rayleigh test is that the sample was derived from a circularuniform distribution, versus the alternative that the distribution is not uniform. A ci rcular uniform distribution would imply no preferred orientation of the phase angle. As discussed by Mardia and Jupp (2000), it is useful to take the Rayleigh test statistic as 22 NR, where N is the number of phase data i and R is the mean resultant length defined by Equation 2-2. 22 111 (cos)(sin)NN ii iiR N (2-2)

PAGE 24

24 Mardia and Jupp (2000) report that the Rayleigh test statistic 22 NR is distributed as chi-squared with two degrees of freedom. The upper tail probabilities of 2NR (the p value of Rayleigh test) can be approximated usi ng Equation 2-3. 22 3 4 2 2224132769 Pr(){1} 4288KKKKKKK NRKe NN (2-3) By setting a significance level for the test (for instance, 0.05), we can decide to accept the uniform distribution hypothesis ( p -value significance level) or reject it ( p value < significance level). The p -value of different datasets ar e calculated and listed in Table 2-2. For phase data that are not uni formly distributed, we can estimate the preferred phase angle or phase angle interval by assuming a Von Mises distribution. Jones (2006) developed a MATLABTM program for the statisti cal analysis of circular data that includes Von Mises distribution fitting Note that results from the Rayleigh tests (Table 2-2) are all based on the assumpti on that there is no age uncertainty in reversals/excursions or in orbital solutions. Monte Carlo Simulation It is obvious that age uncertainty from both the reversal/ excursion timescales and the astronomical solution will di rectly influence the phase val ue of the orbital cycle at the time of the reversal/excursion, and hence the p -value from the Rayleigh tests. As can be seen from Equation 2-3, the size of the population of phase data used in the Rayleigh test will also influence the p -value. The following procedure is designed to estimate these influences. 1. Following Fuller (2006), we assume that phases are preferentially distributed between 0 and 180, which is the decreasing part (maximum to minimum) of orbital cycles.

PAGE 25

25 2. We generate points (numbers of reversals/excurs ions used in Table 2-2, i.e. 8, 9, 22, 37, or 89) of phase data which are evenly distributed between 0 and 180, and calculate the Rayleigh test p-value 3. We choose an age uncertainty level (e.g. kyr) and the orbital cycle for testing (e.g. 41 kyr obliquity cycle). 4. For each phase point in 2), we add the age uncertainty using Equation 2-4. (2-4) (,) randuu means a uniform distri buted random number between u and u. 5. We calculate the Rayleigh test p -value for the phase dataset'(1,2,3,...,)iiN. 6. We repeat steps 4) and 5) 1,000,000 times, so rt the calculated 1,000,000 p values into an increasing series. 7. We find the maximum index of the sorted p -value series, corresponding to p values that are less than +0.05. 8) We calculate the percentage of p -values that are bigger than +0.05 in the sorted p -value series: ((1,000,000 )/1,000,000)100% n where n is the maximum index number acquired from step 7. This percentage value gives an estimate of the likelihood that such an age uncertainty would cause a change of the Rayleigh test p -value by >0.05. Percentage values for different age uncertainty levels, and for different numbers of data points in different orbital cycles, are calc ulated and listed in Table 2-3. Results In Figure 2-3, we show the phases of ac tual orbital obliquity corresponding to ages for: (1) the eight best-established excursions in the Brunhes and Matuyama Chrons (Table 2-1); (2) the last 9 reversals (i.e. base of Brunhes, top and base of Jaramillo, top and base of Cobb Mountain, top and base of Olduvai, and top and base of Runion) in CK95 (Cande and Kent, 1995) and ATNTS2004 (Lour ens et al., 2004); (3) reversals of the last 5 Myr in CK95 and ATNTS2004; (4) re versals in the 12.5-8.5 Ma interval in CK95, ATNTS2004, Hsing et al., (2007), and Ev ans et al., (2007); (5) reversals in the

PAGE 26

26 25-15 Ma interval in CK95, ATNTS2004, Billups et al. (2004), and Plike et al. (2006); and (6) reversals of the last 25 Myr in CK95 and ATNTS2004. Assuming no age uncertainty in reversal/excursion ages or in the astronomical solution, the p -values of the Rayleigh tests (Table 2-2) indicate that none of the data groups show any preferred phase distribution in the obliquity cycle at the 5% significance level, although reversal ages for the last 5 Myr have borderline significance ( p -value=0.157 or 0.103 depending on timescale used, see Table 2-2). The Monte Carlo simulation indicates that a reversal age uncertainty of 5-15 kyr (depending on the number of reversal ages) causes large changes in the percentage values (Table 2-3) implying that reversal/excursion ages would have to be known within these tight cons traints in order for a phase relationship to be resolvable. This conclusion is intuitively obvious in view of t he brevity of the 41-kyr cycle relative to a reversal age uncert ainty of 5-15 kyr. Uncertainties in reversal/excursion ages in current timescales exceed 5-15 kyr (with the exception of the two excursions known to have occurred in the last 50 kyrs) in part because the duration of the reversal transition itse lf probably exceeds 5 kyr. It is, therefore, very unlikely that a relationship between reversal age and orbita l obliquity, were it to exist, would be resolvable. A similar calculation has been carried out for orbital eccentricity (Figure 2-4) assuming no age uncertainty in reversal/exc ursion ages or astronomical solution. The last 9 reversals seem to preferentially occu r during the increasing part of eccentricity cycles (Figure 2-4B and C) and the distribution of phases (p-value=0.009 or 0.036 depending on timescale used) is non-uniform at the 5% significance level (Table 2-2). This result is, however, not consistent with the Rayleigh test results from the compiled

PAGE 27

27 excursion ages or any other groupings of reversal ages (Table 2-2), for which no preferred phase angle is observed. The p -values of Rayleigh tests for phases of eccentricity and phases of the maximum ob liquity envelope corresponding to different numbers of reversal ages (last 9 to la st 36 reversals) are listed for CK95 and ATNTS2004 (Table 2-4). The preferred phase dist ribution in the eccentricity cycle at the 5% significance ceases when adding even one more reversal age beyond the last 9 reversals in CK95, or adding three more re versal ages in ATNTS2004 timescale. The Monte Carlo simulations for eccentricity cycl es indicate that large changes in the percentage values (Table 2-3) occur when the age uncertainty is in the 10-40 kyr range, depending on the number of revers al/excursion ages in the simulation. For less than 10 reversal ages (the last 3 Myrs), age uncertainties of 10-20 kyr are sufficient to inhibit the recognition of a relationship between reversal/excursion age and phase of eccentricity. This is deduced from the change in the percentage values as reversal/excursion age uncertainty increases (Table 2-3). As a test, we replace the Runion ages (both top and base) inCK95 and ATNTS2004, derived from the cylostratigraphies in the Italian sections (Zijderveld et al., 1991; Lourens et al., 1996), with more recent Runion ages from ODP Site 981 (Channell et al., 2003). The results indicate that p -values for eccentricity phases corresponding to the last 9 reversals change from 0.036 to 0.147 for CK95, and from 0.009 to 0.033 us ing ATNTS2004. This indicate s the sensitivity of the Rayleigh tests to estimates of the age of (Runion) reversals that differ by 5-25 kyr. According to the simulations, when larger populations of reversal ages back to 25Ma are considered, age uncertainties up to 40 kyr are sufficient to inhibit the recognition of any phase relationship that may be present. It is unlikely that reversal ages older than 5

PAGE 28

28 Ma in current timescales are known with uncertainties less than 40 kyr. For example, differences between polarity chron ages in ATNTS2004 (Lourens et al., 2004) and the later Billups et al. (2004) timescale exceed 200 kyr in t he Early Miocene (Figure 2-1). Any relationship between reversal age and the phas e of orbital eccentricity is unlikely to be resolvable, at least beyond the last 5 Myrs. For the last nine reversals, the Rayleigh test ( p -value=0.009 or 0.036, Table 2-2) indicate s a preferred phase distribution in the eccentricity cycle at the 5% significance level, however, in view of the simulations, this implies that reversal ages for the last 3 Myr are known to within ~15 kyr. To analyze the relationship between reversal/excursion age and ~1.2 Myr modulation envelope of the orbital obliquity, the envelope data were smoothed using the Savitzky-Golay smoothing filter (Savitzky an d Golay, 1964), a time-domain smoothing based on a least squares polynomial fit acro ss a moving window applied to the dataset. From Table 2-2 and Figure 2-5, consideri ng no age uncertainty in reversal/excursion ages or orbital solution, we see that the phases corresponding to the last 9 reversals (Figure 2-5B and C) and reversals in the la st 5 Myr (Figure 25D and E) are not uniformly distributed (i.e. have pr eferential phase) at the 5% significance level (Table 22). A preferred relationship wit h the maximum obliquity envelope is also indicated for the eight best-established excursion ages (Table 2-1) by the relatively low p -value (0.080) from the Rayleigh test (Table 2-2) although it is not significant at the 5% level. Note that even a manually generated 8-point phase dataset that is evenly distributed between 0 and 180 has a Rayleigh test p -value of 0.088 (Table 2-3). Th e compiled excursion ages (Table 2-1) might not be appropriate for expl oring a phase distribution in the maximum obliquity envelope because: 1) these excu rsion ages only span the 1.115-0.033 Ma

PAGE 29

29 interval, with a duration that is even shor ter than a maximum obliquity envelope cycle; 2) these excursions were chosen based on their age quality, with no excursions included in the 0.850-0.211 Ma interval. In this case, in the Monte Carlo simulations, the percentage of simulations that have p>0.05 at the 50 kyr age uncertainty level (Table 23) is small (~13.5% for 8 data points, ~7 .0% for 9 data points, and ~0% for 22 data points). According to the simulations, the re sult is not influenced by age uncertainties unless these age uncertainties exceed 50 kyr (for the 8 excursi ons or the last 9 reversals, Table 2-3) or 200 kyr for last 5 Myrs (Table 2-3). Assuming a Von Mises distribution for phases corresponding to revers als during the last 5 Myrs (in ATNTS2004 timescale), a mean phase of 103.5, and a 95% confidence interval between 56.0 and 151.1 is obtained using the MATLAB protocol (Jones, 2006). In contrast to Fuller's conclusion (Fuller, 2006) that reversals preferentially occurred when the average amplitude of the obliquity signal is lower than the mean, this result implies that, in the last 5 Myrs, reversals preferentially occurr ed during decrease of the maximum obliquity envelope. The results of t he Rayleigh test do not hold, however, when we consider reversal ages back to 25 Ma or other groupi ngs of reversal ages (Figure 2-5F~O). Results from Table 2-4 indicate that the pr eferential distribution of reversal ages with phase of the maximum obliquity envelope fo r the last 5 Myr breaks down when adding even one more reversal age using either CK95 or ATNTS2004. The inconsistency could be attributed to larger than expected reversal age uncertainties beyond 5 Ma combined with a link between reversal age and phase of the obliquity envelope cycle. Note that the maximum difference between reversal ages for the 25-15 Ma interval in the ATNTS2004 (Lourens et al., 2004) and Billups et al. (2004) timescales reaches 230 kyr

PAGE 30

30 (Figure 2-1). The Monte Carlo simulations (T able 2-3) indicate t hat, for the larger populations of reversal ages in the 25-15 Ma interval (37 reversals), N300 kyr age uncertainties would drive the simulated p -value to values indicative of a uniform distribution of phases. Conclusions Fuller (2006) made several observations linking the paleomagnetic records to orbital solutions for obliquity: (1) Severa l paleointensity minima in the Sint-800 paleointensity stack (Guyodo and Valet, 1999) correlate with indi vidual obliquity minima, (2) The durations of polarity chrons in t he ATNTS2004 timescale (Lourens et al., 2004) display a peak at 30-40 kyr. (3) The last 9 reve rsals occurred at a preferred phase in the obliquity cycle. (4) The last 17 reversals o ccurred preferentially during minima in the orbital obliquity envelope. A relationship between paleointensity lows in the Sint-800 stack and the obliquity minima is difficult to estab lish due to uncertainties in the chronology of the stack that must approach the obliquity period (G uyodo and Valet, 1999; McMillan et al., 2004). Correlations of the obliquity signal with prominent lows in individual paleointensity records from ODP Site 983 (Channell, 1999 ; Channell and Kleiven, 2000; Channell et al., 1997) and Site 984 (Channell, 1999; Cha nnell et al., 2004), that have oxygen isotope age control, do not show any obvious pattern (Figure 2-6). The distribution of polarity chron durations in the ATNT S2004 timescale (point 2, above) can be attributed to the Poisson di stribution of polarity chron durations, combined with truncation of low duration values in the timescale as a result of the resolution of the MMA data on which the templa te for polarity reversal is largely based. The pattern of polarity chrons in ATNTS2 004 is essentially inherited from MMA data

PAGE 31

31 where the practical lower lim it of duration for polarity chron recognition is ~30 kyr (Cande and Kent, 1995). The Rayleigh test is used to determi ne the likelihood of a relationship of reversal/excursion ages to the phases of obliquity, eccentricity and obliquity envelope. Assuming no age uncertainty in reversal/excur sion ages or astronomic al solution, small p -values in the Rayleigh test indicate a nonuniform distribution of phases (bold in Table 2-2). Although there is a significant relati onship (at 5% level) between the last 9 reversals and phase of eccentricity cycles, the relationship breaks down when adding 1 or 3 additional reversal ages, depending on pol arity timescale used (Table 2-4). The relationship was not observed for any other groupings of reversal ages, or for a compilation of excursion ages (Figure 2-4). Monte Carlo simulations demonstrate that these tests are very sensitive to revers al age uncertainties and may be biased by reversal age uncertainties as low as 5 kyr in the case of obliquity at low reversal populations, to 40 kyr for eccentricity at higher reversal populations extending back to 25 Ma. A conservative estimate for revers al age uncertainties beyond 5 Ma is 40 kyr (one obliquity cycle) and for the last 5 Myr the reversal age uncertainties certainly exceed 10 kyr. For this reason, we consider that any relationship between reversal age and the phase of obliquity or eccentricity would not be resolvable due to imprecision in reversal ages. Considering no age uncertainty in reversal/excursion ages or astronomical solution, the phase of the maximum obliquity envelope at times of the last 9 reversals, and reversals in last 5 Myr, are not uniforml y distributed at the 5% significance level. Polarity reversals younger than 5 Ma prefer entially occur during decrease in amplitude

PAGE 32

32 of the envelope, rather than when the obliqui ty is lower than the mean (as deduced by Fuller (2006)). This significant relationship for the last 5 Myr does not hold when adding even one additional reversal age to the test (Table 2-4), or w hen applied to other groupings of reversal ages (Figure 2-5). Th e Monte Carlo simulations indicate that uncertainties in reversal/ excursion age and/o r orbital solution for the small reversal population back to 5 Ma would ha ve to exceed 50-100 kyr to a ccount for the test result in the presence of a phase relationship simila r to that advocated by Fuller (2006). The reversal age uncertainties would have to lie in the 300-500 kyr range to account for the test results for reversal populat ions extending back to 25 Ma (Table 2-3). We would not expect reversal ages to be sufficiently imprec ise to influence this result for the last 5 Myr, however, the difference between the ATNTS2004 (Lourens et al., 2004) and Billups et al. (2004) reversal ages reach 230 kyr in the Early Miocene (Figure 2-1), indicating that reversal age uncertainties may reach several hundred kyrs for reversals older than 5 Myr. Inconsistency of the re lationship between phase of obliquity envelope and reversal age for the last 5 Myr, and for other time intervals, could be attributed to larger than expected reversal age uncer tainties beyond 5 Ma and a link between reversal age the obliquity envel ope, or, more probably, the fortuitous occurrence of a low probability relationship ov er the last 5 Ma that has no mechanistic implication.

PAGE 33

33 Table 2-1. Ages of the 8 best-established excursions in the Brunhes and Matuyama Chrons Excursion name Estimated age (ka) Principal references Mono Lake 33 Benson et al. (2003) Laschamp 41 Laj et al. (2000) Blake 120 Tric et al. (1991) Iceland Basin 188 Channell (1999), Channell et al. (1997) Pringle Falls 211 Singer et al. (in press) Kamakatsura 850 Channell et al (2002), Singer et al. (2004) Santa Rosa 932 Channell et al (2002), Singer et al. (2004) Punaruu 1115 Channell et al. (2002), Singer et al. (2004)

PAGE 34

34 Table 2-2. p -value of Rayleigh test for phase data in Figure 2-3, Figure 2-4, and Figure 2-5 Datasets Actual obliquity Actual eccentricity Maximum obliquity envelope 8 best established excursions (Table 2-1) 0.231 0.922 0.080 Last 9 reversals in CK95 (Cande and Kent, 1995) 0.138 0.036a 0.039a Last 9 reversals in ATNTS2004 (Lourens et al., 2004) 0.570 0.009a 0.023a Reversals in CK95 (last 5 Myr) (C ande and Kent, 1995) 0.103 0.365 0.036a Reversals in ATNTS2004 (last 5 Myr) (Lourens et al., 2004) 0.157 0.847 0.031a Reversals in CK95 (12.5-8Ma) (Cande and Kent, 1995) 0.751 0.741 0.095 Reversals in ATNTS2004 (12.5-8Ma) (L ourens et al., 2004) 0.719 0.675 0.141 Hsing et al. (2007) timescale (12.5-8 Ma) 0.105 0.744 0.142 Evans et al. (2007) timescale (12-9 Ma) 0.131 0.211 0.098 Reversals in CK95 (25-15Ma) (Cande and Kent, 1995) 0.880 0.855 0.298 Reversals in ATNTS2004 (25-15Ma) (Lour ens et al., 2004) 0.857 0.943 0.954 Billups et al. (2004) timescale (25-15 Ma) 0.926 0.262 0.291 Plike et al. (2006) timescale (25-15 Ma) 0.824 0.340 0.393 Reversals in CK95 (last 25 Myr) (C ande and Kent, 1995) 0.124 0.813 0.424 Reversals in ATNTS2004 (last 25 Myr) ( Lourens et al., 2004) 0.317 0.519 0.980 a p -values that are less than 0.05.

PAGE 35

35 Table 2-3. Monte Carlo simulation of influence of age unc ertainties and number of data points on Rayleigh test p -value Obliquity cycle (41 kyr) Data points Original p -value Age uncertainty level ( kyr ) .5 .515.520 8 0.08846.7%59.1%68.5% 76.4%81.9%85.1%86.2% 9 0.05643.1%58.0%69.2% 78.1%84.5%88.1%89.4% 10 0.03638.5%55.9%68.9% 79.0%86.0%90.1%91.4% 22 0.0000.2%9.3%a34.3% 62.0%81.8%91.9%94.9% 37 0.0000.0%0.2%7.2% a 34.6%69.0%89.3%94.9% 114 0.0000.0%0.0%0.0% 0.4%a19.4%73.0%94.6% Eccentricity cycle (100 kyr) Data points Original p -value Age uncertainty level ( kyr ) 15 25 8 0.08819.2%40.8%52.9% 61.9%69.4%75.7%84.0% 9 0.05611.9%36.1%50.7% 61.5%70.2%77.4%86.9% 10 0.0365.9%a30.0%47.2% 59.9%70.1%78.2%88.6% 22 0.0000.0%0.0%2.2% a 15.2%37.3%59.9%88.5% 37 0.0000.0%0.0%0.0% 0.7%9.0%a31.9%82.6% 114 0.0000.0%0.0%0. 0% 0.0%0.0%0.3%a49.7% Maximum obliquity envelope (~1.2 Myr) Data points Original p -value Age uncertainty level ( kyr ) 50 8 0.0880.1%a13.5%35.2% 56.1%69.5%79.2%84.7% 9 0.0560.0%7.0%a29.5% 54.5%70.2%81.4%87.7% 10 0.0360.0%2.5%a22.6% 51.8%70.1%82.7%89.6% 22 0.0000.0%0.0%0.0% 5.1%a37.1%72.5%90.8% 37 0.0000.0%0.0%0.0% 0.0%9.0%a51.5%87.1% 114 0.0000.0%0.0%0. 0% 0.0%0.0%3.9%a65.0% Note: Original phase data are generated evenly between 0 and 180. Using various populations of data points (corresponding to numbers of reve rsals/excursions in datasets of Table 2-2) 1,000,000 Monte Carlo simulations for each level of age uncertainty provided the percentage of simulations that have >0.05 difference in p -value from the original p value. a critical thresholds where small changes in the corresponded age uncertainty lead to large changes in p -value.

PAGE 36

36 Table 2-4. p -value of Rayleigh tests for phases of eccentricity and ma ximum obliquity envelope co rresponding to different number of reversal ages (last 9 to last 36 revers als) in CK95 (Cande and Kent, 1995) and ATNTS2004 time scale (Lourens et al., 2004) Reversal numbers Eccentricity Obliquity envelope Reversal numbers Eccentricity Obliquity envelope CK95 ATNTS04 CK95 ATNTS04 CK95 ATNTS04 CK95 ATNTS04 9 (2148) 0.036 0.0090.0390.023 23 (5235)0.5040.9300.0860.077 10 (2581) 0.105 0.0160.1210.088 24 (6033)0.6890.8330.0550.080 11 (3032) 0.146 0.0270.2050.155 25 (6252)0.5440.9120.0830.155 12 (3116) 0.218 0.1040.2170.159 26 (6436)0.6340.8050.1590.236 13 (3207) 0.376 0.2020.1380.101 27 (6733)0.7990.7820.1690.164 14 (3330) 0.448 0.1580.0700.049 28 (7140)0.9020.9110.1090.165 15 (3596) 0.359 0.3090.0830.067 29 (7212)0.8110.8020.0890.185 16 (4187) 0.548 0.4780.0750.058 30 (7251)0.8700.9110.0770.206 17 (4300) 0.743 0.5310.0440.032 31 (7285)0.9110.8000.0690.226 18 (4493) 0.885 0.3630.0220.016 32 (7454)0.9780.7040.0880.304 19 (4631) 0.734 0.5530.0130.010 33 (7489)0.9120.5800.1110.392 20 (4799) 0.563 0.5530.0150.011 34 (7528)0.9420.6900.1460.489 21 (4896) 0.517 0.7460.0210.016 35 (7642)0.8460.5710.2160.620 22 (4997) 0.365 0.8470.0360.031 36 (7695)0.7610.5870.3190.751 Note: Inside the parentheses are reversal ages (in kyr) from ATNTS2004 timescale corresponding to the reversal number to the left of the parentheses.

PAGE 37

37 Figure 2-1. Comparing differences in diffe rent reversal time scales (Lourens et al., 2004; Cande and Kent, 1995; Billups et al ., 2004; Plike et al., 2006; Hsing et al., 2007; Evans et al., 2007; Sha ckleton et al., 1995) relative to ATNTS2004 time scale (Lourens et al., 2004). P/M denotes the PlioceneMiocene boundary at ~5.332 Ma, and M/O denotes Miocene-Oligocene boundary at ~23.030 Ma.

PAGE 38

38 Figure 2-2. Definition of phase of orbital cycl es corresponding to a reversal/excursion.

PAGE 39

39 Figure 2-3. Circular plot of phase of actual obliquity (Laskar et al ., 2004) for: A) 8 bestestablished excursions (Table 2-1). B) Last 9 reversals in CK95 (Cande and Kent, 1995). C) Last 9 reversals in ATNTS2004 (Lourens et al., 2004). D) Reversals in CK95 (last 5Myr) (C ande and Kent, 1995). E) Reversals in ATNTS2004 (last 5 Myr) (Lourens et al., 2004). F) Reversals in CK95 (12.5-8 Ma) (Cande and Kent, 1995). G) Reversals in ATNTS2004 (12.5-8 Ma) (Lourens et al., 2004). H) Hsing et al. ( 2007) timescale (12.5-8 Ma). I) Evans et al. (2007) timescale ( 12-9 Ma). J) Reversals in CK95 (25-15 Ma) (Cande and Kent, 1995). K) Reversals in ATNT S2004 (25-15 Ma) (Lourens et al., 2004). L) Billups et al. (2004) timescale (25-15 Ma). M) Plike et al. (2006) timescale (25-15 Ma). N) Reversals inCK95 (last 25 Myr) (Cande and Kent, 1995). O) Reversals in ATNTS2004 (last 25 Myr) (Lourens et al., 2004). Phase angles of 0 (or 360), 90, 180, 270 are marked by lines clockwise from top of each circular plot.

PAGE 40

40 Figure 2-4. Circular plot of phase of actual eccentricity (Laskar et al., 2004) for: A) 8 best-established excursions (Table 2-1). B) Last 9 reversals inCK95 (Cande andKent, 1995). C) Last 9 re versals in ATNTS2004 (Lourens et al., 2004). D) Reversals in CK95 (last 5 Myr) (C ande and Kent, 1995). E) Reversals in ATNTS2004 (last 5 Myr) (Lourens et al., 2004). F) Reversals in CK95 (12.5-8 Ma) (Cande and Kent, 1995). G) Revers als inATNTS2004 (12.5-8 Ma) (Lourens et al., 2004). H) Hsing et al. ( 2007) timescale (12.58 Ma). I) Evans et al. (2007) timescale ( 12-9 Ma). J) Reversals in CK95 (25-15 Ma) (Cande and Kent, 1995). K) Reversals in ATNT S2004 (25-15 Ma) (Lourens et al., 2004). L) Billups et al. (2004) timescale (25-15 Ma). M) Plike et al. (2006) timescale (25-15 Ma). N) Reversal s inCK95 (last 25 Myr) (Cande and Kent, 1995). O) Reversals in ATNTS2004 (last 25 Myr) (Lourens et al., 2004). Phase angles of 0 (or 360), 90, 180, 270 are marked by lines clockwise from top of each circular plot.

PAGE 41

41 Figure 2-5. Circular plot of phase of ma ximum obliquity envelope for: A) 8 bestestablished excursions (Table 2-1). B) Last 9 reversals in CK95 (Cande and Kent, 1995). C) Last 9 reversals in ATNTS2004 (Lourens et al., 2004). D) Reversals in CK95 (last 5 Myr) (C ande and Kent, 1995). E) Reversals in ATNTS2004 (last 5 Myr) (Lourens et al., 2004). F) Reversals in CK95 (12.5-8 Ma) (Cande and Kent, 1995). G) Reversals in ATNTS2004 (12.5-8 Ma) (Lourens et al., 2004). H) Hsing et al. ( 2007) timescale (12.58 Ma). I) Evans et al. (2007) timescale (12-9 Ma). J) Reversals inCK95 (25-15 Ma) (Cande andKent, 1995). K) Reversals inATNT S2004 (25-15 Ma) ( Lourens et al., 2004). L)Billups et al. (2004) timescale (25-15 Ma). M) Plike et al. (2006) timescale (25-15 Ma). N)Reversals inCK95 (last 25 Myr) (Cande and Kent, 1995). O) Reversals in ATNTS2004 (last 25 Myr) (Lourens et al., 2004). Phase angles of 0 (or 360), 90, 180, 270 are marked by lines clockwise from top of each circular plot.

PAGE 42

42 Figure 2-6. Comparing relati ve paleointensity records with orbital obliquity during the last 800 kyr. Top panel: Sint-800 record (Guyodo and Valet, 1999) in green, ODP 983 paleointensity record (Channe ll, 1999; Channell and Kleiven, 2000;Channell et al., 1997) in blue, and ODP 984 paleointensity record (Channell, 1999; Channell et al., 2004) in red. Bottom panel: orbital obliquity signal from Laskar et al. (2004).

PAGE 43

43 CHAPTER 3 ORIGIN OF ORBITAL PERIODS IN THE SEDIMENTARY RELATIVE PALEOINTENSITY RECORDS Introduction The possible connection between the geom agnetic field and or bital parameters has been controversial for nearly 60 years. Geo-dynamos driven by precessional forces were discussed by Bullard (1949) and were then advocated in the 1960s (e.g. Malkus, 1968). Theoretical studies by Rochester et al. (1975) and Loper (1975), based on the energetics of laminar flow conditions, howev er, concluded that t he energy available from precession is insufficient to drive the geodynamo. Since then, Gubbins and Roberts (1987) and Kerswell (1996) considered turbulent dissipation and suggested that the energy due to orbital precession could be of the order required to power the geodynamo. More recent theoretical (e.g. Ch ristensen and Tilgner, 2004; Tilgner, 2005, 2007; Wu and Roberts, 2008) and experimenta l work (e.g. Vanyo and Dunn, 2000) have been interpreted to favor the possibilit y of a precessionally powered geodynamo. Orbital forcing has not been taken into a ccount in numerical simulations of the geodynamo (Glatzmaier and Roberts, 1996; Glatzm aier et al., 1999) in part because the time step in any precession model is necessarily so small that useful simulations require unreasonably long run times, which taxes even high-speed computers. By comparing the timing of reversals with or bital solutions over the last few million years, Fuller (2006) proposed a link between orbital obliquity and polarity reversal, thereby providing further evidence of orbita l forcing of the geodynamo. Kent and Carlut (2001), however, found no discernable tendency for reversals or excursions to occur at a consistent amplitude or phase of obliquity or eccentricity. A recent analysis over a longer time period also failed to confirm such a relationship and pointed out that

PAGE 44

44 imprecision in current polarity timescales does not, at present, allow a relationship to be firmly established between reversal age and orbita l obliquity or orbital eccentricity (Xuan and Channell, 2008). Normalized records of sedimentary nat ural remanent magnet ization (NRM) are often interpreted as representing the relative paleointensity (RPI) of the geomagnetic field, and criteria have been proposed for selecting sediments suitable for paleointensity determinations (Johnson et al., 1948; Levi and Banerjee, 1976; King et al., 1983; Tauxe, 1993). Normalization is generally carried out by using a rock magnetic parameter such as anhysteretic remanent magnetization (ARM), isothermal remanent magnetization (IRM), or magnetic susceptibility ( ) to compensate for changes in magnetic concentration of remanence carrying grains. In the last few decades, many RPI records with variable resolution have been obtained from marine sedimentary sequences covering portions of the last several million years (e.g. Valet and Meynadier, 1993; Tauxe and Shackleton, 1994; Channell et al., 1997; Guyodo et al., 2001; Stoner et al., 2003; Thouveny et al., 2004). Reliability of these normalized remanence records as RPI records comes from the observed similari ties among RPI records from contrasting environments in the worlds oceans (e.g Guyodo and Valet, 1996, 1999; Laj et al., 2000; Stoner et al., 2002; Yamazaki and Oda, 2005; Channell et al., in press), as well as from the similarity of marine RPI record s with lacustrine records (Peck et al., 1996) and with records obtained from marine magnetic anomalies re corded at fast spreading centers (Gee et al., 2000). In addition, the adequate ma tch with cosmogenic isotope records from ice cores over the last 100 kyr (e.g. Finkel et al ., 1997; Baumgartner et al., 1998; Wagner et al., 2000; Mu scheler et al., 2005) or sedimentary cores over the

PAGE 45

45 last several hundred kyrs (Frank et al., 1997; Carcaillet et al., 2004) indicates that RPI records offer a proxy record of the strength of the main dipol e field over these intervals. RPI records not only refine our understanding of temporal geomagnetic field variations and provide data to constrain geodynamo models, they also provide a global stratigraphic tool that augment s oxygen isotope records. Our understanding, however, of the proce sses responsible for the magnetization of sediments is far from being complete. In addition to field strength, depositional remanence (DRM) is related to mineralogy, concentration and grain size of the magnetic phases, properties of the non-magnetic matrix, and pore water chemistry (Tauxe, 1993). The physical theory concerni ng particle alignment and magnetization lock-in depends on sediment characteristics t hat are poorly constrained such as grain flocculation and alignment efficiency (see, e. g. Tauxe et al., 2006). It has long been realized that rock magnetic and other lithological variations induced by paleoclimatic changes can significantly affect or contro l the NRM, and hence bias RPI records (Kent, 1982). Franke et al. (2004) used a three-mem ber regression model to demonstrate that nonmagnetic matrix effects such as opal cont ent, terrigenous content and kaolinite/illite ratio influence RPI records from the South Atlantic. Attempts have been made to reduce envir onmental influence on the RPI records by adding, or subtracting a fraction of the normalizer ( ARM, or IRM) to/from the normalized intensity record so that the coherence between paleointensity records and their normalizers is minimized. Using this method, the correlation between the paleointensity records obtained with the diffe rent normalizers was improved (Mazaud, 2006). A correction function based on the linear relationship between the normalized

PAGE 46

46 intensity and the median destructive field of the NRM (considered as grain size proxy) was used by Brachfeld and Banerjee (2000) to reduce the grain size dependence in the RPI records. The Recognition and Interpretation of Orbital Cycles in RPI Records Orbital cycles (with 43 kyr period) were found in the normalized intensity record of the Brunhes Chron from a deep-sea sediment piston co re by Kent and Opdyke (1977) using power spectrum analysis, and were interpreted as evidence for orbital forcing on the geodynamo. In recent years, orbital cycles (with 100 kyr and/or 41 kyr periods) have also been detected in a number of high-resolution sedimentary relative paleointensity (RPI) records using power s pectra (e.g. Channell et al., 1998; Yamazaki, 1999; Thouveny et al., 2004; Yamazaki and Oda, 2005) and wavelet spectra (e.g. Guyodo et al., 2000; Yokoyama and Yamazaki 2000), as well as in the inclination record using power spectra (Yamazaki and Oda, 2002), and have often been attributed to orbital control on the geodynamo. An orbital cycle with 100 kyr period was also reported in a 300 kyr long 10Be/9Be record, considered to be a proxy for geomagnetic field strength, from the Port uguese margin (Carcaillet et al ., 2004). From three different areas of the Pacific Ocean, Yokoyama et al. (2007) studied the correlations (in terms of correlation coefficients for the entire durati on of the records) among the 100-kyr period components extracted from the wavelet tr ansforms of RPI records and rock magnetic parameters. The authors found that RPI variations in the thre e cores exhibit significant correlation (a single value, 0.55, was used as the threshold for determining the significance of the correlation coefficient s), while rock magnetic parameters do not.

PAGE 47

47 Therefore, the authors attributed the 100-kyr period in RPI to orbital forcing on the geodynamo. Orbital periods reported in RPI records have also been attributed to lithologic/climatic contaminati on. Kok (1999) found that rela tive paleointensity stacks derived from both normalized NRM (Guyodo and Valet, 1996) and 10Be records (Frank et al., 1997) show coherent features wit h the oxygen isotope records, and suggested that agreement of the two st acked paleointensity records may stem from similar influences of climate variation rather than from geomagnetic intensity variations alone. Wavelet analyses in the 0-1.1Ma interval in paleomagnetic records from ODP Site 983 indicated that orbital periods are present in the normalizer records over the same time intervals as in the RPI record, implying that orbital periods embedded in the RPI record are due to the influence of lithologic variat ions (Guyodo et al., 2000). Roberts et al. (2003) showed that the 100-kyr signal observed by Yamazaki and Oda (2002) in the inclination record from Core MD982185 fr om the western Caroline Basin is not statistically significant for the entire record, and is not modulated by the 404-ky eccentricity component as might be expected if the inclination record was influenced by orbital eccentricity. Furthermore, based on their coherence analysis (in the frequency domain), highly variable phase relationships and no statistically significant coherence were observed between the inclination record and the orbital signal. In addition, Horng et al. (2003) found no orbital cycles in their RPI records from the Western Philippine Sea (Core MD972143) using wavelet methods. No stable orbital periodicities were found in the global RPI stack (SINT800, Guyodo and Valet, 1999), although this could be accounted for by the low time resolution of the stack.

