Magnetostratigraphy of Chinese loess–paleosol sequences

Magnetostratigraphy of Chinese loess –paleosol sequences

Qingsong Liu a ,?,Chunsheng Jin a ,Pengxiang Hu a ,Zhaoxia Jiang a ,Kunpeng Ge a ,Andrew P.Roberts b

a State Key Laboratory of Lithospheric Evolution,Institute of Geology and Geophysics,Chinese Academy of Sciences,Beijing 100029,China b

Research School of Earth Sciences,The Australian National University,Canberra,ACT 0200,Australia

a b s t r a c t

a r t i c l e i n f o Article history:

Received 30January 2015

Received in revised form 10July 2015Accepted 17July 2015

Available online 26July 2015Keywords:Chinese loess

Magnetostratigraphy Geomagnetic reversals Geomagnetic excursions

As one of the longest and most continuously deposited terrestrial sedimentary archives in the world,Chinese loess –paleosol sequences record paleoclimatic and paleomagnetic variations at a range of time scales.Magnetostratigraphic studies provide a ?rst-order chronological framework for Chinese loess sequences.In this review,we highlight recent developments in loess magnetostratigraphy,including pedostratigraphy based on magnetic susceptibility variations.We highlight progress in understanding the mechanisms by which the nat-ural remanent magnetization (NRM)is acquired and discuss in detail the ?delity of paleomagnetic recording in loess records,including the recording of magnetic polarity reversals,excursions,and relative paleointensity var-iations.Finally,we discuss future prospects for studies of loess NRM.

?2015Elsevier B.V.All rights reserved.

Contents 1.Introduction ..............................................................1402.

Overview of loess pedo-and magnetostratigraphy .............................................1412.1.Magnetic susceptibility variations in the Chinese loess ........................................1412.2.Paleomagnetic polarity reversals recorded in Chinese loess......................................1423.Origins and complexities of NRM acquisition in the Chinese loess .....

.................................1433.1.Magnetic mineral assemblages in Chinese loess/paleosol sequences..................................1433.2.NRM acquisition during loess deposition and pedogenesis ......................................1443.3.Methods for identifying loess ChRM ................................................1443.4.Fidelity of the loess ChRM.....................................................1443.5.Continuity of the Chinese loess record ...............................................1464.Paleomagnetic polarity reversals ....................

.................................1464.1.Matuyama –Brunhes (MB)reversal.................................................1464.2.Stratigraphic position and age of the MBB .............................................1464.3.Stratigraphic thickness of the MBB.................................................1504.4.Morphology of the transitional ?eld associated with the MBB.....................................1504.5.Other polarity reversal boundaries.................................................1515.Geomagnetic excursions in Chinese loess ................

.................................1535.1.Mono Lake and Laschamp excursions ...............................................1565.2.Blake excursion .........................................................1575.3.Other possible excursions.....................................................1596.RPI records from Chinese loess......................................................1597.Future studies .............................................................162Acknowledgments..............................................................162References ................................

.................................

163

Earth-Science Reviews 150(2015)139–167

?Corresponding author.

E-mail address:liux0272@fe1f8ab82f60ddccda38a0ef (Q.

Liu).

fe1f8ab82f60ddccda38a0ef/10.1016/j.earscirev.2015.07.0090012-8252/?2015Elsevier B.V.All rights

reserved.

Contents lists available at ScienceDirect

Earth-Science Reviews

j o u r n a l h om e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /e a r s c i r e v

1.Introduction

Chinese loess sequences consist of accumulations of wind-blown dust deposits on the Chinese Loess Plateau(CLP)(abbreviation refers the Table1),which are referred to as“Huangtu”in Chinese(yellow earth). Loess is produced via a composite‘loessi?cation’process that involves bi-ological,physical,and chemical processes associated with dust produc-tion,transportation,deposition,and post-depositional modi?cation of material.The dust is deposited in arid and semi-arid regions,and is then affected by rain,groundwater,freeze–thaw processes,snow,and biotur-bation in a weakly alkaline medium under oxidizing conditions.Carbon-ate precipitation then turns the dust into a buff-colored,mottled,porous and homogeneous deposit(Liu,1985).

The vast expanses of the Chinese loess(areal distribution of ~4.4×105km2and up to300m in thickness)(Fig.1)make these deposits unique continental recorders of paleoclimate variations over different time scales since at least the late Miocene(Heller and Liu, 1986;Kukla et al.,1988;Maher and Thompson,1991,1992,1995; Banerjee et al.,1993;An and Porter,1997;Liu et al.,1999;Ding et al., 2002),but in places since the early Miocene(b~22Ma,Guo et al., 2002b)or even the late Oligocene(Qiang et al.,2011).Paleoclimatic patterns recorded in the CLP can be well correlated to global paleocli-matic signals,e.g.,from marine sedimentary and ice core records (Heller and Liu,1986;Bloemendal et al.,1995;Porter and An,1995; Ding et al.,2002;Nie et al.,2008a,b),not only at orbital time scales but also at millennial time scales(Chen et al.,1997;Sun et al.,2010, 2011).Therefore,Chinese loess–paleosol sequences play an important role in understanding the global climate system and especially for better understanding dynamic coupling between atmosphere,hydrosphere and lithosphere(An,2000;Maher et al.,2010;Hao et al.,2012),for ex-ample,between terrestrial and marine systems(Porter and An,1995; Guo et al.,1998b;Bailey et al.,2011;Roberts et al.,2011),and between tectonic events and climatic responses(Fang et al.,1999b;An et al., 2001;Nie et al.,2014a,b).

For the uppermost sediments that span the last glacial–interglacial cycle,an integrated approach has been adopted to establish an accurate chronological framework,which includes14C(e.g.,Lang et al.,2003), thermoluminescence(e.g.,Wintle,1990;Berger et al.,1992),and opti-cally stimulated luminescence(OSL)dating(e.g.,Lai et al.,2007;Lu et al.,2007).In some cases,identi?cation of geomagnetic excursions (e.g.,Zhu et al.,1994a;Reinders and Hambach,1995)and peak matching of normalized remanence signals with global relative geo-magnetic intensity reference curves(Hambach et al.,2008;Zeeden et al.,2009)have been used to constrain age models,along with more conventional correlation between paleoclimatic proxies(e.g.,magnetic susceptibility and grain size)and standard paleoclimate target curves (Clemens and Prell,1990;Maher and Thompson,1992).However,for long loess sequences,magnetostratigraphy is almost the only method that enables development of a?rst-order chronological framework. Once a robust magnetostratigraphy has been obtained,peak matching of loess magnetic records to marine oxygen isotope records(Heller and Liu,1986;Porter and An,1995;Ding et al.,2002)or orbital tuning (Sun et al.,2006)can be conducted.

Liu and Chang(1961)proposed the?rst detailed Chinese loess petrostratigraphy at the6th International Union for Quaternary Re-search(INQUA)Congress in Poland,but they provided no accurate age model.Subsequent magnetostratigraphic studies were conducted on loess outcrops or drill cores,such as at Wucheng,Luochuan,and Longxi (e.g.,Li et al.,1974;An et al.,1977;Wang et al.,1980;Wang and Li, 1982).These pioneering studies provided a preliminary chronological framework for the Chinese loess,although there were limitations due to the measurement techniques used and uncertainties in the strati-graphic pisions proposed.Heller and Liu(1982)provided the?rst reliable magnetostratigraphy for the Luochuan section,from the central CLP,which extended to below the Olduvai subchron.A phase of exten-sive investigation of paleomagnetic signals recorded by Chinese loess then ensued(Heller and Evans,1995;Evans and Heller,2001),which included analyses of the Gauss/Matuyama(G/M)(Zhu et al.,2000a), Matuyama/Brunhes(M/B)(Sun et al.,1993;Zhu et al.,1993,1994b; Guo et al.,2001;Spassov et al.,2001;Wang et al.,2007;Yang et al., 2007b,2010;Jin and Liu,2010,2011b),upper Jaramillo(Zhu et al., 1994b;Guo et al.,2002a),and upper Olduvai reversals(Yang et al., 2008),along with analysis of short-period geomagnetic excursions (Zhu et al.,1994a,1999,2006b,2007;Zheng et al.,1995;Fang et al., 1997;Pan et al.,2002;Yang et al.,2004;Wang et al.,2010),and at-tempts to determine geomagnetic relative paleointensity variations (Pan et al.,2001;Liu et al.,2005a).