PAGE 48

48 Coherence analysis (in the frequency domai n) between the RPI record and the normalizer has been employed (e.g. Tauxe a nd Wu, 1990; Tauxe, 1993) as a way to determine whether paleointensity records are c ontaminated by lithologic/climatic factors. The method has been extensively used to assess the quality of individual RPI determinations (e.g. Channell et al., 1998; Channell, 1999; Yamazaki, 1999; Channell and Kleiven, 2000; Yamazaki and Oda, 2005). However, as Valet (2003) pointed out, this type of coherence analysis deals with t he entire record and generally does not rule out the possibility that some specific intervals of the two re cords could exhibit significant coherence on certain frequencies (e.g. orbita l frequencies). More sophisticated timefrequency based methods such as cross-wavelet transform and squared wavelet coherence analysis with robust significance te sts, has been made available by Torrence and Compo (1998), Torrence and Webster (1999), and by Grinsted et al. (2004). These methods are used by Heslop (2007) to in vestigate the relationship among orbital, paleoclimatic, and paleointensity changes. The author used the low resolution SINT800 stack (Guyodo and Valet, 1999) and the marine magnetic anomaly paleointensity record of Gee et al. (2000). For these records, it was found that while the paleointensity proxies and orbital variations exhibit common power at certain periods in certain time intervals, they do not exhibit a consistent phase relationship or significant squared wavelet coherence, suggesting no direct physical link between them. Data and Methods In this paper, we use seven previously pub lished high-resolution RPI records from different regions of the world, spanning the past 2Myr with good age control, along with rock magnetic records, climate records, and orbital parameters to investigate the origin of orbital periods in these RPI reco rds. The RPI records were first analyzed by

PAGE 49

49 local wavelet power spectra (LWPS), with signi ficance tests, to ascertain the presence of orbital periods. Following methods out lined by Torrence and Compo (1998), Torrence and Webster (1999), and by Grinsted et al (2004), cross-wavelet power spectra (|XWT|), and squared wavelet coherence (WTC), with robust statistical significance tests, were then calculated between RPI records, in which orbital periods exist, and other relevant records such as orbital solutions, benthic oxygen isotope records, normalizer records, grain size proxy records, and physical pr operty records, in order to determine the origin of the orbital periods in the RPI records. Attempts were made to estimate the degree of influence from contam ination on the RPI records by comparing RPI records from different regions of the world in both the time domain and timefrequency space after optimally correlating t he RPI records to one of the RPI records (ODP Site 984). Prior to the 1980s, piston cores recover ed from the deep sea rarely exceeded a few tens of meters in length (e.g. Kent and Opdyke, 1977). The use of the hydraulic piston corer (HPC) by the Deep Sea Drilling Project (DSDP) and subsequently by the Ocean Drilling Program (ODP), as well as the Calypso coring syst em of the Marion Dufresne, have revolutionized our ability to collect long relatively undisturbed sediment cores. Sedimentary sequences up to se veral hundred meters in length can now be recovered by HPC, and this has provided a valuable archive for monitoring detailed changes in the paleomagnetic fiel d. In the last 20 years, many long continuous highresolution RPI records have become available from the Atlantic (e.g. Channell et al., 1997, 2008; Channell, 1999, 2006; Lund et al., 2001a,b; Channell a nd Raymo, 2003; Stoner et al., 2003), from the equatorial Pacific (e.g. Yamaza ki and Oda, 2002; Horng et

PAGE 50

50 al., 2003; Yamazaki and Oda, 2005) as well as lower resolution RPI records from North Pacific (Yamazaki, 1999). For this analysis, we have chosen seven high-resolution RPI records with good age control from different regions of the world spanning the past 2Ma (Table 3-1 and Figure 3-1). Age models for these records were constructed by correlation of the oxygen isotope records or rock magnetic proxies (e.g. ARM, volume susceptibility, or S-ratio) to a reference oxygen isotope curve. Orbital parameters used in this analysis are represented by the ETP curv e (Eccentricity + Tilt Precession) using normalized values from the orbital solutions of Laskar et al. (2004). The most commonly used way to find frequencies/periods in a time series is power spectrum analysis. The disadvantage of power spectral analysis is that it has only frequency resolution and no time resolution. In addition, it implicitly assumes that the underlying processes are stationary in time, wh ich is generally not true for geophysical processes or their proxy records. In compar ison, wavelet transforms expand time series into time frequency space and can therefore det ect localized intermittent periodicities, although the spectral estimate with the wa velet method are biased towards the longwavelength periods (Lau and Weng, 1995; Torr ence and Compo, 1998; Grinsted et al., 2004). The Morlet wavelet, a plane wave modul ated by a Gaussian envelope, provides a good balance between time and frequency. A Mo rlet wavelet with a non-dimensional frequency of 6 is used here in our analyses. Compared with traditional coherence analys is methods (e.g. Tauxe and Wu, 1990; Tauxe, 1993), which reveal phase relationship and coherency between two signals within particular frequency bands (in the fr equency domain), cross-wavelet transform and squared wavelet coherence (WTC) are more powerful methods for testing proposed

PAGE 51

51 linkages between two time series (Torrence and Compo, 1998; Torrence and Webster, 1999; Grinsted et al., 2004). The cross-wavelet transform reveals common power and relative phase between two time series in the time-frequency domain. For there to be a simple cause and effect relationship between the phenomena recorded in two time series, we would usually expect the oscill ations to be phase locked. Squared wavelet coherence can further measure how coherent the cross-wavelet transform is in timefrequency space. The definition of squared wave let coherence closely resembles that of a traditional correlation coefficient, and it is useful to think of the squared wavelet coherence as a localized correlation coeffici ent in time frequency space. Compared with Fourier squared coherency, which is used to identify frequency ba nds within which two time series are co-varying, the squared wa velet coherency is used to identify both frequency bands and time intervals within which the two time series are co-varying (Liu, 1994; Torrence and Webster, 1999). It is often necessary to address the issue of significance tests for these waveletbased analyses to distinguish statistically si gnificant results from those due to random chance. Statistical significance of wave let power can be tested by assuming a 2 distribution of the wavelet power spectra and a certain model of the background noise (Torrence and Compo, 1998). For RPI reco rds, a white noise background has been considered to be inadequate (e.g. Bart on, 1982; Lund and Keigwin, 1994). Power spectra of RPI records often show continuous power decay with increasing frequency that fit the first order autor egressive (AR1) spectrum. Hence, we use AR1 to model the background noise of the RPI data. Using an AR1 noise model, statistical significance levels for the cross-wavelet power can be deriv ed from the square root of the product of

PAGE 52

52 two 2 distributions, and significance levels of the squared wavelet coherence can be estimated using Monte Carlo methods (Tor rence and Webster, 1999; Grinsted et al., 2004). The records (Table 3-1) were analyzed using these wavelet-based methods to detect intermittent cyclical behavior in the records, and possibly make connections between RPI records and other records in ti me-frequency space. As a test of these wavelet methods, LWPS (Figure 3-2A and B) of the calculated ETP curve and of the global benthic oxygen isotope stack (LR04 from Lisiecki and Raymo, 2005), and |XWT| (Figure 3-2C) and WTC (Figure 3-2D) between the two records were performed. Due to the modulating effect of the 404-kyr cycle, eccentricity is only significant (at the 5% significance level) during limited time interval s of the ETP curve (F igure 3-2A). LWPS of the LR04 stack (Figure 3-2B) shows the well -known mid-Pleistocene climate transition of significant power from a dominant 41-kyr period to a 100-kyr period at 800 ka. Since the LR04 stack is a global climate (ice volu me) record controlled by orbital forcing, Figure 3-2C and D provide a clear demonstr ation of how the |XWT| and WTC should appear if the two records have a dire ct physical linkage at particular frequencies/periods. At these frequencies/periods, as expected, we see significant common power on the |XWT| map, signific ant squared wavelet coherence on the WTC map, and a consistent phase relationship (indicated by arrows on |XWT| and WTC maps) where significant squared wavele t coherence was found (Figure 3-2D). Significance levels of all WTC maps in this study were determined using 500 Monte Carlo runs of randomly generated pairs of red noise records that have the same estimated AR1 parameters as the original data. It should be noted that, according to the

PAGE 53

53 definition of the significance levels of these wavelet maps even a purely random time series produces significant peaks above the 5% significance level and makes up 5% of the total area on these wavelet maps (Figure 4 in Torrence and Compo, 1998). In other words, 5% of all the power on the wavelet maps will exceed that threshold just by chance. To test whether significant ar eas on WTC maps exceed random levels, we calculated the percentages of significant area at different significance levels for WTC maps, and compiled the results in a figure (Fi gure 3-3) similar to Figure 5 of Heslop (2007). We ignored data under the cone of influence (COI, region on wavelet maps where edge effects make the analyses unreliabl e, see caption in Figure 3-2) when we count the significant area and the total area on WTC maps. As is shown in Figure 3-3A, the percentage of signific ant area on WTC map between ETP curve and LR04 dramatically exceeds the random levels (shaded regions in Figure 3-3) at all significance levels, indicating a link between these two signals. Results and Discussion Local wavelet power spectra with robust si gnificance tests are calculated for the seven RPI records (Table 3-1) from differ ent regions (Figure 34). Prior to wavelet analyses, all records have been linearly interpolated into 1-kyr increment time series, linearly detrended, and normalized to have zero mean and unit variance. As can be seen, orbital periods are not significant in RPI records fr om equatorial Pacific cores MD982185 (Figure 3-4F) and MD972143 (Figure 3-4G) through time. This result is consistent with analyses by Horng et al (2003) and Roberts et al. (2003). The variegated character of the LW PS map at periods less than 20 kyr, particularly for Core MD972143 (Figure 3-3F), is probably a manifestation of the low resolution (sedimentation rate) of this core. Orbital per iods of 100 kyr and/or 41 kyr are significant

PAGE 54

54 intermittently in RPI record s from North Atlantic (ODP Sites 983, 984, 919, and IODP Site U1308, Figure 3-4A ~D). For instance, the 100 kyr period is significant during the 300-700 ka interval in the ODP Site 983 RPI record (Figure 3-4A), and is significant mostly during the 500-800 ka interval in the ODP Si te 984 RPI record (Figure 3-4B). The 100 kyr period is shifted to shorte r periods and is limited within the 550 ka time interval in the IODP Si te U1308 RPI record (Figure 3-4D ). Except for the ODP Site 983 RPI record, where it is clearly evident, the 41 kyr period is significant only intermittently in time frequency space of t he other RPI records. For ODP Site 983 RPI record, in which the orbital periods (bot h 100-kyr and 41-kyr) ar e most evident, the characteristic mid-Pleistocene climate transition of significant power from a dominant 41-kyr period to a 100-kyr period is observed at 750 ka (Figure 3-4A), probably implying a climatic origin of the orbital periods for this RPI record. In the case of the South Atlantic (ODP Site 1089) RPI record, 100-kyr periods are significant only in the younger part close to the COI (Figure 3-4E). From Figure 3-4, we can also see that time intervals where significant power at orbi tal periods (either 100 kyr or 41 kyr) were detected in the RPI records are not comparab le among different RPI records. Although uncertainties in the age models may modulat e the wavelet power distribution, they should not be responsible for the absence of or bital periods as long as 41 kyr or 100 kyr. A simple test shows that although the RPI record from equatorial Pacific Core MD972143 shows no significant power at the 100-kyr orbital period (Figure 3-4F), LWPS applied to the oxygen isotope record of this core, using the same age model (Horng et al., 2002), clearly shows significant periods of 100 kyr. Differences in LWPS

PAGE 55

55 among these RPI records (Figure 3-4) may impl y that orbital periods in the RPI records probably do not have a common origin su ch as direct orbital forcing. As a further test of whether orbital periods in RPI records are directly due to orbital forcing, |XWT| and WTC between the ETP curve and RPI records from ODP Sites 983 and 984 were calculated (Figure 3-5A, D, G and J). Although significant common power was observed at orbital periods between the ETP curve and the RPI records, phase relationships (displayed as arrows on |XWT | and WTC maps, see caption in Figure 3-2 for explanation) between the records vary through time (Figure 3-5A and G). The WTC maps (Figure 3-5D and J) further suggest t hat the RPI records are not significantly coherent with the orbital param eters at orbital periods. T he percentages of significant areas on these two WTC maps are lower t han that generated just by random chance (Figure 3-3A). Therefore, orbital periods in RPI records were not caused directly by orbital forcing. It should be noted that wavelet analyses utilized in this study can only reveal the existence of a direct (linear) lin k between two time seri es similar to that between the ETP curve and LR04. The possibi lity of a non-linear relationship between EPT and RPI cannot be excluded. Similarly, |XWT| and WTC were calculated between benthic oxygen isotope records and the RPI records from ODP Sites 983 and 984. The results, however, show significant squared wavelet coherence (Figure 3-5E and K) and consistent in-phase relationships along with significant common power at orbital periods between the records (Figure 3-5B and H). The percentages of significant area on the WTC maps between RPI records and the o xygen isotope records clearly exceed random levels (Figure 3-3A), indicating orbita l periods in the RPI records are most likely due to climatic contamination.

PAGE 56

56 To understand the origin of the apparent cont amination of RPI records, we test various non-magnetic and magnetic parameters. The first c andidate would be the rock magnetic concentration parameters used to normalize the NRM records, for example, anhysteretic remanent magnetization (ARM), isothermal remanent ma gnetization (IRM), or the susceptibility ( ) record. We test whether orbi tal periods in RPI records are directly due to the normalizer records. A test for a synthetic RPI record and a synthetic normalizer record show how the |XWT| and WTC map are expected to appear if the orbital periods in RPI records are directly due to the normalizer records (Figure 3-6). The synthetic RPI record is calculated usi ng a synthetic NRM record, which is modeled by red noise plus a 100-kyr period during the 500-1000 ka interval, normalized by a synthetic ARM record (normalizer), which includes a 41-kyr period during 500-1000 ka interval, a 100-kyr period during 0-500 ka inte rval plus a red noise background. The results indicate that orbital periods in the RPI record could come from either the normalizer record (41 kyr period during 500-1000 ka interval, and 100 kyr period during 0-500 ka interval in the synthetic RPI record see Figure 3-6A) or the NRM record (100 kyr period during 500-1000 ka interval in the syn thetic RPI record, s ee Figure 3-6A). If the orbital periods in the RPI record are completely and directly due to the normalizer record, we might expect to see consistent phase relationships (anti-phase) and significant squared wavelet coherence (F igure 3-6D) along with significant common power (Figure 3-6C) between the RPI record and the normalizer record at these orbital periods. |XWT| and WTC between RPI records and t heir corresponding normalizer records from ODP Sites 983 and 984 (F igure 3-5C, F, I and L) i ndicate significant common

PAGE 57

57 power at orbital periods between RPI record s and their normalizers (Figure 3-5C and I). The records are significantly coherent (Figur e 3-5F and L) during certain time intervals, and the percentages of significant area in the WTC maps are also above the random levels (Figure 3-3A). However, phase rela tionships between the RPI records and their normalizer (ARM) records at orbital periods are quite variable (arrows on Figure 3-5F and l), implying that contaminati on (at orbital periods) is not directly or at least not completely due to the normalizers. It shoul d be noted that RPI records from ODP Sites 983 and 984 were published in a series of papers (i.e. Channell et al., 1997, 1998, 2002; Channell, 1999; Channell and Kleiven, 2000), and the normalizer used to generate the RPI records varied downcore (Table 3-1), although ARM was generally used. For these two sites, the two RPI records (normalized using ARM and IRM) are very similar to each other, hence we feel co nfident using ARM (only) as the normalizer record for the purposes of this test. Since the NRM record is a combinat ion of geomagnetic field and lithologic variations forced by environment/climate, it is not surprising that orbital periods are significant in the NRM records. This is confirmed by the LWPS of NRM records from ODP Sites 983 (Figure 3-4H) and 984 (Figure 3-4I). As has been pointed out (e.g. Tauxe et al., 2006), several factors need to be considered when translating DRM into a RPI record: (1) physical theory concerning par ticle alignment in a viscous medium; (2) compensation for variations in the magnetizability (the choice of normalizer); (3) temporal resolution issues involving the dept h at which the magnetization is fixed (lockin depth) and the degree of smoothing. We ignore the possible influence of temporal resolution issues, and simplify (1) as a param eter called efficiency of alignment which

PAGE 58

58 summarizes all factors that may control par ticle alignment. Further, assuming a uniform magnetic mineralogy (magnetite) in the se diments, normalized intensity record R PI can be expressed by Equation 3-1. FIMCMGAL MCMGSSSE NRM RPI ARMSS (3-1) In Equation 3-1, NRM represents natural re manent magnetization, A RM represents anhysteretic remanent magnetization, which is the normalizer here, FIS represents magnetic field intensity variation, M CS represents concentration of magnetic minerals (magnetite), M GS represents population of magnet ic grains contributing to NRM, ALE represents efficiency of alignment of magnetic grains contributing to NRM, M CS represents concentration of magnetic mi nerals (magnetite) contributing to A RM, M GS represents population of magnetic grains contributing to A RM. We might expect that: M CMCSS so, Equation 3-1 can be reorganized to Equation 3-2. FIMGALMG ALFI MGMGSSES R PI ES SS (3-2) Therefore, if there were any contamination in the RPI record, we would expect it to be from the incompletely normalized component: One way to find the origin of the contaminat ion would be by looking at the |XWT| and WTC between the RPI records and other proxy records that ma y potentially represent the incompletely normalized component, for instance, grain size proxies (may mimic variation) or physical properties (could influence ). For ODP Site 983, the WTC between the RP I record and physica l properties, such as carbonate content and GRA (gamma-ray attenuation) density, and between the RPI

PAGE 59

59 record and grain size proxies, such as ARM/ ( ARM is the anhysteretic remanent susceptibility, and is the volume susceptibility) and ARM/IRM were calculated (Figure 3-7A~D). Although percentages of signific ant area on the WTC maps between RPI records and physical properties from ODP Si te 983 are above the random levels (Figure 3-3B), the phase relationships vary through time, contamination in RPI records is hence not directly related to physical properti es such as GRA density (Figure 3-7B) or carbonate content (Figure 3-7A). This indicates that either these physical properties are not good proxies of or has little contribution to contamination. Similarly, contamination is not directly related to grain size proxy ARM/ at ODP Site 983 (Figure 3-7C and Figure 3-3B). However, signi ficant squared wavelet coherence and a consistent phase relationship (in-phase) at orbital periods between the RPI record and the grain size proxy ARM/IRM (Figure 3-7D) indicate that contamination in the RPI record appears to be reflected by ARM/IRM. We further tested the WTC between RPI records and grain size proxy ARM/IRM for ODP Site 984 and ODP Site 919. The results (Figure 3-7E and F) are consistent with resu lt for ODP Site 983. Significant squared wavelet coherence and consistent in-phase re lationship at orbital periods between RPI record and grain size proxy ARM/IRM, t ogether with significant common power were observed for records from these two sites. The percentages of significant area on the WTC maps between RPI records and ARM/IRM records from these sites are dramatically above random le vels (Figure 3-3B). ARM/IRM ratios are widely employed as grai n size indicators for magnetite (e.g. Meynadier et al., 1995; Rousse et al., 2006; Venuti et al., 2007). Because ARM is more effective in activating finer magnetite grains than IRM, small (large) particles lead to

PAGE 60

60 higher (lower) values of the ratio. The dependence of various ro ck magnetic parameters such as ARM, SIRM, and on magnetite grain size has been studied by many researchers (e.g. Maher, 1988; Dunlop an d Argyle, 1997; Dunlop and zdemir, 1997; Egli and Lowrie, 2002). ARM, SIRM, and activate different gr ain size populations. The ARM/IRM ratio depicts the relative population of magnetic grains contributing to ARM to the population of magnetic grains contributi ng to IRM in the sediments. The ARM/IRM ratio appears to influence the RPI proxies (e.g. NRM/ARM) while the ARM/ ratio may not because is sensitive to magnetic grains wit h much larger grain size that are unlikely to carry appreciable remanence. Changes in fraction of certai n grain sizes (i.e. sortable silt with grain size ranging 10-63 m) have been proposed as reliable tracer s of the bottom current behavior (McCave et al., 1995). Prins et al. (2002) dem onstrated that, for some North Atlantic sites on the Reykjanes Ridge, well-sorted cl ay to fine silt fractions with a peak distribution <10 m provide a better proxy for the behavior of bottom currents because of an important silt fraction in ice ra fted debris (IRD). Snow ball and Moros (2003) analyzed cores that are also from the Re ykjanes ridge and noted that the magnetic grain size distribution of the sediments lies within the size range considered by Prins et al. (2002), and hence is suitable for reconstruc tion of the near-botto m current velocity. Coarse (fine) grains within this grain size fraction would generally imply enhanced (reduced) bottom current strength, presumably re lated to the formati on of North Atlantic Deep Water (NADW). As ODP Site 919 ( Clausen, 1998), and ODP Sites 983 and 984 (Bianchi and McCave, 2000) are located on the flow path of NADW, it is likely that variations in bottom current behavior at t hese sites correspond to climatic changes on

PAGE 61

61 orbital time scales. Significant power at 100 kyr period is clearly observed in ARM/IRM records from these sites that is coherent with the oxygen isotope records. See Figure 34J~L for LWPS of the ARM/IRM record, Fi gure 3-7G~I for WTC maps between the oxygen isotope records and ARM/IRM records, and Figure 3-3C for results of testing the significant areas on WTC maps against r andom chance. Lack of significant WTC between the oxygen isotope and ARM/IRM reco rds from ODP Site 919 (Figure 3-7I) may be a reflection of melt water related per turbations on the planktic oxygen isotope record at this site (St. John et al., 2004). It is likely that orbital periods in the RPI records were introduced into the NRM records (and have not been normalized) through climatic control on bottom current velocity, which in turn controls the magnetic grain size distribution (corresponds to the component in Equation 32) of the sediments. Changes in bottom current velocity in the Nort h Atlantic are accompanied by variation in grain size and concentration of magnetic parameters (e.g. Kissel et al., 1999; Ballini et al., 2006). As floc size is controlled by s hear stress (current velocity) and particle concentration (Winterwerp and Kesteren, 2004), changes in bottom current velocity at orbital periods may introduce additional effe cts on the NRM record that amplify the contamination at orbital periods (i n terms of effici ency of alignment, in Equation 3-2). Since contamination exists in the RPI records, one critical question is whether the contamination is debilitating to the RP I records as a geomagnetic signal, and as a useful stratigraphic correlation tool. First of all, as can be seen from WTC between RPI records and benthic oxygen isotope records from ODP Sites 983 an d 984 (Figure 3-5e and k), climatic contamination (area with significant common power, consistent inphase relationship, and significant squar ed wavelet coherence on |XWT| and WTC

PAGE 62

62 maps) exists only at orbital periods during so me time intervals. Secondly, a comparison was made between RPI records from different regions of the world (ODP Sites 983 and 984 from North Atlantic, and cores MD97214 3 and MD982185 from equatorial Pacific). It appears that these RPI records are highly comparable with each other in the time domain (Figure 3-8A). To reduce discrepancies attributable to age models, RPI records have been first optimally correlated to the OD P Site 984 RPI record using the Match protocol of Lisiecki and Lisiecki (2002). We also compared the RPI records from ODP Sites 983 and 984 before and after filtering the orbital periods from the RPI records. A bandstop filter has been applied to the RPI records with stop bands located at (1/120) (1/80) kyr 1 frequencies and at (1/50) (1/30) kyr 1 frequencies. It appears that filtering does not alter the RPI records very much (Fi gure 3-8A). This is not surprising when we look at the WTC maps of these RPI record s (Figure 3-8B~D, also see Figure 3-3C for significance tests for the WTC maps). The c ommon features (indicated by significant WTC values and in-phase relationship) in these RPI records span a large range of periods ( 16-400 kyr). These results suggest that c ontamination, although it exists in the RPI records (at orbital periods), does not strongly affect the char acteristic features of the paleointensity records, and hence is not debilitating to these RPI records as a global signal that is primarily of geomagnetic origin. Conclusions The present study confirms previous analyses that orbital periods are not significant in RPI records from equatorial Pacific cores MD982185 and MD972143 through time. Orbital periods are significant only during limited time intervals in RPI records from North Atlantic (ODP Sites 919, 983, 984, IODP Site U1308), and South

PAGE 63

63 Atlantic (ODP Site 1089). The characteri stic mid-Pleistocene c limate transition is observed in the ODP Site 983 RPI record, in which the orbital periods are most evident, indicating a possible climatic origin of the or bital periods for this RPI record. The fact that time intervals where orbital periods are significant are not comparable among RPI records from different regions provides evidence that orbita l periods in the RPI records are not directly due to orbital forcing. This is further indicated by the |XWT| and WTC between ETP curve and RPI records from ODP Sites 983 and 984. The |XWT| and WTC between benthic oxygen isotope records and the RPI records from ODP Sites 983 and 984, however, show significant coherence and consistent in-phase relationships at orbital periods and hence imply that orbital periods in the RPI records are most likely due to climatic contamination. Phase relationships between RPI reco rds and their corresponding normalizer records from ODP Sites 983 and 984 exclude t he possibility that contamination (at orbital periods) is directly or completely due to the normalizers. As suggested by the analysis of a synthetic RPI record and a synt hetic normalizer record we may expect to see consistent anti-phase relationship along with significant squared wavelet coherence between RPI and the normalizer if orbital peri ods in the RPI record are completely and directly due to the normalizer record. Orbita l periods are significant in the NRM records from ODP Sites 983 and 984. D eduction from the theoretical RPI model attributes the contamination in RPI records to incomp letely normalized component of the NRM records. Further tests of re cords from ODP Site 983 indica te that contamination is apparently not directly related to physical pr operties such as density or carbonate content, and the grain size proxy ARM/ WTC between RPI records and grain size

PAGE 64

64 proxy ARM/IRM from ODP Sites 919, 983, and 984 imply that ARM/IRM may mimic the contamination in the RPI records at t hese sites. Orbital periods were probably introduced into the NRM records (and have not been normalized when calculating RPI records) through orbital control on the botto m current velocity that regulated NADW formation, which in turn controls magnetic grai n size distribution at the site, and possibly also the degree of particle flocculation. Four lines of evidence indicate that cont amination, although it exists in these RPI records, is not debilitating to these RPI record s as a global signal that is primarily of geomagnetic origin: (1) contamination exists only at orbital periods during limited time intervals. (2) RPI records from different regions of the world (ODP Sites 983 and 984 from North Atlantic, and co res MD972143 and MD982185 from equatorial Pacific) are highly comparable with each other in the ti me domain. In view of the contrasting climatic/lithologic characteristics of the No rth Atlantic and equatorial Pacific, we would not expect to observe this level of similari ty if the RPI records were dominated by lithologic contamination. (3) F iltering orbital periods from the RPI records does not alter the RPI records appreciably other than reducing the amplitude of certain features in the records. (4) WTCs among these RPI records show large areas (s panning a large range of periods: 16-400 kyr) characterized by significant coherence and in-phase relationships.

PAGE 65

65 Table 3-1. Location, water depth, lengt h, sedimentation rate, normalizer, and es timated age for relative paleointensity records used in this study Records Location Water depth (m) Record length (m) Sedimentation rate (cm/kyr) Normalizer used Oldest age (Ma) Reference ODP 983 60.40N, 23.64W 1983 260 14.7 (2.9-34.6) ARM, IRM 1.9 [1-4] ODP 984 61.48N, 24.18W 1660 260 13.3 (2.7-41.2) ARM, IRM 2.15 [2, 4, 5] ODP 919 62.67N, 37.46W 2088 70 15.6 (7-28) ARM 0.5 [6] IODP 1308 49.87N, 24.23W 3871 85 7.3 (3.8-16) ARM 1.2 [7] ODP 1089 40.94S, 9.89E 4620 88 ~17.5 (5-33) ARM 0.58 [8] MD982185 3.08N, 135.00E 4415 42 ~1.5 (1.3-4.9) ARM 2.25 [9, 10] MD972143 15.87N, 124.65E 2989 38 ~1.5 (0 .6-4.6) Susceptibility 2.14 [11] Note: In the sedimentation rate column mean sedimentation rate is followed by the range of sedimentation rate in parentheses. In the normalizer column, listed are normalizers fo r RPI data analyzed in this study Note that for ODP Sites 983 and 984 the normalizer used to generate the RPI records varied downcore, although ARM was generally used. See Figure 3-1 for location map of these records. References: [1]: Channell et al., 1997; [2]: Channell et al., 1998; [3]: Channell and Kleiven, 2000; [4]: Channell et al., 2002; [5]: Channell, 1999; [6]: Channell, 2006; [7]: Channell et al., 2008; [8]:Stoner et al., 2003; [9]:Y amazaki and Oda, 2002; [10]:Yamazaki and Oda, 2005; [11]:Horng et al., 2003.

PAGE 66

66 Figure 3-1. Location of sites discussed in this study.

PAGE 67

67 Figure 3-2. A) Local wavelet power spectrum (LWPS) of the ETP curve calculated from normalized value of the Laskar et al (2004) solution, B) LWPS of global benthic oxygen isotope stack LR04 fr om Lisiecki and Raymo (2005), C) cross-wavelet power spectrum (|XWT|) of the ETP curve and LR04, D) squared wavelet coherence (WTC) between the ETP curve and LR04. Values of normalized wavelet power, cross-wavelet power, and squared wavelet coherence are indicated using different colors on LWPS, |XWT|, and WTC maps (with blue to red indicating incr easing values). The 5% significance level against red noise is shown as thi ck contours in all figures. The cones of influence (COI) where edge effects make the analyses unreliable are marked by areas of crossed lines. In C) and D) the relative phase relationship is shown as arrows (with in-phase poin ting right, anti-phase pointing left, and orbital signal leading oxygen isotope signal 90 pointing straight up). Orbital periods of 100 kyr, 41 kyr, and 23 kyr are marked by white dashed lines from bottom to top on the vertical axes.

PAGE 68

68 Figure 3-3. Percentage of significant area at different significance levels for A) squared wavelet coherence (WTC) plots in Figure 3-2 and Figure 3-5, B) WTC plots in Figure 3-7A~F, and C) WTC plots in Figure 3-7G~I and Figure 3-8. Significance levels of WTC were dete rmined using 500 Monte Carlo runs of randomly generated pairs of red noise that have the same estimated AR1 parameters as the origi nal data. COI was ignored when counting the significant area and the total area on WTC plot. The shaded region indicates results where the percentage of signific ant area is less than that expected by random chance.

PAGE 69

69 Figure 3-4. Local wavelet power spectra (LW PS) of RPI records fr om A) Ocean Drilling Program (ODP) Site 983 (Channell et al., 1997, 1998, 2002; Channell and Kleiven, 2000), B) ODP Site 984 (C hannell et al., 1998, 2002; Channell, 1999), C) ODP Site 919 (Channell, 2006), D) Integrated Ocean Drilling Program (IODP) Site U1308 (Channel l et al., 2008), E) ODP Site 1089 (Stoner et al., 2003), F) Equatoria l Pacific core MD982185 (Yamazaki and Oda, 2002, 2005), G) Equator ial Pacific core MD972143 (Horng et al., 2003); LWPS of NRM records (from 25mT demagnet ization step) from H) ODP Site 983 and I) ODP Site 984; and LWPS of AR M/IRM records from J) ODP Site 983, K) ODP Site 984 and L) ODP Site 919. ARM/IRM records are calculated using ARM and IRM records from 35 mT demagnetization step. See caption of Figure 3-2 for description of wavelet maps.

PAGE 70

70 Figure 3-5. Cross-wavelet power spectra (|XWT|) of A) ODP Si te 983 RPI record and ETP curve, B) RPI record and benthic ox ygen isotope record from ODP Site 983, C) RPI record and ARM record fr om ODP Site 983, G) ODP Site 984 RPI record and ETP curve, H) RPI re cord and benthic oxygen isotope record from ODP Site 984, I) RPI record and ARM record from ODP Site 984; and squared wavelet coherence (WTC) between D) ODP Site 983 RPI record and ETP curve, E) RPI record and benthic ox ygen isotope record from ODP Site 983, F) RPI record and ARM record from ODP Site 983, J) ODP Site 984 RPI record and ETP curve, K) RPI record and benthic oxygen isotope record from ODP Site 984, L) RPI record and AR M record from ODP Site 984.

PAGE 71

71 Figure 3-6. Local wavelet power spectra (LWPS) of A) synthetic RPI record and B) synthetic ARM record, C) cross-wavelet power spectrum (|XWT|) of the synthetic RPI record and synthetic ARM record, D) squared wavelet coherence (WTC) between the synthetic RPI record and synthetic ARM record.

PAGE 72

72 Figure 3-7. Squared wavelet coherence (WTC) between A) RPI record and carbonate content record from ODP Site 983, B) RPI record and GRA (gamma-ray attenuation) density record from OD P Site 983, C) RPI record and ARM/ record from ODP Site 983, D) RPI re cord and ARM/IRM record from ODP Site 983, E) RPI record and ARM/IRM re cord from ODP Site 984, and F) RPI record and ARM/IRM record from ODP Site 919, G) benthic oxygen isotope record and ARM/IRM record from ODP Site 983, H) benthic oxygen isotope record and ARM/IRM record from ODP Si te 984, (i) planktic oxygen isotope record and ARM/IRM record from ODP Site 919. ARM/IRM is calculated using ARM and IRM records from 35 mT demagnetization step. Carbonate content data of ODP Site 983 are from Ortiz et al. (1999). GRA density data of ODP Site 983 are from Shipbo ard Scientific Party (1996).

PAGE 73

73 Figure 3-8. Comparing RPI records in A) the time domain: RPI records are from the equatorial Pacific (MD982185 and MD972143) and North Atlantic (ODP Sites 983 and 984), also shown are RPI records of ODP Sites 983 and 984 with orbital periods filtered ( bandstop filtering was applied with stop bands located at (1/120) (1/80) kyr 1 frequencies and at (1/50) (1/30) kyr 1 frequencies); comparing RPI records in the time frequency domain: B) WTC between RPI records from ODP Sites 984 and 983, C) WTC between RPI records from ODP Site 984 and equatorial Pacific Core MD972143, and D) WTC between RPI records from ODP Site 984 and equat orial Pacific Core MD982185. To reduce discrepancies from age models, RPI records have been first optimally correlated to ODP Site 984 RPI record us ing the Match protocol of Lisiecki and Lisiecki (2002). In A) RPI scale has been shifted for each record for better display. See caption of Figure 3-2 for description of the WTC maps, and caption in Figure 3-4 for references for the RPI record s. The additional dashed white lines on the bottom of B), C), and D) mark the 404-kyr period on the vertical axes.

PAGE 74

74 CHAPTER 4 STACKING PALEOINTENSITY AND OXYGEN ISOTOPE DATA FOR THE LAST 1.5 MYR (PISO-1500) Introduction The quest for improved stratigraphic correlation remains one of the great challenges in paleoceanography. Since Shac kleton (1967) made th e case for oxygen isotopic data as a monitor of global ice volume, benthic 18O has been the hallmark of marine stratigraphy, however, oxygen isotop e changes in seawater are not globally synchronous on (millennial) timescales associated with the mixing time of the oceans (e.g. Skinner and Shacklet on, 2005). Although oceanic mean 18O will reflect global ice volume change over multimil lennia, an individual benthic 18O record will reflect both global glacio-eustatic and local hydrogr aphic signals (temperature and deepwater 18O) to varying degrees (Skinner and Shackleton, 2006). There would be great advant age in coupling oxygen isot opes with an independent stratigraphic tool that is global in nat ure and devoid of environmental influences. Traditional magnetic stratigr aphy, the observation of polar ity zones in sedimentary sections, has become the backbone of geologic timescales partly because polarity reversal is a geophysical phenomenon attributable to the main dipole field, and therefore provides global timelines for pr ecise correlation at t he time of reversal. Accumulation of relative paleointensity (RPI) data in the last 10 yr holds the promise of stratigraphic correlation within polarity ch rons, possibly at millennial scale. The association of some RPI minima with brief magnetic excursions means that there is a manifestation of RPI stratigraphy in the paleomagnetic directional record. A first step in the utilization of paleointensity records in stratigraphy is the development of a calibrated template. As for benthic oxygen isotopes, a template based

PAGE 75

75 on multiple records rather than an individual record, has the advantage of increasing the signal to noise ratio in that local or regi onal change recorded by individual records may be averaged out by the stacking process. The disadvantage of stacks, on the other hand, is that the stacking process inevitabl y reduces the resolution of the output, compared to the input of individual reco rds. Several paleointensity stacks have been produced in the last 10 yr including Si nt-800 and Sint-2000 (Guyodo and Valet, 1999; Valet et al., 2005) covering the last 800 and 2000 kyr, respectively. For the last 75 kyr, regional stacks for the North At lantic (NAPIS) (Laj et al ., 2000) and South Atlantic (SAPIS) (Stoner et al., 2002), and a global sta ck (GLOPIS) (Laj et al., 2004), have been generated. The EPAPIS stack fr om the western equatorial Pa cific covers the 0.75.0 Ma interval (Yamazaki and Oda, 2005). Here we present a stack that differs from earlier stacks in that the stacking process is conduc ted simultaneously on both oxygen isotope and RPI data. The stack includes, therefore, only RPI records that have accompanying oxygen isotope records. Most of the coupled RPI-isotope records included here (Table 4-1) have not been included in previously der ived RPI stacks, at least not in their entirety. The new RPI/isotope stack (named PISO-1500) provides a new template for correlation and calibration of RPI and isotope records. The fidelity of the accompanying oxygen isotope stack can be tested by com parison with a widely-used calibrated oxygen isotope stack (Lisiecki and Raymo, 2005). This exercise can gauge the level of consistency of the two stratigraphies, and possibly determine whether the RPI correlations enhance isotopic correlations w hen the records are used in tandem.

PAGE 76

76 The Match and the Stack The potential of RPI for high-resolution co rrelation stems from the high rate of change of the Earth's dipole fi eld intensity, that has decr eased by ~5% over the last 100 yr in the historical record, and by ~20% ov er the last 1000 yr in the archaeological record (Korte and Constable, 2005; Valet et al., 2008). In comparison, the rate of change of global ice volume (the basis for benthic 18O stratigraphy) tends to be greatest at glacial terminations but relatively slow otherwise. Benthic 18O shows comparatively little variability during ma rine isotopic stage (MIS) 2-4 (20 ka) in contrast to RPI variations in this interv al. Previous RPI stacks involve some form of visual matching of features in RPI records that can be arbitrary, and poorly constrained. To make signal correlation mo re objective, we use the Match protocol (Lisiecki and Lisiecki, 2002) that utilizes dynamic progra mming to find the optimal fit between record pairs. The method was used by Lisiecki and Raymo (2005) to cons truct their benthic oxygen isotope stack (LR04).We apply the prot ocol simultaneously to isotope and RPI records for 12 published records (Table 4-1, Figur e 4-1) to optimize their fit to IODP Site U1308 records. We use Site U1308 as the refe rence record for the stack, because the site has high-resolution benthic isotope and RPI data for the entire 1.5 Myr interval (Channell et al., 2008;Hodell et al., 2008). The age model for Site U1308 was constructed by fit of the benthic oxygen is otope record to LR04 (Hodell et al., 2008). At ODP Sites 983 and 984 (Figure 4-1, Table 4-1) the RPI records have higher resolution than the accompanying isotope records which ar e limited by paucity of foraminifera prior to 1.1 Ma (Channell, 1999; Channell et al., 2002, 1997; Channell and Kleiven, 2000). The only other record that spans the full 1. 5Myr interval is Core MD97-2143 (Horng et al., 2002, 2003); however, this record has lower mean sedimentation rate, than ODP

PAGE 77

77 Sites 983/984, by almost an order of magnitude (Table 4-1) Other records included in the stack cover parts of the 1.5Myr inte rval: ODP Site 980 (Channell and Raymo, 2003) extends back to 1.3 Ma, ODP Site 919 (Cha nnell, 2006) extends back to 540 ka, ODP Site 1089 (Stoner et al., 2003) extends back to 580 ka, MD99-2227 (Evans et al., 2007) extends back to 430 ka, andMD95-2039 (Thouveny et al., 2004; Thomson et al., 1999) extends back to 317 ka. The coupled isotope/RPI record from Core MD97-2140 covers the 560-1300 ka interval (Carcaillet et al ., 2003, 2004; De Garidel-Thoron et al., 2005). Three records that cover the last glacia l cycle only: MD95-2024 (Stoner et al., 2000), the Somali Basin record (Meynadier et al., 1992), and PS2644 (Laj et al., 2000; Voelker et al., 1998) are included to strengthen the stack in this interval (Table 4-1). The use of the Match protocol here differs from its use in a recent paper (Channell et al., 2008) where we applied the protocol to RPI records only, and forced compatibility with oxygen isotope records by inserting hard tie points (usually at terminations) derived from the marine isotope records. By applying the Match protocol to the RPI and isotope records in tandem, we make better use of the isotope record and provide an improved test of the compatibility of RPI and isotope records. Each RPI and isotope record (Table 4-1) is matched to the IODP Site U1308 records, after normalizing each RPI or isotope record to have zero mean and one standard deviation. The starting points for each match are the original (publis hed) age models (references in Table 4-1). In the Match protocol, each RPI and isotope record is divided in to time intervals (with initial ~1 kyr duration in our case), and each ~1 kyr interva l is matched to interval(s) in the Site U1308 records. The quality of the fit of a parti cular interval to interval(s) at Site U1308 is gauged by the sum of the squares of the differences for the data points within

PAGE 78

78 each interval pair. The optimal correla tion of the two records is determined by minimization of the squares of the differences through the dynamic programming procedure (Lisiecki and Lisiecki, 2002), with penal ty functions limiting the likelihood of abrupt sedimentation rate c hanges both within an individual record and between record pairs. The data points from one interval may lie anywhere along the interpolated line connecting data points in the i nterval with which it is compared. The sequential order of data points within an interva l, and the order of intervals in each time series, must be maintained. When matching multiple data types (e.g. RPI and oxygen isotopes), the oxygen isotope data are linearly interpolated to the same age scale as the RPI records, and Match computes the sum of the squares of the differences for each data type to find the optimal match. Unlike routine eyeball sig nal correlations, the Match correlations are repeatable and unbiased, and less likely to be diverted by local solutions. The quality of fit repr esented by the Match output (Fi gure 4-2) is checked visually for obvious discrepancies, and then formally test ed in two ways. Firstly, the correlation coefficients between IODP Site U1308 records and the other records are calculated before and after matching RPI and isotope dat a (Table 4-2). RPI and isotope data from each record were interpolated to the age scale of IODP Site U1308 before calculation of correlation coefficients. The before match correlation coefficient s correspond to the match of record pairs on their published age models (Table 4-1).Note that, in all cases, the correlation coefficients are increased, oft en substantially, after matching (Table 4-2). Secondly, we test whether the matching proc ess has indeed improved the correlation in time frequency space by calculating the squared wavelet coherence (WTC) before and after matching. The definition of the WTC resembles that of a traditional correlation

PAGE 79

79 coefficient, and can be thought of as a localiz ed correlation coefficient in time-frequency space. It can be used to identify both fr equency bands and time intervals within which the two time series are co-varying and statis tical significance levels can be determined using Monte Carlo methods (Torrence and Compo, 1998; Grinsted et al., 2004). The RPI and oxygen isotope data were linearly in terpolated at 1-kyr increments prior to calculation of squared wavelet coherence (W TC). As an example, we plot the WTC before and after matching of RPI and isot ope data for ODP Site 983 and IODP Site U1308 (Figure 4-3). For both RP I and isotope data from these two sites, we observe an improvement in coherence of RPI and isot ope records after matching (more area with significant WTC and constant phase relationsh ip), compared to the original age models (Figure 4-3). The improvement is evident in the 10-200 kyr period range, indicating lower record sensitivity for periods below10 kyr. In Table 4-2, the improved correlation to IODP Site U1308 after matc hing is expressed as the perc entage of significant (at 5% level) area on WTC maps (red area bounded by thick black line in Figure 4-3) before and after matching. Data within the cone of influence (where edge effects make the analyses unreliable) were ignored when counti ng the significant areas. In all but one case, both correlation coefficients and percentages of significant WTC indicate that the Match protocol improves the fit of both RP I and isotope data to the Site U1308 records (Table 4-2). The one exception is the MD 95-2024 (planktic) isotope record where although the correlation coefficients indicate improvement of fit after matching, the percentage of significant WTC indicates no significant improv ement for the isotope records (Table 4-2).

PAGE 80

80 The stack is constructed for RPI and oxygen isotopes by re-sampling each matched RPI and isotope record, and the reference records from IODP Site U1308, at 1-kyr sample spacing. The ar ithmetic mean, determined at each point, constitutes the stacked record (Figure 4-4). The standard error ( ) is determined by (bootstrap) calculations from 1 million random samplings of data points associated with each point in the stack. The half-width of the error envelope in Figure 4-4 equals 2 The PISO1500 stacks can be found in Supplementary Data as two files. Comparison with other Stacks The ~100-kyr scale features in the new st ack are largely compat ible with the Sint2000 stack (Valet et al., 2005) (Figure 4-4) and enhanced detail in the new stack can be attributed to higher sedimentatio n rate records included in the new stack as well as the use of coupled isotope/RPI matching that reduces age discrepancies among the stacked records. The Sint-2000 stack is an extension of the earlier Sint-800 (Guyodo and Valet, 1999), and comprises records from the world's oceans including the ODP Leg 138 record and the younger parts of ODP Sites 983 and 984. The resolution of the Sint-2000 stack is limited by low sedimentation rates and poor age control for some of the records, and smoothing inherent in the st acking process. In Figure 4-5, we compare the new stack with the EPAPIS Pacific sta ck (Yamazaki and Oda, 2005), the equatorial Pacific record from ODP Leg 138 (Valet and Meynadier, 1993) and the record of paleointensity derived from t he East Pacific Rise (EPR) magnetic anomaly data (Gee et al., 2000). Whereas the age calibration of the EPR record is loosely constrained by spreading rate assumptions, the ODP Leg 138 reco rd is rather precisely calibrated by astrochronology (Shackleton et al., 1995) al though low mean sedimentation rates (1 cm/kyr) limit the resolution of the record. Mean sedim entation rates in the EPAPIS

PAGE 81

81 Pacific stack are also b2 cm/kyr (Yam azaki and Oda, 2005). Th e EPAPIS age model relies on the correlation of lows in magnetic concentration parameters (susceptibility and anhysteretic remanence intensity) with gl acial intervals in a reference oxygen isotope record, in effect relying on the IndoPacific carbonate pattern of high carbonate content during glacial intervals. Interestingly, the age of features in the EPAPIS stack appear to be displaced to older ages relative to correlative features in the PISO-1500 stack (Figure 4-5), implying an offset in carbonate (and magnetic) concentration and isotope stages in the EPAPIS cores. This ty pe of offset, resulting from carbonate concentration variations lagging benthic 18O, has been documented in the IndoPacific carbonate pattern of the sub-Anta rctic South Atlantic (Hodell et al., 2001). Discussion Benthic 18O stratigraphy has been the gold standard for correlation of Quaternary marine sequences, but this tool has import ant limitations, especially for shorter timescales (Skinner and Shackleton, 2005, 2006). Simultaneous matching of isotope and RPI signals (from the same cores) prov ides additional cons traints on oxygen isotope correlations and may allow detection of subtle leads and lags among aspects of the climate system. Simultaneous correlation of RPI and isotope records reduces the degree of freedom associated with correlations using RPI or isotope records alone, and serves as a test for the consistency of RPI and isotope correlations. The PISO-1500 RPI stack is based on 13 coupled RPI/isot ope records where record matching is accomplished by simultaneous optimization of RPI/isotope records using the Match protocol (Lisiecki and Lisiecki, 2002). The fidelity of the PISO1500 isotope stack can be gauged by comparison with the LR04 isotope stack (Lisiecki and Raymo, 2005) (Figure 4-4). Local wavelet power

PAGE 82

82 spectra (LWPS) for the PISO-1500 isotope stack (Figure 4-6D) and LR04 (Figure 4-6B) indicate very similar distributions of significant orbital pow er in the two stacks, implying that the resolution of the two stacks is comparable. Note that the mid-Pleistocene climate transition from 41 kyr to 100 kyr powe r, at about 800 ka, is well represented in LWPS maps of both the LR 04 and PISO-1500 isotope sta cks (Figure 4-6B and D). Non-axial-dipole (NAD) com ponents in the historical field vary on centennial timescales (Hulot and Le Moul, 1994; Hongre et al., 1998; Valet et al., 2008) and if similar repeat times hold in the geologic past, paleointensity records from cores with sedimentation rates less than ~15 cm/kyr are unlikely to record anything but the axial dipole field, and therefore shoul d represent a global signal. In the last 10 yr, there has been considerable debate ov er the observation and interpretation of orbital periods in RPI data (Channell et al., 1998; Yamazaki, 1999; Yokoyama and Yamazaki, 2000; Guyodo et al., 2000; Yamazaki and Oda, 2002; Roberts et al., 2003; Yokoyama et al., 2007; Thouveny et al., 2008; Xuan and Channell, 2008a,b). Some authors consider ed that the orbital power in RPI records can be attributed to the geodynamo (e.g. Yokoyama et al., 2007) while others considered that it is due to lithologic/climatic contamination of some RPI re cords (e.g. Xuan and Channell, 2008b). Although there are patches of significant power at orbital periods in the PISO-1500 RPI stack (Figure 4-6C) and Sint-2000 (Figure 4-6A), there is much less significant orbital power in the stacks than in some RPI individual records (see Xuan and Channell, 2008b). The distribution of significant power in these plots (Figure 4-6A and C) is dependent on the estimated level of (AR1) noise, which is set at 0.9264 for both stacks

PAGE 83

83 (estimated using the PISO -1500 RPI stack). The squared wavelet coherence map (Figure 4-6E) indicates significant c oherence between the PISO -1500 RPI and isotope stacks during limited time intervals (e .g. 660 ka and 1120 ka) and in variable frequency bands. The phase relationships betw een the two records, however, vary during these time intervals (see Figure 4-6E) implying either a non-linear relationship between the two records or, more probably, a fortuitous occurrence of significant coherence of two unrelated records that have power in similar frequency bands. A visual comparison between the PISO-1500 RPI and 18O stacks shows positive correlations at ~660-810 ka (shaded area in Figure 4-9) and negative correlations at ~1120-1240 ka (shaded area in Figure 4-9), c onsistent with the WTC map (Figure 46E). The patch of significant orbital power (at ~90 kyr) at ~600800 ka in the PISO-1500 RPI stack (Figure 4-6C) appears significantly coherent with the is otope stack (Figure 46E), implying (climate-related lit hological) contamination in th is part of the RPI stack. The inventory of magnetic excursions (bri ef quasi polarity chrons with duration less than ~10 kyr) has developed over the last 20 yr to the point where about 7 magnetic excursions are reasonably well documented in the Brunhes Chron, with about the same number having been adequately recorded in the la ter part of the Matuyama Chron (see Laj and Channell, 2007). The brie f (a few kyr) duration of directional excursions makes their observation problematic even for sediments with mean sedimentation rates above 10 cm/kyr because of smoothi ng (filtering of higher frequency remanence variations) associated with the remanence acquisition, the stochastic (non-uniform) nature of sediment deposition, and the effe cts of bioturbation. The paleointensity lows associated with magnetic excursions (and polarity reversals), on th e other hand, have greater

PAGE 84

84 duration than the associated di rectional excursions and ar e often clearly observed in individual RPI records, even when the direct ional excursion is no t. In the PISO-1500 RPI stack, the ages of excursions coinci de with paleointensity minima in the stack (Figure 4-4). In order to calibrate the new RPI stack and arrive at temporal va riations in virtual axial dipole moment (VADM),we adopted the value of 7.4622Am2 for the time averaged VADM for the last 800 kyr from Va let et al. (2005), based on an analysis of the volcanic paleointensity (PINT03) databas e of Perrin and Schnepp (2004). We then scaled the stack to this mean value (for the last 800 kyr) and assigned an intensity of 7.5 T at IODP Site U1308 (the reference re cord) for the minimum RPI value in the stack, that lies within the Cobb Mountain Subchron (Figure 44). This procedure follows Constable and Tauxe (1996) who suggested scali ng RPI records by using this minimum intensity value, based on the modern field, th at corresponds to the likely value of the residual field after total demise of the axial dipole (at times of reversal). The resulting scaling (Figure 4-7) yields a VADM value appropriate for the recent field (~7.522 Am2), maxima of ~1522 Am2 at ~660 ka and between the Jaramillo and the Cobb Mountain subchrons, and an apparent th reshold VADM value of ~2.522 Am2 associated with excursions and reversals. Co mparison of Figs. 4 and 7 indicates that almost all VADM values below this threshold are associated with either excursions or reversals. Note that the Cobb Mount ain Subchron (at ~1200 ka) appears as an exceptionally long interval (~30 kyr) of particu larly low field intensity. The VADM values for the last 1.5 Myr do not show an increas e in paleointensity at the onset of the Brunhes Chron analogous to that seen in Si nt-2000 (see Figure 2A in Valet et al.,

PAGE 85

85 2005). The mean VADM is higher fo r the Brunhes Chron (7.522 Am2) than for the late Matuyama (6.8022 Am2); however, this difference in mean VADM is less than the mean value of the standard error associated with the stack (0.77), and therefore the two mean values are not statistica lly distinguishable. Note t hat the mean Brunhes VADM is close to the apparent modern field value, whic h is close to its actual modern value although this value is not set as part of the scalin g process. The range of VADM values associated with PISO-1500, reaching more than 1522 Am2 (Figure 4-7), is consistent with the range of values for the last few m illion years in the volcanic PINT06 database (see Figure 18 in Tauxe and Yamazaki, 2007). Following Fuller (2006), Thouveny et al. (2008) have made the case for a link between RPI and orbital obliquity in which RP I lows are associated with a particular phase (minima or decreasing obliquity angles) in the obliquity cycle. This relationship has been thought to be due to common orbital forcing for RPI and oxygen isotope values (Thouveny et al., 2008). Although there is no statistical evidence for a relationship between the obliquity angle and ages of excursions and reversals on longer timescales (Xuan and Channell, 2008a), the new stack provides us with the opportunity to test the relationship between the obliq uity angle (and other orbital cycles) and RPI. Following the methods described by X uan and Channell (2008a), and assuming no uncertainty in the ages of the RPI minima or in the astronomical solution, we compute the phases, as defined by Xuan and Channell (2008a), of the orbital eccentricity, obliquity, and precession at the time of the prominent RPI mi nima (as labeled in Figure 4-7). The results are shown in Figure 4-8 with p -values associated with the Rayleigh test, indicating that we cannot reject a circular uniform distribution of RPI minima

PAGE 86

86 relative to the phases of any of the orbi tal parameters (even at a 30% significance level). In other words, ther e is no tendency for RPI minima in the stack to occur at particular phases of orbital variations such as lows or decreasing values of obliquity. Visually, the timing of the prom inent RPI lows in the stack (red dots in Figure 4-9) do not appear to preferentially occur at obliquity lows or in the dec reasing part of the obliquity cycles. Conclusions The PISO-1500 RPI stack provides a template for geomagnetic paleointensity for the last 1.5 Myr. The stack has higher tem poral resolution than earlier stacks such as Sint-2000 (Valet et al., 2005) due to the addi tion of new RPI/isotope records and the use of coupled isotope and RPI records that allow the simultaneous correlation of isotope and RPI records using the Match al gorithm (Lisiecki and Lisiecki, 2002). The advantage of this procedure is that it limits the degree of freedom associated with individual signal correlations, is repeatable, a nd allows assessment of the compatibility of RPI and isotope records. The tandem stacks pr ovide a powerful stratigraphic tool that can be used to correlate among marine sedim ent records, and link them to polar ice cores via cosmogenic nuclide flux. RPI data from marine cores have been correlated to modeled paleointensity based on cosmogenic nucli de flux in Greenland ice cores (e.g. Muscheler et al., 2005), and to 10Be/9Be data derived from the sediments themselves (e.g. Carcaillet et al., 2004). Once a continuous record of comogenic isotope abundance has been completed for the 800-kyr record fr om EPICA-Dome C (Jouzel et al., 2007), correlation to the new PISO-1500 stack will pe rmit synchronization of marine and ice cores. This will eventually allow transfer of ice core chronologies, such as those derived by correlation of N2/O2 to local insolation forcing (Kawamura et al., 2007), to the marine

PAGE 87

87 record (and vice-versa). Such an indepen dent marine chronology could eventually eliminate difficulties in herent in oxygen isotope stratigraphy such as the assignment of phase lag between forcing and response implicit in insolation-driven ice-sheet models (Imbrie and Imbrie, 1980), and the assumption that 18O is purely a glacial ice volume signal. In the future, the timing and phase re lationships of com ponents of the climate system may be resolvable independent of ox ygen isotope stratigraphy, thereby advancing our understanding of the mechanisms of glacial interglacial climate change.

PAGE 88

88 Table 4-1. Records used in the c onstruction of the PISO-1500 stacks. Record Loc. Lat. () Long. () WD (m) L (m) Age range (ka) Ave. sed. rate (cm/ky) Ave. time step (yr) Iso Ref. IODP U1308 N.Atl. 49N 24 3871 109 5-1500 7.3 137 B (1 ODP 983 N.Atl. 60N 23W 1983 199 5-1500 13.3 75 B (2) ODP 984 N.Atl 61N 24W 1648 183 5-1500 12.2 82 B (3) ODP 919 N. Atl 62N 37 W 2088 83 9-540 15.4 65 P (4) ODP 980 N.Atl 55N 14W 2168 116 20-1300 8.9 112 B (5) PS 2644 N.Atl 67N 21W 777 9.2 12-76 12.1 83 P (6) MD95-2024 N.Atl 50N 45W 3448 27.3 5-110 24.8 40 P (7) MD99-2227 N.Atl 58N 48W 3460 43 1-430 10.0 100 P (8) MD95-2039 N.Atl 40N 10W 3381 35.7 30-317 11.3 89 B (9) ODP 1089 S.Atl 40S 9E 4624 90 20-580 15.5 65 B (10) MD97-2140 W.Pac 2N 141E 2547 12.5 560-1300 1.7 600 B (11) MD97-2143 W.Pac 15N 124E 2989 23.3 7-1500 1.6 1250 B (12) Somali Ind 0S 46E 4020 6.8 12-140 4.9 500 Bulk (13) The Iso column refers to the planktic (P), benthi c (B) or bulk source of oxygen isotope data. References: (1) Channell et al., 2008; Hodell et al., 2008 (2) Channell et al., 1987, 2002; Channell, 1999; Channell and Kleiven, 2000 (3) Channell, 1999; Channell et al., 2002, 2004 (4) Channell, 2006 (5 ) Channell and Raymo, 2003 (6) Laj et al., 2000; Voelker et al., 1998 (7) Stoner et al., 2000 (8) Evans et al., 2007 (9) T houveny et al., 2004; Thomson et al., 1999 (10) Stoner et al., 2003 (11) Carcaillet et al., 2003, 2004; Garidel-Thoron et al., 2005 (12) Horng et al., 2002, 2003 (13) Meynadier et al., 1992.

PAGE 89

89 Table 4-2. Correlation coefficient and percentage of signifi cant area (at 5% level, on WTC maps) between IODP U1308 records (both paleointensity and oxygen isotope record s) and other collected records before and after matching Records Correlation coefficient Percentage of significant area RPIs before matching RPIs after matching ISOs before matching ISOs after matching RPIs before matching RPIs after matching ISOs before matching ISOs after matching ODP 983 0.2816 0.7083 0. 6279 0.8017 0.432 0.7 807 0.6625 0.7092 ODP 984 0.2728 0.6742 0. 5597 0.8091 0.4739 0.7 193 0.5824 0.6515 ODP 919 0.2644 0.6091 0. 4275 0.6934 0.3921 0.6 499 0.5048 0.5906 ODP 980 0.2255 0.5661 0. 7068 0.8537 0.5078 0.7 086 0.6503 0.7051 ODP 1089 0.4347 0.7144 0. 823 0.8817 0.5536 0.7 429 0.7079 0.6938 MD97-2143 0.3272 0.6891 0. 6908 0.8277 0.463 0.7 204 0.5775 0.6747 MD95-2024 0.3169 0.5639 0. 7061 0.8129 0.4949 0.5 748 0.5679 0.5461 MD95-2039 0.3512 0.6141 0. 8502 0.8726 0.368 0.5 634 0.6872 0.7228 MD99-2227 0.2616 0.6565 0. 7175 0.8501 0.4301 0.7 311 0.5612 0.5616 PS 2644 0.6432 0.7592 0.3732 0.4724 0.5857 0.7869 0.3805 0.4582 Somali Basin 0.4893 0.6058 0.7423 0.7838 0.5066 0. 5647 0.2052 0.4707 MD97-2140 0.2553 0.6676 0. 6431 0.7558 0.5588 0.7 321 0.6599 0.6832

PAGE 90

90 Figure 4-1. Location of the 13 coupled isot ope and relative paleointensity records used in this analysis (see Table 4-1).

PAGE 91

91 Figure 4-2. Oxygen isotope and relative pa leointensity (RPI) dat a from the 13 sites (color coded in Table 4-1) used to c onstruct the RPI and isotope stacks, after optimal simultaneous matching of the RPI and isotopes to the Site U1308 records (in red).

PAGE 92

92 Figure 4-3. Squared wavelet coherence (WTC ) between the relative paleointensity (RPI) and oxygen isotope records from ODP Site 983 and IODP Site U1308 before and after simultaneous matching usin g the Match protocol. Values of squared wavelet coherence are indicated us ing different colors (with blue to red indicating increasing values). The 5% significance level against red noise is shown as thick contours. The cones of influence (COI) where edge effects make the analyses unreliable are marked by areas of crossed lines. The relative phase relationship is shown as arrows (with in-phase pointing right, anti-phase pointing left, and first signal le ading second by 90 pointing down). Orbital periods of 404 kyr, 100 kyr, 41 kyr, and 23 kyr are marked by white dashed lines.

PAGE 93

93 Figure 4-4. Oxygen isotope and relative paleoi ntensity (RPI) stacks (both in red) based on data in Figure 4-2. Half -width of the error envelope in both cases is 2 (2standard error) computed using the bootstrap method for 1 million samplings. The oxygen isotope stack (red) is compared wit h the LR04 benthic isotope stack (blue) (Lisiecki and Raym o, 2005), and the RPI stack (red) is compared with Sint-2000 stack (blue) (Valet et al., 2005). In this plot, the Sint2000 data have been subtract ed from their mean and divided by their standard deviation. Paleointensity minima in the stack correspond to established ages of magnetic excursions and chron/subchr on boundaries: LALaschamp, BL: Blake, IB: Iceland Basin, PR: Pringle Falls, BLT: Big Lost, LP: La Palma, ST17: Stage 17, B/M: Brunhes-Matuyama boundary, KA: Kamikatsura, SR: Santa Rosa, JA: Ja ramillo Subchron, PU: Punaruu, CB: Cobb Mt. Subchron, BJ: Bjorn, GA: Gardar.

PAGE 94

94 Figure 4-5. The new paleointensity stack (PISO-1500, red) compared with the EPAPIS Pacific stack (black) (Yamazaki an d Oda, 2005), with the inversion of magnetic anomaly data from the East Pacific Rise (green) (Gee et al., 2000) and the RPI record from ODP Leg 138 (b lue) (Valet and Meynadier, 1993). In this plot, these records have been subtra cted from their means and divided by their standard deviations.

PAGE 95

95 Figure 4-6. Local wavelet power spectrum (LWPS) of: A) the Sint-2000 relative paleointensity (RPI) stack (Valet et al., 2005), B) the LR04 benthic oxygen isotope stack of Lisiecki and Raymo ( 2005), C) the new PISO-1500 RPI stack and D) the PISO-1500 ox ygen isotope stack, and E) squared wavelet coherence (WTC) between the PISO-1500 RPI and oxygen isotope stacks. In E), the relative phase relationship is show n as arrows (with in-phase pointing right, anti-phase pointing left, and first signal leading second by 90 pointing down. Orbital periods of 404 kyr, 100 kyr, 41 kyr, and 23 kyr are marked by white dashed lines.

PAGE 96

96 Figure 4-7. Virtual axial dipole moment (VADM) calibration of the PISO-1500 relative paleointensity stack (red). See text for ca libration procedure. Half-width of the error envelope is 2 (2standard error) computed using the bootstrap method for 1 million samplings. The VADM values for the Sint-2000 stack (Valet et al., 2005) are in blue. Red dots indicate paleointensity minima that have ages close to those attributed to geomagnetic excursions and reversals (see Figure 4-4 for excursion/reversal labels). T he dashed line represents the threshold value of 2.522 Am2 that appears to trigger ex cursions and reversals.

PAGE 97

97 Figure 4-8. Circular plot of phase of orbi tal eccentricity, obliquity and precession (Laskar et al., 2004) corresponding to 17 relative paleointensity (RPI) lows labeled in Figure 4-7. The p -values associated with the Rayleigh test are indicated for each distribution (see Xuan and Channell, 2008a, for further details).

PAGE 98

98 Figure 4-9. Comparison of the PISO-1500 relative paleointens ity stack (red) with accompanying oxygen isotope stack (green) as determined in this work, and with the solution for orbital obliquity (blue) (from Laskar et al., 2004). Red dots indicate dominant RPI minima labeled in Figure 4-7, with dashed lines indicating the timing of those minima. Shaded areas indicate time intervals where RPI and isotope stacks appear to correlate with each other both positively (~ 660-810 ka) and negatively (~1120 ka).

PAGE 99

99 CHAPTER 5 UPMAG: MATLAB SOFTWARE FOR VIEWING AND PROCESSING U-CHANNEL OR OTHER PASS-THROUGH PALEOMAGNETIC DATA Introduction The need to acquire remanent magnetizatio n data at high resolution led to the development of pass-through cryogenic magnet ometers for long core measurements (e.g., Goree and Fuller, 1976). Aboard the R/V Joides Resolution a system for measuring whole/half core sections has been operating for over 25 years, enabling magnetostratigraphic data to be generated from sediment core s. The u-channel sample (which is typically enclosed in a tr ansparent plastic tube with square 2x2 cm2 crosssection, and up to 150-cm length), was first ad vocated by Tauxe et al. (1983) to improve the spatial resolution associ ated with traditional discrete sampling. U-channel samples have been increasingly employ ed in paleomagnetic studies on deep-sea and lake sediments since high-resolution pick-up coils ( with narrow width of response functions) and small diameter pass-thr ough magnetometers were devel oped by 2G Enterprises and first installed at the paleomagnetic labora tory at Gif-sur-Yvette in France (Weeks et al., 1993). U-channel sample meas urement provides a resolution close to that obtained with back-to-back 8 cm3 discrete cubic samples, after deconvolution, with the advantage of greater measurement speed and efficiency (Nagy and Valet, 1993; Guyodo et al., 2002; Roberts, 2006). For u-channel samples, the natural remanent magnetization (NRM) before and after stepwise demagnetization is routinely measured to recover directional variations of the geomagnetic field and to establish high re solution magnetostratigraphies. The NRM normalized by laboratory-induced magnetizat ions such as the anhysteretic remanent magnetization (ARM), or isot hermal remanent magnetization (IRM), is often used to

PAGE 100

100 estimate relative paleointens ity (RPI) variation, although grain-size variation, shape and flocculation effects may complicate interpre tations (e.g., Levi and Banerjee, 1976; King et al., 1983; Tauxe, 1993). Attempts have been made to reduce environmental (lithologic) influences using different normalization methods (e.g., Brachfeld and Banerjee, 2000; Mazaud, 2006). It is r outine to measure laboratory-induced magnetizations (ARM and IRM) and to compar e the normalized intensities to find an optimal normalizer and RPI estimate. Additionally, these rock magnetic measurements provide records of the paleo-environmen t, through concentration and grain-size sensitive magnetic parameters. At the University of Florida (UF) we routinely m easure NRM during demagnetization, followed by susceptibi lity, ARM during demagnetization and acquisition, and IRM during demagnetizat ion for each u-channel. The ARM is completely removed by demagnetization prior to the ARM acquisition experiment. Each type of data is checked prior to proceedi ng to the next step. Over 5000 data points (depending on number of demagnet ization steps) can be accumulated for a single uchannel in ~10 h, while hundreds of u-c hannel samples may need to be processed for one deep-sea drilling site. Checking and proc essing these data, displaying orthogonal projections (Zijderveld, 1967) for each measur ement position, calculating or optimizing the component magnetization directions by principal component analysis (PCA) (Kirschvink, 1980), and estimating RPI are ti me-consuming in the absence of software designed for these purposes. Mazaud (2005) developed Microsoft Ex cel software that incorporates micro-commands to treat NRM data from long co re sediment samples. The software requires introduction of data column s from the respective measurement data

PAGE 101

101 files to certain regions of the Excel f ile, and is capable of calculating component directions and median destructive fields, as well as drawing orthogonal projections for each measurement position. The new UPmag software described here cont ains three graphical user interfaces that are designed to view and process rout inely measured u-chann el paleomagnetic data such as NRM, ARM and IRM data. The software allows the user to open and check through the measurement data, and corrects for flux -jumps in the data if necessary. In addition to interactively calc ulating (optimizing) and displaying PCA values, orthogonal projections, and stepwise intensity data for any measurement position, UPmag also enables data visualiz ation in VGP maps and equal area projection plots. Using UPmag, RPI estimates can be evaluated and compared using two methods: the slope of the best-fit line between NRM and normaliz er demagnetization data (e.g., Channell et al, 2002), and the average of NRM/normalizer ratios (e.g., Channell et al., 1997). Description of UPmag UPmag is developed in MATLABTM (version R2008b) on a PC with a Windows XP operating system. UPmag consists of three MATLABTM graphical user interfaces: UVIEW, UDIR, and UINT. Each interface fac ilitates certain aspects of processing uchannel paleomagnetic data. UVIEW displays t he measurement data, corrects for fluxjumps, and prepares data for further tr eatment. UDIR focuses on calculation and visualization of directional data such as PCA calculations and VG P data visualization. UINT calculates relative paleointensity esti mates. These programs can be activated by typing UVIEW, UDIR, and UINT (all in capital letters) followed by a return key in the MATLABTM command window. The UPmag progr am folder should be under the

PAGE 102

102 MATLABTM current directory or should be added to the MATLABTM search path when running UPmag. The UPmag progr ams, data format files, and test data can be found at http://earthref.org/cgi -bin/er.cgi?s=erda.cgi?n=985. A ty pical flow path for processing uchannel paleomagnetic data using UPmag is summarized in Figure 5-1. UVIEW UVIEW reads NRM, ARM or IRM measur ement files created by the u-channel magnetometer measurement syste m. Contents of a typical measurement file generated at UF, by the LABVIEWTM software accompanying the 2G Enterprises u-channel magnetometer, are listed in Tabl e 5-1. After selecting Load Measurement File from the File menu (Figure 5-2), UVIE W asks for the type of the dat a (e.g., NRM or ARM) and whether the data are in default UF format (T able 5-1). The data type is used later to label the exported file s. A measurement file format table (i.e., Measurement_File_Format.txt, which is also available for download) must be modified and loaded to UVIEW first, if the measurement file is not in the default format. After importing the measurement file, information about u-channel samples and demagnetization procedures are automatically listed in the File Information panel (Figure 5-2A). Initially, the remanence in tensity data for the re spective orthogonal sample axes (Xint, Yint, and Zint. See Table 51) from the first de magnetization step of the first u-channel sample are displayed (Fi gure 5-2A). The list of parameters available for plotting is grouped in the Plot Parameters panel. To view a para meter, choose the radio button associated with t he parameter. As an example, Figure 5-2B contains plots of intensity (Int), declination (Decl), and inclination (Incl) data for the same demagnetization step for the same u-channe l shown in Figure 5-2A. The Now Checking panel hosts components that allow the user to check through data and plots

PAGE 103

103 (for a selected parameter) for different demagnetization steps and for different uchannel samples. Multiple demagnetization st eps for a u-channel sample can be plotted together (Figure 5-2C) by togg ling the View Multiple Steps of This Sample button. The range of steps is adjustable using the two pop-up menus in the Now Checking panel. Occasionally, a flux-jump may occur dur ing the pass-through measurement, in which a sudden (abrupt) change in magnetizat ion appears at a certain measurement position for a particular axis. Flux-jumps are generally rare, r andom, non-predictable, and yet they greatly distort the data. Although re-measuring the sample would usually be ideal, re-measuring may oft en not be feasible if the fl ux-jump was not immediately recognized. In UVIEW, if flux-jumps are observed at any stage when scrolling through the measurement data, a correction can be in itiated by clicking on Correct Current Data in the Flux-Jump panel. Flux-jum ps can be located either manually or automatically. A threshold value can be set for the automatic detection, and different interpolation methods are ava ilable for correcting flux-jumps After correction, a diagram comparing the measurement dat a before and after flux-jump correction will be displayed (Figure 5-2D), and all data related to the flux-jumps will be updated if the correction is accepted by the user. After checking and co rrecting for flux-jumps for all u-channel samples, a new measurement file can be generated using the Save Corrected Data button in the Flux Jump panel. All plots displayed in UVIEW can be m odified and exported in various formats such as jpg, bmp, pdf, or eps. For example, the user ca n zoom and pan the plots, add legends, titles, grids, or data value markers. The Hold on/off tool facilitates comparison of different data plots when toggl ed. The Print tool allows the user to save the whole

PAGE 104

104 figure, including plots and information listed in various panels, to a pdf file. UVIEW also exports two types of data file: *.dir and *.int. The two files truncate data from the leader and trailer (see Table 5-1), and re-group the measurement data by position in the uchannel. The *.dir file features declination, in clination, and intensit y data (typically from NRM data), reorganized by measurement position, for further direction-related investigation in UDIR. The *.int file re-for mats intensity data by measurement position for RPI estimation in UINT. UDIR UDIR opens *.dir files generated by UVIEW or by other sources. The default UFformatted *.dir file is represented in Table 5-2. If the *.dir file is not in this format, the Dir_File_Format.txt file (available for download) must be modified accordingly and loaded to UDIR prior to loading the *.dir file. After successful loadi ng of a *.dir file, component directions for the entire u-channel are calculated using PCA (Kirschvink, 1980), with a default sample orientation opt ion and demagnetizati on range. Component declination and inclination, and maximum angul ar deviation (MAD) values, determined from the PCA calculation, and intensity data (for a single demagnetization step) for the entire u-channel are displayed at the top of the UDIR interfac e, with an orthogonal projection and a stepwise intensit y plot for the first measur ement position at the bottom of the display windo w (Figure 5-3A). The sample orientation and demagnetizat ion range (including the option of anchoring the best fit through the origin of the plot) for PCA calculations can be changed from the Coordinates and the PCA Values panels (Figure 5-3A), respectively. Related data and plots will then be updated accordingly. Similarly, the range of demagnetization steps for plotting the demagnetization data (bottom left), can

PAGE 105

105 be controlled for each measurement posit ion using the Plot Range panel. The demagnetization data for any position can be plotted in three styles: a north/up orthogonal projection (Figure 5-3A, bottom left) or west/up orthogonal projection (Figure 5-3D, bottom left), or in a 3-D vector plot (F igure 5-3B, bottom left). These styles can be switched using the Plot Control panel. The Choose Top Plot panel allows the user to switch the uppermost plots with the following options: intensity (for a single demagnetization step) and PCA dat a for all measurement pos itions (Figure 5-3A), demagnetization data for the selected measurement posi tion and PCA data for all positions on an equal area stereographic pr ojection (Figure 5-3B), and virtual geomagnetic pole (VGP) positions (which require s input of the sampling site location, declination correction if appropr iate, and map projection, Figure 5-3C, D). The following options are available for scrolling th rough the data and plots from different measurement positions: 1) going to a specific position in the Go to Position panel by inputting the position and clicking on Go!; 2) manually controlling the position using the Previous and Next buttons; 3) automatically displaying the data for each position at a user-defined speed by toggling the Move/Stop button in the Auto Display panel. Similar to UVIEW, all plots displayed in UDIR can be modified and exported in various formats. Component direction data and VGP data calculated using a uniform demagnetization range for the entire u-channel can be exported by clicking on the Uniform Range PCAs and Export VGP Data butt ons in the Export Data/Plots panel. It is sometimes convenient to optimize dema gnetization ranges for PC A calculations at a particular measurement posit ion, and then add the optimized PC A values to a txt file using the Current Opt. PCA button.

PAGE 106

106 UINT UINT sequentially reads *.int files generated from NRM and laboratory-induced magnetization measurement data (e.g. ARM). The intensity data are displayed immediately to facilitate determination of the desired data range and demagnetization range for RPI calculation. Warning messages prevent selection of invalid data or demagnetization ranges. After clicking on the Apply button in the Data Range Selection panel, UINT updates the NRM and normalizer intensity plots with data in the selected range, calculates and displays RPI estimates using two methods: the slope of the best-fit line and the average ratio between NRM and the respective normalizer. The correlation coefficient (R) values for the best-fit line, and standard deviations of the ratios, are also calculated (Figure 5-4). These values provid e a measure of the quality/uncertainties of/in the RPI estimation. The respective in tensity data within the selected range of demagnetization steps, the best-fit line, and the ratios between the NRM and the respective normalizer are displayed on the left-hand side of the interface. The most appropriate normalizer activates the same gr ains that carry the NRM (e.g. Levi and Banerjee, 1976). The upper left-hand plot is a display of the coercivity spectra of the NRM and the normalizer, which aids in the choice of an appropriate normalizer for RPI estimation. Similar to UDIR, users can check through the data row-by-row of the two files manually or automatically. All plots presented in UINT can be modified and exported in various formats. RPI estimates calculated in UINT can be saved for other operations. It should be noted that, although designed for RPI estimation, UINT can also be used to

PAGE 107

107 estimate grain-size proxies such as ARM /IRM, by simply loading ARM and IRM *.int files sequentially as the NRM *.int File and Normalizer *.int File, respectively. Conclusions UPmag aims to improve the efficiency of viewing and proc essing u-channel or pass-through paleomagnetic data such as NRM, ARM and IRM data. The software consists of three graphical user interf aces called UVIEW, UDIR, and UINT. Each interface focuses on a specific aspect of treating u-channel paleomagnetic data. UVIEW allows users to open and scroll through output data files from the u-channel magnetometer measurement system as well as to correct for detected flux-jumps in the data, and to export files for further treat ment. UDIR calculates, presents, and saves component direction data (i.e., declination, inclination, and MA D values) using the standard PCA method for either a uniform demagnetization range for the entire uchannel or for an optimized range of demagnetiz ation steps, anchored to the origin of the orthogonal projection or no t. UDIR also calculates, disp lays, and saves VGP data, displays data on equal area projections, and conveniently scrolls through data for different measurement positions. UINT es timates, compares, and exports RPI data using the slopes of the best-fit line or the averages of ratios between the NRM and a normalizer within a selectable range of demagnetization steps, with linear correlation coefficients (of slopes) and standard deviations (of ratios) calculated to monitor the quality of the respecti ve RPI estimates.

PAGE 108

108 Table 5-1. The default UF measurement data file format Columns in file Column number Format string Explanation Sample IDa 1 %s Name of U-channel sample Positiona 2 %f Measurement position in U-channel Deptha 3 %f Depth of the position in the sedimentary sequence AF Xa, b 4 %f Demagnetization peak field on X measurement axis AF Ya, b 5 %f Demagnetization peak field on Y measurement axis AF Za, b 6 %f Demagnetization peak field on Z measurement axis Declination: Unrotateda 7 %f Declination from corrected X and Y intensity Inclination: Unrotateda 8 %f Inclination from correct ed X, Y, and Z intensity Intensitya 9 %f Intensity from corrected X, Y, and Z intensity X intensitya 10 %f X magnetic moment X meana 11 %f X volume magnetization X corra 12 %f X mean after drift and tray correction Y intensitya 13 %f Y magnetic moment Y meana 14 %f Y volume magnetization Y corra 15 %f Y mean after drift and tray correction Z intensitya 16 %f Z magnetic moment Z meana 17 %f Z volume magnetization Z corra 18 %f Z mean after drift and tray correction ARM Gauss 19 %s Bias field used for generating ARM ARM axis 20 %s Direction of the applied bias field Orientation 21 %s Archive or working half of core Leader lengtha 22 %f Positions measured before 1s t position of u-channel Trailer lengtha 23 %f Positions measured after last position of u-channel Drift corrected 24 %s Whether drift correction applied Tray corrected 25 %s Whether tray correction applied Sample Timestamp 26 %s Time stamp for u-channel measurement Tray Timestamp 27 %s Time stamp for last tray measurement Run number 28 %s Run (ID) number assigned to measurement a Items that have to be included in a meas urement file. '%s' or '%f' indicates co lumns composed of strings or numbers. b In the default UF fo rmat, after data reduction, X, Y, and Z measurement axes correspo nd to north, east, and vertical components, respectively.