Among these studies,the most serious questions raised concern the ?delity of paleomagnetic recording in Chinese loess(Zhou and Shackleton,1999).These studies pointed out that the M/B boundary (MBB)is recorded in Chinese loess unit L8(a cold period),but in marine isotope stage(MIS)19in marine sediments(a warm period)(Tauxe et al.,1996).Inconsistency between the positions of the MBB and paleo-climatic boundaries may indicate that the MBB recorded in Chinese loess has been displaced downward(by a few tens of cm to over 300cm)due to magnetization lock-in processes(Hyodo,1984)if it is assumed that the paleoclimatic stratigraphy of loess and marine sedi-ments is consistent(Zhou and Shackleton,1999).Alternatively,it has been proposed that there exists a large phase lag between continental and marine climate by assuming global consistency in the recorded po-sition of the MBB(Zhu et al.,1998).Understanding complexities associ-ated with lock-in of paleomagnetic signals has remained a key issue in magnetostratigraphic studies of Chinese loess.

In this review,we summarize how magnetostratigraphy has contrib-uted to understanding paleoclimatic and paleomagnetic information carried by the Chinese loess over different time scales.In the following sections,we overview Chinese loess magnetostratigraphy,natural rem-anent magnetization(NRM)acquisition mechanisms in Chinese loess, paleomagnetic polarity reversals(especially the characteristics of the MBB)and relative paleointensity(RPI)records from Chinese loess. Finally,we point to important future research directions for ongoing magnetostratigraphic studies on the fe1f8ab82f60ddccda38a0efpared to previous reviews

Table1

List of major abbreviations used in this paper.

ARM Anhysteretic remanent magnetization

ChRM Characteristic remanent magnetization

CRM Chemical remanent magnetization

DRM Depositional remanent magnetization

IRM Isothermal remanent magnetization

NRM Natural remanent magnetization

pDRM Post-depositional remanent magnetization

SIRM Saturation isothermal remanent magnetization

VRM Viscous remanent magnetization

χLow-?eld magnetic susceptibility,mass-speci?c

κLow-?eld magnetic susceptibility,volume-speci?c

AF Alternating?eld

CBD Citrate–bicarbonate–dithionite

CLP Chinese Loess Plateau

G/M Gauss/Matuyama

GMB Gauss/Matuyama boundary

GPTS Geomagnetic polarity time scale

LE Laschamp Excursion

LTC Low-temperature cycling

M/B Matuyama/Brunhes

MBB Matuyama/Brunhes boundary

MIS Marine isotope stage

MLE Mono Lake Excursion

OSL Optically stimulated luminescence

PSD Pseudo-single domain

RPI Relative paleointensity

SAR Sediment accumulation rate

SD Single domain

SML Surface mixed layer

SP Superparamagnetic

VGP Virtual geomagnetic pole

140Q.Liu et al./Earth-Science Reviews150(2015)139–167

of loess magnetism (e.g.,Heller and Evans,1995;Evans and Heller,2001;Liu et al.,2007),we provide an up-to-date and comprehensive re-view of loess magnetostratigraphy,including new developments over the last decades.Especially,we highlight progress in assessing and de-termining the mechanisms by which the NRM is acquired in Chinese loess.Mismatches between the positions of paleomagnetic reversal boundaries (e.g.,the MBB)and the corresponding pedostratigraphy (i.e.,the stratigraphic and spatial relationships of surface and buried soils)are discussed in detail.

2.Overview of loess pedo-and magnetostratigraphy

Magnetostratigraphy is a branch of stratigraphy in which strati-graphic pisions are made on the basis of changes in magnetic signals (e.g.,magnetic susceptibility,paleomagnetic polarity)through time (Opdyke and Channell,1996).In a narrow sense,magnetic stratigra-phy is often used to refer to the magnetic polarity stratigraphy as in-dicated by the geomagnetic polarity pattern recorded by rocks or sediments.The polarity is de ?ned to be normal if paleomagnetic records parallel the modern ?eld direction,while it is de ?ned to be reversed if the ?eld was in an antiparallel orientation.In a broader sense,magnetostratigraphy can also be used to describe other mag-netic variations,such as magnetic property,geomagnetic secular variation and relative geomagnetic paleointensity.Magnetic polarity stratigraphy has been widely used in the Earth sciences because geo-magnetic reversals are globally synchronous and well determined,

and because it is suitable for constraining the age of long sequences back to at least ~165Ma (Opdyke and Channell,1996;Ogg,2012).In addition to providing chronological information,paleomagnetic results provide useful information about the deep-Earth dynamo that generates the geomagnetic ?eld.Below,we brie ?y summarize the pedostratigraphic subpision of the Chinese loess,which can be well de ?ned on the basis of magnetic susceptibility variations.We then provide a detailed overview of published paleomagnetic results from the Chinese loess.

2.1.Magnetic susceptibility variations in the Chinese loess

Loess sequences from the central CLP consist of ?ve units from top to bottom:Blake Loam (S0,Holocene),Malan Loess (L1,late Pleistocene),Lishi Loess (S1–L15,middle Pleistocene),Wucheng Loess (S15–L33,middle Pleistocene to late Pliocene),and the underlying red clay (late Pliocene to early Miocene)(Liu,1985;An,2000;Sun et al.,2006)(Fig.2).Traditionally,loess and paleosol units are termed “Ln ”and “Sn ”,respectively.Within loess and paleosol units,sub-loess and sub-paleosol units are termed LnLLm and LnSSm,respectively.The numbers n and m indicate the sequential number of the unit from top to bottom.The low-?eld magnetic susceptibility (χ,mass-speci ?c;κ,volume-speci ?c)from CLP sequences undergoes periodic stratigraphic variations.Generally,paleosol units are magnetically enhanced due to neoformation of nano-sized ferrimagnets during pedogenesis (e.g.,Zhou et al.,1990;Maher and Thompson,

1991;

Fig.1.Schematic map of the geographic distribution of Chinese loess and the locations of loess sections mentioned in the text (circles).

141

Q.Liu et al./Earth-Science Reviews 150(2015)139–167

Banerjee et al.,1993;Liu et al.,2007).The fact that susceptibility peaks occur in paleosols due to pedogenic magnetic mineral forma-tion during warm/humid periods and low susceptibility values occur in loess units that were deposited during cool/arid phases means that paleosols correspond to interglacial (odd-numbered)MISs and loess units correspond to glacial (even-numbered)stages (Martinson et al.,1987).

The last two paleosol units S0and S1can be correlated unambigu-ously to MIS 1and 5,fe1f8ab82f60ddccda38a0efing the magnetic enhancement of paleosol units,S2,S3,and S4can be reasonably correlated to MIS 7,9,and 11,respectively.However,this approach is not suitable for older paleosols.Chronological analysis has,therefore,long been a signif-icant issue for Quaternary Chinese loess investigations.Although a ?rst-order chronological framework can be well-established using magnetic polarity stratigraphy (Fig.2),it can be dif ?cult to determine accurate ages for each loess and/or paleosol unit.In an early loess study,an age model was constructed using loess bulk density (Liu,1985),but this ap-proach was not used more widely.Following identi ?cation of similarity between loess susceptibility and the marine stable oxygen isotope curve (Heller and Liu,1984),a loess susceptibility timescale has been con-structed since the middle Pleistocene based on linear interpolation be-tween tie points of known age,assuming a constant loess deposition rate between tie points (Kukla et al.,1988).This means that loess accu-mulation rates can be assessed as a function of time (Kukla et al.,1988).Susceptibility-based age models have been widely adopted in subse-quent loess studies and susceptibility has been tuned astronomically to constrain accurately the loess stratigraphy (Lu et al.,1999;Heslop et al.,2000).Age models have also been developed using loess grain size (Ding et al.,1994,2002;Porter and An,1995;Sun et al.,2006).Hao et al.(2012)correlated loess magnetic susceptibility curves to

marine oxygen isotope records based on a grain-size age model to dis-cuss Arctic ice sheet evolution.