PAGE 109

109 Table 5-2. The default UF *.dir file format Items in *.dir file Column number Format string Explanation Positiona 1 %f Position in the u-channel (usually in cm) Treatmenta 2 %f AF or thermal demagnetization level IC 3 %s Instrumental code, e.g. cryogenic CD 4 %f Circular standard deviation (for discrete sample) Intensitya 5 %f Measured intensity DECL (Sample) a 6 %f Declination data without any orientation INCL (Sample) a 7 %f Inclination data without any orientation DECL (Geographic) a 8 %f Declination data of the sample in situ INCL (Geographic) a 9 %f Inclination data of the sample in situ DECL (Bedding) a 10 %f Bedding corrected declination data INCL (Bedding) a 11 %f Bedding corrected inclination data SUSC 12 %f Susceptibility data V/M 13 %f Volume or mass information a Items that have to be included in a *.di r file. '%s' or '%f' indi cates columns composed of strings or numbers.

PAGE 110

110 Figure 5-1. A flow chart summarizing the logic and major steps in the UPmag software.

PAGE 111

111 A B Figure 5-2. The UVIEW graphical user interf ace illustrating the major features of the software: A) Xint, Yint, and Zint data for a selected u-channel sample at a selected demagnetization step; B) intensity (Int), declination (Decl), and inclination (Incl) data for the same u-channel for a uniform demagnetization interval; C) intensity (I nt) data from multiple de magnetization steps for a selected u-channel sample; D) Ymean data for a u-channel sample for a certain demagnetization step before and after correct ion for a flux-jump.

PAGE 112

112 C D Figure 5-2. Continued

PAGE 113

113 A Figure 5-3. The UDIR graphical user interface illustrating the major features of the software: A) orthogonal projection with No rth/Up convention (bottom left) and stepwise intensity plot (bottom right) fo r the currently vis ualized measurement position, with intensity (at a si ngle demagnetization step) and PCA data (declination, inclination, and MAD) fo r the entire u-channel shown in the upper two panels; B) demagnetization data in 3-D (bottom left, in green: trajectory of the total vector; in red: projection of the vector end-points on the vertical plane, and in blue: projection of the vect or end-points on the horizontal plane) and equal area projection (top left) with stepwise intens ity data (bottom right) for the currently visualized measurement position, and PCA data for the entire u-channel on an equal area pr ojection (top right, in red: data with positive inclination; in cyan: data with negative inclination); C) the same data as in b) with VGP data plotted on a Hammer pr ojection (top), and input information window (bottom left) for VGP calculation; D) the same data as in b) with VGP data plotted on a 3-D g lobe projection, and dem agnetization data shown on a West/Up convention or thogonal projection.

PAGE 114

114 B C Figure 5-3. Continued

PAGE 115

115 D Figure 5-3. Continued

PAGE 116

116 Figure 5-4. The UINT graphical user inte rface illustrating NRM and normalizer (ARM) intensity data from one u-channel, RPI re sults estimated using slopes of the best-fit line and average ratio (see te xt), and detailed information for the currently visualized measurement position in the u-c hannel data files.

PAGE 117

117 CHAPTER 6 QUATERNARY PALEOMAGNETIC RECORD FROM DIATOM RICH SEDIMENTS AT IODP SITE U1304 (SOUTHERN GARDAR DRIFT, NORTH ATLANTIC) Introduction Integrated Ocean Drilling Program (IODP) Site U1304 (Fi gure 6-1) was drilled by the R/V JOIDES Resolution with the Advanced Piston Corer (APC) in October 2004, during the IODP Expedition 303. The site (53.40 N, 33.78 W; water depth, 3024 m) lies in a semi-enclosed basin at the southern edge of the Ga rdar Drift, to the north of the Charlie Gibbs Fracture Zone. Shipboard data (Exped ition 303 Scientists, 2006) and initial post-cruise results indicate that the 263.8-meter composite depth (mcd) sediments from the four dr illed holes (A-D) at Site U1304 recorded a continuous sequence including the Brunhes Chron and part of the Matuya ma Chron that includes the Jaramillo Subchron (see Mazaud et al., 2009), the Cobb Mountain Subchron, and the top of the Olduvai Subchron. Site U1304 contains some of the highest-resolution (i.e., greatest sedimentation rate) deep-sea records yet recovered from the North Atlantic. Mean sedimentation ra tes of ~17.8 cm/kyr are esti mated for the last 0.78 Ma, and 12.2 cm/kyr for the interval from 0.78 to 1.77 Ma, with an overall mean sedimentation rate of 14.9 cm/kyr. The sediments at Site U1304 consist of inter-bedded diatom and nannofossil oozes with clay and silty clay, in highly vari able grayish colors. The thinly laminated diatomaceous ooze (LDO) deposition comprises mainly the planktonic, araphid, needlelike species Thalassiothrix longissima and appears to be episodic and discontinuous, but strikingly present throughout the whole re covered intervals that date back to ~1.8 Ma (Shimada et al., 2008). Documentation of LDO in the North Atlantic is rare. The

PAGE 118

118 best known occurrence is from conventional piston core (EW9303-17) collected ~500 km NNW of Site U1304 in which LDO was recorded during the last interglacial (Bodn and Backman, 1996). The occurrence of LDO in th is core was attributed to its proximity to the sub-arctic convergence zone (Bodn and Backman, 1996). Magnetic properties of sediments from IOD P Site U1304, comp rising diatom-rich and clay-rich intervals, were studied for the 0 224 mcd interval, corresponding to the last ~1.5 Ma. High resolution paleomagnetic directional and intensity records from the site are presented, with an age model acquired by correlating t he relative paleointensity (RPI) record of IODP Site U1304 to the PISO-1500 RPI st ack record (Channell et al., 2009). In view of the unusual characterist ic of LDO occurrences in the sediment sequence, this chapter will also address whether diatom-richness distorts the preservation of paleomagnetic in formation, by statistically comparing magnetic records between diatom-rich intervals and non-diatom intervals, and by comparing the RPI record at the site to other RPI records. Power spectral analysis and wavelet analysis methods will be employed to investigate whether orbital periods exists in the Site U1304 RPI record. Sampling and Methods U-channel samples were collected from the archive half of each core section within the IODP Site U1304 composite splice (Expedition 303 Scient ists, 2006) except for occasional intervals (i.e., 177.74-182.18 mcd) where samples from the splice appear to be disturbed by drilling. The u-channel samples are enclosed in plastic containers with a 2 cm cross-section and the same lengt h as the core section (usually 150 cm), with a clip-on plastic lid that allows the samples to be sealed to retard dehydration and

PAGE 119

119 other chemical/physical alteration. U-channel sampling almost completely covered the 0-224.36 mcd interval for IODP Site U1304, with a gap in the 197.89-201.41 mcd interval where thick, continuous, diat om mats made sampling very difficult. Natural remanent magnetization (NRM) of the u-channel samples was measured at 1-cm spacing before demagnetization, and after 12 alternating field (AF) demagnetization steps in the 20 100 mT peak field range, using a 2G Enterprises cryogenic magnetometer designed to measure u-channel samples. Note that shipboard AF treatment of archive halves of core sections (from which the u-channel samples were collected) was carried out at peak fi elds not exceeding 20 mT (Expedition 303 Scientists, 2006). After completion of t he NRM measurements of each u-channel sample, susceptibility measurements of the samples were carried out using a susceptibility track designed for measuring u-channel samples (Thomas et al., 2002). ARM was acquired in a peak AF of 100 mT and a 50 T DC bias field, and measured prior to demagnetization and after demagnetiz ation at 9 steps in the 20-60 mT peak field range. ARM was then reacquired stepw ise in peak alternating fields of 20 60 mT, in 5 mT steps, and in a constant 50 T DC bias field. Isothermal remanent magnetization (IRM) was acqui red in a DC field of 300 mT, and was measured before demagnetization, and after demagnet ization in the same steps applied to the ARM. Hysteresis parameters of the sediments were m easured using a Princeton Measurements Corp. vibrating sample magnetometer (VS M) on core section-averaged samples, rests from cleaning the uchannels during the u-channel sampling. LDO distributions in the sequence were re corded in the sampling notes during the u-channel sampling, by visual observation of u-channel samples themselves, and from

PAGE 120

120 shipboard images of core sections. Diatom oo ze has a characteristic oatmeal texture and sharp contacts with interbedded silts and clays that result in characteristic laminations. The number was assigned to eac h 1-cm interval that comprise mainly diatom ooze, and the number .5 was assi gned to 1-cm intervals that consist of disseminated diatom ooze. For 1-cm interv als that contain silts and clays with no observable diatom ooze, the number was assigned. Rock Magnetic Properties of the Sediments The ratio of saturation magnetization to saturation remanence (Mr/Ms), and the ratio of coercivity of remanence to coercivity (Hcr/Hc) are calculated using the hysteresis measurements. These ratios are grouped in the pseudo-single (PSD) domain field in the Day et al. (1977) hysteresis ratio plot (Figure 6-2), indicating uniformity in magnetic grain sizes. The distribution of h ysteresis ratios is elongated along a typical magnetite grain size mixing line. The anhysteretic susceptibility ( ARM) of the sediments is computed by normalizing the ARM intensity (after AF demagnetization at peak field of 35 mT) by the DC bias field (i.e., 50 T) used to apply the ARM. A plot of ARM versus susceptibility ( ) can be used to estimate the grai n size distribution of magnetite. Because ARM is mainly carried by smalle r (single domain to pseudo-single domain) magnetic particles, whereas susceptibility is linked more to larger (multidomain) particles, higher values of ARM/ (larger slope on the ARM versus plot) would indicate finer grains of magnetite. Following the calibration of ARM/ by King et al., (1983), the magnetic (magnetite) grai n size at IODP Site U1304 is largely restricted to the 1-5 m grain-size range (Figure 6-3), consistent with the observation in hysteresis ratio plot (Figure 6-2). The slopes from diatom-rich in tervals (red and blue circles in Figure 6-3) yield finer mean grain sizes (red and blue lines in Figure 6-3) than that (green line in

PAGE 121

121 Figure 6-3) of samples from non-diatom inte rvals (green circles), although the range of grain sizes is large, particularly for the diat om rich intervals (red circles). The distance from the origin of the plot corresponds to magnetic concentration which is greatest for the samples from non-diatom interv als (green circles in Figure 6-3). Diatom distributions plotted against the mcd show several thick diatom-rich intervals in 45-65 mcd, 85-120 mcd, 140-165 mcd, and 190-224 mcd (Figure 6-4), consistent with the observation of Shimada et al. (2008). The magnetic concentration parameters such as susceptibility, ARM, ARM acquisition (ARMAQ), and IRM from IODP Site U1304 mimic one another (Figure 6-4) and there is an obvious correlation between LDO distribution and the magnetic conc entration parameter values. In diatomrich intervals, magnetic concentration param eters are about two orders of magnitudes lower than for non-LDO intervals. LDO inte rvals are probably accompanied by elevated deposition rates that dilute the magnetic conc entration in those intervals. The correlation between diatom distribution and the magnetic concentrati on parameters means that these magnetic concentration parameters may be used as diat om proxies at IODP Site U1304. Natural Remanent Magnetization and RPI Proxies Component NRM directions were comput ed at 1-cm spacing using the standard principal component analysis method (Kirschvink, 1980). The demagnetization interval used to compute the characterist ic magnetization component was 2080 mT, and the origin of orthogonal projections was not us ed for the component direction calculations. The maximum angular deviation (MAD) values associated with the ca lculation provide a means of monitoring the uncertainty associated with component directions. The MAD

PAGE 122

122 values are largely below 1 for samples fr om non-diatom intervals, and are generally below 5 for samples from diatom-rich inte rvals (Figure 6-5), indicating very well and well defined component directions for samples from non-diatom and diatom-rich intervals, respectively. Component declinations were corrected using tensor core orientation data where available (green circ les in Figure 6-5), and using rotation of entire cores (red circles in Figure 6-5) so t hat the mean declination is oriented North or South for positive and negative inclination intervals, res pectively. There are apparent differences between the declination values corrected using the two methods (Figure 65), implying that tensor tool s may not always be reliable for co re orientations. Note that the first three cores (down to ~30 mcd) ar e routinely not oriented due to the increased risk associated with the time delay for orienting the cores. The green symbols for the 030 mcd interval (Figure 6-5) therefor e represent unorient ed declinations. The component inclination and the optimally correct ed component declination unambiguously reveal the Brunhes/Matuyama boundary, the Jaramillo subchron, and the Cobb Mountain subchron at IODP Site U1304 (Figure 6-5). NRM demagnetization data of non-diatom interval samples plotted on orthogonal projections (Figure 6-6) indicate that the maximum peak demagnetization field at 100 mT reduced the NRM to < 5% of the NRM intensity prio r to demagnetization, and the demagnetization data (after demagnetiz ation) fall on a perfect line that indicates a very well defined one-component NRM. There a ppears to be more (occasionally >10 %) NRM left after the 100 mT peak field demagnet ization for samples from diatom-rich intervals (i.e., 103.3 mcd in Figure 6-6), indica ting a higher coercivity remanence. This is consistent with the ARM versus plot, which imply finer gr ains in the diatom-rich

PAGE 123

123 intervals, and finer grains of magnetite us ually have higher coercivity, although a high coercivity remanence carrier such as hemat ite cannot be ruled out. The NRM for the 2080 mT demagnetization range from the diatom-rich intervals still indicate a well defined one-component NRM. An additional test was performed to inve stigate whether there is any significant difference between componen t inclinations (in the Brunhes Chron) recorded by samples from diatom-rich in tervals and by samples from non-diatom intervals. A consistent difference between t he two groups of samples may debilitate the reliability of NRM acquired fr om the diatom-rich interval s. The histogram of the component inclinations (in the Brunhes) calc ulated using samples from non-diatom and diatom-rich intervals were displayed in Fi gure 6-7. Although histogram of inclinations recorded by samples from diatom-rich interval s are slightly more scattered, inclination values of the two groups of samples we re both centered around 65, close to the inclination value (69) expect ed at this site for a geocent ric dipole field, and the two histograms show generally comparable dist ribution patterns, indicating diatom abundance in the sediments does not produce significant alteration of the NRM. Normalized intensity records of sedim entary natural remanent magnetization (NRM) are often used as the relative paleoin tensity proxies of the geomagnetic field. Normalization is usually carried out by using ARM, ARMAQ, IRM, or magnetic susceptibility to compensate for changes in magnetic concentration of remanence carrying grains. Two RPI proxies were ca lculated for the IODP Site U1304 records: slopes of the best-fit line of the NRM versus ARM demagnetization data for 20-60 mT peak field range, and NRM dem agnetization data for 20-60 mT peak field range versus ARMAQ acquired in 20-60 mT peak field r ange. The slope calculation is accompanied

PAGE 124

124 by linear correlation coefficients (R values) that indicate the quality of linear fit of each slope. RPI estimates were made at 1-cm in tervals throughout the sediment sequence. The two RPI proxies are very similar to each other, with associated linear correlation coefficients close to 1 (Figure 6-8). The in tervals where R values depart from 1 appear to correlate with diatom-rich intervals. This is consistent with the observation that MAD values associated with the component direction calculations for diatom-rich intervals are higher than that for no n-diatom interval samples (Figure 6-5), indicating slightly lower efficiency of NRM recording for samples from the diatom-rich intervals. The overall small departures (<0.01) from 1 for almost all samples indicate that even for samples from diatom-rich intervals, high quality RP I estimates were obtained. The R values associated with the NRM versus ARMAQ slopes (r ed circles in Figure 6-8) are generally closer to 1 than R values fo r the NRM versus ARM slopes (b lue circles in Figure 6-8), indicating NRM versus ARMAQ slopes repres ent a higher quality RPI estimates. The NRM versus ARMAQ slopes are t herefore used as the RPI reco rd for IODP Site U1304. Two magnetic excursion events were re corded at IODP Site U1304 during the Matuyama Chron in the 149.22-149.38 mcd and 220.83-221.01 mcd intervals (Figure 65 and 6-6). Orthogonal projections of the NRM demagnetization da ta in the two intervals clearly show positive inclination component s at high peak field demagnetization steps (Figure 6-6). The excursion in the 149.22-149. 38 mcd interval occurs between the top of Jaramillo (164.5 mcd) and the Brunhes/Mat uyama boundary (139.1 mcd, Figure 6-5). Only two magnetic excursions were well documented within this time interval: the Santa Rosa excursion and the Kamikatsura excu rsion (see Laj and Channell, 2007). This excursion apparently corresponds to a dramatic RPI minimum that is the younger of the

PAGE 125

125 two dominant RPI minima between the top of Jaramillo and the Brunhes/Matuyama boundary (Figure 6-8). This excursion is hence likely to be the Kamikatsura excursion. Although clearly manifest in the orthogonal projections (Figure 6-6), the Kamikatsura excursion is not particularly obvious in the declination/inclination plot of Figure 6-5, because the component directions for th is figure are calculated using a uniform demagnetization interval (20-80 mT) that does not adequately capture the component representing the excursion. Assuming a uniform sedimentation rate between the top of Jaramillo and the Brunhes/Matuyama boundar y, and utilizing the ages of 990 ka and 775 ka from Channell et al. (2009) for the top of Jaramillo and the Brunhes/Matuyama boundary, respectively, would yield an age of ~861 ka for the this excu rsion, close to the Kamikatsura excursion ages of 866 ka and 899 ka reported by Singer et al. (1999, 2004), respectively. The excursion in the 220.83-221.01 mcd interval occurs below the Cobb Mountain subchron (Figure 6-5), and is also accompanied by a dramatic paleointensity minimum. Assuming uniform sedi mentation rate between the top of Cobb Mountain (187.8 mcd) and the top of Olduv ai (262 mcd, base on shipboard data), and adopting the ages of 1185 ka (Channell et al., 2009) and 1770 ka for the two reversals, yield an age of 1446 ka for this excursio n, comparable to the ages of 1472-1478 ka reported for the Gardar excu rsion in ODP Sites 983 and 984 (Channell et al., 2002). Paleointensity-Based Age Model and Det ection of Orbital Periods in the RPI Record Preferably, the age model for IODP Site U1304 would be established using the combination of stable oxygen isotope record (in process by Prof. Dave Hodell at Cambridge) and the paleomagnetic directional and intensity records. While we await the oxygen isotope record, a tentative age model can be established by correlating the RPI

PAGE 126

126 record of IODP Site U1304 to the recent ly constructed PISO-1500 RPI stack record (Channell et al., 2009). The correlation was made automatically fo llowing the dynamic programming method by Lisiecki and Lisiecki (2002). The RPI records were normalized to have 0 mean and standard deviation of 1 prio r to the matching. The following ties points were used for the correlation. From the available oxygen isotope data for IODP Site U1304 (Hodell et al., 2009): 17.74 mcd is tied to 118 ka and 22.57 mcd is tied to 128 ka. From the component inclination record: the Brunhes/Matuyama boundary (139.1 mcd), the Kamikatsura excursion ( 149.3 mcd), top (164.5 mcd) and base (176 mcd) of Jaramillo subchron, the top (187. 8 mcd) of Cobb Mount ain subchron, and the Gardar excursion (220.9 mcd) were tied to the corresponding ages in Channell et al. (2009), i.e., 775 ka, 888 ka, 990 ka, 1071 ka, 1185 ka, and 1463 ka, respectively. The resulting age model yields a RPI record comparable to the PISO-1500 RPI stack record (Figure 6-9), even in diatom-rich intervals such as that between 550-600 ka, further indicating that diatom-richness does not debi litate the reliability of paleointensity information from the sediments. According to this paleointensity based age model, the sedimentation rate at IODP Site U1304 vari es between ~5-45 cm/ kyr, with a mean value close to 20 cm/kyr (Figure 6-10). High se dimentation rate appear to correlate with interglacial intervals, and di atom-rich intervals seem genera lly accompanied by elevated sedimentation rates (Figure 6-10). In recent years, orbital periods, wit h ~100 kyr and/or ~41 kyr periods, have been detected in RPI records from both the Atl antic (e.g., Channell et al., 1998) and the Pacific oceans (e.g., Yamazaki, 1999; Ya mazaki and Oda, 2002) using either power spectral methods or wavelet analyses, and have been considered evidence for orbital

PAGE 127

127 influence on the geodynamo. Teanby and Gubbins (2000) discussed the possibility of orbital periods in RPI being caused by alia sing of the geomagnetic signal by coarse sampling of the field. Wavelet analyses of records from ODP Site 983 indicated that orbital periods in the RPI re cord are probably due to the influence of lithologic variations (Guyodo et al., 2000). Furthermore, Xuan and Channell (2008b) sugges t that orbital periods may have been introduced into the NRM records (and not adequately normalized in RPI calculations) through orbi tal control on the bottom current velocity, which in turn controls magnetic grain size di stribution at the sites. Assuming a red noise background, spectral analysis was perfo rmed directly on the unevenly spaced IODP Site U1304 RPI record, us ing REDFIT developed by Schulz and Mudelsee (2002). The 95% and 99% confidence levels (CL) were estimated using both the theoretical firstorder autoregressive model (dashed green and red lines in Figure 6-11) and Monte Carlo (1000 runs) methods (solid green and r ed lines in Figure 6-11). Clearly, none of the orbital frequencies (vertical dashed lines in Figure 6-11) appears to be significant in the IODP Site U1304 RPI record. Wavelet analyses with significance tests were performed on IODP Site U1304 RPI record and the PISO-1500 RPI stack record to find out whether there are any signifi cant orbital periods during certain time intervals of the RPI records, and how coherent the two RPI re cords are in time-frequency space. The RPI record from Site U1304 was interpolated to have same time axis as the PISO-1500 RPI stack record. Local wavelet power spectr a of the IODP Site U1304 RPI record and the PISO-1500 RPI stack record show very similar features (Figure 6-12A and B), with no significant power on any orbital periods and no significant common power on orbital periods (Figure 6-12C). The squared wavelet coherence (WTC) between the two RPI

PAGE 128

128 records indicate that the two RPI records are significantly coherent on the ~10-500 kyr period range for almost the whole time span (F igure 6-12D). The lower limit (~10 kyr) of this period range may provide a measure of ac hievable resolution for correlating these RPI records. Conclusions Rock magnetic studies indicate that magnet ic concentrations from diatom-rich intervals are about two magnitudes lower t han for the non-diatom intervals, and magnetic grain sizes tend to be finer in diatom-rich intervals. U-channel NRM measurements at IODP Site U1304 yield a high resolution paleomagnetic record for almost the entire sediment sequence down to ~224 mcd (~1.5 Ma). The generally low (<5) MAD values associated with the component direction calculation, the similarity between distribution patterns of component in clinations (in the Brunhes) calculated for samples from diatom-rich and non-diatom inte rvals, the overall small departures of R values from 1 that accompany the RPI es timates, and the comparability between RPI records from diatom-rich intervals and the PISO-1500 RPI stack record from the same time interval, indicate that diatom richness does not debili tate the reliability/efficiency of acquiring paleomagnetic directional and intensity records from the sediments deposited at IODP Site U1304. Component direct ion records unambiguously reveal the Brunhes/Matuyama boundary, the Jaramillo subchron, and the Cobb Mountain subchron at IODP Site U1304, with two additional events that we interpret as representing the Kamikatsura excursion and the Gardar excu rsion. According to the paleointensity based age model, diatom-ri ch intervals appear to be generally accompanied by elevated sedimentation rates. Spectral analysis and wavelet analyses fail to reveal significant orbital periods in the RPI record. The IODP Site U1304 RPI

PAGE 129

129 record is significantly coherent with the PISO-1500 RPI stack in the ~10-500 kyr period range. The lower limit of this period range possibly indicating achievable resolution for correlating RPI records with these mean, but very variable, sedimentation rates.

PAGE 130

130 Figure 6-1. Location of IODP Site U1304 and other sites drilled during IODP Expedition 303.

PAGE 131

131 Figure 6-2. Hysteresis ratios measured from IODP Site U1304 sediments plotted on a Day plot (Day et al., 1977). SD: si ngle domain; PSD: pseudo-single domain; MD: multi-domain.

PAGE 132

Figure 6 6 -3. Anhys t t eretic sus c c eptibility ( 132 ARM) plotte d d against v olume sus c c eptibility ( ).

PAGE 133

133 Figure 6-4. Cores splice image, diatom distribution, and magnetic concentration measurements including volume susc eptibility, ARM intensity (after demagnetization at peak field of 35 mT), ARM acquisition intensity (acquired in demagnetization peak field of 35 mT with a 50 T DC bias field), and IRM intensity (after demagnetization at peak fi eld of 35 mT). Disseminated diatom intervals are represented by half length of diatom intervals.

PAGE 134

134 Figure 6-5. Component NRM inclinat ion and declination with maximum angular deviation (MAD) calculated for the 20-80 mT peak alternating field demagnetization range, compared to diat om distribution. Declinations were corrected using tensor core orientation data where available (green circles), and using rotation of entire cores (red circ les) so that the mean declination is oriented North or South for positive and negative inclinat ion intervals, respectively.

PAGE 135

135 Figure 6-6. Orthogonal proj ections of NRM demagnetizat ion data for representative samples from clay-rich intervals and di atom-rich intervals, and for samples from the two recorded magnetic excursion intervals (i.e., Kamikatsura and Gardar). Circles (red) and squares (b lue) denote projection on vertical and horizontal planes, respectively. Unit for intensity scale is mA/m.

PAGE 136

136 Figure 6-7. Histograms of component inclin ation values within the Brunhes Chron for samples from diatom-rich intervals (ri ght) and non-diatom intervals (left).

PAGE 137

137 Figure 6-8. Slopes and accompanying linear correlation coefficients (R values) for NRM versus ARM demagnetization data (blue) and for NRM versus ARMAQ demagnetization data (red), compared to diatom distribution.

PAGE 138

138 Figure 6-9. Volume susceptibility, com ponent NRM inclination, and slopes of NRM versus ARMAQ (as a RPI proxy) of IO DP Site U1304, compared to the PISO1500 RPI and oxygen isotope stack records (Channell et al., 2009).

PAGE 139

139 Figure 6-10. Depth-age curve and sedimentat ion rates for IODP Site U1304 according to the paleointensity-based age model, com pared to diatom distribution at the site and the PISO-1500 oxygen isotope st ack record (Channell et al., 2009).

PAGE 140

140 Figure 6-11. Power spectrum of the IODP Site U1304 relative paleointensity record (blue). REDFIT (Schulz and Mudelsee, 200 2) is used for spectral analyses of the unevenly spaced RPI record a fter applying the age model. The 95% (green) and 99% (red) confidence levels (CL) were estimated using both the theoretical first-order autoregressive model (dashed lines) and Monte Carlo (1000 runs) methods (solid lines). Orbi tal frequencies at 1/400, 1/100, 1/41, and 1/23 kyr-1 are marked by vertical dashed lines.

PAGE 141

141 Figure 6-12. A) Local wavelet power spec trum (LWPS) of t he IODP Site U1304 RPI record, B) LWPS of the PISO-1500 RPI record (Channell et al., 2009), C) cross-wavelet power spectrum (|XW T|) of the IODP Site U1304 and the PISO-1500 RPI records, D) squared wa velet coherence (WTC) between the IODP Site U1304 and the PISO-1500 RPI records. Values of normalized wavelet power, cross-wavelet power, and squared wavelet coherence are indicated using different colors on LWPS, |XWT|, and WTC maps (with blue to red indicating increasing values). The 5% significance level against red noise is shown as thick contours in all figures. The cones of influence (COI) where edge effects make the analyses unreliable are marked by areas of crossed lines. In C) and D) the relative phase rela tionship is shown as arrows (with inphase pointing right, anti-phase pointi ng left, and first signal leading the second 90 pointing straight up). Orbital per iods of 400kyr, 100 kyr, 41 kyr, and 23 kyr are marked by white dashed lines from bottom to top on the vertical axes.

PAGE 142

142 CHAPTER 7 SELF-REVERSAL AND APPARENT MA GNETIC EXCURSIONS IN ARCTIC SEDIMENTS Introduction Piston cores from the Arctic and No rwegian-Greenland Sea have presented a stratigraphic challenge due to the lack of bi ogenic carbonate (foraminifera) for isotope analyses and poorly constrained biostratigr aphies. As explained by Backman et al. (2004), the Arctic Ocean was considered a sediment starved basin with mm/kyr scale sedimentation rates based lar gely on the interpretation of magnetostratigraphic records in which the Matuyama-Brunhes boundary was o ften placed ~1mbelow seafloor (e.g., Steuerwald et al.,1968; Clark,1970). Si nce the 1980s, higher (cm/kyr scale) sedimentation rates have become evident throughout the Arctic oceans from radiocarbon dates, and biostratigraphic observations (e.g ., Markussen et al., 1985). These revised ages resulted in radical chan ge in magnetostratigraphic interpretations, with records from the Arctic Ocean and No rwegianGreenland Sea being interpreted to exhibit intervals of negative inclination (exc ursions) within the Brunhes Chron (Lvlie et al., 1986; Bleil and Gard, 1989; Nowaczyk and Baumann, 1992; Nowaczyk et al., 1994; Nowaczyk and Antonow, 1997; No waczyk and Knies, 2000). Excursion ages, derived from outside the Arctic, hav e then been adopted as age control points in Arctic cores. The paucity of corroborating stratigraphic information in the Arctic and Norwegian Greenland Sea results in a high degree of fr eedom in labeling the apparent magnetic excursions. In 2004, the Integrated Ocean Drilling Progr am (IODP) Arctic Coring Expedition (ACEX) focused the resources of IODP on drilling on the crest of the Lomonosov Ridge (Figure 7-1) (Backman et al., 2006). In a recent ACEX age m odel (Backman et al.,

PAGE 143

143 2008), the authors broke with tr adition and constructed their age model without using paleomagnetic data, apart fr om one (Paleocene) polarity reversal. Backman et al. (2008) were circumspect about the magnetic data and concluded that: the occurrence of these high frequency polarity changes in the Neogene ACEX sediment sequence may represent either distortions of the paleomagnetic record or the genuine behavior of the geomagnetic field in this part of the Arctic Ocean. The AC EX NRM inclination record (after 40 mT peak field dem agnetization) indicate s an upper 4.6 m of predominantly steep positive inclinations inte rspersed with few cm scale events where the inclination values are negative (O'Regan et al., 2008). The few cm thick events were correlated to the Mono Lake excursion (~33 ka ), the Laschamp excursion (~41 ka), and the Norwegian-Greenland Sea excursions at 55 and 66 ka. These age assignments are consistent with four radi ocarbon ages and the presence of a few specimens of E. huxleyi at 1.91 mcd (meters composite depth), interp reted as indicative of MIS5 (Cronin et al., 2008). Below this uppermost 4.6minterv al, well-defined zones of steep positive and steep negative inclinations are interspersed, with the uppermost negative inclination interval associated with the Biwa II magnetic excursion at ~240 ka (O'Regan et al., 2008). In comparing the ACEX inclination record below ~4.6 m depth with the inclination record in neighboring Cores PS-2186-6 and 96/12-1PC, O'Regan et al. (2008) state: the overall similarity in the excursion patte rns at these and other Arctic sites suggests they record synchronous variations in the gl obal dipole field. However, once these cores have been stratigraphically aligned using se diment physical properties, it becomes apparent that inclination changes ar e not necessarily synchronous.

PAGE 144

144 In the last 20 years, the revelation of numerous polarity excursions within the Brunhes and Matuyama Chrons has been one of the most important developments in paleomagnetism. Following Champion et al. (1988), who made the case for 8 excursions within the Brunhes Chron, the number of excu rsions in the Brunhes Chron has proliferated to 12-15 although only abo ut seven Brunhes-aged excursions have been adequately documented and age calibrated ( Lund et al., 2006; Laj and Channell, 2007). The recording of magnetic excursions is uncommon because their brief duration restricts them to sedim entary sequences characterized by high fidelity magnetic recording and sedimentation rates in exce ss of ~10 cm/kyr (Roberts and Winklhofer, 2004). In the Arctic Ocean and Norwegi an-Greenland Sea, cores with mean sedimentation rates that are an order of magnitude lower have apparently recorded numerous excursions. For example, the ACEX mean sedimentation rate for the Brunhes Chron is estimated to be 1.8 cm/kyr (O'Regan et al., 2008) with a Brunhes Chronozone thickness of 14.3 m. Outside the Arctic and Norwegian-Gr eenland Sea, the thickness of the Brunhes Chronozone exceeds 80 m in se ctions where Brunhes excursions have been observed. The better estimates of excursi on duration from the No rth Atlantic lie in the few (1-3) thousand-year range (e.g. Laj et al., 2000; Lund et al., 2005; Channell, 2006). Sediment cores from the Arctic Oc ean and Norwegian-Greenland Sea usually yield longer duration estimates for geomagnetic excursions. For example, assuming the labeling of excursions in the Arctic r egion is correct and adopting excursion ages determined outside the region, the Mono Lake, Laschamp and Blake excursions have apparent durations of 15, 20 and 26 kyr, respectively, in Yermak Core PS2212 (Nowaczyk et al.,1994). A related anomaly common to these high latitude cores is that

PAGE 145

145 the cumulative percentage thickness of zones of negative inclination is far greater than expected. For example, zones of negative in clination occupy ~50% of the recovered sedimentary sequence in Fram Strait Core PS1535 (Nowaczyk and Baumann, 1992; Nowaczyk and Frederichs, 1999) and N 50% for the top 4 m of th e section interpreted to represent the last 120 kyr in Core PS2212 from the Yermak Plateau (Nowaczyk et al.,1994). Fortuitous fluctuations in sedim entation rates have to be invoked to explain these amplified excursion records. Magnetic Properties of Core HLY0503-6JPC The natural remanent magnet ization (NRM) of one of the cores (Core HLY05036JPC at 78 17.6N and 176 59.2W) tak en from the Mendeleev Ridge (Figure 7-1) during the Healy-Oden Trans-Arctic Ex pedition 2005 (HOTRAX) expedition is characterized by negative component inclinat ions within the uppermost few meters of the core (Figure 7-2). The sediments comprise brown to yellow-gray silts and silty-clays with occasional sandy layers. The chronology of sedimentation on the Mendeleev Ridge is based on amino acid racemization with accompanying radiocarbon ages, and sparse biostratigraphic data. The mid-Brunhes Chron at 300 ka is estimated to lie at ~5mdepth in Core HLY0503-6JPC and other HOTRAX co res collected from the Mendeleev Ridge (Polyak et al., 2004; Kaufman et al., 2008), consistent with few-cm/kyr scale sedimentation rates (Backman et al., 2004). The important point for the purposes of this paper is that the entire 5 m represented in Figure 7-2 cons titutes sediment deposited during the Brunhes Chron. The NRM data were partly acquired from u-channel (2 cm3) samples. The NRM of each u-channel was measured at 1 cm intervals before demagnetization and after alternating field (AF) demagnetizati on at 14 steps in the 10-100 mT peak field

PAGE 146

146 range. For each step, the samples were measured after two demagnetization sequences for the X, Y and Z sample axes The first XYZ demagnetization sequence was followed by measurement, and then the ZXY demagnetization sequence was followed by repeat measurement at the same demagnetizati on step. The purpose of the procedure was to monitor any spurious anhysteretic remanence (ARM) acquisition during AF demagnetization. AF-derived NRM component m agnetization directions include several intervals of negative component inclination, with inclinations generally lower than the expected inclination (84) for a geocentric axial dipole field at the site (Figure 7-2). The component magnetization directions, is olated in the 20-80 mT peak field range using the standar d procedure (Kirschvink, 1980), are reasonably welldefined (Figure 7-3) as indi cated by maximum angular deviation (MAD) values often less than 10 (Figure 7-2). Choosing a diffe rent demagnetization range, other than 2080 mT, would not appreciably change the result, and by using a uniform demagnetization range, the MAD values serve to indicate the variation in definition of the component direction. Note that t he directions are not dependent on the demagnetization sequence (XYZ or ZXY), ruli ng out appreciable influence from spurious magnetizations acquired during the AF demagnetization proce ss (Figure 7-2). Ten discrete samples (approximately 2 cm3) were collected from both positive and negative inclination intervals of the u-channel used for AF demagnetization experiments (Figure 7-2). Isothermal remanent magnetizations (IRM) acquired in DC fields of 1.2 T, 0.5 T, and 0.1 T were imposed along three orthogonal axes of the discrete samples (method of Lowrie, 1990). Thermal demagnetization of each orthogonal component indicates that the sa mples are dominated by soft and medium-

PAGE 147

147 coercivity IRM components (Figure 7-4). The lowand medium-coercivity IRMs, acquired in 0.1 T and 0.5 T fields, respective ly, have a maximum blocking temperatures (580 C) indicative of magnetit e. An abrupt drop in intensit y for the medium-coercivity fraction below 300 C (Figure 7-4) is interpreted to indicate the presence of (titano)maghemite, that loses its remanence due to inversion at temperatures above 250 C (Readman and O'Reilly,1970,1972; zdemir ,1987). An alternative explanation, the presence of magnetic iron sulfides, c an be discarded based on the oxidized nature of these red-brown sediments, the lack of evidence for iron sulfides from X-ray diffraction, scanning electron microscopy (SEM), and gas chromatography for the detection of sedimentary sulfur performed at Old Dominion University by C. Lingle and D. Darby using the very sensitive methods described by Cutter and Oatts (1987). Samples from intervals characterized by negative NRM inclination components (red in Figure 7-4) have a higher proportion of the medium-coercivity (titanomaghemite) magnetic component. Magnetic moment in a 0.5 T field was measured as a func tion of temperature, in helium atmosphere, on a vibrating sample magnetometer (Figure 7-5A) and during thermal cycling (Figure 7-5B). The decreas e in moment below 300 C and below 580 C (Figure 7-5A) can be associated with the pr esence of titanomaghemite and magnetite, respectively. The increase in magnetic mom ent on cooling from temperatures above 400 C (Figure 7-5B) is associated with invers ion of titanomaghemites to intergrowths of Ti-poor spinel (magnetite) with a Ti-rich rhombohedral phase near ilmenite (Readman and O'Reilly, 1970, 1972; zdemir and Baner jee, 1984; zdemir, 1987; Krasa and Matzka, 2007; Soubrand-Colin et al., 2009). The inversion of titanomaghemite to a more

PAGE 148

148 magnetic (magnetite) phase in th is temperature range is more likely than the possibility of a restricted titaniumcontent in tit anomagnetite as the cause for the observed decrease in magnetization at ~300 C (Figure 7-4 and 5). On monitoring of magnetization at low te mperatures (Figure 7-6), evidence for a suppressed Verwey transition at ~120 K, in samples from both positive and negative inclination intervals, is consistent wit h the presence of magnet ite and maghemite. The maghematization of magnetite suppresses t he Verwey transition relative to its appearance in the unoxidized magnetite, and the tr ansition is likely to be smeared over a wider range of temperatures (say 70-120 K) than for stiochiometric magnetite (zdemir et al., 1993). Thermal demagnetization of the NRM was carried out on discrete (8 cm3) samples, collected alongside the u-channel samples in plastic cubes. The NRM was measured without demagnetizati on, and then rom the plasti c containers, and wrapping in Al foil for thermal treatment. Magnetization directions were found to be consistent before and after drying and wrapping. Thermal demagnetization revealed NRM components with negative inclinat ion that have blocking temperatures largely, but not entirely, below 300 C (Figure 7-3). Th is negative inclination component is superimposed on a higher blocking temper ature component with maximum blocking temperatures up to 600 C that usually has positive inclination and may represent the direction of the geomagnetic field close to the time of sediment deposition. For one sample shown in Figure 7-3 (at 0.91 m), the negative inclinati on component appears to have blocking temperatures up to 500 C. Thes e observations implies that the negative

PAGE 149

149 inclination components resolved by AF demagnet ization (Figs. 2 and 3) are carried by the mineral (titanomaghemite) that has blocking temperatur es largely below 300 C. As the sediment was deposited in the Brunhes Chron, the negative inclination component, apparently carried by titanom aghemite, may have been generated by partial self-reversal. The thermo-magnetic experim ents (Figs. 4 and 5) indicate that the presence of titanomaghemite is not restri cted to zones characterized by negative inclinations (Figure 7-2). Note that t he zones of negative inclination correspond to intervals of low NRM intensity (Figure 7-2) implying the presence of antiparallel magnetization components in these intervals. In summary, zones with negative component inclination as revealed by AF demagnetization generally show, on thermal demagnetization, positive inclination comp onents with higher bloc king temperatures superimposed on negative incl ination components with lowe r blocking temperatures. The coercivity spectra of titanomagnetite and titanomaghemite overlap, but blocking temperature spectra of the two phases are distinct. XRD and SEM Observations Magnetic extracts were made by dispersing a small amount of sedi ment (~30 g) in a sodium metaphosphate solution, and using a magnetic finger (comprising a string of rare earth magnets in a glass test tube) to repeatedly separat e magnetic particles. Centrifuging with a heavy liquid (sodium polyt ungstate with density 2.84 g/cc) was used to refine the extract. Extracts were was hed using water, then methanol, and dried for scanning electron microscopy (SEM) usi ng a Zeiss EVO microscope and EDAX (Genesis) energy dispersive analysis. Images and EDAX analyses indicate the presence of micron-sized titanomagnetite/tit anomaghemite grains with varying amounts of titanium and compositions in the x=0. 35.65 range, with anci llary quartz and sphene

PAGE 150

150 (Figure 7-7). The larger FeTi rich parti cle in Figure 7-7 shows indications of compositional zoning. X-ray diffraction (XRD) for unheated magnet ic separate, and magnetic separate heated to 700 C, indicate peaks, calibrated using a silica standard, corresponding to magnetite/maghemite and quartz (Fi gure 7-8). The XRD peak at 2 = 35.63 for the unheated separate lies close to the peak for (titano)maghemite deriv ed from the [113] diffraction plane. The 2 value (35.63) is larger than the 2 value for magnetite standards (35.42), and the peak is shifted to lower values of 2 (by ~0.3) after heating the separate, consistent wit h (titano)maghemite in the unheated sample being inverted to magnetite in the heated sample. The diffraction peaks close to 2 =57.0.5 and 62.5.5 (Figure 7-8) are derived from the [115] and [440] diffraction planes of magnetite/maghemite, respectively. Based on comparison with standar ds for magnetite (where 2 =56.943 and 62.515) and maghemite (where 2 =57.271 and 62.925), the peaks are consistent with mixtures of ti tanomaghemite and titanom agnetite. Both peaks are shifted to lower values of 2 on heating (again by about 0.3) consistent with the inversion of (titano)maghemite to lo w-Ti magnetite (Figure 7-8). The asymmetric shape of the three diffr action peaks A, B and C (Figure 7-8) is interpreted to indicate the co-existence of (titano)maghemite and (titano)magnetite in both the unheated and heated samp les. We attempt to m odel the diffraction peaks, using pseudoVoigt functions, to derive two components which best fit each of the three diffraction peaks (Table 7-1). For the unheated extract, the average lattice parameters for the two modeled components (8 .3388 and 8.3773 ) indicate high oxidation states with zN0.9 (Readman and O'Reilly, 1972). For the heated extract, higher average lattice

PAGE 151

151 parameters (8.3781 and 8. 4230 ) indicate lower oxidati on states (z=0.7 for x=0.6), interpreted to be due to partial inversion of titanomaghemite to Ti-poor magnetite on heating. Self-reversal in Titanomaghemite Individual single-domain titanomagnetite grains ca rrying thermal remanent magnetization (TRM) can be statistically aligned in the ambient field during or shortly after sediment deposition to yield a detrita l remanent magnetization (DRM). Seafloor oxidation of the titanomagnetite grains results in the TRM of individual grains being partially or completely transformed to a chemical remanent magnetization (CRM) carried by titanomaghemite. T he oxidation of ti tanomagnetite to titanomaghemite on the seafloor is restricted to sedimentary env ironments characterized by low accumulation rates, such as the red-clay facies of t he central North Pacific (Kent and Lowrie, 1974; Johnson et al., 1975), where sl ow burial enables oxidati on at the sedimentwater interface. The CRM of the titanomaghemite is usually considered, based on experimental data, to paralle l the original TRM of the titanomagnetite grain (Johnson and Merrill, 1974; zdemir and Dunlop, 198 5), however, the oxidized phase can become magnetized anti-paralle l to the host phase due either to a change in the balance of anti-parallel moments in the ferrimagnetic subla ttices of the spinel structure during oxidation, or alternatively by negative magnetostatic interactions between magnetic phases. Titanohematites and titanomaghemites have been central to models of the selfreversal phenomena since the 1950s. The best documented model of self-reversal involves high temperature ex solution of phases of titan ohematite with bulk composition near y=0.5. Titanohematites with th is bulk composition are rare but include the famous

PAGE 152

152 Haruna dacite (Nagata et al., 1952; Uyeda, 1958) and more recently investigated examples (e.g. Bina et al., 1999). Models of self-reversal involving titanomaghemite can be traced to the ionic orderi ng models of Verhoogen (1956, 1962) in which highly oxidized titanomaghemites (with high values of x and z) lead to partial self-reversal of chemical remanence (CRM) relative to the remanence direction in the host titanomagnetite (O'Reilly and Banerj ee, 1966; Schult, 1968, 1971; Ozima and Sakamoto, 1971; Readman and O'Reilly, 1970, 1972). The process of low-temperature oxidation of titanomagnetite to titanomaghemite can be accomplished by diffusion of Fe2+ ions from the B (octahedral) sublattice s beginning at the surface of the grain (O'Reilly, 1984; Dunlop and zdemir, 1997). The inverse spinel structure is unchanged during oxidation but vacancies develop at the octahedral (B sublattice) sites of leached Fe2+ ions. The leached Fe2+ ions are converted to Fe3+, and may be exchange coupled at the surface of the grai n. During high degrees of lo w-temperature oxidation, ferrimagnetic titanomagnetite with a magnet ica lly dominant B (octahedral) sublattice can be transformed to a tit anomaghemite with a dominant A (t etrahedral) sublattice moment, aligned anti-parallel to the B subl attice moment. This change is accompanied by a decrease in the Fe/Ti ratio. In mid-ocean ridge basalts, Doubrovi ne and Tarduno (2004) associated selfreversed NRM components, with blocki ng temperatures below 325 C, to titanomaghemite formed by low-tem perature seafloor oxidation. In the case of the Arctic sediments studied here, a detrital remanent magnetization (DRM) carried by singledomain titanomagnetite appears to have been conv erted by sea floor oxidation to a titanomaghemite carrying a partially se lf-reversed CRM. The apparent stratigraphic

PAGE 153

153 correlation of intervals of negative inclination within the Arctic Ocean (e.g. Spielhagen et al., 2004; O'Regan et al., 2008) implies that t he alteration of detrital titanomagnetite to titanomaghemite is lithol ogically controlled. Organic ca rbon records from the Arctic Ocean and NorwegianGreenland Sea indicate the dominance of terrigenous organic matter, presumably because sea-ice coverage inhibited marine productivity (Ikehara et al., 1999; Stein et al., 2003; Expedition 303 Sci entists, 2006). The refractory nature of the (terrigenous) organic matter in the Arctic regi on results in the low activity of (sulfate) reducing microbes, thereby maintaini ng oxidizing diagenetic conditions. Conclusions In Core HLY0503-6JPC from the Mendeleev Ridge, the evidence for self-reversal lies in the identification of an authigenic carrier of NRM (titanomaghemite) and the observation that this mineral carries an NRM component, isolated by thermal demagnetization that is approximately ant i-parallel to the NRM of primary titanomagnetite. AF demagnetizat ion does not reveal this same antiparallelism (Figure 7-3), presumably due to overlapping coercivi ty spectra of the tw o remanence carriers. AF demagnetization generally implies a primary magnetization component carrying positive and negative inclinations that mimic polarity zones or excursions (Figure 7-2). The presence of titanomaghemite with negative inclinations in this core may be relevant to the long-standing problem of interpreting magnetostratigraphies from the Arctic Ocean and Norwegian-Greenland Sea. Up to now, AF demagnetization has been used exclusively in magnetic studies of marine sediments from the Arctic Ocean and Norwegian-Greenland Sea. Due to the sim ilar coercivities of titanomagnetite and titanomaghemite, AF demagnetizat ion of these Arctic sediments does not reveal that positive and negative inclinati on components are carried by different mineral phases. As

PAGE 154

154 the global record of geomagnetic excursions has improved, the propensity for, and duration of, excursions in the Arctic Ocean and NorwegianGreenl and Sea has become increasingly anomalous, and requires special char acteristics of the geo magnetic field at high latitudes for which there is little evidence in the modern field or in numerical models of field behavior. For this reason, the evidenc e for partial self-reversed CRM in Arctic sediments from the Mendeleev Ridge has implications for Arctic sediments in general, and for diagenetic alteration of titanomagnetite DRM in oxidizing.

PAGE 155

155 Table 7-1. 2 values and estimated lattice parameters (L.P.) for Component 1 and Component 2 derived by modeling of the three asymmetric X-ray diffraction peaks (Figure 7-8). Samples Components Diffraction plane Average L.P. () [113] [115] [440] 2 () L.P. () 2 () L.P. () 2 () L.P. () Heated (700C) magnetic extract Comp. 1 35.3124 8.4230 56.7404 8.4230 62.4455 8.4060 8.4173 Comp. 2 35.5130 8.3768 57.0652 8.3793 62.6478 8.3818 8.3793 Unheated magnetic extract Comp. 1 35.5050 8.3787 57.1686 8.3652 62.5946 8.3879 8.3773 Comp. 2 35.6909 8.3367 57.4112 8.3331 62.9357 8.3467 8.3388

PAGE 156

156 Figure 7-1. Location of Core HLY05036JPC and the IODP Expedition 302 (ACEX) sites. LR denotes Lomonosov Ridge, MR denotes Mendeleev Ridge.

PAGE 157

157 Figure 7-2. Core HLY0503-6JPC: Component declination and inclin ation with maximum angular deviation (MAD) statistic calculat ed for the 20-80 mT peak alternating field range using two demagnetization s equences for the three sample axes: XYZ (blue) and ZXY (red). NRM intensity is shown after demagnetization at peak fields of 10 mT for the XYZ sequence (blue) and ZXY sequence (red). Note that declination values are arbitrary as core was not oriented in azimuth. Section breaks are indicated.

PAGE 158

158 Figure 7-3. Orthogonal projection of AF demagnetizat ion (above) and thermal demagnetization (below). Peak demagnetizing field ranges are 10-60 mT in 5 mT steps then 60-100 mT in 10 mT st eps. Temperature ra nges are 25-600 C in 25 C steps. Circles (red) and squares (blue) denote projection on vertical and horizontal planes, respectively. Decli nation values are arbitrary as core was not oriented in azimuth. Unit for intensity scale is mA/m. Meter levels correspond to meter leve ls in Figure 7-2.

PAGE 159

159 Figure 7-4. Thermal demagnetization of three-axis isothermal remanent magnetizations (IRM) imposed orthogonally and sequentially in DC fields of 1.2 T (hard), 0.5 T (medium) and 0.1 T (soft), for sample s collected from intervals showing positive (blue) and negative (red) componen t inclinations. Refer to Figure 7-2 for positions of the samples.

PAGE 160

160 Figure 7-5. A) Magnetization in a 0.5 T applied field derived from hysteresis loops measured at increasing temperatur es (in 25 C steps in an helium atmosphere) using a vibrating sample magnetometer (VSM). Values have been normalized to the room temperatur e measurement. B) Magnetization in a 0.5 T applied field measured during thermal cycling where red and blue lines denote heating and cooling, respectively. The sample was heated in 100 C increments during measurement at ~1 C steps, then cooled by 100 C during further measurement, then heated through 100 C without measurement to the onset temperatur e of the next heating cycle. Refer to Figure 7-2 for positions of samples.

PAGE 161

161 Figure 7-6. A) Saturation remanent magnetizat ion (SIRM), acquired in a 2.5 T field at room temperature, on cooling and warming; m easured using a magnetic properties measurement system (MPMS). Measurements have been normalized to the room te mperature magnetization of the sample on cooling. B) Field cooled (2.5 T), and zero fi eld cooled, low-temperature SIRM measured on warming. Measurements have been normalized to the field cooled magnetization of t he sample at room temperature. Blue and red curves denote samples from positive and negative inclination intervals, respectively (see Figure 7-2).

PAGE 162

162 Figure 7-7. Energy dispersive (EDAX) elem ental mapping of micr on-sized grains from an unheated magnetic extract. The grains in the image comprise titanomagnetite/titanomaghem ite, Ti-poor magnetit e/maghemite (circled), sphene/titanite (Sp) and quartz (Qtz). The total X-ray spectrum acquired during the mapping is also displayed, where carbon (C) and aluminum (Al) are attributed to background noise from the carbon tape and aluminum stub.

PAGE 163

163 Figure 7-8. X-ray diffraction (XRD) resu lts for unheated magnetic extract (blue) and heated (to 700 C) magnetic extract (red), wi th higher resolution scans around the three peaks (A, B, C) associated with titanomaghemite and titanomagnetite. The positions and magnitudes of XRD peaks for magnetite (green) and maghemite (orange) standards are indicated. Ancillary peaks are associated with quartz (Qtz). Diffraction peaks close to 2 = 35.7, 57.4 and 62.9 are interpreted as com posite peaks due to mixtures of (titano)maghemite and (tit ano)magnetite. The displacement of the peaks to lower values of 2 on heating is interpreted as being due to partial inversion of titanomaghemite to Ti-poor magnetite.

PAGE 164

164 CHAPTER 8 ORIGIN OF APPARENT MAGNETIC EXC URSIONS IN DEEP-SEA SEDIMENTS FROM MENDELEEV-ALPHA RIDGE (ARCTIC OCEAN) Introduction The significance of the Arctic Ocean in understanding climate change has brought increasing geological investigations to the Arctic since the 1960s. Limited biogenic productivity and high carbonate dissolution in the Arctic Ocean present a challenge for the construction of conventional biostratigraphy and isotopi c stratigraphy. Magnetic stratigraphy has been important for providing age constrai nts for Arctic deep-sea sediments. Magnetic polarity patterns of sediment cores retrieved from the Northwind Ridge (e.g., Poore et al., 1993; Phillip s and Grantz, 1997), the Mendeleev-Alpha Ridge (e.g., Steuerwald et al., 1968; Hunkins et al., 1971; Herman, 1974; Clark et al., 1980, 1984; Witte and Kent, 1988), and the Lomonosov Ridge (e.g., Morris et al., 1985; Spielhagen et al., 1997) often yielded sedimentation rates on t he mm/kyr scale (see Figure 8-1 for location of these studied cores). In these studies, negative NRM inclinations often occurred at a depth of 12 meters below sea fl oor (mbsf), and were commonly interpreted as represent ing the Brunhes/Mat uyama boundary. The amino acid epimerization dating from a Mendeleev Ridge core suggested cm/kyr scale sedimentation rates (Sejr up et al., 1984) inconsistent with earlier paleomagnetic data from the same core by Herman (1974). The cm/kyr scale sedimentation rate was later supported by ages provided by radiocarbon dating (e.g., Darby et al., 1997), correlating manganese and color cycles to low latitude 18O records (Jakobsson et al., 2000), and optically st imulated luminescence dating (Jakobsson et al., 2003), as well as improved amino acid racemization methods (Kaufman et al., 2008). These methods were accompanied by poorly constrained biostratigraphic

PAGE 165

165 markers, such as t he sporadic appearance of E. huxleyi indicative of the late Brunhes Chron. It appears that cm/kyr scale sedi mentation rates are evident throughout the Arctic Ocean (Backman et al., 2004; Spie lhagen et al., 2004; Polyak et al., 2009). Accordingly, magnetic excursions of Br unhes age have often been invoked to explain the negative inclinations observ ed in the top several meters of the Arctic sediment sequences (e.g., Lvlie et al., 1986; Bleil, 1987; Nowa czyk and Baumann, 1992; Jakobsson et al., 2000; Nowaczyk et al., 2001; Spielhagen et al., 2004; O'Regan et al., 2008), and magnetic excursion ages derived from outside the Arctic Ocean were often adopted as age control points fo r Arctic sediments. However, there are obvious problems associated with the excursion interpretation. Due to limited duration of ma gnetic excursions and smoothing effects of the magnetization lock-in process, even s ediments with deposition rate >10 cm/kyr rarely preserve magnetic excursions (Roberts and Winklhofer, 2004). The Arctic sediments appear to be particularly effici ent in recording magnetic excursions. Furthermore, negative inclination intervals in the Arctic sediments often reach tens of centimeters thick implying excursion duratio ns of >10 kyr (e.g., Backman et al., 2008; Channell and Xuan, 2009). Excursion durations estimated ou tside the Arctic region, however, are usually < 5 kyr (see Laj and Channell, 2007), compar able with the ~3 kyr timescale for diffusive field changes in the Earths solid inner core (Gubbins, 1999). Fortuitous variations in sedimentation rate have to be invoked to explain the amplified excursions, and excursions do not always corre late after stratigraphic alignment using sediment physical properties (e.g., O'Regan et al., 2008). Jak obsson et al. (2000) noted that below their inferred Brunhes/Matuy ama boundary in Core 96/12-1pc from the

PAGE 166

166 Lomonosov Ridge, inclination pa tterns cannot be correlated to the geomagnetic polarity time scale without introducing major disc ontinuities in sedimentation rate. Paleomagnetic studies (C hannell and Xuan, 2009) on Core 06JPC recovered by Healy-Oden Trans-Arctic Expedition 2005 (H OTRAX05) to the Mendeleev-Alpha Ridge indicated that titanomagnet ite and titanomaghemit e are the magnetic carriers in the sediments. The authors proposed that negative inclinations in the sediments could have resulted from a partially self-reversed chemical remanent magnetization (CRM) acquired during the oxidation of host titanomagnet ite grains to titanomaghemite. In this paper, we extend the study of Channell and Xuan (2009) to Cores 08JPC, 10JPC, 11JPC, and 13JPC (Figure 8-1) along the Mendeleev-Alpha Ridge of the Arctic Ocean to further understand the origin of the apparent magnetic excursion in these sediments. NRM Measurements U-channel samples (typically 2 cm3) were collected from Cores 08JPC, 10JPC, 11JPC, and 13JPC recovered by HOTRAX05 to the Mendeleev Ridge and Alpha Ridge (Figure 8-1, Table 8-1). The sediments of these cores generally consist of dark brown to yellowish or grayish silts and silty-clays with occasional coarse sand layers. Natural remanent magnetization (NRM) of each u-channel sample was measured at 1 cm intervals before demagnetiz ation and after alternating field (AF) demagnetization at 10-14 steps in the 10-100 mT peak field range. For each step, samples from Cores 08JPC, 10JPC, and 11J PC were measured after two separated demagnetization sequences for the X, Y and Z sample axes to monitor any spurious anhysteretic remanence (ARM) acquisition du ring AF demagnetization. The first XYZ demagnetization sequence was followed by measurement, and then the ZXY demagnetization sequence was followed by repeat measurement at the same

PAGE 167

167 demagnetization step. For each demagnetization step, only the XYZ demagnetization sequence was applied to samples from Core 13JPC. Component magnetization directions of these cores were calculated from the 20-80 mT demagnetization peak field range using the principle component analysis (PCA) method (Kirschvink, 1980). For the uppermost 5 meters, component inclinations of these cores are characterized by several negative inclination intervals that reach thicknesses of several tens of centimeters (Figure 8-2). The maximum angul ar deviation (MAD) values associated with the PCA calculations are o ften lower than 10, especially for Cores 08JPC and 10JPC, indicating reasonably well defined compone nt directions (Figure 8-3). However, component inclinations of these cores ar e generally tens of degrees lower than the expected inclinations for a geocent ric axial dipole field at the coring sites (vertical green lines in Figure 8-2). Although a slump or gr avity-flow deposit may distort the record (e.g., 2.7-3.5 m in Core 08JPC, Adler et al., 2009), slumps are generally not seen in these cores. No significant differences were observed between component directions calculated using data from the two separate demagnetization sequences (XYZ and ZXY, blue and red curves in Figure 8-2 re spectively), precluding the possibility of negative inclinations being caused by spurious ARM acquisition during AF demagnetization. Age control for Core 08JPC was obtained from radiocarbon dating (uppermost ~70 cm), amino acid racemization methods (K aufman et al., 2008), correlation with cores from the Lomonosov Ridge, and correlati on of supposed glacial intervals with glaciations of the Eurasian Ar ctic margin (Adler et al., 2009). The uppermost 5 m of Core 08JPC was estimated to span the last ~250 kyr. This and other sediment cores

PAGE 168

168 from the Mendeleev-Alpha Ri dge have been correlated using paleomagnetic inclination patterns (of uncertain origin), detrita l carbonate abundance, and the top of the predominantly brown sediment with an estima ted age of ~500 kyr (Polyak et al., 2009). This marked color change was recovered at 8-10 mbsf in cores on the southern Mendeleev Ridge (e.g., Core 08JPC), and at 4-6 mb sf in cores from the interior of the western Arctic Ocean (e.g., Cores 10JPC, 11JPC, and 13JPC). The resulting sedimentation rate of ~1-2 cm /kyr is consistent with the cm/kyr scale sedimentation rate that has been suggested for this area (e.g., Backman et al. 2004; Sp ielhagen et al., 2004). In summary, the top 5 m of the studi ed cores can be constrained to the Brunhes Chron, restricting the observed negative inclinations to m agnetic excursions of Brunhes age. The NRM of discrete (2 cm3) samples collected in cubic plastic boxes alongside the u-channel samples was meas ured during thermal demagnetization. The discrete samples were dried in a magnetically shielded space with flowing helium gas before being taken out from the plastic containers and wrapped in Al foil for thermal treatment. Magnetizations we re monitored before and after the drying and wrapping, and no significant differences were found. M agnetizations of the discrete samples were then measured after thermal treatments in 25C steps in the 50-600C temperature range. For samples that show steep positive AF-derived inc linations, samples from the same depth level show similar thermal dem agnetization behavior (e.g., sample 10JPC 2.03 m in Figure 8-3). For samples characte rized by negative AF-derived inclinations, thermal demagnetization often reveals multiple NRM components with negative inclination components having blocking te mperatures largely below 350C, but

PAGE 169

169 occasionally reaching 500C (e.g., sample 10JPC 1.91 m in Figure 8-3). The negative inclination component is superimposed on a hi gher blocking temperature (up to 600C) component with positive inclinat ion that may represent the direction of the geomagnetic field close to the time of sediment deposit ion. It appears that t he negative inclination components, resolved by AF demagnetization (F igures 8-2 and 8-3), are carried by a magnetic mineral(s) that has blocking te mperatures largely below 350C. A soft component with positive inclinations is al so apparent in the 50-175C demagnetization range. This low blocking temperature component could originate from either a viscous remanence (VRM) or remanence carried by (mul ti-domain) grains of the original magnetic phase. Rock Magnetic Studies To identify the magnetic minerals in these sediments, rock magnetic experiments have been conducted at high (r oom temperature to 700C) and low (20 K to room temperature) temperatures at the Institute of Ro ck Magnetism (IRM) at the University of Minnesota. Hysteresis loops were measur ed at 25C temperatur e steps ranging from room temperature up to 700C, in a he lium atmosphere, on a vibrating sample magnetometer (VSM) for sele cted samples from Cores 08JPC and 10JPC (see Figure 8-2). Samples were freeze dried, powdered, cemented, and stuck to a ceramic holder, prior to the VSM measurem ents. Saturation magnetizat ion (Ms) derived from the hysteresis loops after slope correction, reveals abrupt drops below 300C and below 600C (Figure 8-4A), indicating two magnetic phases. The coercivity of remanence (Hcr) derived from the hysteres is loops show a minimum at ~250C, followed by a slight increase and a small peak at ~400C (Figure 8-4B). For two samples from Core 08JPC, magnetization was monitored at 1C steps during cooling from 700C to room

PAGE 170

170 temperature, in a 0.4 T fiel d, after heating during hyster esis loop measurements. The magnetization acquired during co oling is about one order of magnitude higher than the magnetization prior to heating (Figure 8-4C). This dramat ic increase of magnetization implies that a strongly magnetic phase has been produced during the heating. For selected freeze dried bulk sediment samp les (see Figure 8-2), room temperature saturation isothermal remanences (RT-SIRM), ac quired in a 2.5 T field, were monitored using a Quantum Designs Magnetic Proper ties Measurement System (MPMS) on cooling to 20 K and warming to room temperatur e (Figure 8-5A). In addition, field cooled (FC) and zero-field cooled (ZFC) low-te mperature SIRMs were measured on warming from 20 K to room temperature (Figure 8-5B). The FC re manences were measured after cooling in a 2.5 T field. For the ZFC remanenc es, samples were cooled in a zero field, and a 2.5 T field was applied at 20 K and t hen turned off prior to measurement on warming. The Ms against temperature (Ms(T)) data from Cores 08JPC and 10JPC sediments (Figure 8-4A) are si milar to those observed from sediments of Core 06JPC (Figure 5 in Channell and Xuan, 2009). The abrupt drop of Ms just below 600C indicates the presence of magnetite. This magnetite phase is probabl y a mixture of the original magnetite in the sediments and magnetite that has been produced during the heating. A much smaller contribution of magnetite to the IRM remanence (Figure 4 in Channell and Xuan, 2009), and t he huge increase of magnetization observed during the cooling (Figure 8-4C), as well as the suppr essed Verwey transition in low temperature data (Figure 8-5) suggest that the drop of Ms just below 600C is mainly due to the magnetite produced during heati ng. Although titanomagnetite could exhibit the abrupt

PAGE 171

171 drop in Ms at ~300C, the ti tanium content and grain size of titanomagnetite in the sediments would need to be very restricted to yield repeatabl e drops in Ms in the same temperature interval (i.e., ~300C) for different samples. Titanomagnetite in deep-sea sediments usually has a range of titanium content and grain size leading to a range of unblocking temperatures. We ther efore interpret the abrupt dr op of Ms at ~300C to the inversion of titanomaghemite. Titanomaghemi te is metastable when heated above 250300C, and, when heated up to 600C, the inversion product typically leads to increased Ms due to intergrowths of magnetite ilmenite and other minerals (Readman and O'Reilly, 1972; O'Reilly, 1983; zdemir, 1987; Dunlop and zdemir, 1997; Krsa and Matzka, 2007), explaining the observations in Figure 8-4C. The fa ct that the abrupt increase of magnetization during cooling occurs in a temper ature interval (~400-580C) lower than the temperature associated wit h the magnetization decrease on heating (~500-600C) is consistent with the two stage inversion observed in synthetic titanomaghemite samples durin g heating and cooling (zdemir, 1987). The minimum in Hcr at ~250C probably indicates the start of the titanomaghemite inversion (Figure 84B). The changes of Hcr during ~250-500 C may represent a blend between the generation of the new magnet ic phase and the unblocking of pre-existing magnetic grains. Partial oxidation of magnetite is known to smear the Verwey transition in magnetite from an abrupt drop at ~120 K to a wider temperat ure range (e.g., zdemir et al., 1993), consistent with our observation in the Arctic sedim ents (Figure 8-5). Titanomaghemite in the Arct ic sediments might be produc ed through low temperature oxidation of titanomagnetite. The low sediment accumulation rate (cm/kyr scale) and

PAGE 172

172 low concentration of labile organic matter appear to provide the conditions for diagenetic oxidation. To further understand the magnetic mineralogy of t he sediments, magnetic extracts were made from Cores 08JPC and 10JPC using sediments from intervals characterized by negative AF-derived NRM in clinations (see Figure 8-2). The extracts were made by dispersing ~2 cm3 sediment (~45 g) ta ken from the u-channel samples in a sodium metaphosphate solution of 4 wt.%, and using an automated extraction system to separate magnetic particles. The extraction system is equipped with a peristaltic pump so that the sediment slurry can re peatedly flow next to a glass tube filled with strong rare earth magnets. The acquired extrac ts were then intermittently washed into a glass container using dist illed water until no further extraction was achieved. A string of rare earth magnets in a glass test tube was then used to refine the extract. Extracts were again washed using distilled water, then methanol, and dried for isothermal remanence magnetization (IRM) acquisition measurements. IRM acquisitions by the magnetic extracts were measured at one hundred equidistant field steps on a logarithmic scale ranging from ~7 mT to 1 T, using an alternating gradient magnetomet er (AGM) at the University of Florida. Due to the expected logarithmic distribut ion of (magnetic) grain-size, IRM acquired by natural samples can be approximated by linear addition of cumulative log Gaussian functions (e.g., Robertson and France, 1994). Decomposit ion of the gradient of IRM acquisition curves provides a non-destructive tool fo r discriminating magnetic phases in natural samples (e.g., Kruiver et al., 2001; Heslop et al., 2002). IRM acquisition data of the magnetic extracts were analyzed using the me thod of Heslop et al (2002). Gradients of

PAGE 173

173 the IRM acquisition data plotted on a log scale show apparent asymmetry in shape, indicative of multiple coercivity com ponents. Fairly good fitting was achieved by modeling the gradient of t he IRM acquisition data using two magnetic coercivity components (Figure 8-6). Data from magnetic extract of Core 10JPC sediments yields two components with mean coercivity of 55.6 mT and 106.6 mT, comparable to 55.5 mT and 101.5 mT obtained from Core 08JPC magnetic extracts. The low and high coercivity components are consistent with the presence of (titano)magnetite and titanomagh emite in the samples, respectively. Theoretical modeling has shown that grain-size thres holds for superparamagnetic (SP) to single domain (SD), SD to pseudo-single domai n (PSD), and PSD to multi-domain (MD) behaviors increase with increasing oxidat ion state (Moskowit z, 1980). Consequently, grains with certain size may change their domain state from MD to PSD or SD during the oxidation, thereby incr easing their coercivities. Elevated coercivity has been observed for synthetic SD titanomagnetite grai ns with increasing oxidation parameter z up to ~0.5 (zdemir and O'Reilly, 1982). Fu rther oxidation seems to cause a decrease of coercivity for the synthetic samples. In nature, coerci vity of titanomaghemites is higher than that of unoxidiz ed titanomagnetites for high z values (Dunlop and zdemir, 1997). Increases in coercivity observed in h ysteresis loops (e.g., Marshall and Cox, 1972; Beske-Diehl and Soroka, 1984) or in median destructive field (MDF) from AF demagnetization (e.g., Ryall et al., 1977; Peterson and Vali, 1987) are generally seen for natural titanomaghemite in ocean basalts wit h increasing oxidation sate. It is also likely that the oxidation of titanomagnetite to ti tanomaghemite preferenti ally occurs in finer grains that generally have larger surface to volume ratio, leaving the larger grains

PAGE 174

174 less altered. Therefore, the titanomaghemite component may tend to have smaller grain-size and hence higher coercivity than the titanomagnetite component. SEM and XRD Analyses Magnetic extracts spread on a carbon tape were examined under a Zeiss EVO scanning electron microscopy (SEM) equippe d with Genesis X-ray energy-dispersive spectroscopy (EDS) at the University of Fl orida. Elemental maps were collected on micron-sized grain clusters for up to 20 hours using the EDS. The results show a number of grains that ar e rich in O, Ti, and Fe, confirming the presence of (titano)magnetite and/or tit anomaghemite. Calculation us ing the total image spectra acquired during the mapping yields mean com position paramet ers with x values of 0.35 and 0.45 for magnetic extracts from Core s 08JPC and 10JPC, respectively. The Ti concentration apparently varies widely from one grain to another (Figure 8-7), with calculated Ti composition x ranging from 0.08 and 0.85 (Figure 8-7). The huge C peak is due to the carbon tape, and the Si, Al, and O rich grains can be attributed to quartz and clay minerals that were not completely removed during the extraction. The collected X-ray spectrum and the elemental map for magnetic extracts from Core 08JPC are consistent to those of Core 10JPC, implying similar magnetic mineralogy for sediments in these cores. X-ray diffraction (XRD) analyses were perfo rmed on freeze dried bulk sediment powders and the magnetic extracts using a Ri gaku Ultima IV X-ray di ffractometer at the University of Florida. Magnetic extracts were placed on a zero background sample holder, and the focusing beam option was us ed for all measurements. High resolution diffraction patterns were collected from 10-90 2 range for the bulk sediment powders, and 5-90 2 range for magnetic extracts, using a 0.02 step-size and a 10-second

PAGE 175

175 count-time for each step. The results for bul k sediments were dominated by quartz and dolomite peaks (Figure 8-8A). Detrital ca rbonate layers with high dolomite contents have been noted as a lithostratigr aphic feature for sediments across the western Arctic Ocean (e.g., Polyak et al., 2009). Magnetic phases in the bulk sediments are not detectable from the XRD data due to low concentrati ons. The two magnetic extract samples show very similar diffraction patte rns. The major peaks are consistent with a synthetic magnetite standard, and the tit anomaghemite standard r eported from pillow basalts of mid-Atlantic Ocean (Xu et al., 1997). Two closely overlapping yet distinct peaks that correspond to magnetite (blue lin es in Figure 8-8) and titanomaghemite (red lines in Figure 8-8) can be recognized for all dominant diffraction planes (Figure 8-8B and 8C). Diffraction patterns of magnetic ex tracts also show peaks corresponding to quartz and clay minerals such as kaolinite, consistent with the EDS observation. Higher resolution diffraction patterns were collect ed for magnetic extracts within four 2 intervals (i.e., 29-31, 34-37, 42-44.5, 5265) that cover the dominant peaks of magnetite and titanomaghemite associat ed with the [2 2 0], [3 1 1], [4 0 0], [4 2 2], [5 1 1], and [4 4 0] diffraction planes. A 0.01 step-size and a 20-second count-time were used for the analyses. It is clear that the observed peaks on these diffraction planes comprise two distinct peaks that fit (tit ano)magnetite and (titano) maghemite (Figure 88D~G). Calculated lattice parameters (Table 82) using the higher resolution XRD data give an average value of 8.3945 for the (titano)magnetite component and 8.3647 for the titanomaghemite component, indicating high oxidation states with z > 0.9 (Readman and O'Reilly, 1972; Nishitani and Kono, 1983).

PAGE 176

176 Discussion Thermal demagnetization of the NRM implie s that negative inclination components resolved by AF demagnetization are carried by a magnetic phase that has blocking temperatures largely below 350C. Rock magnetic studies, SEM and EDS observations, and XRD analyses on sediment s and magnetic extracts re veals the presence of titanomaghemite and (tit ano)magnetite in these Arctic sediments, consistent with the results from Core 06JPC (Channell and Xuan, 2009). The negativ e NRM inclination component is apparently carried by the tita nomaghemite that has slightly higher coercivity (AF demagnetization) than (t itano)magnetite, and blo cking temperatures (thermal demagnetization) largely below 350 C due to inversion upon heating. Typical AF and thermal demagnetization behaviors observ ed in Figure 8-3 (e.g. that of sample 10JPC 2.83 m) can be explained using a conceptual model in which coercivity and blocking temperature spectr a for titanomaghemite and (t itano)magnetite components are represented as being distributed as gaussian functions (Figure 8-9). Note that the low-coercivity and the low-blocking temper ature part on the spectr a may be susceptible to VRM. Titanomaghemite is known to have a propensity for acquisition of VRM (e.g., zdemir and Banerjee, 1981), a nd Arctic sediment has been noted for their ability to acquire VRM that often resists moderately high peak field (up to >20 mT) AF demagnetization (e.g., Lvlie et al., 1986; Witte and Kent, 1988). The low coercivity (generally <20 mT) and low blocking tem perature (<175C) component with positive inclinations (Figure 8-3) may be a VRM or detrital remanent magnetization (DRM) carried by the original (titano)magnetite with la rger grain-sizes (lower coercivities), or more likely, a mixture of t he two types of remanences.