2.2.Paleomagnetic polarity reversals recorded in Chinese loess

Loess magnetostratigraphic studies have developed signi ?cantly in the decades since the earliest magnetic polarity results were presented for the Wucheng (Li et al.,1974),Luochuan (An et al.,1977),and Longxi sections (Wang and Li,1982).Earlier magnetostratigraphic results contained major uncertainties,such as concerning the stratigraphic po-sitions of geomagnetic reversal boundaries.Heller and Liu (1982)?rst established a robust magnetic polarity pattern and a ?rst-order chrono-logical framework for Quaternary Chinese loess deposits.Since then,magnetic stratigraphies have been reported from various CLP sections,such as at Xifeng (Kukla,1987;Liu et al.,1988)and Luochuan (Ge,1984;Liu and An,1984;Liu,1985;Heller et al.,1987;An et al.,1989;Liu et al.,2010)in the central CLP,at Weinan (Ge et al.,1991;Pan et al.,2002)and Baoji (Rutter et al.,1990,1991;Evans et al.,1991)in the southern CLP,at Mangshan (Zheng et al.,2007)and Sanmenxia (Wang et al.,2005)in the southeastern CLP,at Jingyuan (Yue et al.,1991),Jiuzhoutai (Burbank and Li,1985;Rolph et al.,1989)and Xining (Zeng et al.,1993)in the western CLP,as well as for sections outside the CLP,such as the Fanshan section near Beijing (Xiong et al.,2001),the Linxia (Li et al.,1997)and Dazhaicun sections in west Qinling (Fang et al.,1999a ),the Ganzi section in west Sichuan Province (Qiao et al.,2006),and the Dongwan section in the Tianshan area (Fang et al.,2002).Loess in the southern and western CLP and outside of the CLP has generally accumulated since the middle Pleistocene,but in the southern and central CLP,loess is preserved throughout the Quaternary.Nevertheless,the stratigraphic positions of geomagnetic reversal boundaries are spatially consistent to ?rst order across all studied regions.

The base of the Chinese loess has been traditionally designated to occur at the base of loess unit L33(~2.6Ma),which is close to the Pli-ocene –Pleistocene boundary (Gibbard et al.,2010).However,two units (S33and L34)have been identi ?ed between L33and the up-permost Pliocene “red clay ”that underlies the Quaternary loess at Lingtai,Jingchuan,Baoji,and Lantian (Yang et al.,2010).The pedostratigraphy of units S33and L34is similar to that of overlying loess –paleosol sequences,and is different from that of the underly-ing “red clay ”.Thus,the base of the Chinese loess could extend to 2.8Ma,which is shortly before the Gauss –Matuyama boundary (GMB),and below the recently modi ?ed position of the Pliocene –Pleistocene boundary (2.58Ma),as proposed by the International Commission on Stratigraphy (ICS)and INQUA (Gibbard et al.,2010).

Magnetostratigraphic studies of western CLP loess sequences have also led to extension of the age of earliest eolian dust accumu-lation on the CLP to the early Miocene,and possibly even to the late Oligocene in the so-called Tertiary red clay.Such work also sug-gests an earlier origin for the East Asian paleomonsoon,arid conti-nental dust source regions and energetic westerly winter monsoon winds (Guo et al.,2002b;Qiang et al.,2011;Nie et al.,2014a ).A red clay section at Baoji with a thickness of 50to 60m was ?rst re-ported and dated back to 4or 5Ma (Evans et al.,1991).The Lantian red clay section near Xi'an was then dated to 5Ma (Zheng et al.,1992).The red clay sections at Xifeng (Bajiazui section,nearly 66m in thickness)and Lantian (nearly 50m in thickness)were dated back to 6.6Ma and 6.8Ma,respectively,by magnetostratigraphy (Sun et al.,1997).The Zhaojiachuan red clay section in the Xifeng area was dated to 7.6Ma (Sun et al.,1998b )(Fig.3b).The Lingtai red clay section (nearly 130m in thickness)in the Pingliang area was then dated back to 7.05Ma (Ding et al.,1998b,1999b )and to 7.2Ma (Sun et al.,1998a )(Fig.3a),respectively.The Jingchuan red clay section,which is also in the Pingliang area and has a similar thickness (126m)to the Lingtai section,was dated to 7.7Ma (Ding et al.,2001)(Fig.3d).The Jiaxian red clay section (60m in thickness)

r u

n h e a t u y a m i l b e r

l o e s s -p a l e o s o l r e d c l a y

Wucheng loess

Lishi loess

Black Loam Loess Paleosol

Reddish Loess Reddish Paleosol

Fig.2.Pedostratigraphy (left),magnetostratigraphy (right),and depth plots of magnetic susceptibility (blue lines)and mean grain size of bulk sediment (black line)and quartz (red line)for a loess –paleosol sequence from Lingtai (modi ?ed from Sun et al.(2006)).The dashed line de ?nes the boundary between the late Neogene red clay formation and the Quaternary loess –paleosol sequence.On the polarity log,black =normal polarity,and white =reversed polarity.Abbreviations for geomagnetic chrons are:Gi =Gilbert,G =Gauss,M =Matuyama,and B =Brunhes;and subchrons are:O =Olduvai and J =Jaramillo.(For interpretation of the references to color in this ?gure legend,the reader is referred to the web version of this article.)

142Q.Liu et al./Earth-Science Reviews 150(2015)139–167

in Shaanxi Province was dated back to 5.23Ma by Ding et al.(1998a)and to 7.2Ma by Qiang et al.(2001)(Fig.3e),respectively.The Chaona section in the Lingtai area from the central CLP was identi ?ed as the so-called “oldest ”red clay section,with a thickness of 128m and dated back to 8.1Ma (Song et al.,2000,2001)(Fig.3c).This “oldest ”age for the red clay sequences was soon surpassed by the Qin'an red clay in Gansu Province (western CLP),for which an oldest age of 22Ma (early Miocence)was reported (Guo et al.,2002b;Hao and Guo,2007),and by the Shilou red clay in Shanxi Province,which has a reported basal age of 11Ma (Xu et al.,2009)(Fig.3f).Therefore,there is no ?rm consensus on the maximum age of the red clay.These magnetostratigraphic studies provide a generally convincing ?rst-order chronological framework and were usually constructed by cor-relating the measured polarity to the geomagnetic polarity time scale (GPTS)of Cande and Kent (1995).

In addition to chronological studies based on magnetic polarity stra-tigraphy,polarity reversal processes have been widely studied from the Chinese loess for the MBB,the upper and lower Jaramillo and Olduvai boundaries,and the GMB.These studies concern the stratigraphic position of the respective reversal boundaries and the duration and morphology of the transitional ?eld.The validity of the results of these studies depends crucially on paleomagnetic recording ?delity and remanence acquisition mechanism(s)in the Chinese loess.Ambiguities remain in inpidual studies mostly because of discrepancies in assigning the exact stratigraph-ic position of paleomagnetic reversal boundaries (e.g.,the MBB)(Liu et al.,2008).We consider in more detail below the origins and complexities as-sociated with NRM acquisition in Chinese loess.

3.Origins and complexities of NRM acquisition in the Chinese loess 3.1.Magnetic mineral assemblages in Chinese loess/paleosol sequences Magnetic mineral assemblages in Chinese loess/paleosol sequences mainly contain magnetite,maghemite,hematite and goethite (see review of Liu et al.,2007).Magnetic iron sul ?des have not been reported because these sequences predominantly have oxic depositional and post-depositional environments.Magnetic minerals in Chinese loess/paleosol sequences and the underlying red clay units have detrital (eo-lian)and pedogenic origins (Evans and Heller,1994;Mishima et al.,2001;Spassov et al.,2003b;Liu et al.,2007;Nie et al.,2014a,b ).In par-ticular,the coarsest magnetic particles,including pseudosingle domain (PSD)magnetite,have a eolian origin (Maher and Thompson,1992;Spassov et al.,2003b;Hu et al.,2013).Due to low-temperature oxida-tion,partially oxidized magnetite particles have a stoichiometric magnetite core surrounded by a maghemite rim.The lattice mismatch between the magnetite core and maghemite rim increases the coerciv-ity of remanence of the partially oxidized magnetite,and stabilizes its remanence (Cui et al.,1994;van Velzen and Dekkers,1999;Liu et al.,2003a ).Other than this maghemite rim,most maghemite par-ticles formed during pedogenesis and occur in the superparamagnetic (SP)and stable single domain (SD)states (Verosub et al.,1993;Heller and Evans,1995;Liu et al.,2003a,2007;Nie et al.,2013;Yang et al.,2013).