PAGE 177

177 The excessively large thickness (up to >40 cm) of negative inclination (AF-derived) intervals in the studied cores and the lack of a comparable patterns from core to core (Figure 8-2) means that the observed negative inclinations cannot be easily ascribed to geomagnetic excursions. The negative inclination component carried by titanomaghemite could originate as a self-reversed CRM formed during the lowtemperature oxidation of the original detrital (titano)ma gnetite grains to titanomaghemite (Channell and Xuan, 2009). The possibility of self-reversed CRM in titanomaghemite has been discussed since 1950s. It has been suggested (Verhoogen, 1956, 1962; O'Reilly and Banerjee, 1966) t hat self-reversal can be acco mplished by ionic reordering of the two sublattices in ti tanomagnetite during extr eme low-temperature oxidation. Prior to oxidation, the B (octahedr al) sublattice in titanomagnetite has a higher spontaneous magnetization than the A (tetrahedral) sublattice. During th e oxidation, cation vacancies only form in B sublattice, and the inverse magnetization of the A sublattice could eventually become the stronger, causing the self-reversal. Partial and complete selfreversals have been reported in oceanic basalts (Doubrovi ne and Tarduno 2004, 2006a), carried by titanomaghemite with N-type thermomagnetic properties, and were attributed to CRM acquired durin g the oxidation of titanomagnetite by ionic reordering. Theoretical models of the maghemitization process (O'R eilly and Banerjee, 1966) and studies on compositions of titanomaghemite with/ without self-reversal observations (Doubrovine and Tarduno, 2005, 2006a, 2006b) suggest that hi gh oxidation states ( z 0.9) and relatively high Ti contents ( x 0.6) are required to produce natural selfreversed components. Self-reversal has also been reported in continental basalts containing oxidized titanomaghemite (Krsa et al., 2005). Reproduction of the self-

PAGE 178

178 reversals in laboratory thermoremanence l eads the authors to preclude any long term process such as ionic reordering as the responsible mechanism. The authors explained the observed self-reversals by magnetic coupling between the two magnetic phases with different blocking temperatures (i.e., titanomagnetite and oxidized titanomaghemite), in a close side-by-side assemblage. However, such a mechanism often requires specific geometry and magnetic properties of the associated magnetic phases, and the formed self-reversed component is carried by the magnetically softer titanomagnetite that has lower Curie temperatures, inconsist ent with the observations in the Arctic sediments studied here. Although relatively low mean Ti content s with estimated mean x values of 0.35 and 0.45 were obtained from t he EDS analyses, the existenc e of a population of Ti-rich iron oxide grains, presumably ti tanomaghemite, with elevated x values up to 0.85 is also apparent (Figure 8-7). Lattice parameters of the titanomaghemit e in these Arctic sediments, estimated from XRD data (Figure 8-8, Table 8-2), indicate high oxidation state (z > 0.9). These observations appear to favor the explanation of the self-reversed component carried by titanomaghem ite being caused by ionic reordering. It is likely that the population of low-temper ature oxidized titanomaghemite meets the requirements associated with the ionic reordering mec hanism, and therefore could have formed a self-reversed CRM from the thermal rem anent magnetization (TRM) carried by the original (titano)magnetite that contributed to the original DRM of the sediments. Statistically, the grains that carry the self-reversed CRM w ould apparently alter (distort) the original DRM to a degree that depends on the abundance of affected grains, causing the observed negative inclinati on intervals and the generally shallow

PAGE 179

179 inclinations. Stratigraphic correlations of the negative inclinat ion intervals among different cores from the Arct ic Ocean (e.g., Spielhagen et al., 2004; Backman et al, 2004; Kaufman et al., 2008; Adle r et al., 2009; Polyak et al., 2009) indicate a primary lithological control on the degree of alterati on, and hence on the position of zones of negative inclination. Conclusions Several lines of evidence lead to the recognition of titanomaghemite and (titano)magnetite in these Arct ic sediments: 1) Ms(T) curv es show abrupt drop below 300C and 600C; 2) a dramatic increase of magnetization during cooling of samples heated to 700C; 3) manifestation of a suppre ssed Verwey transition in low temperature measurements including room temperature SIRM on cool ing to 20 K and subsequent warming to room temperature, and FC ZFC curves measured on warming from 20 K to room temperature. Decomposit ion of the gradient of IRM acquisition curves measured from magnetic extract samples is consist ent with the presence of titanomaghemite and (titano)magnetite that hav e slightly different coercivities Elemental maps collected using SEM and EDS on micron-sized grains of magnet ic extracts confirm the existence of titanomaghemite /(titano)magnet ite with various Ti contents. High resolution X-ray diffraction patterns of the magnetic extracts fit well with titanomaghemite and magnetite standard samples. Thermal demagnetization of NRM indicates that negativ e inclination components in these Arctic sediments are carried by titanomaghemite that has blocking temperatures largely below 350C due to in version. SEM and EDS observation of high Ti content titanomaghemite /( titano)magnetite grains with es timated x values of up to 0.85 and high oxidation state with z > 0.9 estimated using lattice parameters calculated

PAGE 180

180 from the XRD data, provide the conditions for a certain proportion of titanomaghemite grains in the Arctic sedim ents to undergo self-reversal by ionic reordering. Negative inclination intervals and the generally shallow in clinations in these studied Arctic cores resulted from modifications of the original DRM by a self-reversed CRM carried by titanomaghemite formed during diagenesis from host (titano)magnetite grains.

PAGE 181

181 Table 8-1. Location, length, wa ter depth, and age model information for cores studied in this chapter. Core Location Latitude Longitude Length Water depth Age constraints HLY0503-08JPC MR 79.593N 172.502W 11.88 m 2792 m [1-3] HLY0503-10JPC MR 81.226N 177.194W 12.72 m 1865 m [3] HLY0503-11JPC AR 83.144N 177.194W 10.19 m 2644 m [3] HLY0503-13JPC AR 84.306N 160.680W 12.00 m 1400 m [3] MR denotes Mendeleev Ridge, AR denotes Alpha Ridge. [1]: Kaufuman et al., 2008; [2]: Adler et al., 2009; [3] Polyak et al., 2009.

PAGE 182

182 Table 8-2. 2 values and estimated lattice parameters (L.P.) for (titano)magneti te and titanomaghemite derived by modeling of the 6 high-resolution X -ray diffraction peaks in Figure 8-8D~G. Components Parameters Diffraction plane Mean [2 2 0] [3 1 1] [4 0 0] [4 2 2] [5 1 1] [4 4 0] (titano)magnetite 2 () 30.0595 35.4414 43.0767 53.4418 56.9510 62.5559 L.P. () 8.4015 8.3933 8.3926 8.3924 8.3948 8.3926 8.3945 titanomaghemite 2 () 30.1830 35.5376 43.2078 53.7190 57.1725 62.7973 L.P. () 8.3679 8.3713 8.3683 8.3523 8.3650 8.3636 8.3647

PAGE 183

183 Figure 8-1. Location of Cores 08JPC, 10JPC, 11JPC, and 13JPC retrieved by the HOTRAX05, in comparison with location of previously studied nearby cores. LR denotes Lomonosov Ridge, MR denotes Mendeleev Ridge, and AR denotes Alpha Ridge. Base map data is from international bathymetric chart of Arctic Ocean (IBCAO, Jakobsson et al., 2008). Map is processed using the GeoMapApp software. References for pr eviously studied cores listed on the map are as following. NW5: Poore et al., 1993; Core 4: Phillips and Grantz, 1997; T3-67-11: Herman, 1974; Sejrup et al., 1984; T3-67-6 and T3-67-12: Hunkins et al., 1971; Witte and Kent, 1988; FL224: Steuerwald et al., 1968; Clark et al., 1980; FL270 and FL228: Clark et al., 1980; FL196 and FL199: Clark et al., 1984; PS2180 and PS2178: Nowaczyk et al., 2001; LOREX-B8 and LOREX-B24: Morris et al., 1985; 96/12-1PC: Jakobsson et al., 2000; PS2185-6: Spielhagen et al., 1997; ACEX Sites: Backman et al., 2008.

PAGE 184

184 Figure 8-2. Component in clination and declination wit h maximum angular deviation (MAD) calculated for the 20-80 mT peak alternating field range for Cores 08JPC, 10JPC, 11JPC, and 13JPC. Results calculated using data from XYZ and ZXY demagnetization sequence for th e three sample axes are in blue and red, respectively. Note that Core 13JPC samples were measured using only the XYZ demagnetization sequence. De clination values are arbitrary as cores were not oriented in azimuth. Gray shaded areas indicate intervals where magnetic extracts were made. R ed triangles represent depth levels where thermally demagnetized samples (Figure 8-3) were collected. Green triangles indicate depth levels where lo w temperature data (F igure 8-5) were acquired and orange triangles show dept h levels where high temperature data (Figure 8-4) were obtained. Vertical green lines are expected inclinations for a geocentric axial dipole field at the coring sites.

PAGE 185

185 Figure 8-3. Orthogonal projection of t hermal demagnetization (orange heading) for discrete cubic (~2x2x2 cm3) samples from Cores 08JPC, 10JPC, 11JPC, and 13JPC, compared to or thogonal projection of alternating field demagnetization (blue heading) for u-channel intervals from the same depth level. Circles (red) and squares (b lue) denote projection on vertical and horizontal planes, respectively. Declinatio n values are arbitrary as cores were not oriented in azimuth. Unit for in tensity scale is mA/m. Meter levels correspond to meter levels of each core as in Figure 8-2.

PAGE 186

186 Figure 8-4. A) Saturation magnetization (Ms) and B) Coercivity of remanence (Hcr) derived from hysteresis loops (saturation field of 0. 4 T) measured at increasing temperatures (in 25 C steps in an helium atmosphere) using a vibrating sample magnet ometer (VSM). Values have been normalized to saturation magnetization at room temperature. C) M agnetization in a 0.4 T applied field measured for two samples from Core 08JPC during heating and cooling. Magnetizations dur ing heating were derived fr om the hysteresis loops measured at 25-600C in 25C steps. M agnetizations during cooling were measured at each ~1C step. Refer to Figure 8-2 for positions of samples from each core.

PAGE 187

187 Figure 8-5. A) Saturation isothermal rem anent magnetization (SIRM), acquired in a 2.5 T field at room temperature, on cooling (open symbols) and warming (closed symbols); measured using a Quantum Designs Magnetic Properties Measurement System (MPMS). Meas urements have been normalized to the room temperature magnetizat ion of the samples on coo ling. B) Field (2.5 T) cooled (FC, open symbols), and zero field cooled (ZFC, closed symbols), lowtemperature SIRM measured on warming. Measurements have been normalized to the field cooled magnet ization of the samples at room temperature.

PAGE 188

188 Figure 8-6. IRM acquisition curves for magnet ic extracts from A) Core 08JPC and C) Core 10JPC, measured using an alter nating gradient magnet ometer (AGM), and 2-component modeling of the IRM gr adient data for magnetic extracts from B) 08JPC and D) 10JPC using the method of Heslop et al. (2002).

PAGE 189

189 Figure 8-7. Scanning electron microscopy (SEM) and X-ray energy-dispersive spectroscopy (EDS) analyses for micron-si zed grains of magnetic extracts from A) Cores 08JPC and B) 10JPC sediments. The image comprises (titano)magnetite and/or ti tanomaghemite grains with various Ti contents with x values up to 0.85, and clay mineral s. The total image spectrum acquired during the mapping is also displayed, where the car bon peak is attributed to the carbon tape background.

PAGE 190

190 Figure 8-8. X-ray diffraction (XRD) results for A) freeze dried bulk sediment powders from Core 08JPC (black), and magnetic extracts from B) Cores 10JPC (gray) and C) 08JPC (green), with (D~G) higher resolution scans around the major peaks associated with tit anomaghemite and (titano)magnetite. The 2 positions and magnitudes of XRD peaks for synthetic magnetite and titanomaghemite standards are indicated by vertical blue and red lines respectively.

PAGE 191

191 Figure 8-9. Conceptual models for the coer civity spectra (left) and the blocking temperature spectra (right) that can explain typical alternating field demagnetization behavior exhibited by or thogonal projections from Figure 83. In this model, the (titano)magnetite carries a primary magnetization (DRM) and the titanomaghemite carries a partially self-reversed CRM.

PAGE 192

192 CHAPTER 9 PALEOMAGNETIC AND ROCK MAGNETIC STUDIES ON DEEP-SEA SEDIMENTS FROM LOMONOSOV RIDGE AND YERMAK PLATEAU Introduction Conventional piston cores from the Arctic Ocean often re cord intervals of negative magnetic inclinations, particularly in top several meters of the sediment cores. The thicknesses of the negative incli nation intervals typically reach tens of centimeters. The negative inclination intervals have often been in terpreted as polarity chrons recorded in the sediments implying mm/kyr scale deposition rates (e.g., Steuerwald et al., 1968; Clark, 1970; Hunkins et al., 1971; Herman, 1974). When it became clear that cm/kyr sedimentation rates were more likely, the negat ive inclination intervals were attributed to magnetic excursions within the Brunhes Chron (e .g., Sejrup et al., 1984; Lvlie et al., 1986). Unusually long durations (>10 kyrs) for these apparent magnetic excursions and the absence of correlative magnetic excu rsions in sediments outside the Arctic region with much higher deposit ion rates implies that the observed negative inclinations represent special behavior of t he magnetic field in the Arctic area, or distortions of the paleomagnetic record. Studies on sediments from Cores 06, 08, 10, 11, and 13 retrieved by the 2005 Healy-Oden Trans-Arctic Expedition (H OTRAX05) to the Mendeleev-Alpha Ridge recognized titanomaghemite and (titano)magnet ite as the magnetic remanence carriers (Channell and Xuan, 2009; Xuan and Channell, 2010). Thermal demagnetization of the natural remanences (NRM) indicates that low and negative NRM inclinations in these sediments are partially carried by titanom aghemite that could hav e formed during sea floor oxidation from original titanomagnetite. The aut hors proposed that negative inclinations in cores from the Mendeleev-Alpha Ridge represent partially self-reversed

PAGE 193

193 chemical remanent magnetization (CRM). High Ti content s and high oxidation states indicated by the X-ray energy-dispersive spectroscopy (EDS) and X-ray diffraction (XRD) data seems to provide the conditions required for partial self-reversal by ionic reordering during diagenetic m aghemitization. This process appears to have affected all HOTRAX05 cores collected from the Mendeleev-Alpha Ridge. Thick negative NRM inclination intervals are also commonly observed in top several meters of cores recovered from the central Arctic Ocean and Lomonosov Ridge (e.g., Jakobsson et al., 2000; Nowaczyk et al., 2001; O'Regan et al., 2008), and close to the Yermak Plateau (Nowaczyk et al., 1994; Nowaczyk and Knies, 2000; Nowaczyk et al., 2003). These negative inclin ation intervals have often been interpreted as magnetic excursions. It is important to understand whether the negative inclinations reported in Brunhes-aged cores from these two areas represent genuine magnetic field behavior or whether they are due to partial self-reversal processes similar to that observed in the Mendeleev-Alpha Ridge cores. This issue affe cts the ACEX core (IODP Expedition 302) from Lomonosov Ridge where the chronology was partially built on the recognition of magnetic excursions (O'Regan et al., 2008). In this chapter, paleomagnetic and rock magnetic measurements as we ll as EDS and XRD analyses were employed to study HOTRAX05 Core 20 from the Lomonosov Ridge and Core 22 from the Yermak Plateau, to understand the origin of the negative incli nations in these sediments and the regional importance of the possible parti al self-reversal mechanism. Materials and Methods The 10.58-m long Jumbo Piston Core (JPC) 20 was retrieved from the Lomonosov Ridge at 88 48.36N, 163 34.78E, in 2654 meter water depth during the HOTRAX05 in September 2005 (Figure 9-1). Core 20 is most ly composed of brown to dark brown and

PAGE 194

194 lighter colored yellowish (olive, grayish) brown silty clays, clays, and sandy clays, with occasional ice rafted debr is (IRD). The 13.31-m long JPC 22 was recovered by the HOTRAX05 to the Yermak Plateau at 80 29.39 N, 7 46.14E, in 798 meter water depth (Figure 9-1), and the core mainly consists of grayish/greenish to dark grey silty clays. Continuous u-channel samp les (typically 2 cm3) were collected from working halves of Cores 20 and 22. NRM of u-channel samples was measured at 1-cm spacing before demagnetization and after alter nating field (AF) demagnetization at 14 steps in the 10-100 mT peak field range, using a 2G Enterprises cryogenic magnetometer designed to measure u-channel samples. For each demagnetization step, samples from Core 20 were m easured after two s eparate demagnetization sequences for the X, Y and Z sample axes to monitor any spurious anhysteretic remanence (ARM) acquisition during AF demagnetization. The first XYZ demagnetization sequence was followed by measurement, and then the ZXY demagnetization sequence was followed by repeat measurement. As no discernable affect was observed with differing or der of demagnetization, only the XYZ demagnetization sequence was applied to sa mples from Core 22. Susceptibility measurements were carried out using a susc eptibility track desi gned for measuring uchannel samples (Thomas et al., 2002). ARM wa s acquired in a peak AF of 100 mT and a 50 T DC bias field, and measured prior to demagnetization and after demagnetization at 9 steps during the 20-60 mT peak field range. Discrete (2 cm3) samples were collected fr om Cores 20 and 22 in cubic plastic boxes alongside the u-channel samples for NRM measurements during thermal demagnetization. The discrete samples were dr ied in a magnetically shielded space

PAGE 195

195 with flowing helium gas before being taken out from the plastic containers and wrapped in Al foil for thermal treatment. Magnetizations of the discrete samples were then measured after thermal treat ments in 25C steps in the 50-600C temperature range. Ten discrete samples (approximately 2 cm3) were subsampled from the uchannels of Cores 20 and 22, after u-channel measurements were completed. The samples were from intervals that are char acterized by typical positive and negative AFderived NRM inclinations. Samples were firs t dried and wrapped in Al foil in preparation for a 3-axis isothermal remanent magnet izations (IRM) thermal demagnetization experiment, following the method of Lowrie (1990). IRM was acquired in DC fields of 1.2 T, 0.5 T, and 0.1 T sequentially along three orthogonal axes of the samples, and then the composite IRM was measured before t he thermal treatment and after thermal treatments in 25C steps in the 50-600C tem perature range. Magnetic extracts were made using sedime nts from the 0.91-1. 02 m, 3.64-3.74 m, 3.93-4.05 m, and 4.86-4.97 m depth interval s of Core 20, and from the 0.67-0.77 m, 1.01-1.13 m, 1.93-2.06 m, and 3.43-3.55 m depth intervals of Core 22. The extracts were made by dispersing ~2 cm3 sediment (~45 g) ta ken from the u-channels (after finishing u-channel measurements) in a sodium metaphosphate solution of 4 wt.%, and using an automated extraction system to separate magnetic particles. The extraction system is equipped with a peristalt ic pump so that the sediment slurry can repeatedly flow next to a glass tube filled wi th strong rare earth magnets. The acquired extracts were then intermittently washed into a glass container using distilled water until no further extraction was achieved. A string of rare earth magnets in a glass test tube

PAGE 196

196 was then used to refine the extract. Extracts were again washed using distilled water, then methanol. A proportion of the magnetic extract samp les were used for the IRM acquisition experiments. IRM of the ex tracts was measured at one hundr ed equidistant field steps on a logarithmic scale ranging from ~7 mT to 1 T, using an alternating gradient magnetometer (AGM). Select ed magnetic extracts were spread on a carbon tape and examined under a Zeiss EVO scanning el ectron microscopy (SEM) equipped with Genesis EDS. Elemental maps were collected on micron-sized grain clusters in the extract samples for up to 20 hours using the EDS. XRD analyses were performed on the magnetic extracts using a Rigaku Ultima IV X-ray diffractometer. The extracts were placed on a zero background sample holder, and diffraction patterns were collected within four 2 intervals (i.e., 29-31, 34-37, 42-44.5, 52-65) that cover the dominant peaks of standard magnetite and ti tanomaghemite associated with t he [2 2 0], [3 1 1], [4 0 0], [4 2 2], [5 1 1], and [4 4 0] diffraction planes. A 0.01 step-size and a 20-second count-time were used for the analyses. Susceptibility of the magnet ic extract samples were monitored on heating from room temperature to 700C and subsequent cooling to room temperature, in an ar gon gas environment, using a KLY-3S Susceptibility Bridge equipped with heater. For two extract samples (i.e., ex tracted from 0.91-1.02 m and 4.86-4.97 m depth intervals) from Core 20, a second heating and cooling curves were measured. Results and Discussions From u-channel measur ements, component declinat ion and inclination with maximum angular deviation (MAD) were calc ulated from the NRM measurements for the 20-80 mT demagnetization peak field range, using the principal component analysis

PAGE 197

197 method (Kirschvink, 1980). For Core 20 from the Lomonosov Ridge, MAD values are generally less than 10, indicati ng that the component directions are reasonably well defined (Figure 9-2). The inclinat ion record of Core 20 is characterized by several thick (up to ~50 cm) intervals of lo w/negative inclinations in top 10 m of the core, with the first significant inclination drop occurring at ~3.4 m depth level. In addition, component inclinations of Core 20 are generally tens of degrees lower than t he expected inclination (vertical green line in Figure 9-2) for a geocent ric axial dipole field at the location of Core 20. There are no significant differences in component directions calculated using data from the XYZ (blue lines in Figure 9-2) and the ZXY (red lines in Figure 9-2) demagnetization sequences. Theref ore, low/negative inclinati ons are not be caused by spurious ARM acquisition during AF demagnet ization. These observations are very similar to those of the Mendeleev-Alpha Ri dges cores studied by Channell and Xuan (2009), and by Xuan and Channell (2010). For Co re 22 from the Yermak Plateau, MAD values associated with the component direction calculation are mostly less than 5 (Figure 9-2), which is much lower than that of Core 20 and the Mendeleev-Alpha Ridge cores. Component inclination of Core 22 appears to be less shifted from the expected geocentric dipole field value, however, four 10-30 cm thi ck shallow/negativ e inclination intervals can be observed at ~1 m, ~2.5 m, ~3.5 m, and ~5.5 m depth level of this core (Figure 9-2). Component inclinations of Core 20 appear to be comparable to inclinations observed in other Lomonosov Ridge cores includ ing 96/12-1 (Figure 3 of Jakobsson et al., 2000), 2185 (Figure 3 of Spielhagen et al., 2004), and ACEX (Figure 3 of O'Regan et al., 2008). Ages of sediments from the ce ntral Arctic Ocean provided by correlating

PAGE 198

198 manganese and color cycles to lower latitude 18O records (Jakobsson et al., 2000), optically stimulated luminesc ence dating (Jakobsson et al., 2003), as well as sporadic biostratigraphic markers such as E. huxleyi that has late Brunhes age, usually yield >1 cm/kyr sedimentation rates of the area during the late Brunhes. This is consistent with the long-term (last 17 Ma) >1 cm/kyr sedimentation rates derived from the10Be/9Be age model (Backman et al., 2008; Frank et al., 2008) Therefore, it is reasonable to expect top several meters (e.g., 5 m) of Core 20 to be within the Brunhes Chron, restricting the low/negative inclinations to Brunhes aged magnetic excursions or distortion of the paleomagnetic record. For instance, the first apparent drop in inclination at ~3.4 m in Core 20 was commonly observed in other central Arctic cores, and was often interpreted as the Biwa II excursion that has an assumed age of ~240 ka (e.g. O'Regan et al., 2008). Component inclinations in Yermak Plateau Core 22 mimic those of nearby cores 1533 and 2212 (Figure 4 of Nowaczyk et al ., 1994), and core 2138 (Figure 4 of Nowaczyk and Knies, 2000; Figure 3 of Nowa czyk et al., 2003), which have been dated using radiocarbon, oxygen isotope stratigrap hy, and relative paleointensity correlations. In these cores, 3-4 magnetic excursions were reported with durations often reaching 1020 kyrs (i.e., Figure 11 of Nowaczyk et al., 1994) after adopting the age models. These durations far exceed excursi on durations estimated from outside the Arctic where excursion durations are typically < 5 kyrs (see Laj and Channell, 2007). The magnetic grainsize proxy, ARM/ the ratio of ARM susceptibility (ARM intensity divided by the ARM bias field) to susceptibility, was ca lculated for Core 22 using u-channel ARM and susceptibility measurements (Figure 9-2). The ARM/ curve of Core 22 apparently

PAGE 199

199 resembles an oxygen isotope record, with finer grains (larger ARM/ ) occurring in interglacial intervals. For instance, the ARM/ peak intervals at ~25 cm and ~5.6 m could be correlated to late Holocene and marine is otope stage (MIS) 5, respectively. A similar correlation of the ARM/ magnetic grain size proxy to 18O has previously been noticed by Nowaczyk and Knies (2000) in nearby core 2138. A tentative age model was constructed by correlating Core 22 ARM/ curve to the PI SO-1500 oxygen isotope stack record (Channell et al., 2009) for t he last ~160 kyr (Figure 9-3), using the automated dynamic correlation method by Lisi ecki and Lisiecki (2002). It seems clear that the four low/negative incl ination intervals in Core 22 can be constrained to the last ~130 kyrs, and low/negat ive inclination intervals appear to occur at paleointensity minima (Figure 9-3). For both Cores 20 and 22, thermal and AF demagnetization of the NRM for samples from the sample depth level show comparable behavior (Figure 9-4), with characteristic components generally revealed after 50-175C thermal treatments or after 20-30 mT peak field AF demagnetization, res pectively. Thermal demagnetization data usually show hints of overlapping component s on high temperature steps (300-600C) for samples that have either negative (e.g., 22JPC 1.98 m, 20JPC 7.62 m) or shallow positive (e.g., 22JPC 2.15 m) AF-derived inc linations. Furthermore, NRM intensities monitored during thermal demagnetization (bottom plots in Figure 9-4) clearly show that the negative inclination or s hallow positive inclination components are often largely unblocked below ~300C. These observations i ndicate that shallow/ negative inclinations in Cores 20 and 22 are carried by a magnetic phase that has unblocking temperatures largely below ~300C. Similar results were reported by Channell and Xuan (2009), and

PAGE 200

200 by Xuan and Channell (2010) for the Mendeleev-Alpha Ridge cores, in which low/negative inclinations were interpreted as being partially carried by a self-reversed CRM carried by diagenetic tit anomaghemite that has unblocking temperatures (due to inversion) largely below ~300C. Thermal demagnetization of the three-orthogonal-axis IRM (Figure 9-5) shows that samples from both Cores 20 (in red) and 22 (in blue) are dominated by soft (< 0.1 T) and medium (0.1-0.5 T) coercivity magnet ic phases. The abrupt drops below ~300C, presumably due to the inversion of titanomaghemite, can be observed on both soft and medium coercivity component IRM, and are dominant on medium coercivity component IRM, indicating that titanomaghemite grains in these cores have various coercivities, possibly due to variations in grain size and/or oxidation states, and that medium coercivity (0.1-0.5 T) magnetic phases are mostly titanomaghemite grains. A difference between samples from intervals of typical positive and negative AF-d erived inclinations is not apparent in either core. This is pr obably because samples from positive (usually shallow) and negative inclination interval s have all been subjected to diagenetic maghemitization, and the incli nation of a sample depends not only on the amount of grains that have experienced oxidation (hence distortion of the NRM directions), but also on the statistic orientation and oxidation states of th ose grains. On the other hand, samples from Core 20 (in red) clearly have higher proportion of the medium coercivity magnetic phase (Figure 9-5), and therefore more titanomaghem ite grains, than Core 22 samples (in blue). This observation is c onsistent with the fact that component inclinations of Core 22 appears to be less sh ifted from the expecte d geocentric dipole field value than those of Core 20, and MAD values associated with component

PAGE 201

201 directions from Core 22 are also much sm aller than those of Co re 20 (see Figure 9-2). Core 22 is from the Yermak Pl ateau, at the edge of the Arctic Ocean, with water depth of 798 m. It is likely that seaf loor oxidation of sediments from Core 22 is less developed at this location of the Arctic Ocean, generating less alteration of the magnetic remanences. IRM acquisition data of magnetic extracts from Cores 20 and 22, plotted on logarithmic scale, show distin ct features (Figure 9-6), i ndicating different magnetic contents in sediments from the two cores. Similar to the Mendeleev-Alpha Ridge cores studied by Xuan and Channell (2010) the gradient of the IRM acquisition data of Core 20 magnetic extracts show apparent asymmetry in shape, and satisfactory fits were achieved by modeling the IRM gradient data using two magnetic coercivity components (method of Heslop et al., 2002). Analyses of three magnetic extract samples from Core 20 are highly comparable to one another (Figure 9-7), yielding mean coercivities of ~55 mT and ~108 mT for the two magnetic component s, consistent with the presence of (titano)magnetite and titanomaghem ite, respectively. As discussed in a previous chapter (Xuan and Channell, 2010), coercivities of ti tanomaghemite are expected to be slightly higher than those of (titano)magnetite. For Core 22 magnetic extracts, the gradient of the IRM data show less asymme try (bottom plots of Figure 97). Reasonable fits seem to be obtained by modeling the IRM gradien t data using one magnetic coercivity component with mean coercivity of ~32 mT although small mismatches are also observable between the IRM gradient data and the fitting curves. The results are repeatable for magnetic extracts made from the three do minant negative inclination intervals in top 4 meters of Core 22 (Figur e 9-7). The absence of apparent asymmetry in

PAGE 202

202 IRM gradient data for Core 22 magnetic extrac t samples is probably due to the following reasons: 1) there is less titanomaghemite in Co re 22 sediments, as suggested by the 3axis IRM thermal demagnetization data; 2) magnetic extracti on can be grainsize selective, and tend to pull out coarser grains while finer grains are more likely to be oxidized titanomaghemite; 3) only a small propor tion (few mg) of the magnetic extracts was used for the IRM acquisition measurem ents and the sample was inhomogeneous. Susceptibility of the eight magnetic extract samples from Cores 20 and 22, monitored on heating from room temperature to 700C, show apparent humps at around 300C followed by an abrupt drop in susceptibil ity (red curves in Figure 9-8A and B). These features are, however, not obs erved in susceptibil ity measured during subsequent cooling from 700C to room tem perature (blue curves in Figure 9-8A and B). The humps are probably related to ti tanomaghemite inversion. As discussed in Chapter 8, titanomaghem ite when heated up to 600C typically leads to intergrowths of minerals including magnetit e and ilmenite, without substantia l increase in susceptibility thereby explaining the absence of humps in the cooling curves. In addition, the humps are not seen in repeated heating and cooling curves of susceptibility (Figure 9-8C and D) measured for two magnetic extract sa mples from Core 20 that had been heated up to 700C and cooled to room temperature. The repeated heating and cooling curves of the two samples show typical magnetite behavio rs and seem to be reversible (Figure 98C and D), consistent with magnetite bei ng the magnetic inversion product from titanomaghemite during the fi rst heating and cooling of the samples. The other observation from the high temper ature susceptibility data (Fi gure 9-8A and B) is that the humps in the heating curves appear to be more apparent in core 20, indicating more

PAGE 203

203 titanomaghemite in this core. This is consistent with the 3-axis IRM thermal demagnetization data, and apparent absence of asymmetry in t he IRM gradient data for Core 22 magnetic extracts. From SEM/EDS observations, the total im age spectra of magnetic extract samples from Cores 20 and 22, collected during the el emental mapping, are comparable to each other with dominant peaks for O, Ti, and Fe (Figure 9-9). The elemental maps clearly show a number of grains that are rich in O, Ti, and Fe, c onsistent with the presence of (titano)magnetite and/or tit anomaghemite in these two cores. Ti compositions ( x) of these grains were estimated using the simple equation: x = 3/(1+(Fe/Ti)), where Fe/Ti is the atomic ratio estimated from the EDS data (see Doubrovine and Tarduno, 2006a). The Ti composition varies widely from one grain to another, with calculated x values ranging from 0.33 to 0.95 for Core 20 extrac t sample (Figure 9-9A), and from 0.43 to 0.89 for Core 22 extract sa mple (Figure 9-9B). From XRD data, magnetic extract samples from Core 20 show two distinct peaks that fit magnetite and tit anomaghemite corresponding to a ll six diffraction planes associated with the dominant peak locations of tit anomaghemite and magnetite standards (Figure 9-10). For ma gnetic extract samples from Core 22, the diffraction patterns are dominated by peaks that fi t themagnetite standard. Titanomaghemite peaks are not apparently observable in the diffraction data from Core 22 magnetic extract samples. The XRD data of the two cores agree well with the results from the other experiments, showing that there is less titanomaghemite in Core 22. The generally shallow inclinat ions (tens of degree lower than the expected dipole value at core locations), and the unusually th ick durations of the low/negative inclination

PAGE 204

204 intervals means that the paleomagnetic reco rd has been distorted. Similar to the explanation for the Mendel eev-Alpha Ridge cores (Channell and Xuan, 2009; Xuan and Channell, 2010), low/negative in clinations may be partially carried by a self-reversed CRM in titanomaghemite formed dur ing the diagenetic oxidation of titanomagnetite. The reason why negative inclination intervals in Core 22 appears to occur at paleointensity minima (Figure 9-3) may be bec ause during the paleointensity minima, only finer grains can be aligned to the weak geomagnetic field, and finer grains are more likely to undergo maghemitization resulting in self-reversed CRMs. Conclusions Several lines of evidence indicate the presence of titanomaghemite and (titano)magnetite in Lomonosov Ridge Core 20 and Yermak Plateau Core 22, although Core 22 contains less titanom aghemite than Core 20: 1) t hermal demagnetization of the 3-axis IRM data for bulk samples show an abrupt drop below ~300C, and Core 22 has lower proportion of the medium coercivity component; 2) gradient of the IRM acquisition data for magnetic extract samples show a symmetries in shape, and the asymmetries are less apparent in Core 22 samples; 3) humps were observed around 300C on heating curves of susceptibi lity and are not seen on cooli ng curves, and humps are less apparent for Core 22 samples; 4) elemental maps collected using EDS on micron-sized grains of magnetic extracts are consist ent with the existence of titanomaghemite /(titano)magnetite with various Ti contents; 5) high resolution XRD patterns of the magnetic extract samples from Core 20 fit well with titanomaghemite and magnetite standards, XRD data of core 22 samples ar e dominated by (tit ano)magnetite peaks and titanomaghemite peaks are not apparently observable.

PAGE 205

205 Thermal demagnetization of the NRMs indica tes that low/negative inclinations in the two cores are carried by a titanomaghemite that has unblocking temperatures largely below ~300C. Low/negat ive inclinations in Core 20 and 22 are probably caused by partially self-reversed CRMs formed dur ing seafloor diagenetic maghemitization, similar to those of the M endeleev-Alpha Ridge cores. Core 22 from the Yermak Plateau contains less titanomaghemite and was less altered, indicating the partial self-reversal processes are less apparent in the Yermak Strait than on the Lomonosov Ridge.

PAGE 206

206 Figure 9-1. Location of Cores 20JPC and 22 JPC (red solid circles) retrieved by the HOTRAX05, in comparison with location of previously studied cores. LR denotes Lomonosov Ridge, MR denotes Mendeleev Ridge, AR denotes Alpha Ridge, and NR denotes Northwind Ridge, YP denotes Yermak Plateau. References for previously studied core s listed on the map are as following. 06: Channell and Xuan, 2009; 08, 10, 11, and 13: Xuan and Channell, 2010; 2178 and 2180: Nowaczyk et al., 2001; 96/12-1: Ja kobsson et al., 2000; 2002; 2003; 2185: Spielhagen et al., 1997; 2004; ACEX: Backman et al., 2008; O'Regan et al., 2008. 1533 and 2212: Nowaczyk et al., 1994; 2138: Nowaczyk and Knies, 2000; Nowaczyk et al ., 2003; 1535: Nowaczyk et al., 2003.

PAGE 207

207 Figure 9-2. Component in clination and declination wit h maximum angular deviation (MAD) calculated for the 20-80 mT peak alternating field range for Cores 20JPC (Lomonosov Ridge) and 22JPC (Yermak Plateau), and magnetic grain size proxy ARM/ for Core 22JPC calculated using u-channel measurements. Results calculated using data from XYZ and ZXY demagnetization sequences for the three sample axes are in blue and red, respectively. Note that Core 22JPC samples were measured using only the XYZ demagnetization sequence. Declination values are arbitr ary as cores were not oriented in azimuth.

PAGE 208

208 Figure 9-3. Component inclination and ARM/ data from Core 22JPC compared to PISO-1500 oxygen isotope and relative paleointensity stack records (Channell et al., 2009) for the last ~ 160 kyr. Age model for Core 22JPC was acquired by correlating ARM/ record from 22JPC to PISO-1500 oxygen isotope stack record using automated correlation method (Lisiecki and Lisiecki, 2002).

PAGE 209

209 Figure 9-4. Orthogonal projection of t hermal demagnetization for discrete cubic (~2 cm3) samples from Cores 22JPC and 20JPC, compared to orthogonal projection of alte rnating field demagnetizat ion (blue labels) for uchannel intervals from the same depth le vel, with NRM intensity variations during thermal demagnetization display ed. Peak demagnetizing field ranges are 10-60 mT in 5 mT steps then 60100 mT in 10 mT steps. Temperature ranges are 50-600C in 25C steps for all samples. Circles (red) and squares (blue) denote projection on vertical and horizontal planes, respectively. Declination values are arbitrary as core s were not oriented in azimuth. Meter levels correspond to meters below sea floor (mbsf) of each core as in Fig. 2. Unit for intensity scale is mA/m.

PAGE 210

210 Figure 9-5. Thermal demagnetization of three-axis isothermal remanent magnetizations (IRM) imposed orthogonally and sequentially in DC fields of 1.2 T (hard), 0.5 T (medium) and 0.1 T (soft), for sample s collected from intervals showing positive (+) and negative (-) component AF -derived inclinations in Core 22JPC and 20JPC. Results from Core 20JPC and 22JPC are in red and blue lines respectively. IRM values of eac h sample were normalized by the room temperature soft component IRM.

PAGE 211

211 Figure 9-6. IRM acquisition data for magnetic extracts from Cores 20JPC (squares linked with red lines) and 22JPC (cir cles linked with green lines).

PAGE 212

212 Figure 9-7. Two-component and one-component modeling of the IRM gradient data for magnetic extracts from Cores 20JPC (top plots) and 22JPC (bottom plots), respectively.

PAGE 213

213 Figure 9-8. Susceptibility monitored on heating from room te mperature to 700C and subsequent cooling to room temperat ure, in an argon gas environment, for magnetic extracts from A) 22JPC and B) 20JPC; and on repeated heating and cooling for magnetic extracts from C) 20JPC 0.9-1.02 m, and D) 20JPC 4.86-4.97 m. Heating (cooling) curves are in red (blue).

PAGE 214

214 Figure 9-9. SEM and EDS analy ses for micron-sized grains of magnetic extracts from A) Cores 20JPC and B) 22JPC s ediments. The image comprises (titano)magnetite and/or ti tanomaghemite grains with various Ti contents with x values up to >0.9. The total image spectrum acquired during the mapping is also displayed, where the carbon peak is attributed to the carbon tape background.

PAGE 215

215 Figure 9-10. High resolution XRD results fo r magnetic extracts from Cores 20JPC and 22JPC around the dominant peaks of standard titanomaghemite and magnetite associated with the [2 2 0], [3 1 1], [4 0 0], [4 2 2], [5 1 1], and [4 4 0] diffraction planes. The 2 positions and magnitudes of XRD peaks for synthetic magnetite and ti tanomaghemite standards are i ndicated by vertical blue and red lines, respectively.