Eolian hematite is present in both loesses and paleosols,and it is generally coarser-grained (micron-size)than pedogenic hematite (less than several 100nm)(Chen et al.,2010;Hu et al.,2013).In ad-dition,eolian hematite has a higher coercivity of remanence (~1T)than pedogenic nano-sized hematite (~130mT)(Hu et al.,2013).Therefore,pedogenic hematite is more prone to dissolution by cit-rate –bicarbonate –dithionite (CBD)treatment (Hu et al.,2013),which is a commonly applied method for separating pedogenic from detrital iron minerals (Mehra and Jackson,1960;Verosub et al.,1993).Aluminum is ubiquitous in natural environments,and can,therefore,be easily incorporated into the crystal structure of pedogenic hematite (Cornell and Schwertmann,2003).Goethite usually co-exists with hematite,although it appears to have a domi-nantly eolian origin and its concentration tends to be independent of pedogenesis (Hu et al.,2013

).

Fig.3.Magnetic polarity logs for Neogene red clay sediments at the:(a)Lingtai section (Sun et al.,1998b ),(b)Zhaojiachuan section in the Xifeng area (Sun et al.,1998a ),(c)Chaona section (Song et al.,2000,2001),(d)Jinchuan section (Ding et al.,2001),(e)Jiaxian section (Qiang et al.,2001),and (f)Shilou section (Xu et al.,2009).The geomagnetic polarity timescale (GPTS)is the astronomically tuned Neogene time scale (ATNTS)of Hilgen et al.(2012).Black =normal polarity;white =reversed polarity.

143

Q.Liu et al./Earth-Science Reviews 150(2015)139–167

3.2.NRM acquisition during loess deposition and pedogenesis

During glacial periods,coarse-grained(micron-sized)eolian he-matite and magnetite was transported from dust source regions, and was then deposited across the CLP.Similar to lake and marine sediments,a depositional remanent magnetization(DRM)or a com-bination of DRM and a post-depositional remanent magnetization (pDRM)is acquired by Chinese loess(Zhao and Roberts,2010), where pDRM acquisition involves a gradual lock-in process(Irving and Major,1964;Kent,1973;Hyodo,1984;Tauxe et al.,1996).Sub-sequently,magnetic particles with small energy barriers will acquire a secondary viscous remanent magnetization(VRM),which grows with the logarithm of time(Stacey and Banerjee,1974)during longer-term exposure to the ambient geomagnetic?eld.

During interglacial periods,pedogenic maghemite and hematite nanoparticles form in situ as soils develop.Pedogenic maghemite spans the SP/SD threshold size(~20–30nm)(Zhou et al.,1990;Liu et al.,2005b;Nie et al.,2013),and thus a portion of this maghemite will have stable SD properties and will carry a relatively stable chemical remanent magnetization(CRM).However,because of their relatively small grain size,the particles should have low unblocking temperatures (Dunlop and?zdemir,1997;Liu et al.,2005b),and maghemite near the SP/SD threshold size will be prone to remagnetization via VRM acquisi-tion in the modern geomagnetic?eld.Therefore,an initially acquired DRM will be contaminated by a CRM and VRM carried by pedogenic maghemite and hematite.Wang and L?vlie(2010)observed that coarse-grained eolian magnetite can be remagnetized by physical rota-tion in the presence of the geomagnetic?eld due to reworking associat-ed with freezing and thawing in winter.Jin and Liu(2011a)also proposed that coarse magnetite particles,e.g.,in speci?c loess intervals such as in L9,can be remagnetized by VRM acquisition because of the weak coercivity of remanence of multi-domain magnetic particles. Zhao and Roberts(2010)proposed that water plays an important role in remanence acquisition in loess because it allows magnetic particles that are poorly oriented during deposition to align more ef?ciently with the geomagnetic?eld before being mechanically locked in place during subsequent drying.Zhao and Roberts(2010),therefore,pro-posed that the NRM of Chinese loess/paleosol sediments is a mixture of DRM,pDRM,CRM,and VRM.For the former two mechanisms, water content provides the major control on which remanence acquisi-tion mechanism is dominant.

Liu and Zhang(2013)observed a positive correlation between the characteristic remanent magnetization(ChRM)intensity of Chinese loess/paleosols and the degree of pedogenesis experienced by samples. This strongly indicates that VRM and CRM could have partially contam-inated the initial DRM,but to different degrees for little-altered loess compared to pedogenically-altered fe1f8ab82f60ddccda38a0efing low-temperature cycling(LTC),Liu et al.(2003b)characterized the ChRM(obtained after thermal demagnetization at300°C)carried by loess and imma-ture/mature paleosols.They observed a strong Verwey transition within loess,which indicates that coarse-grained detrital magnetite is the dominant ChRM carrier for loess.In contrast,the Verwey transition is progressively smeared with increasing degree of pedogenesis.This indi-cates that the ChRM of paleosols is carried by maghemite or highly oxidized magnetite.

3.3.Methods for identifying loess ChRM

Because the NRM of loess is a mixture of different remanent mag-netizations,it is necessary to isolate the ChRM carried by detrital iron oxides to isolate the primary component of magnetization.There are three major demagnetization methods:chemical,alternating?eld (AF),and thermal demagnetization(Schmidt,1993).Pedogenic iron oxides occur as nanoparticles,therefore,chemical demagnetiza-tion would theoretically be ef?cient for separating detrital from sec-ondary pedogenic components(Tan et al.,2003).However,it is practically dif?cult to conduct chemical demagnetization on loess samples.AF demagnetization is often not effective in removing VRM components from Chinese loess(Jin and Liu,2011a)because low-temperature oxidation of detrital magnetite can signi?cantly in-crease its coercivity due to unit cell mismatches between the maghemite rim and magnetite core(Housden and O'Reilly,1990; van Velzen and Dekkers,1999;Liu et al.,2003a,2004a,b,2005a). VRM is a thermally activated process and maghemite can be re-moved by heating to~350°C.Thermal demagnetization has,there-fore,become the most widely used method to remove secondary VRM and CRM overprints and to de?ne the ChRM of loess/paleosol samples by heating to300–550°C(Fang et al.,1997;Pan et al., 2001;Jin and Liu,2011a,b).However,for samples from the western CLP,the NRM is carried dominantly by partially oxidized magnetite (Liu et al.,2004b).Liu et al.(2004b)found that thermal treatment destroys the magnetite–maghemite core–shell structure and re-duces the NRM stability carried by PSD magnetite.Therefore,Liu et al.(2005a)proposed that AF treatment is superior to thermal de-magnetization for isolating the ChRM carried by partially oxidized PSD/MD magnetite.

3.4.Fidelity of the loess ChRM

Post-depositional disturbance is an important factor when consider-ing sedimentary NRM acquisition.For lake and marine sediments,sur-face sediment can be mixed by bioturbation to form a surface mixed layer(SML)(Boudreau,1994,1998).A long-term paleomagnetic signal can only be preserved when mixing tends to zero(at the base of the SML)and magnetic particles can lock in a magnetization.While down-ward displacement of paleomagnetic records in lake and marine sedi-ments is caused both by surface mixing and by pDRM lock-in (Channell and Kleiven,2000;Channell and Guyodo,2004;Roberts and Winklhofer,2004),these two processes have different effects on pDRM acquisition.The lock-in process serves as a low-pass?lter, which attenuates high-frequency signals(Roberts and Winklhofer, 2004).In contrast,surface mixing produces a downward shift of the recorded geomagnetic signal.High-frequency geomagnetic signals are lost by surface mixing,but mixing does not change the shape of the re-corded signal(Liu et al.,2008).For paleoclimatic signals,surface mixing also acts as a low-pass?lter,which attenuates the amplitude of high-frequency signals when increasing the thickness of SML,but the ampli-tude of low-frequency signals remains relatively unaltered(Nie et al., 2008a,2008b;Sun et al.,2010).By introducing a simple surface mixing model for a Chinese loess grain size record,Sun et al.(2010)estimated that the SML in the central CLP is only~5-cm thick.The total downward displacement of the MBB is~20cm at Lingtai and Zhaojiachuan(Liu et al.,2008).Therefore,the lock-in depth should be~15cm after subtracting the thickness of the loess SML.By comparing the timing of the Laschamp Excursion(LE)in the CLP with radiometric ages inferred from a Greenland ice-core(10Be?ux),Sun et al.(2013)found that the depth of downward displacement of the LE(dated at~41ka B.P.)grad-ually increases to the southeast due to the combined effects of pDRM lock-in,surface mixing,and intensi?ed pedogenic alteration(Fig.4).

Post-depositional sedimentary disturbances that signi?cantly affect the paleomagnetic record of loess can be detected using anisotropy of magnetic susceptibility.Zhu et al.(2006a)investigated short-term pa-leomagnetic signals recorded by loess at Datong and observed four paleomagnetic directional anomalies,three of which strongly correlate to anomalies of the principal axis of the inclination of the maximum magnetic susceptibility.These anomalies appear to have been caused by post-depositional disturbances.In contrast,one anomaly had a nor-mal magnetic fabric and was,thus,interpreted to represent an authentic record of the LE(Fig.5).

Although the position of major geomagnetic reversal boundaries such as the MBB can be determined readily from magnetostratigraphic studies,paleomagnetic directional variations within the MBB and

144Q.Liu et al./Earth-Science Reviews150(2015)139–167

other reversals are usually inconsistent both locally and from location to location.Jin and Liu (2010)investigated ten sub-sets of parallel samples from a block sample (about 10×10×2.5cm 3)across the MB reversal from Luochuan and reported inconsistent paleomagnetic directions for these sample sub-sets.The observed inconsistencies are interpreted to have been caused by low alignment ef ?ciency of mainly PSD magnetite grains associated with low ?eld intensities during paleomagnetic rever-sals (see reviews of Merrill and McFadden,1999;Valet and Herrero-Bervera,2011).Therefore,while it appears that loess sequences from the CLP can record geomagnetic reversals and excur-sions,limitations of the recording process mean that it is often not pos-sible to recover high-frequency geomagnetic variations from Chinese loess sequences (e.g.,Zhao and Roberts,2010).In particular,care should be taken when seeking to investigate the morphology of transitional ?elds from Chinese loess (Zhu et al.,1994a,b;Yang et al.,2010).Mea-surement of multiple sets of sub-samples (e.g.,Jin and Liu,2010)is a valuable approach for assessing the reliability of detailed paleomagnetic records from loess sequences.

16

151413

121110987654

3210

45

90

κ1-Inc. (o

)

-90

90

VGP Lat. (o )

05101520

κ (10-5 SI)

(b)(c)-90

90

Inc. (o )

-180

180

Dec. (o )

(f)A B

C (LE)D

45

90

κ3-Inc. (o )

(a)Lithology

Fig.5.Stratigraphic plots of (a)the lithological subpision,(b)magnetic susceptibility,(c)paleomagnetic inclination and (d)declination,(e)VGP latitude,(f)inclination of the maximum axis of susceptibility (K 1-Inc.),and (g)inclination of the minimum axis of susceptibility (K 3-Inc.)for the Datong section (from Zhu et al.,2006a ).The horizontal gray bars associated with the letters A –D mark four observed paleomagnetic directional anomalies.Black and blue dots denote results obtained with AF and thermal demagnetization,respectively.Blue arrows with numbers indicate the age controls.Anomalous magnetic fabrics for anomalies A,B and D indicate that these stratigraphic levels are disturbed and,therefore,do not represent real records of geomagnetic excursions.In contrast,anomaly C has a normal sedimentary magnetic fabric,which suggests that the recorded paleomagnetic anomaly could represent the Laschamp ex-cursion (LE)as indicated by the age control.(For interpretation of the references to color in this ?gure legend,the reader is referred to the web version of this article.)

fe1f8ab82f60ddccda38a0efparison of low-?eld magnetic susceptibility (κ),mean grain size and paleomagnetic inclination for (a)the Gulang section,with magnetic susceptibility and VGP latitude for the (b)Luochuan and (c)Weinan sections (Zhu et al.,2007).Gray bars denote the stratigraphic correlations of paleosol and sub-paleosol units.The green bar indicates the position of the Laschamp excursion in the last glacial loess deposits from each section (after Sun et al.,2013).(For interpretation of the references to color in this ?gure legend,the reader is referred to the web version of this article.)

145

Q.Liu et al./Earth-Science Reviews 150(2015)139–167

3.5.Continuity of the Chinese loess record

Dust deposition is not a continuous process,and wind can re-mobilize sediments deposited during previous dust deposition events.It should,therefore,not be surprising that loess records from the CLP are only quasi-continuous,and that continuity of re-cords is a critical concern when scrutinizing high-frequency paleo-magnetic and paleoclimatic signals.Good correspondence between Chinese loess–paleosol cycles and marineδ18O records indicates that Chinese loess deposition is continuous at orbital time scales (Ding et al.,2002).However,at?ner(e.g.,millennial)timescales, the continuity of loess sequences is debated(Porter and An, 1995;Chen et al.,1997;Zhu et al.,2007).By correlating rapid cli-matic events(Heinrich events and Bond cycles)between the North Atlantic region and the CLP,Chen et al.(1997)demonstrated that loess pro?les from the western CLP were deposited continuously at millennial scale.Sun et al.(2010)demonstrated that loess grain size variations from the western CLP,where sedimentation rates are high,mimic variations observed in high-resolution Greenland ice core records even at sub-millennial timescales.In contrast,based on closely spaced OSL dates for samples from S0 and the uppermost part of L1from the Beiguoyuan section,central CLP,Stevens et al.(2006)proposed that loess was not continuous-ly deposited at around10–15ka.

To further evaluate the continuity of Chinese loess deposition, Zhu et al.(2007)investigated the recording of geomagnetic excur-sions(the Mono Lake Excursion(MLE)and LE)along a west to east transect in the central CLP.Geomagnetic excursions have durations of a few thousand years(Laj and Channell,2007;Roberts,2008), therefore,the presence or absence of short-period excursions can be used to assess the continuity of loess sedimentation at millennial timescales.The Yichuan section is situated near the Yellow River val-ley between Shaanxi and Shanxi Provinces.The absence of the LE at Yichuan indicates highly episodic deposition in the valley where loess accumulation was probably affected by the local environment (Zhu et al.,2007).For the Luochuan and Xifeng sections,the LE was recorded,but the MLE was not.This is probably due to discontinuous deposition on timescales of less than2kyr(Zhu et al.,2007)and could also explain inconsistent recording of the LE at Luochuan, Xifeng,Weinan and Lingtai.Therefore,the continuity of Chinese loess records depends on site location,temporal(stratigraphic)var-iations at a given location,and the timescale under investigation.

To resolve problems associated with continuity of loess records,cre-ation of a composite record from different pro?les is a useful strategy (Ding et al.,2002).This can reduce the effects of missing information from a single pro?le.For example,the exact pedostratigraphy of L9 can be well de?ned using several pro?les.Ambiguities can then be re-duced when linking loess paleoclimatic proxies to marineδ18O records (Jin and Liu,2011a,b).