PAGE 216

216 CHAPTER 10 CONCLUSIONS AND FUTURE WORK Assuming no uncertainty in reversal/excur sion ages and orbital solutions, Rayleigh tests indicate no preferred phase distribution for reversals/excursions on the obliquity cycle at the 5% significance level over the last 3 Myr. There appears to be a statistically significant (at the 5% level) relationship between reversal age and the phase of orbital eccentricity for the last 3 Myr, and reversals seem to preferentially occur during decrease of the maximum obliq uity envelope over the last 5 Myr. However, these relationships break down on adding just a fe w reversals beyond 3 Myr or 5 Myr. Monte Carlo simulations indicate t hat reversal/excursion ages w ould have to be known within 5-10 kyr to resolve a preferred phase in obliquity over the la st 3 Myr. Reversal/excursion ages would have to be known within ~15 kyr to resolve a preferred phase in orbital eccentricity for reversals over the last 3 Myr, and within ~40 kyr for the last 25 Myr. Comparison of astrochronological reversal timescales indicates that reversal age uncertainties exceed these limits, maki ng it unlikely that a relationship of reversal/excursion age to phase of obliquity or eccentricity, which might imply orbital forcing or the geodynamo, would be resolvable. Wavelet analyses reveal significant coher ence between RPI records (that contain orbital periods) and records of the magnetic grain size proxy ARM /IRM (from the same sediment sequences) on orbital periods, indica ting orbital periods in some RPI records are likely due to lithologic contamination re lated to grain size. Orbital periods may have been introduced into the NRM records ( and not normalized when calculating RPI records) through orbital control on bottom cu rrent velocity, which in turn controls magnetic grain size distribut ion at the deposition site or at sites of sediment

PAGE 217

217 provenance. RPI records from the Atlantic and Pacific oceans, and RPI records in which orbital periods have been filter ed by band-pass filters, are highly comparable with each other in the time domain, and are coherent and in-phase in time-frequency space, especially at non-orbital peri ods, indicating that contamination, although present (at orbital periods) is not debilitating to these RPI records as a global signal that is primarily of geomagnetic origin. Calibrated RPI and oxygen isotope stack records (PISO-1500) were developed by simultaneously matching and stacking both RPI and oxygen isotope data for 13 pairs of high-resolution global records. The simultane ous match reduces the degree of freedom associated with correlations using RPI or oxygen isotope records alone. The overall compatibility of RPI and oxygen isotopes i ndicates a dominant global (but independent) component in both signals. The PISO-1500 stack represents a new stratigraphic template that can be used to correlate among marine sediment records. Wavelet analyses on the PISO-1500 RPI stack record failed to show significant orbital periods, and no tendencies were found for RPI minima in the stack to occur at particular phases of orbital variations. UPmag software is introduced for easy and rapid processing of large volumes of paleomagnetic measurements for u-channel or other pass-through samples. The software comprises three MATLABTM graphic user interfaces: UVIEW, UDIR, and UINT. UVIEW allows users to open and check through measurement data from the magnetometer as well as to correct detected fl ux jumps in the data, and to export files for further treatment. UDIR reads file generated by UVIE W, automatically calculates component directions using selectable demagnetization range(s) wit h anchored or free

PAGE 218

218 origin on orthogonal projections, and displays vector component plots and stepwise intensity plots for any posit ion along the u-channel sample. UDIR can also display data on equal area stereographic projections and draw virtual geomagnetic poles on various map projections. UINT provides a convenient platform to evaluate re lative paleointensity (RPI) estimates using the files exported fr om UVIEW. Two methods are used for RPI estimation: the slopes of the best fit line between the NRM and the respective normalizer, and the averages of the NRM/no rmalizer ratios. R values and standard deviations can be calculated simultaneously to monitor the quality of the RPI estimates. All resulting data and plots fr om UPmag can be exported into various file formats. Continuous high resolution Quaternary paleo magnetic records, spanning the last ~1.5 Ma, were acquired from IODP Site U1304 sediments. Com ponent NRM directions clearly record the Brunhes/Matuyama boundary, the Jaramillo subchron, and the Cobb Mountain subchron, as well as the Kamikats ura excursion and the Gardar excursion in the Matuyama Chron. The age model for IODP Site U1304 is constructed by correlating the IODP Site U1304 RPI record to the PISO-1500 RPI stack using an automated dynamic programming method, with a few tie points provided by the available oxygen isotope data and the recorded magnetic reversal s and excursions at the site. The RPI record of IODP Site U1304 correlates well with the PISO-1500 RPI stack in both the time and time-frequency domain. Power spectr al analysis and wavelet analyses failed to detect significant orbital periods in the RP I record. A unique characteristic of the IODP Site U1304 sediments is the episodic deposition of laminated diatom ooze throughout the sediment sequence. The r apid deposition (high sedimentation rates) of laminated diatom ooze dilutes the magnetic concent ration of the sediments to about two

PAGE 219

219 magnitudes lower than that of sediments from non-diatom in tervals. Several lines of evidence indicate that diatom abundance, although yielding slight ly noisier results, does not significantly degrade the reliability of the paleomagnetic direction and intensity records: 1) MAD values are generally less than 5 for directions calculated from diatomrich samples, indicating well defined magnetiz ation components, as is also implied by orthogonal projections; 2) histograms of component inclinations (in the Brunhes) from diatom-rich and non-diatom intervals show si milar distributions; 3) Linear correlation coefficients (R values) associated with the RPI calculations are very close to 1, indicating high quality RPI estima tes; 4) RPI from diatom-ri ch intervals are comparable to the PISO-1500 RPI stack in the same time intervals. Component NRM inclination records of HOTRAX05 cores from the MendeleevAlpha Ridge (JPC 06, 08, 10, 11, and 13), the Lomonosov Ridge (JPC 20), and the Yermak Plateau (JPC 22) reveal several thi ck low and negative inclination intervals in top several meters of these cores that have Brunhes age. Rock ma gnetic experiments, SEM and EDS analyses, as well as XRD studies, on bulk sediments and magnetic extracts recognize titanomaghemite and (titano)magnetite as the magnetic remanence carriers in these sediments. Thermal NRM demagnetization indicates that low and negative NRM inclinations in these sediment s are partially carried by titanomaghemite that could have formed during sea floor ox idation from origin al titanomagnetite. Negative inclinations in cores from the Mendeleev-Alpha Ridge may represent partially self-reversed chemical remanent magnetiz ation (CRM). High Ti contents and high oxidation states indicated by EDS and XRD data seem to provide the conditions required for partial self-reversal by ionic reordering during diagene tic maghemitization.

PAGE 220

220 This process appears to have affected all t he studied HOTRAX05 cores. Results from 3-axis IRM thermal demagnetization on bulk sediments, susceptibility of magnetic extracts monitored on heati ng and cooling, and analyses on t he IRM acquisition data of magnetic extracts, as well as high resoluti on XRD studies, indicate that Core 22 contains less titanomaghemite and was less altered than t he other studied cores, indicating that the partial self-reversal proc esses are manifest to a lesser extent on the Yermak Plateau, located at the edge of the Arctic Ocean. Results from this dissertation work indicate that orbital periods in the RPI records are most likely due to lithologic contaminat ion. In other words, the RPI records, typically estimated by normalizing the NRM in tensity by the intensity of a lab-induced magnetization such as ARM or IRM, are not ideal proxies of paleomagnetic field intensity changes. It has long been realized that, in addition to the geomagnetic field strength and the concentration of magnetic phases, depositional remanence is also related to a number of other factors including magnetic mi neralogy, grain size of the magnetic phases, the nature of the non-magnetic matrix, and salinity/flocculation (e.g., Johnson et al., 1948; Levi and B anerjee, 1976; King et al., 1983; Tauxe, 1993). A purer proxy for paleointensity would not only help in further determining the level of orbital contamination in the RPI records and benefit our understanding of the geodynamo, but also provides better tools for stratigr aphic correlation. As pointed out by Levi and Banerjee (1976), the most appropri ate normalizer should activate the same grains that carry the NRM. Therefore, intensities dur ing demagnetization of NRM and of the ideal normalizer should track one another after being normalized by their respective values from certain demagnetization step (e.g., the first step). For mo st parts of the record at

PAGE 221

221 IODP Sites U1308 and U1304, the demagnetizat ion behavior of different normalizers such as ARM and IRM deviate from NRM in opposite directions when normalized and plotted against demagnetization step. ARM a nd IRM activate different grain size populations (e.g., Maher, 1988; Dunlop and Argyle, 1997; Dunlop and zdemir, 1997; Egli and Lowrie, 2002). ARM is more effective in activating finer magnetite grains than IRM. It is possible that grains that ca rry the NRM are sometimes better represented partly by ARM and partly by IRM. It woul d be reasonable to combine the ARM and IRM data in a way that the resu lting normalizer optimally mimics the NRM data after normalization. The validity of the com bined normalizer method can be tested by comparing the correlations between two (or more) RPI records fr om the same time interval before and after performing the combined normalization. If the 'combined' normalization method does improve the RPI estimates, we would expect to see an increased correlation between the two (or more) RPI records estimated using the 'combined' normalization methods during that time interval. At present, best age models for sedimentary RPI records (e.g., RPI record from IODP Site 1308, Channell et al., 2008) are generally acquired by correlating the oxygen isotope record at the site to a standard ox ygen isotope curve, and hence are subject to orbital assumptions, which may amplify the pres ence of orbital cycles in RPI records. An additional approach to the underst anding of orbital periods in some RPI records is to develop depth-derived age models using both RPI and oxygen isotope data from the sites. As has been suggested that the age of geological events identifiable in multiple stratigraphies may be estimated using me an sediment accumulation rates (e.g., Huybers and Wunsch, 2004; Huybers 2007). The depth-deriv ed age models serve to

PAGE 222

222 divorce RPI and oxygen isotope time-series fr om the possible influences of orbital cycles by not relying on orbital solutions as the basis for age models. The following factors contribute to the qua lity of such a depth-derived age model: the number of high quality individual records in dept h; the number of identifiable events in each record; the synchronicity of each event in all records; and the number of high quality age control points. Compared to depth-der ived age models, constructed using oxygen isotopes alone, the coupling of RPI and oxygen isotope records would introduce events at RPI minima that often feature magnetic reve rsals and excursions, in addition to the terminations and other events recognized from the oxygen isotope records. The larger variability of RPI on short timescales (mill ennial) and its global m anifestation also provide higher resolution and better synchronici ty of the identified events. In addition, magnetic reversals and well documented magnetic excursions of the last few Myr have been dated using radiometric me thods including radiocarbon, 40Ar/39Ar, and 230Th/238U, providing accurate age control points for th e construction of a higher resolution depthderived age model, and potentially a well-c alibrated template for stratigraphic correlation. There has been increasing interest on w hether there is any link between the variations in the Earth's magnetic field and climate change on various timescales (e.g. Courtillot et al., 2007; Pazu r and Winklhofer, 2008). Gallet et al. (2005) suggested a causal link between changes in geomagnetic field intensity and climate (and regime) changes over centennial time scales. St -Onge et al. (2003) suggested that the geomagnetic field may be the cause of millennialand even some centennial-scale cosmogenic nuclide flux variations seen in the Greenland ice cores. Bond et al. (2001)

PAGE 223

223 found a close correlation between changes in production rates of the cosmogenic nuclides and changes in proxies of drift ice measured in deep-sea sediment cores. The geomagnetic field may influence climate cycles through the shielding of galactic cosmic rays, and the influence of galactic cosmic rays on cloudiness and hence on climate. These are very topical questions that ar e at the threshold of an answer. The improvement in estimating pal eointensity variations and t he development of a depthderived age model using paleointensity and oxygen isotope records may allow detection of subtle leads and lags among aspects of the climate system, and between geomagnetic field variations and climate change proxies, and ultimately contribute to the understanding of these topics. RPI records dated using either traditional astronomical tuning or the above depthderived age model will have age uncertainties due to sedimentation rate changes (e.g., Guyodo and Channell, 2002), uncertainties in the age of tie point s, and magnetization lock-in effects (Kent 1973). For instance, according to Lisiecki and Raymo (2005), oxygen isotope stratigraphy has inherent uncerta inties of ~6 kyr from 1-3 Ma, and ~4 kyr since 1 Ma. When used to date an individual record, uncertainties usually propagate to several times that simply due to changes in accumulation rate (Guyodo and Channell, 2002; McMillan et al., 2002). Even close to Terminations, global correlations may be problematic. For example, Skinner and Shackleton (2005) have demonstrated a discrepancy of ~4 kyr between Terminati on I (MIS 1/2 boundary) recorded by 18O in the deep North Atlantic and deep equatorial Pacific, due to local deep-water 18O and deep-water temperature. Uncertainties in ages of the RPI re cord would obviously alter or even destroy the frequency signat ures of the real paleointens ity variation. It would

PAGE 224

224 be useful to estimate the possible range of age uncertainty a ssociated with the above mentioned processes and numerically demonst rate whether certain age uncertainty level(s) inhibit the detection of particular frequencies (i.e., orbital periods), assuming they are significant in the original signal. The UPmag software presented in this dissertation work will be continuously augmented and refined based on suggestions/co mments from the increasing population of UPmag users. For instance, deconvol ution of paleomagnetic measurements has been proved to improve the resolution of the acquired paleomagnetic direction and intensity data, overcoming the smoothing of measurements caused by the magnetometer response function (e.g., Oda and Shibuya, 1996; Guyodo et al., 2002). There have already been several requests fo r introduction of the deconvolution into UPmag. Hirokuni Oda (Geological Survey of Japan) will be visiting the University of Florida in February 2010 to help with this effo rt. It may also be possible to incorporate the new 'combined' normalization method for RPI estimates into the software, if and when the validity of the me thod is established.

PAGE 225

225 LIST OF REFERENCES Adler, R.E., Polyak, L., Ortiz, J.D., Kaufm an, D.S., Channell, J.E.T., Xuan, C., Grottoli, A.G., Selln, E., Crawford, K.A., 2009. Sedi ment record from the western Arctic Ocean with an improved Late Quaternary age resolution: HOTRAX core HLY05038JPC, Mendeleev Ridge. Global Planet. Change 68, 18-29. Backman, J., Jakobsson, M., Frank, M., Sang iorgi, F., Brinkhuis, H., Stickley, C., O'Regan, M., Lvlie, R., Plike, H., Spoffort h, D., Gattacecca, J., Moran, K., King, J., Heil, C., 2008. Age model and core-seismic integration for the Cenozoic Arctic Coring Expedition sediments from t he Lomonosov Ridge. Palaeoceanography 23, PA1S03. doi:10.1029/ 2007PA001476. Backman, J., Jakobsson, M., Lvlie, R., Po lyak, L., Febo, L.A., 2004. Is the central Arctic Ocean a sediment starved bas in? Quat. Sci. Rev. 23, 1435-1454. Backman, J., Moran, K., McInro y, D.B., Mayer, L.A., Expedi tion 302 Scientists, 2006. Proc. IODP. Integrated Ocean Drilling Pr ogram Management International, Inc., Edinburgh, p. 302. 10. 2204/iodp.proc.302.2006. Ballini, M., Kissel, C., Coli n, C., Richter, T., 2006. Deep-water mass source and dynamic associated with rapid climatic va riations during the last glacial stage in the North Atlantic: a multip roxy investigation of the detrital fraction of deep-sea sediments. Geochem. Geoph ys. Geosyst. 7, Q02N01. doi:10.1029/2005GC001070. Barton, C.E., 1982. Spectral-analysi s of paleomagnetic time-series and the geomagnetic spectrum. Philos. Trans R. Soc. Lond. A 306, 203-209. Baumgartner, S., Beer, J., Ma sarik, J., Wagner, G., Mey nadier, L., Synal, H.A., 1998. Geomagnetic modulation of the 36Cl flux in the GRIP ice core, Greenland. Science 279, 1330-1332. Benson, L., Liddicoat, J., Smoot J., Sarna-Wojcicki, A., N egrini, R., Lund, S., 2003. Age of the Mono Lake excursion and associated tephra. Quat. Sci. Rev. 22, 135-140. Beske-Diehl, S.J., Soroka, W.L., 1984. Magnetic properties of variably oxidized pillow basalt. Geophys. Res. Lett. 11, 225-228. Bianchi, G.G., McCave, I.N., 2000. Hy drography and sedimentation under the deep western boundary current on Bjorn and Gardar Drifts, Iceland Basin. Mar. Geol. 165, 137-169. Billups, K., Plike, H., Channell, J.E.T. Zachos, J.C., Shackleton, N.J., 2004. Astronomic calibration of the late Oligocene through early Miocene geomagnetic polarity time scale. Earth Planet. Sci. Lett. 224, 33-44.

PAGE 226

226 Bina, M., Tanguy, J.C., Hoffman, V., Prvot, M., Listanco, E.L. Keller, R., Fedr, K.T.h., Goguitchaichvili, A.T., Punongbayan, R. S., 1999. A detailed magnetic and mineralogical study of self-reversed dacitic pumices from the 1991 Pinatubo eruption (Philippines). Geophys. J. Int. 138, 159-178. Blackett, P.M.S., 1952. A negative experim ent relating magnetism and the Earth's rotation. Philos. Trans. R. So c. Lond. Ser. A 245, 309-370. Bleil, U., 1987. Quaternary high latitude m agnetostratigraphy, Polar Research 5, 325327. Bleil, U., Gard, G., 1989. Chronology and co rrelation of Quaternar y magnetostratigraphy and nannofossil biostratigraphy in Norwegi an-Greenland Sea se diments. Geol. Rundschau 78, 1173-1187. Bodn, P., Backman, J., 1996. A laminated sedi ment sequence from the northern North Atlantic Ocean and its climatic record. Geology 24, 507-510. Bond, G., Kromer, B., Beer, J. Muscheler, R., Evans, M.N., Showers, W., Hoffmann, S., Lotti-Bond, R., Hajdas, I. Bonani, G. 2001. Persistent solar influence on North Atlantic climate during the Holocene. Science 294, 2130-2136. Brachfeld, S.A., Banerjee, S.K., 2000. A new high-resolution geomagnetic relative paleointensity record for the North American Holocene: a comparison of sedimentary and absolute intensity dat a. J. Geophys. Res. 105, 821-834. Bullard, E.C., 1949. The magnetic field within the Earth. Pr oc. R. Soc. Lond., Ser. A 197, 433-453. Cande, S.C., Kent, D.V., 1992. A new geomagnet ic polarity time scale for the late Cretaceous and Cenozoic. J. Geophys. Res. 97, 13917-13951. Cande, S.C., Kent, D.V., 1995. Revised ca libration of the geomagnetic polarity time scale for the Late Cretaceous and Cenoz oic. J. Geophys. Res. 100, 6093. Carcaillet, J., Bourles, D.L., Thouveny, N., Arnold, M., 2004. A high resolution authigenic 10Be/9Be record of geomagnetic moment va riations over the last 300 ka from sedimentary cores of the Portuguese margin. Earth Planet. Sci. Lett. 219, 397-412. Carcaillet, J.T., Bourles, D.L., Thouv eny, N., 2004. Geomagnetic dipole moment and 10Be production rate interc alibration from authigenic 10Be/9Be for the last 1.3 Ma. Geochem. Geophys. Geosysyt. 5, Q05006. doi:10.102 9/2003GC000641. Carcaillet, J.T., Thouveny, N., Bourles, D.L., 2003. Geomagnetic moment instability between 0.6 and 1.3 Ma from cosmonuclide evidence. Geophys. Res. Lett. 30, 1792. doi:10.1029/ 2003GL017550.

PAGE 227

227 Champion, D.E., Lanphere, M.A., Kuntz, M. A., 1988. Evidence for a new geomagnetic reversal from lava flows in Idaho: discu ssion of short polarity reversals in the Brunhes and late Matuyama Chrons J. Geophys. Res. 93, 11667-11680. Channell, J.E.T., 1999. Geomagn etic paleointensity and direct ional secular variation at Ocean Drilling Program (ODP) Site 984 (Bjo rn Drift) since 500 ka: comparisons with ODP Site 983 (Gardar Drift). J. Geophys. Res. 104, 22937-22951. Channell, J.E.T., 2006. Late Brunhes polarity excursions (Mono Lake, Laschamp, Iceland Basin and Pringle Falls) recorded at ODP Site 919 (Irming er Basin). Earth Planet. Sci. Lett. 244, 378-393. Channell, J.E.T., Curtis, J.H., Flower B.P., 2004. The Matuyama-Brunhes boundary interval (500-900 ka) in North Atlantic drift sediments. Geophys. J. Int. 158, 489 505. Channell, J.E.T., Hodell, D. A., Lehman, B., 1997. Relative geomagnetic paleointensity and 18O at ODP Site 983 (Gardar Drift, Nort h Atlantic) since 350 ka. Earth Planet. Sci. Lett. 153, 103-118. Channell, J.E.T., Hodell, D.A., McManus, J., Lehman, B., 1998. Orbital modulation of the Earth's magnetic field intensity. Nature 394, 464-468. Channell, J.E.T., Hodell, D.A ., Xuan, C., Mazaud, A., Stoner J.S., 2008. Age calibrated relative paleointensity for the last 1.5 My r at IODP Site U1308 (North Atlantic). Earth Planet. Sci. Lett. 274, 59-71. Channell, J.E.T., Kleiven, H.F., 2000. Geomagnetic palaeointensities and astrochronological ages for the Matu yama-Brunhes boundary and the boundaries of the Jaramillo Subchron: palaeomagnetic and oxygen isotope records from ODP Site 983. Philos. Trans. R. Soc. Lond. A 358, 1027-1047. Channell, J.E.T., Labs, J., Raymo, M.E. 2003. The Runion Subchronozone at ODP Site 981 (Feni Drift, North Atlantic). Earth Planet. Sci. Lett. 215, 1-12. Channell, J.E.T., Mazaud, A., Sullivan, P., Turner, S., Raymo, M. E., 2002. Geomagnetic excursions and paleointensities in the Matuyama Chron at ODP Sites 983 and 984 (Iceland Basin). J. Geophys. Res. 107, 2114. doi:10. 1029/ 2001JB000491. Channell, J.E.T., Raymo, M.E. 2003. Paleomagnetic record at ODP Site 980 (Feni Drift Rockall) for the past 1.2 Myrs. Ge ochem. Geophys. Geosyst. 4, 1033, doi:10.1029/2002GC000440. Channell, J.E.T., Xuan C., 2009. Self-reversal and apparent magnetic excursions in Arctic sediments. Earth Pl anet. Sci. Lett. 284, 124-131. Channell, J.E.T., Xuan, C., Hodell, D.A., 2009. Stacki ng paleointensity and oxygen isotope data for the last 1.5 Myrs (PISO-1500). Earth Plane t. Sci. Lett. 283,14-23.

PAGE 228

228 Christensen, U.R., Tilgner, A., 2004. Power requirement of the geodynamo from ohmic losses in numerical and laboratory dynamos. Nature 429, 169. Clark, D.L., 1970. Magnetic reversals and sediment ation rates in the Arctic Basin. Geol. Soc. Amer. Bull. 81, 3129-3134. Clark, D.L., Vincent, J-S., Jones, G.A., Morr is, W.A., 1984. Correlation of marine and continental glacial and interglacial event s, Arctic Ocean and Banks Island. Nature 311, 147-149. Clark, D.L., Whitman, R.R., Morgan, K.A., Mackay, S.D., 1980. Stratigraphy and glacialmarine sediments of the Basin, central Ar ctic Ocean. Geol. Soc. Am. S. 181, 1-57. Clausen, L., 1998. Late Neogene and Quaternar y sedimentation on the continental slope and upper rise offshore southeast Gr eenland: interplay of contour and turbidity processes. In: Larsen, H.C., Saunder s, A.D., Wise Jr., S.W. (Eds.), Proc. ODP, Sci. Results, 152. Ocean Drilling Program, College Station, TX, pp. 3-18. Constable, C., Tauxe, L., 1996. Toward s absolute calibration of sedimentary paleointensity records. Earth Planet. Sci. Lett. 143, 269-274. Courtillot, V., Gallet, Y., Le Mouel, J.L., Fluteau, F., Genevey, A., 2007. Are there connections between the Earth's magnetic field and climate? Earth Planet. Sci. Lett. 253, 328-339. Cronin, T., Smith, S.A., Eynaud, F., O' Regan, M., King, J., 2008. Quaternary paleoceanography of the central Arctic based on Integrated Ocean Drilling Program Arctic Coring Expedition 302 foraminiferal assemblages. Palaeoceanography 23, PA1Si 8. doi:10.1029/2007PA001484. Cutter, G.A., Oatts, T.J., 1987. Determination of dissolved sulfide and sedimentary sulfur speciation using gas chromatogr aphy-photionization detection. Anal. Chem. 59, 717-721. Darby, D.A., Bischof, J.F., Jones, G.A., 1997. Radiocarbon dating of depositional regimes in the western Arctic Oc ean. Deep-Sea Res. II 44, 1745-1757. De Garidel-Thoron, T., Rosenthal, Y., Ba ssinot, F., Beaufort, L., 2005. Stable sea surface temperatures in the western Pacific warm pool over the past 1.75 million years. Nature 433, 294-298. Doubrovine, P.V., Tarduno, J.A., 2004. Self-reversed magnetization carried by titanomaghemite in oceanic basalts. Eart h Planet. Sci. Lett. 222, 959-969. Doubrovine, P.V., Tarduno, J.A., 2005. On th e compositional field of self-reversing titanomaghemite: Constraints from Deep Sea Drilling Projec t Site 307. J. Geophys. Res. 110, B11104. doi :10.1029/2005JB003865.

PAGE 229

229 Doubrovine, P.V., Tarduno, J.A., 2006a. Alte ration and self-reversal in oceanic basalts. J. Geophys. Res. 111, B 12S30. doi:10.1029/2006JB004468. Doubrovine, P.V., Tarduno, J.A., 2006b. N-type magnetism at cryogenic temperatures in oceanic basalt. Phys. Eart h Planet. Inter. 157, 46-54. Dunlop, D.J., Argyle, K.S., 1997. The rmoremanence anhysteretic remanence and susceptibility of submicron magnetites: nonlinear field dependence and variation with grain size. J. Geoph ys. Res. 102, 20199-20210. Dunlop, D.J., zdemir, O., 1997. Rock Magnetism: Fundamentals and Frontiers. Cambridge Univ. Press. 573 pp. Egli, R., Lowrie, W., 2002. Anhysteretic remanent magnetization of fine magnetic particles. J. Geophys. Res. 107, 2209. doi:10.1029/2001JB000671. Evans, H.F., Channell, J.E.T., Stoner, J.S., Hillaire-Marcel, C., Wright, J.D., Neitzke, L.C., Mountain, G.S., 2007. Pa leointensity-assisted chronostratigraphy of detrital layers on the Eirik Drift (North Atlantic) since marine isotope stage 11. Geochem. Geophys. Geosyst. 8, Q 11007. doi:10.1029/2007GC111720. Evans, H.F., Westerhold, T., Paulsen, H., Channell, J.E.T., 2007. Astronomical ages for Miocene polarity chrons C4Ar-C5r (9.311.2 Ma), and for three excursion chrons within C5n.2n. Earth Planet. Sci. Lett. 256, 455-465. Expedition 303 Scientists, 2006. Site U1304. Proc. IODP, Vol. 303/306. Available at http://publications.iodp.org/proceedi ngs/303_306/EXP_REPT /CHAPTERS/303_10 4.PDF. doi:10.2204/io dp.proc.303306.104.2006. Finkel, R.C., Nishiizumi, K., Hammer, C.U., Mayewski, P.A., Peel, D., Stuiver, M., 1997. Beryllium 10 concentrations in the Greenlan d Ice Sheet Project 2 ice core from 340 ka. J. Geophys. Res. 102, 26699-26706. Frank, M., Backman, J., Jakobss on, M., Moran, K., O'Regan, M., King, J., Haley, B.A., Kubik, P.W., Garbe-Schnberg, D., 2008. Beryllium isotopes in central Arctic Ocean sediments over the past 12.3 million years: Stratigraphic and paleoceanographic implications. Paleoceanography 23, PA1S02, doi:10.1029/2007PA001478. Frank, M., Schwarz, B., Baumann, S., Kubik, P.W., Suter, M., Mangini, A., 1997. A 200 kyr record of cosmogenic radionuclid e production rate and geomagnetic field intensity from Be-10 in gl obally stacked deep-sea sedime nts. Earth Planet. Sci. Lett. 149, 121-129. Franke, C., Hofmann, D., v on Dobeneck, T., 2004. Does lithology influence relative paleointensity records: a statistical analys is on SouthAtlantic pelagic sediments. Phys. Earth Planet. Inter. 147, 285-296.

PAGE 230

230 Fuller, M., 2006. Geomagnetic fi eld intensity, excursions, reversals and the 41,000-yr obliquity signal. Earth Planet. Sci. Lett. 245, 605-615. Gallet, Y., Genevey, A., Fluteau, F., 2005. Does Earth's magnetic field secular variation control centennial climate change? Ea rth Planet. Sci. Lett. 236, 339-347. Gee, J.S., Cande, S.C., Hildebr and, J.A., Donnelly, K., Park er, R.L., 2000. Geomagnetic intensity variations over the past 780 kyr obtained from near-seafloor magnetic anomalies. Nature 408, 827-832. Glatzmaier, G.A., Coe, R.S., Hongre, L., R oberts, P.H., 1999. The role of the Earths mantle in controlling the frequency of geomagnetic reversals. Nature 401, 885890. Glatzmaier, G.A., Roberts, P.H., 1996. Rota tion and magnetism of Earths inner core. Science 274, 1887-1891. Goree, W.S., Fuller, M., 1976. Magnetometers using RF -driven squids and their applications in rock magnetism and pal eomagnetism. Rev. Geophys. Space Phys14, 591-608. Grinsted, A., Moore, J.C., Jevrejeva, S., 2004. Applicat ion of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlin. Process. Geophys. 11, 561-566. Gubbins, D., 1999. The distinction between geomagnetic excursions and reversals, Geophys. J. Int. 137, F1-F3. Gubbins, D., Roberts, P.H., 1987. Magnetohy drodynamics of the Earths core. In: Jacobs, J. (Ed.), Geomagnetism. Academic Press, London, pp. 1-183. Guyodo, Y., Acton, G.D., Brachfeld, S., Channell, J.E.T., 2001. A sedimentary paleomagnetic record of the Matuyama Chron from the western Antarctic margin (ODP Site 1101). Earth Planet. Sci. Lett. 191, 61-74. Guyodo, Y., Channell, J.E.T., 2002. Effects of variable sedimentation rates and age errors on the resolution of sedimentary paleointensity records. Geochem. Geophys. Geosyst. 3, 8, doi: 10.1029/2001GC000211. Guyodo, Y., Channell, J.E.T., Thomas, R., 2002. Deconvolution of u-channel paleomagnetic data near geomagnetic reversals and short events. Geophys. Res. Lett. 29, 1845. doi:10.1029/2002GL014963. Guyodo, Y., Gaillot, P., Channell, J.E.T. 2000. Wavelet analysis of relative geomagnetic paleointensity at ODP Site 983. Earth Planet. Sci. Lett. 184, 109-123. Guyodo, Y., Valet, J.P., 1996. Relative va riations in geomagnet ic intensity from sedimentary records: the past 200,000 years. Earth Planet. Sci. Lett. 143, 23-36.

PAGE 231

231 Guyodo, Y., Valet, J.P., 1999. Gl obal changes in intensity of the Earth's magnetic field during the past 800 kyr. Nature 399, 249-252. Herman, Y., 1974. Arctic Ocean sediments, mi crofauna, and the climatic record in late Cenozoic time. In: Herman, Y. (Ed.), Marine Geology and Oceanography of the Arctic seas. Springer-Verlag, Berlin, pp. 283-348. Heslop, D., 2007. A wavelet investigation of possible orbital influences on past geomagnetic field intensity. Geochem Geophys. Geosyst. 8, Q03003. doi:10.1029/2006GC001498. Heslop, D., Dekkers, M.J., Krui ver, P.P., van Oorschot, I. H. M., 2002. Analysis of isothermal remanent magnetiz ation acquisition curves using the expectationmaximization algorithm. Geoph ys. J. Int. 148, 58-64. Hilgen, F.J., 1991. Extension of the astronomically calibrated ( polarity) time scale to the Miocene/Pliocene boundary. Earth Planet. Sci. Lett. 107, 349. Hodell, D.A., Channell, J.E.T., Curtis, J.H. Romero, O.E., Rohl, U., 2008. Onset of Hudson Strait Heinrich Events in the eastern North Atlantic at the end of the Middle Pleistocene Transition (~640 ka )? Paleoceanography 23, PA4218. doi:10.1029/ 2008PA001591. Hodell, D.A., Charles, C.D., Sierro, F.J., 2001. Late Pleist ocene evolution of the ocean's carbonate system. Earth Plane t. Sci. Lett. 192, 109-124. Hodell, D.A., Minth, E.K., Cu rtis, J.H., McCave, I.N., Hall, I.R., Channell, J.E.T., Xuan, C., 2009. Variations in Iceland Scotland Over flow Water during the last interglacial period and glacial inception. Eart h Planet. Sci. Lett. 288, 10-19. Hongre, L., Hulot, G., Khokhlov, A., 1998. An Analysis of the geomagnetic field over the past 2000 years, Phys. Earth Planet. Int. 106, 311-335. Horng, C.S., Lee, M.Y., Plike, H., Wei, K.Y., Liang, W.T., 20 02. Astronomically calibrated ages for geomagnetic reversals within the Matuyama chron. Earth Planets Space 54, 679-690. Horng, C.S., Roberts, A.P., Liang, W.T., 2003. A 2.14-Myr astronomically tuned record of relative geomagnetic paleointensity from the western Philippine Sea. J. Geophys. Res. 108, 2059. doi:10.1029/2001JB001698. Hulot, G., Le Moul, J.L., 1994. A statistical approach to the Earth's main magnetic field. Phys. Earth Planet. Int. 82, 167-183. Hunkins, K.L., B, A.W.H., Opdyke, N.D., Ma thieu, G., 1971. The late Cenozoic history of the Arctic Ocean. In: Tu rekian, K.K. (Ed.), The Late Cenozoic Glacial Ages. Yale Univ. Press, pp. 215-237.

PAGE 232

232 Hsing, S.K., Hilgen, F.J., Abdul Aziz, H., Krijgsman, W., 2007. Completing the Neogene geological time scale between 8. 5 and 12.5 Ma. Earth Planet. Sci. Lett. 253, 340-358. Huybers, P., 2007. Glacial variability over t he last two million y ears: an extended depthderived age model, continuous obliquity pacing, and th e Pleistocene progression. Quat. Sci. Rev. 26, 37-55. Huybers, P., Wunsch, C., 2004. A depth-der ived Pleistocene age model: Uncertainty estimates, sedimentation variabi lity, and nonlinear climate change. Paleoceanography 19, 1, doi: 10.1029/2002PA000857. Ikehara, M., Kawamura, K., Ohkouchi, N., Taira, A., 1999. Organi c Geochemistry of Greenish Clay and OrganicRich Sediments Since the Early Miocene from Hole 985A, Norway Basin. Proc. ODP Sci. Results, 162. Ocean Drilling Program, College Station, TX, pp. 209-216. Imbrie, J., Imbrie, J.Z., 1980. Modeling the climatic resp onse to orbital variations. Science 207, 943-953. Jakobsson, M., Backman, J., Murray, A. Lvlie, R., 2003. Optically Stimulated Luminescence dating supports central Arctic Ocean cm-scale sedimentation rates. Geochem. Geophys. Geosyst. 4, 1016. doi:10.1029/20 02GC000423. Jakobsson, M., Lvlie, R., Al-Hanbali, H., Arnold, E., Backman, J., Mrth, M., 2000. Manganese and color cycles in Arctic Ocean sediments constrain Pleistocene chronology. Geology 28, 23-26. Jakobsson, M., Macnab, R., Mayer, M., Anderson, R., Edwards, M., Hatzky, J., Schenke, H-W., Johnson, P., 2008. An im proved bathymetric portrayal of the Arctic Ocean: Implications for ocean modeling and geologica l, geophysical and oceanographic analyses. Geophys. Res. Lett. 35, L07602. doi:10.1029/2008GL033520. Johnson, E.A., Murphy, T., To rreson, O.W., 1948. Prehistory of the Earths magnetic field. J. Geophys. Res. 53, 349-372. Johnson, H.P., Lowrie, W., Kent, D.V., 1975. Stability of anhysteretic remanent magnetization in fine and coarse magnetit e and maghemite particles. Geophys. J. R. Astron. Soc. 41, 1-10. Johnson, H.P., Merrill, R.T., 1974. Low-tem perature oxidation of single domain magnetite. J. Geophys. Res. 79, 5533-5534. Jones, T.A., 2006. MATLAB functions to analyze directional (azimu thal) data-I: singlesample inference. Comput. Geosci. 32, 166.

PAGE 233

233 Jouzel, J., Masson-Delmotte, V. Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G., Minster, B., Nouet, J., Barnola, J.M., Chappellaz, J., Fischer, H., Gallet, J.C., Johnsen, S., Leuenberger, M., Loulergue, L., Luet hi, D., Oerter, H. Parrenin, F., Raisbeck, G., Raynaud, D., Schilt, A. Schwander, J., Selmo, E., Souchez, R., Spahni, R., Stauffer, B., Ste ffensen, J.P., Stenni, B., St ocker, T.F., Tison, J.L., Werner, M., Wolff, E.W., 2007. Orbital and millennial Antarctic climate variability over the past 800,000 years. Science 317, 793-797. Kaufman, D.S., Polyak, L., Adler, R., C hannell, J.E.T., Xuan, C., 2008. Dating late Quaternary planktonic forami nifer Neogloboquadrina pachy derma from the Arctic Ocean by using amino acid racemi zation. Paleoceanography 23, PA3224. doi:10.1029/2008PA001618. Kawamura, K., Parrenin, F., Lisiecki, L., Uemura, R., Vimeux, F., Severinghaus, J.P., Hutterli, M.A., Nakazawa, T., Aoki, S., J ouzel, J., Raymo, M.E., Matsumoto, K., Nakata, H., Motoyama, H., Fujita, S., Go to-Azuma, K., Fujii, Y., Watanabe, O., 2007. Northern Hemisphere forcing of climat ic cycles in Antarctica over the past 360,000 years. Nature 448, 912-916. Kent, D.V., 1973. Post-depos itional remanent magnetisation in deep-sea sediment, Nature 246, 32-34. Kent, D.V., 1982. The apparent correlation of paleomagnetic in tensity and climate records in deep sea sediments. Nature 299, 538-540. Kent, D.V., Carlut, J., 2001. A negative test of orbital control of geomagnetic reversals and excursions. Geophys. Res. Lett. 28, 3561-3564. Kent, D.V., Lowrie, W., 1974. Or igin of magnetic instability in sediment cores from the central North Pacific. J. Geophys. Res. 79, 2987-3000. Kent, D.V., Opdyke, N.D., 1977. Paleomagnetic field intensity variation recorded in a Brunhes epoch deep sea sediment core. Nature 266, 156-159. Kerswell, R.R., 1996.Upper bounds on the energy dissipation in turbulent precession. J. Fluid Mech. 321, 335-370. King, J.W., Banerjee, S.K., Marvin, J., 1983. A newrock-magnetic approach to selecting sediments for geomagnetic paleointensity studies; application to paleointensity for the last 4000 years. J. Geophys. Res. 88, 5911-5921. Kirschvink, J.L., 1980. The least squares lines and plane analysi s of paleomagnetic data. Geophys. J.R. As tr. Soc. 62, 699-718. Kissel, C., Laj, C., Labeyrie, L., Dokken, T ., Voelker, A., Blamar t, D., 1999. Rapid climatic variations during marine isotopi c stage 3: magnetic analysis of sediments from Nordic Seas and North Atlantic Earth Planet. Sci. Lett. 171, 489-502.

PAGE 234

234 Kok, Y.S., 1999. Climatic influence in NRM and 10Be-derived geomagnetic paleointensity data. Earth Planet. Sci. Lett. 166, 105-119. Korte, M., Constable, C., 2005. The geomagnetic dipole mo ment over the last 7000 years new results from a global model Earth Planet. Sci. Lett. 236, 348-358. Krsa, D., Matzka, J., 2007. Inversion of titanomaghemite in oceanic basalt during heating. Phys. Earth Pl anet. Inter. 160, 169-179. Krsa, D., Shcherbakov, V.P., Kunzmann, T., Petersen, N., 2005. Self-reversal of remanent magnetization in basalts due to partially oxidized titanomagnetites. Geophys. J. Int. 162, 115-136. Kruiver, P.P., Dekkers, M.J., Heslop, D., 2001. Quantification of magnetic coercivity components by the analysis of acquisi tion curves of isothermal remanent magnetization. Earth. Plan et. Sci. Lett. 189, 269-276. Laj, C., Channell, J.E.T., 2007. Geomagnetic excursions. In: K ono, M. (Ed.), Treatise on Geophysics. Geomagnetism, vol. 5. Elsevier, Amsterdam, pp. 373-416. Laj, C., Kissel, C., Beer, J., 2004. High resolu tion global paleointensity stack since 75 kyr (GLOPIS-75) calibrated to absolute va lues. In: Channell, J.E.T., Kent, D.V., Lowrie,W., Meert, J.G. (Eds.), Timescales of the Paleomagnetic Field. Geophysical Monograph, vol. 145. Amercian Geophysical Union, Washington, DC, USA. Laj, C., Kissel, C., Mazaud, A., Channell, J.E.T., Beer, J., 2000. North Atlantic paleointensity stack since 75 ka (NAPI S-75) and the duration of the Laschamp event. Phil. Trans. R. Soc. Lond. A 358, 1009-1025. Langereis, C.G.,Dekkers,M.J., de Lange,G.J., Pa terne,M., van Santvoort, P.J.M., 1997. Magnetostratigraphy and astronomical cali bration of the last 1.1 Myr from an eastern Mediterranean piston core and dat ing of short events in the Brunhes. Geophys. J. Int. 129, 75-94. Laskar, J., 1990. The chaotic mo tion of the Solar System: a numerical estimate of the size of the chaotic zo nes. Icarus 88, 266-291. Laskar, J., Joutel, F., Boud in, F., 1993. Orbital, prece ssional, and insolation quantities for the Earth from 220 Myr to 110 My r. Astron. Astrophys. 270, 522-533. Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A., Levrard, B., 2004. A long term numerical solution for the insola tion quantities of the Earth. Astron. Astrophys. 428, 261-285. Lau, K.M., Weng, H., 1995. Climate signal dete ction using wavelet transform: how to make a time series sing. Bull. Am. Meteor. Soc. 76, 2391-2402.

PAGE 235

235 Levi, S., Banerjee, S.K., 1976. On the possibility of obtaining relative paleointensities from lake sediments. Earth Pl anet. Sci. Lett. 29, 219-226. Lisiecki, L.E., Lisiecki, P.A., 2002. App lication of dynamic programming to the correlation of paleoclimate records. Paleoceanography 17, 1049. doi:10.1029/2001PA000733. Lisiecki, L.E., Raymo, M.E., 2005. A P liocene-Pleistocene stack of 57 globally distributed benthic 18O records. Paleoc eanography 20, PA1003. doi:10.1029/2004PA001071. Liu, P.C., 1994. Wavelet spectrum anal ysis and ocean wind waves. In: FoufoulaGeorgiou, E., Kumar, P. (Eds.), Wavelets in Geoph ysics. Academic Press, New York, pp. 151-166. Loper, D.E., 1975. Torque balance and energy budget for the precessionally driven dynamo. Phys. Earth Plan et. Inter. 11, 43-60. Lourens, L., Hilgen, F., Laskar, J., Shackl eton, N., Wilson, D., 2004. The Neogene period. In: Gradstein, F., Ogg, J., et al. (Eds.), A Geologic Time Scale. Cambridge University Press, pp. 409-440. Lourens, L.J., Antonarakou, A., Hilgen, F.J. Van Hoof, A.A.M., Vergnaud-Grazzini, C., Zachariasse, W.J., 1996. Evaluation of the Plio-Pleistocene astronomical timescale. Paleoceanography 11, 391-413. Lvlie, R., Markussen, B., Sejrup, H.P., Thiede, J., 1986. Magnetostratigraphy in three Arctic Ocean sediment cores; argum ents for magnetic excursions within oxygenisotope stage 2. Phys. Earth Planet. Inter. 43, 173-184. Lowrie, W., 1990. Identification of ferromagnetic minerals in a rock by coercivity and unblocking temperature properties. Geophys. Res. Lett. 17, 159-162. Lund, S.P., Acton, G.D., Clement, B.M., Ok ada, M.,Williams, T., 2001a. Paleomagnetic records of Stage 3 excursions. Leg 172. In : Keigwin, L.D., Rio, D., Acton, G.D., Arnold, E. (Eds.), Proc. ODP Sci. Resu lts, 172. Ocean Drilling Program, College Station, TX, pp. 1-20. Lund, S.P., Keigwin, L., 1994. Measurement of the degree of smoothing in sediment paleomagnetic secular variation records an example from late Quaternary deepsea sediments of the Berm uda rise, western North-Atl antic Ocean. Earth Planet. Sci. Lett. 122, 317-330. Lund, S.P., Schwartz, M., Keig win, L., Johnson, T., 2005. D eep-sea sediment records of the Laschamp geomagnetic field excurs ion (~41,000 calendar years before present). J. Geophys. Res. 110, B04101. doi:10.1029/2003JB002943.

PAGE 236

236 Lund, S.P., Stoner, J.S., Channe ll, J.E.T., Acton, G., 2006. A summary of Brunhes paleomagnetic field variability recorded in Ocean Drilling Progr am cores. Phys. Earth Planet. Inter. 156, 194-204. Lund, S.P., Williams, T., Acton, G.D., Clem ent, B.M., Okada, M., 2001b. Brunhes Chron magnetic field excursions recovered fr om Leg 172 sediments. In: Keigwin, L.D., Rio, D., Acton, G.D., Arnold, E. (Eds.) Proc. ODP, Sci. Results, 172. Ocean Drilling Program, College Station, TX, pp. 1-18. Maher, B.A., 1988. Magnetic-properties of some synthetic sub-micron magnetites. Geophys. J. Int. 94, 83-96. Malkus, W.V.R., 1968. Precession of the Earth as a cause of geomagnetism. Science 169, 259-264. Mardia, K.V., Jupp, P.E., 2000. Directional St atistics. Wiley, Chic hester, UK. 429 pp. Markussen, B., Zahn, R., Thiede, J., 1985. Late Quaternary sedimentation in the eastern Arctic Basin: stratigraphy and depositional environment. Palaeogeogr., Palaeoclimatol. Palaeoecol. 50, 271-284. Marshall, M., Cox, A., 1972. Magnetic chan ges in pillow basalt due to sea-floor weathering. J. Geoph ys. Res. 77, 6459-6469. Mazaud, A., 2005. User-friendly software for vector analysis of t he magnetization of long sediment cores. Geochem Geophys. Geosyst. 6, Q12006. doi:10.1029/2005GC001036. Mazaud, A., 2006. A first-order correction to minimize environm ental influence in sedimentary records of relative paleoint ensity of the geomagnetic field. Geochem. Geophys. Geosyst. 7, Q 07002, doi:10.1029/2006GC001257. Mazaud, A., Channell J.E.T., Xuan, C., Stoner, J.S., 2009. Upper and Lower Jaramillo polarity transitions recorded in IODP Ex pedition 303 North Atl antic sediments: implications for transitional field geometry. Phys. Earth Planet. Inter. 172,131-140. McCave, I.N., Manighetti, B., Robinson, S.G., 1995. Sortable silt and fine sediment size composition slicing parameters fo r paleocurrent speed and paleoceanography. Paleoceanography 10, 593-610. McMillan, D., Constable, C., Parker, R., 2004. Assessing the dipolar signal in stacked paleointensity records using a statistica l error model and geodynamo simulations. Phys. Earth Planet. Inter. 145, 37-54. McMillan, D.G., Constable, C.G., Parker, R.L, 2002. Limitations on stratigraphic analyses due to incomplete age control and their relevance to sedimentary paleomagnetism. Earth Planet. Sci. Letters 201, 509-523.

PAGE 237

237 Meynadier, L., Valet, J.P., Grousset, F.E., 19 95. Magnetic-properties and origin of upper Quaternary sediments in the Somali basin. Indian-Ocean. Paleoceanography 10, 459-472. Meynadier, L., Valet, J.P., Weeks, R.J., Sha ckleton, N.J., Hagee, V.L., 1992. Relative geomagnetic intensity of the fi eld during the last 140 ka. Earth Planet. Sci. Lett. 114, 39-57. Morris, T.H., Clark, D.L., Blasco, S.M., 1985. Sediments of the Lomonosov Ridge and Makarov Basin: a Pleistocene stratigraphy fo r the North Pole. Geol. Soc. Am. Bull. 96, 901-910. Moskowitz, B.M., 1980. Theoretical grain si ze limits for single-domain, pseudo-singledomain and multi-domain behavi or in titanomagnetite (x = 0.6) as a function of low-temperature oxidation. Earth Planet. Sci. Lett. 47, 285-293. Muscheler, R., Beer, J., Kublik, P.W., Sy nal, H.A., 2005. Geomagnetic field intensity during the last 60,000 years based on 10Be and 36Cl from the Summit ice cores and 14C. Quat. Sci. Revs. 24, 1849-1860. Nagata, T., Uyeda, S., Akimoto, S., 1952. Self-reversal of thermoremanent magnetization of igneous rocks. J. Geomagn. Geoelectr. 4, 22-38. Nagy, E.A., Valet, J.P., 1993. New advances for paleomagnetic studies of sediment cores using u-channels. Geoph ys. Res. Lett. 20, 671-674. Nishitani, T., Kono, M.,1983. Curie temperature and lattice constant of oxidized titanomagnetite. Geophys. J. R. Astron. Soc. 74, 585-600. Nowaczyk, N.R., Antonow, M., 1997. High reso lution magnetostratigraphy of four sediment cores from the Greenland Sea I. Identificatio n of the Mono Lake excursion, Laschamp and Biwa I/Jamaic a geomagnetic polarity events. Geophys. J. Int. 131, 310-324. Nowaczyk, N.R., Baumann, M., 1992. Combi ned high-resolution magnetostratigraphy and nannofossil biostratigraphy for late Quaternary Artic Ocean sediments. Deep Sea Res. 39, 567-601. Nowaczyk, N.R., Frederichs, T.W., 1999. Geomagnetic events and relative paleointensity variations during the last 300 ka as recorded in Kolbeinsey Ridge Sediments, Iceland Basin indication for a strongly variable geomagnetic field. Int. J. Earth Sci. 88, 116-131. Nowaczyk, N.R., Frederichs, T.W., Eisenhauer, A., Gard, G., 1994. Magnetostratigraphic data from late Q uaternary sediments from the Yermak Plateau, Arctic Ocean: evidence for four geomagnetic polarity events within the last 170 Ka of the Br unhes Chron. Geophys. J. Int. 117, 453-471.

PAGE 238

238 Nowaczyk, N.R., Frederichs, T.W., Kassens, H., Nrgaard-Pedersen, N., Spielhagen, R.F., Stein, R. and Weiel, D., 2001. Sedimentation Rates in the Makarov Basin, Central Arctic Ocean: A Paleoma gnetic and Rock Magnetic Approach. Paleoceanography, 16, 368-389. Nowaczyk, N.R., Knies, J., 2000. Magnetostratigraphic results from the eastern Arctic ocean: AMS 14C ages and relative paleointensity data of the Mono Lake and Laschamp geomagnetic reversal excursi ons. Geophys. J. Int. 140, 185-197. Oda, H., Shibuya, H., 1996. Deconvolution of long-core paleomagnetic data of Ocean Drilling Program by Akaikes Bayesian In formation Criterion minimization, J. Geophys. Res 101, 2815-2834. O'Regan, M., King, J., Backman, J., Jakobss on, M., Palike, H., Mo ran, K., Heil, C., Sakamoto, T., Cronin, T.M., Jordan, R.W. 2008. Constraints on the Pleistocene chronology of sediment s from the Lomonosov Ri dge. Paleoceanography 23, PA1S19. doi:10.1029/ 2007PA001551. O'Reilly, W., 1983. The identification of titanomaghemites: model mechanisms for the maghemitization and inversion processe s and their magnetic consequences. Phys. Earth planet. Int. 31, 65-76. O'Reilly, W., 1984. Rock and Mineral Magneti sm. Blackie & Sons Ltd., Glasgow. 220 pp. O'Reilly, W., Banerjee, S.K., 1966. Oxidat ion of titanomagnetites and self-reversal. Nature 211, 26-28. Ortiz, J., Mix, A., Harris, S., OConnell, S., 1999. Diffuse spectral reflectance as a proxy for percent carbonate contnet in north Atlantic sedi ments. Paleoceanography 14, 171-186. zdemir, 1987. Inversion of titanomaghemites. Phys. Earth Planet. Inter. 46, 184196. zdemir, Banerjee, S.K., 1981. An experimental-study of magnetic viscosity in synthetic monodomain titanomaghemites imp lications for the magnetization of the ocean crust. J. Geoph ys. Res. 86, 1864-1868. zdemir, Banerjee, S.K., 1984. High temperature stabil ity of maghemite (g-Fe2O3). Geophys. Res. Lett. 11, 161-164. zdemir, Dunlop, D.J., 1985. An experimental study of Chemical Remanent Magnetizations of synthetic monodom ain titanomaghemites with initial thermoremanent magnetizations. J. Geophys. Res. 90, 11513-11523. zdemir, Dunlop, D.J., Mosko witz, B.M., 1993. The effect of oxidation on the Verwey transition in magnetite. Geoph ys. Res. Lett. 20, 1671-1674.

PAGE 239

239 zdemir, O'Reilly, W., 1982. Magnetic hysteresis properties of synthetic monodomain titanomaghemites Earth Pl anet. Sci. Lett. 57, 437-447. Ozima, M., Sakamoto, N., 1971. Magnetic properties of syn thesized titanomaghemite. J. Geophys. Res. 76, 7035-7046. Plike, H., Laskar, J., Shackleton, N.J. 2004. Geologic constraints on the chaotic diffusion of the Solar S ystem. Geology 32, 929-932. Plike, H., Norris, R.D., Herrl e, J.O., Wilson, P.A., Coxall, H.K., Lear, C.H., Shackleton, N.J., Tripati, A.K., Wade, B.S., 2006. The heartbeat of the Oligocene climate system. Science 314, 1894-1898. Pazur, A., Winklhofer, M., 2008. Magnetic effect on CO2 solubility in seawater: A possible link between geomagnetic field vari ations and climate. Geophys. Res. Lett. 35, 16, doi: 10.1029/2008GL034288. Peck, J.A., King, J.W., Colman, S.M., Kravchinsky, V.A., 1996. An 84-kyr paleomagnetic record from the sediments of Lake Baik al, Siberia. J. Geophys. Res. 101, 11365-11385. Perrin, M., Schnepp, E., 2004. IAGA paleointensity database: distribution and quality of the dataset. Phys. Earth Planet. Int. 147, 255-267. Petersen, N., Vali, H., 1987. Observat ion of shrinkage cracks in ocean floor titanomagnetites. Phys. Eart h Planet. Int. 46, 197-205. Phillips, R. L., Grantz, A ., 1997. Quaternary history of sea ice and paleoclimate in the Amerasia basin, Arctic Ocean, as recor ded in the cyclical strata of Northwind ridge. Geol. Soc. Am Bull. 109, 1101-1115. Polyak, L., Bischof, J., Ortiz, J.D., Darby, D.A., Channell, J.E.T., Xuan, C., Kaufman, D.S., Llie, R., Schneider, D.A. Eberl, D.D., Adler, R.E. and Council, E.A., 2009. Late Quaternary stratigraphy and sediment ation patterns in the western Arctic Ocean. Global Planet. Change 68, 5-17. Polyak, L., Curry, W.B., Darby, D.A., Bishof, J., Cronin, T. M., 2004. Contrasting glacial/ interglacial regimes in the western Arct ic Ocean as exemplified by a sedimentary record from the Mendeleev Ridge. Palaeogeogr. Palaeoclimatol. Palaeoecol. 203, 73-93. Poore, R.Z., Phillips, R.L., Rieck, H.J., 1993. Paleoclimate Record for Northwind Ridge, Western Arctic-Ocean. Pal eoceanography 8, 149-159. Prins, M.A., Bouwer, L.M., Beets, C.J., Troelstra, S.R .,Weltje, G.J., Kruk, R.W., Kuijpers, A., Vroon, P.Z., 2002. Ocean ci rculation and iceberg discharge in the glacial North Atlantic: inferences from unmixing of sediment size distributions. Geology 30, 555-558.

PAGE 240

240 Readman, P.W., O'Reilly, W., 1970. The synthesis and inversion of non-stoichiometric titanomagnetites. Phys. Eart h Planet. Int. 4, 121-128. Readman, P.W., O'Reilly, W., 1972. Magnetic pr operties of oxidized (cation deficient) titanomagnetites (Fe, Ti, )3 O4. J. Geomagn. Geoel ectr. 24, 69-90. Roberts A., 2006. High-resolution magnetic analysis of sediment cores: strengths, limitations and strategies for maximizing the value of long-core magnetic data. Phys. Earth Planet. Int. 156, 162-178. Roberts, A.P., Winklhofer, M., 2004. Why are geomagnetic excursions not always recorded in sediments? Constraint s from post-depositional remanent magnetization lock-in modelling. Earth Planet. Sci. Lett. 227, 345. Roberts, A.P., Winklhofer, M. Liang, W., Horng, C., 2003. Testing the hypothesis of orbital (eccentricity) influence on Earth's m agnetic field. Earth Planet. Sci. Lett. 216, 187-192. Robertson, D.J., France, D.E. 1994. Discrimination of re manence-carrying minerals in mixtures, using isothermal remanent magnetis ation acquisition curves. Phys. Earth Planet. Int., 84, 223-234. Rochester, M.G., Jacobs, J.A., Smylie, D. E., Chong, K.F., 1975. Can precession power the geomagnetic dynamo? Geophys. J. R. Astron. Soc. 43, 661-678. Rousse, S., Kissel, C., Laj, C., Eiriksson, J., Knudsen, K.L., 2006. Holocene centennial to millennial-scale climatic variability : evidence from high-resolution magnetic analyses of the last 10 cal kyr off No rth Iceland (core MD99-2275). Earth Planet. Sci. Lett. 242, 390-405. Ryall, P.J.C., Hall, J.M., Clark, J., Milligan T., 1977. Magnetization of oceanic crustal layer 2 results and thoughts after DSDP Leg37. Can. J. Earth Sci. 14, 684-706. Savitzky, A., Golay, M.J.E ., 1964. Smoothing and differentia tion of data by simplified least squares procedures. Anal. Chem. 36, 1627-1639. Schult, A., 1968. Self-reversal of magnetization and chemical composition of titanomaghemites in basalts. Eart h Planet. Sci. Lett. 4, 57-63. Schult, A., 1971. On the strength of exchan ge interactions in titanomagnetite and its relation to self-reversal of magnetizat ion. Zeitschrift fr Geophysik 37, 357-365. Schulz, M., Mudelsee M., 2002. REDFIT: estimating red-noise spectra directly from unevenly spaced paleoclimatic time se ries. Comput. Geosci. 28, 421-426. Sejrup, H.P., Gifford, H.M., Brigham-Grette, J., Lvlie, R. Hopkins, D., 1984. Amino acid epimerization implies rapid sedimentation rates in Ar ctic Ocean cores. Nature 310, 772-775.

PAGE 241

241 Shackleton, N.J., 1967. Oxygen isotope analyses and Pleistocene temperature reassessed. Nature 215, 15-17. Shackleton, N.J., Berger, A., Peltier, W.R., 1990. An alter native astronomical calibration of the lower Pleistocene timescale based on ODP Site 677. Trans. R. Soc. Edin. Earth Sci. 81, 251-261. Shackleton, N.J., Crowhurst, S., Hagelberg, T., Pisias, N.G., Schneider, D.A., 1995. A new Late Neogene time scale: application to Leg 138 sites. In: Pisias, N.G., Mayer, L.A., Janecek, T.R., Palmer-Julson, A., van Andel, T.H. (Eds.), Proc. ODP, Sci. Results, vol. 138. Ocean Drilling Program, Coll ege Station, TX, pp. 73. Shimada, C., Sato, T., Toyoshima, S. Yamasaki, M., Tanimura, Y., 2008. Paleoecological significance of laminat ed diatomaceous oozes during the middleto-late Pleistocene, North Atlantic Oc ean (IODP Site 1304). Mar. Micropaleontol. 69, 139-150. Shipboard Scientific Party, 1996. Site 983. In: Jansen, E., M. Raymo, P. Blum, et al. (eds.), Proc. ODP, Init. Repts., 162. Oc ean Drilling Program, College Station, TX, pp. 139-167. Singer, B.S., Jicha, B.R., Kirby, B.T., Geissman, J.W., Herrero-Bervera, E., 2008. 40Ar/39Ar dating links Albuquerque Volcanoes to the Pringl e Falls excursion and the geomagnetic instability time scale. Earth Planet. Sci. Lett. 267, 584-595. Singer,B.S., Brown, L.L.,Raba ssa, J.O., Guillou, H., 2004.40Ar/39Ar chronology of late Pliocene and Early Pleistocene geomagnet ic and glacial events in southern Argentina. In: Channell, J.E.T., et al. (Eds.), Timesca les of the Paleomagnetic Field. AGU Geophysical Monograph Series, vol. 145, pp. 175-190. Skinner, L.C. and N.J. Shackleton, 2005. An Atlantic lead over Pcific deep-water change across Termination I: implications for the application of the marine isotope stage stratigraphy. Quat. Sci. Revs. 24, 571-580. Skinner, L.C., Shackleton, N.J., 2005. An At lantic lead over Pacific deep-water change across Termination I: implications for t he application of the marine isotope stage stratigraphy. Quat. Sci. Revs. 24, 571-580. Skinner, L.C., Shackleton, N.J., 2006. Deco nstructing Terminations I and II: revisiting the glacioeustatic paradigm based on deep-water temperature estimates. Quat. Sci. Rev. 25, 3312-3321. Snowball, I., Moro s, M., 2003. Saw-tooth pattern of North Atlantic current speed during Dansgaard-Oeschger cycles revealed by the magnetic grain size of Reykjanes Ridge sediments at 59 degrees N. Paleoceanography 18, 1026. doi:10.1029/2001PA000732.

PAGE 242

242 Soubrand-Colin, M., Horan, H., Courtin-Nomade, A., 2009. Mineralogical and magnetic characterization of iron titanium oxides in soils developed on two various basaltic rocks under temperate climate. Geoderma 149, 27-32. Spielhagen, R.F., Baumann, K.H., Erlenkeuser, H., Nowaczyk, N.R., NrgaardPedersen, N., Vogt, C., Weiel, D., 2004. Ar ctic Ocean deep-sea record of northern Eurasian ice sheet history. Quat. Sci. Rev. 23, 1455-1483. Spielhagen, R.F., Bonani, G., Eisenhauer, A ., Frank, M., Frederichs, T., Kassens, H., Kubik, P.W., Mangini, A., Nrgaard-Peder sen, N., Nowaczyk, N.R., Sch.aper, S., Stein, R., Thiede, J., Tiedemann, R., W ahsner, M., 1997. Arctic Ocean evidence for Late Quaternary initiati on of northern Eurasian ice sheets. Geology 25, 783786. St. John, K., Flower, B.P., Krissek, L., 2004. Evolution of iceberg melting, biological productivity, and the record of Icelandic vo lcanism in the Irminger basin since 630 ka. Mar. Geol. 212, 133-152. Stein, R., Schubert, C.J., MacDonald, R.W., Fohl, K., Harvey, H.R., Weiel, D., 2003. The central Arctic Ocean: distribution, sources, variability, and burial of organic carbon. In: Stein, R., MacDonald, R.W. (Eds.), The organic carbon cycle in the Arctic Ocean. Springer-Verlag, Berlin, pp. 295-314. Steuerwald, B.A., Clark, D.L., Andrew, J. A., 1968. Magnetic stratigraphy and faunal patterns in Arctic Ocean sediments. Earth Planet. Sci. Lett. 5, 79-85. Stoner, J.S., Channell, J.E.T., Hillaire-Ma rcel, C., Kissel, C., 2000. Geomagnetic paleointensity and environmental record from Labrador Sea Core MD95-2024: global marine sediment and ice core chronos tratigraphy for the last 110 kyr. Earth Planet. Sci. Lett. 183, 161-177. Stoner, J.S., Channell, J.E.T., Hodell, D.A., C harles, C., 2003. A 580 kyr paleomagnetic record from the sub-Antarctic South Atl antic (ODP Site 1089). J. Geophys. Res. 108, 2244. doi:10. 1029/2001JB001390. Stoner, J.S., Laj, C., Channell, J.E.T., Kissel, C., 2002. South Atlantic and North Atlantic geomagnetic paleointensity stacks (0-80 ka); implications for interhemispheric correlation. Quat. Sc i. Rev. 21, 1141-1151. St-Onge, G., Stoner, J.S., Hillaire-Marcel C., 2003. Holocene paleomagnetic records from the St. Lawrence Est uary, eastern Canada: centennialto millennial-scale geomagnetic modulation of cosmogenic isot opes. Earth Planet. Sci. Lett. 209, 113-130. Tauxe, L., 1993. Sedimentary reco rds of relative paleointensit y of the geomagnetic field: theory and practice. Rev. Geophys. 31, 319-354.

PAGE 243

243 Tauxe, L., LaBrecque, J.L., Dodson, R., Fuller, M., Dematteo, J.,1983. U-channels a new technique for paleomagnetic analysis of hydraulic piston cores. EOS Trans. AGU 64, 219. Tauxe, L., Shackleton, N.J. 1994. Relative palaeointensit y records from the OntongJava Plateau. Geophys. J. Int. 117, 769-782. Tauxe, L., Steindorf, J.L., Harris, A., 2006. Depositional remanent magnetization; toward an improved theoretical and experi mental foundation. Earth Planet. Sci. Lett. 244, 515-529. Tauxe, L., Wu, G., 1990. Normalized remanence in sediments of the western Equatorial Pacific: relative paleointensity of the geomagnetic field? J. Geophys. Res. 95, 12337-12350. Tauxe, L., Yamazaki, T., 2007. Paleointensities. In: Kono, M. (Ed.), Treatise on Geophysics. Geomagnetism, vol. 5. Elsevier, Amsterdam, pp. 509-563. Teanby, N., Gubbins, D., 2000. The effect of aliasing and lock-in processes on paleosecular variation records from sedim ents. Geophys. J. In t. 142, 563-570. Thomas, R., Guyodo, Y., Channell, J.E.T. 2003. U-channel track for susceptibility measurements. Geochemistry. Geophys. Geosys. 4, 1050, doi:10.1029/2002GC000454. Thomson, J., Nixon, S., Summerhayes, C.P., Schonfeld, J., Zahn, R., Grootes, P., 1999. Implications for sedim entation changes on the Iberian margin over the last two glacial/interglacial transitions from (230Th excess) systemat ics. Earth Planet. Sci. Lett. 165, 255-270. Thouveny, N., Bourles, D.L., Saracco, G., Carcaillet, J.T. Bassinot, F., 2008. Paleoclimatic context of geom agnetic dipole lows and excu rsions in the Brunhes, clue for an orbital influence on the geodynam o? Earth Planet. Sci. Lett. 275, 269284. Thouveny, N., Carcaillet, J ., Moreno, E., Leduc G., Nerini, D., 2004. Geomagnetic moment variation and paleom agnetic excursions since 400 kyr BP; a stacked record from sedimentary sequences of the Portuguesemargin. Earth Planet. Sci. Lett. 219, 377-396. Tilgner, A., 2005. Precession driv en dynamos. Phys. Fluids 17, 034014. doi:10.1063/1.1852576. Tilgner, A., 2007. Kinematic dynamos with prec ession driven flow in a sphere. Geophys. Astrophys. Fluid Dyn. 101, 1-9. Torrence, C., Compo, G.P., 1998. A practi cal guide to wavelet analysis. Bull. Am. Meteor. Soc. 79, 61-78.

PAGE 244

244 Torrence, C., Webster, P.J., 1999. In terdecadal changes in the ENSO-monsoon system. J. Climat e 12, 2679-2690. Tric, E., Laj, C., Valet, J., Tucholke, P., Pa terne, P., Guichard, F., 1991. The Blake geomagnetic event: transition geometry dynamical characteristics and geomagnetic significance. Earth Pl anet. Sci. Lett. 102, 1-13. Uyeda, S., 1958. Thermo-remanent magnetizat ion as a medium of paleomagnetism, with special reference to reverse thermo -remanent magnetism. Jpn. J. Geophys. 2, 1-123. Valet, J.P., 2003. Time variations in geomagn etic intensity. Re v. Geophys. 41, 1004. doi:10.1029/2001RG000104. Valet, J.P., Herrero-Brevera, E., LeMouel, J.L ., Plenier, G., 2008. Secular variation of the geomagnetic dipole during the past 2000 years. Geochem. Geophys. Geosyst. 9, Q01008. doi: 10.1029/2007GC001728. Valet, J.P., Meynadier, L., 1993. Geomagnetic field intensity and reversals during the past four million year s. Nature 366, 234-238. Valet, J.P., Meynadier, L., Guyodo, Y., 2005. Geomagnetic dipole strength and reversal rate over the past two million years. Nature 435, 802-805. Vanyo, J.P., Dunn, J.R., 2000. Core prece ssion: flow structures and energy. Geophys. J. Int. 142, 409-425. Venuti, A., Florindo, F.,Michel, E., Hall, I.R ., 2007.Magnetic proxy for the deep (Pacific) western boundary current variability across t he mid-Pleistocene climate transition. Earth Planet. Sci. Lett. 259, 107-118. Verhoogen, J., 1956. Ionic order ing and self-reversal in impur e magnetites. J. Geophys. Res. 61, 201-209. Verhoogen, J., 1962. Oxidation of iron-titanium oxides in igneous rocks. J. Geol. 70, 168-181. Voelker, A.H.L., Sarnthein, M., Grootes, P.M., Erlenkeuser, H., Laj, C., Mazaud, A., Nadeau, M.J., Schleicher, M. 1998. Correlation of marine 14C ages from the Nordic Seas with GISP2 isotope record: implications for 14C calibration beyond 25 ka BP. Radiocarbon 40, 517-534. Wagner, G., Beer, J., Laj, C., Kissel, C., Masa rik, J., Muscheler, R., Synal, H.A., 2000. Chlorine-36 evidence for the Mono Lake event in the Summit GRIP ice core. Earth Planet. Sci. Lett. 181, 1-6.

PAGE 245

245 Weeks, R., Laj, C., Endignoux, L., Fuller, M., Roberts, A., Manganne, R., Blanchard, E. and Goree, W., 1993. Improvements in l ong-core measurement techniques applications in paleomagnetism and paleoc eanography. Geophys. J. Int., 114: 651-662. Winterwerp, J.C., van Kesteren, W.G.M., 20 04. Introduction to the physics of cohesive sediments in the marine environment. In: De velopments in Sedimentology, vol. 56. Elsevier, Amsterdam. Witte, W.K. and Kent, D.V ., 1988. Revised magnetostratigraphies confirm low sedimentation rates in Arctic Ocean cores. Quat. Res. 29, 43-53. Wu, C.C., Roberts, P.H., 2008. A precessi onally-driven dynamo in a plane layer. Geophys. Astrophys. Fluid Dyn. 102, 1-19. Xu, W., Peacor, D.R., Dollase, W.A ., Van Der Voo, R., Beaubeuf, R., 1997. Transformation of titanomagnet ite to titanomaghemite: A sl ow, two-step, oxidationordering process in MORB. Am. Mineral. 82, 1101-1110. Xuan, C., Channell, J.E.T. 2008a. Testing the relationship between timing of geomagnetic reversals/excursions and phase of orbital cycles using circular statistics and Monte Carlo simulations. Earth Planet. Sci. Lett. 268, 245-254. Xuan, C., Channell, J.E.T., 2008b. Origin of orbital periods in the sedimentary relative paleointensity records. Phys. Earth Planet. Int. 169, 140-151. Xuan, C., Channell, J.E.T., 2009. UPmag: MATLAB software for viewing and processing u-channel or other pass-through pal eomagnetic data. Geochem. Geophy. Geosyst. 10, Q10Y07, doi:10.1029/2009GC002584. Xuan, C., Channell, J.E.T., 2010. Origin of apparent magnetic excursions in deep-sea sediments from Mendeleev-Alpha Ridge (Arctic Ocean). Geochem. Geophy. Geosyst. in press, doi:10.1029/2009GC002879. Yamazaki, T., 1999. Relative paleointensit y of the geomagnetic field during Brunhes Chron recorded in North Pacific deep-sea se diment cores: orbital influence? Earth Planet. Sci. Lett. 169, 23-35. Yamazaki, T., Oda, H., 2002. Orbital influe nce on Earths magnetic field: 100,000-year periodicity in inclination. Science 295, 2435-2438. Yamazaki, T., Oda, H., 2005. A geomagnetic paleointensity sta ck between 0.8 and 3.0 Ma from Equatorial Pacific sediment cores. Geochem. Geophys. Geosyst. 6, Q11H20. doi:10.1029/2005GC001001. Yokoyama, Y., Yamazaki, T., 200 0. Geomagnetic paleointensity variation with a 100 kyr quasi-period. Earth Planet. Sci. Lett. 181, 7-14.

PAGE 246

246 Yokoyama, Y., Yamazaki, T., Oda, H., 2007. Geomagnetic 100-kyr variation extracted from paleointensity records of the equatorial and north pacific sediments. Earth Planets Space 59, 795-805. Zijderveld, J.D.A., 1967. A. C. demagnetization of rocks: Analysis of results. In: Collinson, D.W., Creer, K.M. Runcorn, S.K. (Eds.), Methods in Paleomagnetism. Elsevier, New York. pp. 254-286. Zijderveld, J.D.A., Hilgen, F.J., Langereis, C .G., Verhallen, P.J.J.M., Zachariasse, W.J., 1991. Integrated magnetostratigraphy and bi ostratigraphy of the upper Pliocenelower Pleistocene from the Monte Sing a and Crotone areas in Calabria, Italy. Earth Planet. Sci. Lett. 107, 697-714.

PAGE 247

247 BIOGRAPHICAL SKETCH Chuan g Xuan was born in Anhui, China, in November 1982. He finished his primar y school and middle school in a small town in central Anhui, China. In 1998, Chuang graduated from Feidong First High School and began his studies at the National Training Base of Geology at the China University of Geosciences in Wuhan, Chin a. Chuang obtained his Bachelor of Science degree in June 2002 with excellence, and was recommended admission to the graduate school of China University of Geosciences free from examinations. He then spent three years in the program of applied mathematics in the School of Physics and Mathematic s and was awarded the Master of Science degree in June 2005. From August 2005 to May 2 010, Chuang was supported by the Alumni Fellowship and the Gibson Dissertation Fello w ship of the University of Florida, and was pursuing a Ph.D. degree in the Department of Geological Sciences at the University of Florida. Chuang is a member of the American Geophysical Union and the Geological Society of America. He has been an author of thirteen peer reviewed research articles on top earth science journals, and fifteen abstracts and reports presented at international academic meetings. In addition to the chapters documented in this dissertation, Chuang have contributed to, and is a co author on, the following publications that are related to his dissertation work (1) Kaufman, D.S., Polyak, L., Adler, R., Channell, J.E.T., Xuan, C., 2008. Dating late Quaternary planktonic foraminifer Neogloboquadrina pachyderma from the Arctic Ocean by using amino acid racemization. Paleoceanography 23, PA3 224, doi: 10.1029/ 2008PA001618. (2) Channell, J.E.T., Hodell, D.A., Xuan C., Mazaud, A., Stoner, J.S., 2008. Age calibrated relative paleointensity for the last 1.5 Myr at IODP Site U1308 (North Atlantic). Earth Planet. Sci. Lett. 274, 5971. (3)

PAGE 248

248 Mazaud, A., Channell J.E.T., Xuan, C., Stoner, J.S. 2009. Upper and Lower Jaramillo polarity transitions recorded in IODP Expedition 303 North Atlantic sediments: implications for transitional field geometry. Phys. Earth Planet. Inter. 172, 131140. (4) Polyak, L., Bischof, J., Ortiz, J.D., Darby D.A., Channell, J.E.T., X uan, C., Kaufman, D.S., Lvlie, R., Schneider, D., Eberl, D.D., Adler, R., Council, E.A., 2009. Late Quaternary stratigraphy and sedimentation patterns in the western Arctic Ocean. Global Planet Change 68, 517. (5) Adler, R., Polyak, L., Ortiz, J.D., Kaufman, D., Channell, J.E.T., Xuan, C., Fornaciari, E., Grottoli, A., Sellen, E., 2009. The first sediment record of Late Quaternary environments in the western Arctic Ocean with a true millennial resolution: HOTRAX core HLY05038JPC, Mendeleev Ridge. Global Planet Change 68 1829. (6) Hodell, D.A., Minth, E.K., Curtis, J.H., McCave, I.N., Hall, I.R., Channell, J.E.T., Xuan, C., 2009. Variations in Iceland Scotland Overflow Water during the last interglacial period and glacial inception, Earth Planet. Sci. Lett. 288, 1019.