4.Paleomagnetic polarity reversals

4.1.Matuyama–Brunhes(MB)reversal

Among all paleomagnetic polarity reversals,the MB reversal,which occurred at about0.78Ma(Shackleton et al.,1990),was the last geo-magnetic reversal,and is recorded in a wide range of geological archives (Love and Mazaud,1997;Clement,2004;Singer et al.,2005,and refer-ences therein).The MB transition is the most studied polarity reversal and is an important age marker for dating geological events.Although the MB transition has been widely identi?ed through high-resolution magnetic polarity analysis of CLP sections,there remain inconsistencies in these MBB records in terms of its stratigraphic position at different lo-calities,its duration,and the inferred morphology of the transitional ?eld,which are discussed successively below.4.2.Stratigraphic position and age of the MBB

The MBB is usually recorded in loess unit L8(corresponding to a gla-cial period)at locations with the least pedogenic alteration,such as at Weinan(Zhu et al.,1994a;Pan et al.,2002),Lantian(Zheng et al., 1992),Lingtai(Ding et al.,1998b;Sun et al.,1998a;Spassov et al., 2001),Baoji(Rutter et al.,1991;Spassov et al.,2001;Yang et al., 2004),Jingbian(Ding et al.,1999a;Guo et al.,2002a)and Xifeng(Liu et al.,1988;Sun et al.,1993;Zhu et al.,1993)(Fig.6a–f).However,at the Luochuan(Heller and Liu,1982;Liu et al.,2010)(Fig.6h)and Sanmenxia sections(Wang et al.,2005)(Fig.6g),the MBB has been re-ported to occur within the upper part of paleosol unit S8(an interglacial period).In marine sediments,the MBB is located in interglacial MIS19 (e.g.,deMenocal et al.,1990;Tauxe et al.,1996;Channell et al.,2010). To reconcile this inconsistency,Zhou and Shackleton(1999)proposed that paleomagnetic signals recorded by Chinese loess are misplaced be-cause of a large magnetic lock-in depth.Deep but variable lock-in(tens of cm to3m)has been proposed for the MBB from units S7to L8or S8in different areas(Zhou and Shackleton,1999),although this interpreta-tion has been contested in many other loess studies in which shallow lock-in depths have been inferred(Zhu et al.,1994a,1998,2006b;Pan et al.,2002;Wang et al.,2006;Liu et al.,2008;Yang et al.,2008, 2010).Sediment redeposition experiments involving eolian dust from Chinese loess suggest that ChRM carriers can be more ef?ciently aligned with the ambient?eld after initial wetting(Wang and L?vlie,2010; Zhao and Roberts,2010).If water is available via rainfall on the CLP shortly after deposition,shallow ChRM acquisition is apparently possi-ble.Such shallow lock-in is consistent with the recording of short-period geomagnetic events in Chinese loess,such as excursions (e.g.,Zhu et al.,1994b,1999,2006b,2007;Zheng et al.,1995;Fang et al.,1997;Pan et al.,2002;Yang et al.,2007a).Thus,downward dis-placement of polarity boundaries in Chinese loess may be limited to depths of only several cm,where the surface mixed layer(SML)has a limited thickness(~5cm)(Sun et al.,2010),and where shallow NRM lock-in appears to have occurred(Liu et al.,2008).These lines of evi-dence indicate that measured reversal boundaries in Chinese loess can be considered reliable to?rst order and useful for providing age control that can be correlated to marine sedimentary records.The chronological framework for Chinese loess based on large pDRM lock-in depths (e.g.,Heslop et al.,2000)should,therefore,be reconsidered and alterna-tive explanations should be sought to account for the apparent record-ing of the MB reversal in glacial L8sediments rather than in interglacial sediments as is the case in the deep sea.

In contrast to the above?ndings,Spassov et al.(2003a)proposed a combined model that includes both conventional pDRM lock-in and CRM acquisition.By assuming that chemical alteration of magnetic min-erals can occur at any depth,large downward displacements of the MBB can be simulated.However,the combined pDRM+CRM model needs a considerable CRM contribution,and is dif?cult to reconcile with the fact that most MBB records from the CLP have been obtained from areas with relatively weaker pedogenesis in L8.Therefore,deep CRM acquisi-tion apparently cannot account for the seemingly large lock-in depth suggested by Zhou and Shackleton(1999).

Tauxe et al.(1996)compiled19marine sedimentary records of the MBB with age control based on benthic and planktonicδ18O,and other paleoclimatic proxies,and concluded that the MBB is recorded in the middle of MIS19without apparent downward displacement.The com-bined effects of lock-in and sur?cial mixing on the NRM were,therefore, concluded to be weak for marine sediments.Previously,deMenocal et al.(1990)correlated the stratigraphic location of the MBB within ma-rine sediments in MIS19to sediment accumulation rate(SAR).By ex-trapolating a linear trend to zero SAR,a consistent downward displacement of the MBB was estimated at~16cm below the sediment surface.A limitation of the approaches of deMenocal et al.(1990)and Tauxe et al.(1996)is that they used a range of paleoclimate proxies to de?ne the boundary between MIS19and18and did not consider the

146Q.Liu et al./Earth-Science Reviews150(2015)139–167

potentially important in ?uence of millennial scale variability in age of different oceanic water masses (e.g.,Skinner and Shackleton,2005),which can introduce unwanted age uncertainties into such analyses.A

better approach is to compare parallel records of the MBB that lie below the same water mass with age control for both records from ei-ther planktic or benthic δ18O (Liu et al.,2008).This substantially reduces the number of records from which lock-in depths can be analyzed meaningfully.Liu et al.(2008)used the same approach as deMenocal et al.(1990)but separately compared paired benthic and planktonic δ18O records from the northeast Atlantic Ocean and western equatorial Paci ?c Ocean,respectively.MBB offsets for marine sediment records from these two regions were consistently estimated at about fe1f8ab82f60ddccda38a0efing the same approach,the MBB offset for the Chinese loess is also de-termined to be ~20cm (Liu et al.,2008).Thus,it appears that the real stratigraphic position of the MBB is not in the middle of MIS 19,but in the upper part of MIS 19or close to the boundary between MIS 19and 18after considering the ~20cm MBB offset (Liu et al.,2008;Channell et al.,2010;Suganuma et al.,2015)(Fig.7e –h).

For Chinese loess records,the stratigraphic position of the MBB has been determined mainly on the basis of two factors:interpretation of susceptibility records and paleomagnetic determination of the position of the MBB.Previous studies have shown that susceptibility is strongly enhanced by neoformation of maghemite nanoparticles via pedogenesis (Zhou et al.,1990;Maher and Thompson,1991;Liu et al.,2007,2013;Yang et al.,2013),where paleoprecipitation played the most important role in determining the degree of pedogenesis (Maher and Thompson,1995;Han et al.,1996;Maher et al.,2003;Geiss et al.,2008;Balsam et al.,2011;Liu et al.,2013).The spatial pattern of paleoprecipitation was not uniform (Hao and Guo,2005),and although paleoprecipitation estimates have been made through time for given locations,ambiguities exist concerning the resolution of paleoprecipitation estimates on sub-orbital timescales (Heslop and Roberts,2013)and with respect to de ?n-ing the exact positions of paleoclimatic boundaries (Liu et al.,2004a ).Liu et al.(2008)observed that if sharp susceptibility variations are used to de ?ne the boundary between L8and S8,the MBB is located at the bottom of L8at Lingtai,but in the upper part of S8at Zhaojiachuan (Fig.7c,d).Instead,when using quartz grain size,which is insensitive to pedogenic alteration,to de ?ne the paleoclimatic boundary between S8and L8,MBB records from these two pro ?les both occur in the upper part of S8,which is consistent with the position of the MBB in ma-rine records (Fig.7a,b).The MBB at these two sections is reported as an abrupt reversal boundary due to low sampling resolution (across a 20-cm stratigraphic interval in Liu et al.(1988)and Sun et al.(1998a)).The MB transition zone at Luochuan and Mangshan has been demon-strated to occur within the stratigraphic transition between L8and S8(Jin and Liu,2010),as is the case at Sanmenxia (Wang et al.,2006)and Xifeng (Yang et al.,2010)(Fig.6f –h).Determination of the actual stratigraphic position of the MBB in the Chinese loess requires more de-tailed multi-proxy studies of the paleoclimate stratigraphy of L8and S8,and identi ?cation of an entire and accurate MB transition zone.At pres-ent,integration of results from the southeastern and central CLP indi-cates that the MB transition zone is consistently located across the pedostratigraphic and climatostratigraphic transition between S8and L8(e.g.,Jin and Liu,2011b ).Using multiple subsets of measurements,Jin et al.(2012)identi ?ed the MBB precursor event (cf.Kent and Schneider,1995;Hartl and Tauxe,1996)at the bottom of paleosol unit S8.By combining both the MBB and its precursor,Jin et al.(2012)corre-lated loess unit L8and the underlying paleosol unit S8to MIS 18and 19,respectively,which greatly improves our understanding of loess petrostratigraphy in this important stratigraphic interval.

740

750

760

770

780

790800810

MBB

Fig.7.(a,b)Pro ?les of quartz grain size for the Zhaojiachuan (ZJC)and Lingtai (LT)sec-tions,respectively,from the central CLP.(c,d)Magnetic susceptibility pro ?les for the ZJC and LT sections,respectively.Data are from Sun et al.(2006).(e,f)Depth plots of δ18O for cores V28-238and V28-239,respectively (from Shackleton and Opdyke,1973,1976).(g,h)Age plots of δ18O for ODP sites 982(Venz et al.,1999)and 983(Channell and Kleiven,2000),respectively.The thick dashed line marks the adjusted position of the MBB (Liu et al.,2008).The solid thick lines mark the paleoclimatic boundaries.The ?gure is after Liu et al.(2008).

148Q.Liu et al./Earth-Science Reviews 150(2015)139–167

On the basis of the above correlations,it appears that the MBB is lo-cated at the boundary between MIS18and19,rather than in the middle of the MIS19.Therefore,the orbitally-tuned age(780ka)for the MBB (Shackleton et al.,1990;Tauxe et al.,1996)should be adjusted to 770ka.Suganuma et al.(2010,2011)used an alternative approach in which they compared relative paleointensity variations and?uxes of meteoric10Be for marine sediments,which can be further synchronized using ice core records.They observed an apparent downward offset of paleointensity minima associated with the MBB and its precursor rela-tive to the associated10Be?ux anomalies.Similar to Liu et al.(2008), Suganuma et al.(2010,2011)reported an~15cm pDRM lock-in depth.They also concluded that the authentic age of the MBB (~770±6ka,2σ)should be about10ka younger than was previously thought,which is in agreement with the estimation of Channell et al. (2010).It,therefore,appears that we are converging toward consistent estimations of the real position of the MBB in Chinese loess and marine sedimentary archives.

A?nal important issue concerning MBB records in the Chinese loess is the relationship between the MBB and associated microtektites.Li et al.(1993)claimed that microtektite particles occur around the MBB on the CLP,and assumed that the microtektites are coeval with Austral-asian counterparts.Liu et al.(2008)argued that the geochemical com-position of the Chinese microtektites is inconsistent with those of Australasian microtektites,and thus that microtektites cannot be

used

Fig.8.The MB transition from the L8to S8interval in the Luochuan section for ten parallel sets of samples(from Jin and Liu,2010).Paleomagnetic declination(Dec.),inclination(Inc.),and VGP latitude are shown in the left,middle,and right-hand columns,respectively.The directional transition is indicated in green.(For interpretation of the references to color in this?gure legend,the reader is referred to the web version of this article.)

149

Q.Liu et al./Earth-Science Reviews150(2015)139–167

as a reliable marker to correlate paleoclimatic records between the Chi-nese loess and marine sediments.Instead,the isochronous nature of the MBB indicates that microtektites in the Chinese loess have a different origin from Australasian microtektites.

4.3.Stratigraphic thickness of the MBB

The thickness of the MBB transition varies across the CLP,and ranges from tens of cm to more than2m.For example,the stratigraphic thick-ness of the MBB is30cm at Duanjiapo in Xi'an(Guo et al.,2001) (Fig.6b),160cm at Xifeng(Zhu et al.,1993),40cm at Lingtai (Spassov et al.,2001)(Fig.6c),25–30cm at Weinan(Zhu et al., 1994a)(Fig.6a),~60cm at Sanmenxia(Wang et al.,2006)(Fig.6g), 78cm at Baoji(Yang et al.,2007b,2010)(Fig.6e),and156cm(Yang et al.,2010)(Fig.6f)and223cm(Sun et al.,1993)in the Xifeng area. This variability mainly re?ects variations in sedimentation rate,but it could also re?ect variable depths of pedogenic alteration that contrib-utes to complex recording of the MB transition.

4.4.Morphology of the transitional?eld associated with the MBB

The number of directional oscillations recorded during the MB transi-tion differs from site to site on the CLP.For example,5oscillations are re-corded at Xifeng(Zhu et al.,1993),Weinan(Zhu et al.,1994a)(Fig.6a),and Duanjiapo in the Xi'an area(Guo et al.,2001)(Fig.6b),7at Lingtai (Spassov et al.,2001)(Fig.6c),9at Sanmenxia(Wang et al.,2006) (Fig.6g),and15at both Baoji and Xifeng(Yang et al.,2010)(Fig.6e,f). The number of directional oscillations documented during the MBB de-pends on the sampling resolution,but it can also differ for high-resolution analyses within the same area,such as at Baoji and Xifeng.Fur-thermore,virtual geomagnetic pole(VGP)positions through the MB tran-sition are inconsistent for different records.For the Xifeng section, transitional VGPs follow longitudinal bands through the Americas,the At-lantic Ocean,or Africa(Sun et al.,1993).In contrast,records from another site in the Xifeng area have distinct features,with transitional VGP posi-tions mainly distributed around the north or south poles with some inter-mediate positions when moving from one pole to another(Zhu et al., 1993).At Weinan,VGPs during the MB transition move along two longi-tudinal sectors situated over East Asia and Australia(Zhu et al.,1994a).At Duanjiapo,VGPs are con?ned over Africa(Guo et al.,2001).To test the re-liability of NRM records from Chinese loess,large numbers of parallel sub-samples were investigated across the MBB from the Luochuan(Jin and Liu,2010;Jin et al.,2012)and Mangshan sections(Jin and Liu,2011b). ChRM results from Luochuan for ten sets of parallel samples within the MB transition are scattered(green shading in Fig.8),while parallel sam-ples outside of the transition have consistent paleomagnetic directions. This difference is attributed to low ef?ciency alignment of detrital magnetite within a weak paleomagnetic?eld during the

polarity

Fig.9.Schematic correlations of estimated relative paleointensity variations from the L8to S8interval in the Luochuan section,marine sediments and an Antarctic ice core(from Jin et al., 2012).(a)Green and black lines indicate Antarctic temperature change and geomagnetic paleointensity inferred from the deuterium content(Jouzel et al.,2007)and from10Be?ux(which is inversely related to geomagnetic paleointensity)from the EPICA Dome C ice core with the chronology of Dreyfus et al.(2008),respectively.(b)Magnetic susceptibility variations(green) compared to stacked estimated relative paleointensity(NRM300/SIRM300in black)calculated using the bootstrap method(Tauxe et al.,1991)for5parallel sample sets from the Luochuan section.The error bars represent95%con?dence intervals.(c)Normalized relative paleointensity for12marine sediment cores reported by Hartl and Tauxe(1996).DIP1and DIP2,which correspond to the MB precursor and the MB reversal,are shaded.(For interpretation of the references to color in this?gure legend,the reader is referred to the web version of this article.) 150Q.Liu et al./Earth-Science Reviews150(2015)139–167

transition.This detailed analysis suggests that it is dif?cult to treat high-frequency paleomagnetic variations observed across the MBB as reliable records of?eld behavior(green shading in Fig.8)and that limitations associated with recording?delity should always be considered when interpreting high-frequency paleomagnetic varia-tions from the Chinese loess(e.g.,Jin and Liu,2010,2011b;Zhao and Roberts,2010;Jin et al.,2012).

A precursor to the M

B reversal,during which the?eld decreased to an intensity minimum(Kent and Schneider,1995;Hartl and Tauxe, 1996),is an additional feature that is worth investigating across the same time interval.This precursor occurred just prior to the MB rever-sal,and has been observed in marine sediments(Kent and Schneider, 1995;Hartl and Tauxe,1996;Guyodo et al.,1999;Yamazaki and Oda, 2001,2005;Dinarès-Turell et al.,2002;Carcaillet et al.,2003,2004; Kissel et al.,2003;Macríet al.,2010;Suganuma et al.,2010,2011), lava?ows(Quidelleur et al.,2002;Singer et al.,2002,2005;Brown et al.,2004,2009;Petronille et al.,2005;Gratton et al.,2007),and an Antarctic ice core(Dreyfus et al.,2008).The MB precursor is a geomag-netic instability that has been interpreted as an independent magnetic feature(Kent and Schneider,1995),or as an event during which mag-netic?ux diffused from Earth's solid inner core to suf?ciently destabilize the geomagnetic?eld to allow the subsequent MB reversal to proceed (Singer et al.,2005).A paleomagnetic anomaly with low relative geo-magnetic paleointensity and scattered magnetic directions was identi-?ed~21kyr prior to the MBB in S8at Luochuan(Fig.9b)(Jin et al., 2012),which is consistent with observations from marine sediments (Fig.9c),an ice core(Fig.9a),and lava?ows.This anomaly in S8is, therefore,considered to represent the MB precursor.This is the?rst sig-ni?cant evidence for the MB precursor in terrestrial sediments,which attests to the global nature of this event.

4.5.Other polarity reversal boundaries

The upper boundary of the Jaramillo subchron has been identi?ed in L10at most loess sections,such as at Weinan(Zhu et al.,1994a;Pan et al.,2002),Xifeng(Liu et al.,1988),Baoji(Rutter et al.,1990), Jingchuan(Ding et al.,2001),and Lingtai(Ding et al.,1999b).Different positions have also been reported,such as in S10at Luochuan(Liu et al.,2010),Lantian(Zheng et al.,1992)and Sanmenxia(Wang et al., 2005).The position of the lower Jaramillo boundary has been reported in S11at Baoji(Rutter et al.,1990)and Lingtai(Ding et al.,1999b), L12at Jingchuan(Ding et al.,1998b),Weinan(Pan et al.,2002)and Jingbian(Guo et al.,2002a),S12at Lantian(Zheng et al.,1992),and even in the upper part of L13at Luochuan(Liu et al.,2010)and Sanmenxia(Wang et al.,2005).Moreover,the position of the Olduvai subchron has been reported within L24–L27at Jingchuan(Ding et al., 2001)and Lingtai(Ding et al.,1999b),in L25–L27at Baoji(Rutter et al.,1990),in S25–L29at Luochuan(Liu et al.,2010),in S26–S29at Lantian(Zheng et al.,1992),and even in L25–S26at Baoji(Yang et al., 2008).In contrast to the variable positions of these reversal boundaries, the position of the GMB has been consistently reported to occur within L33,such as at Jiaxian(Ding et al.,1998a),Jingchuan(Ding et al.,2001), Baoji(Rutter et al.,1990),Lingtai(Ding et al.,1999b),and Lantian (Zheng et al.,1992).

Similar to the problems with reliable determination of the posi-tion of the MBB(Liu et al.,2007,2008),inconsistencies in the report-ed stratigraphic positions of other reversal boundaries could also be caused by ambiguous physical de?nition of paleoclimatic bound-aries,the relatively low sampling resolution of some studies,incon-sistent and/or incorrect stratigraphic subpisions for early Pleistocene loess–paleosol sequences in some studies,and/or unknown NRM lock-in depths.For example,consider the upper Jaramillo boundary.At the Luochuan section,this reversal was locat-ed in S10(Fig.10)(Liu et al.,2010).However,S9has been tradition-ally pided into two paleosol layers,termed S9-1and S9-2 (Fig.10a),where S9-1correlates to MIS25and MIS27contains two relatively weaker interglacial periods.It can be confusing to discrim-inate between S9-2and S10and to correlate the loess/paleosol re-cord to MIS27(Fig.10).Alternatively,constrained by both the magnetostratigraphy and climate stratigraphy de?ned

by

Fig.10.Correlation between the Chinese loess across the L8–S11interval and global marine benthicδ18O isotopes.(a)The dark green line is susceptibility and the black line is median grain size(Md)for the Luochuan section(Sun and Liu,2000).(b)Md for the Baoji(orange line)and Pingliang(green line)sections are from Ding et al.(2002);(c)LR04is a stack of57globally distributed benthic marineδ18O records(Lisiecki and Raymo,2005).S8and S10are tied to marine oxygen isotope stages19and27,respectively,based on correspondence of the positions of the MBB and the upper Jaramillo reversals at these positions for both Chinese loess and marine sediments.(For interpretation of the references to color in this?gure legend,the reader is referred to the web version of this article.)

From Jin and Liu(2011a).

151

Q.Liu et al./Earth-Science Reviews150(2015)139–167

susceptibility and grain size,Jin and Liu(2011a)proposed a new di-vision of S9and S10,and de?ned S9-2as S10-1(Fig.10a).This means that S10(made up of S10-1and S10-2)can be more easily correlated to the two weak interglacials in MIS27(Jin and Liu,2011a).This new subpision also largely resolves the problem of inconsistent strati-graphic positions for the upper Jaramillo reversal.Nevertheless,fur-ther high-resolution paleomagnetic and pedostratigraphic studies that aim to determine the precise stratigraphic positions of paleo-magnetic reversal boundaries,such as the lower and upper Jaramillo and Olduvai boundaries,are still needed to provide a consistent chronostratigraphic framework for Chinese loess sequences,which is currently lacking in detail.

Compared to the MBB,there are fewer studies of other reversal boundaries in Chinese loess sequences.A continuous6-m interval in the Jiuzhoutai section was sampled to analyze the lower Jaramillo boundary by Rolph(1993)using both thermal and AF demagnetization. The reversal duration was estimated to be approximately5kyr,while the?eld strength was much reduced for at least9kyr.During the rever-sal,longitudinally con?ned VGPs progressed from the west coast of Africa to the equator,and then swung west to the Caribbean before moving north through North America to complete the reversal.Howev-er,as stated by Rolph(1993),“the stability of the?eld leading up to the reversal in direction is poor,with a number of apparent short duration ‘reversals’”.As indicated above,such records should not be considered high-?delity records for understanding transitional?eld behavior.Zhu et al.(1994a)investigated geomagnetic?eld behavior through the upper Jaramillo transition in L10at Weinan.Directional changes through the transition occur across an11-cm interval(13intermediate directions)(Fig.11a),with VGPs swinging from high northern latitudes to equatorial latitudes,over the Paci?c and then returning to northern latitudes before moving to the south,with longitudinally con?ned?uc-tuations.Guo et al.(2002a)reported transitional directions for the upper and lower Jaramillo boundaries at Jingbian(Fig.11b).For the upper Jaramillo,the transition from normal to reversed polarity con-tains three reversed and four normal polarity intervals over a strati-graphic interval of3.92m.The reversed polarity interval between

28.84and28.36m was considered by Guo et al.(2002a)to represent

a short polarity interval within the Jaramillo subchron,but this interpre-tation is dif?cult to reconcile with records from the Weinan section (Fig.11a)or with paleomagnetic results from deep-sea and

continental Fig.11.Detailed paleomagnetic records of the upper Jaramillo transition from the(a)Weinan(Zhu et al.,1994a)and(b)Jingbian sections(Guo et al.,2002a,b).

152Q.Liu et al./Earth-Science Reviews150(2015)139–167

sequences,and with marine magnetic anomaly patterns.Although Guo

et al.(2002a)concluded that “it is possible that this short polarity

interval represents an interval with delayed remanence acquisition (ac-

quired during the late Matuyama Chron)”,a lack of detailed records of

the Jaramillo subchron from the Chinese loess suggests that the origin

of this short reversed polarity interval should be tested further.For the

lower Jaramillo transition,a smooth progressive change from reversed

to normal polarity directions was observed with no intermediate direc-

tions.This observation may be a result of a relatively low sampling res-

olution rather than re ?ecting geomagnetic ?eld behavior.

Detailed paleomagnetic results have been reported for the Olduvai

subchron by Yang et al.(2008)from the Baoji section,followed by

Spassov et al.(2011)from the Lingtai section.Two sections separated

by about 300m,which span S24to S26,were continuously sampled

(Fig.12a,b)(Yang et al.,2008).For the upper Olduvai boundary,the

normal-to-reversed polarity transition contains an alternation of 8re-

versed and 8normal polarity intervals,with an estimated duration of

24–31kyr.Each episode is estimated to have had durations from 0.3

to 2.1kyr.Although several arguments have been proposed to support

the reliability of these directional ?uctuations,it is dif ?cult to interpret

these episodes as re ?ecting real geomagnetic ?eld behavior rather

than paleomagnetic recording artifacts,for reasons outlined above,es-

pecially for intervals that are de ?ned by only one sample.

Detailed paleomagnetic analysis of the GMB in Chinese loess has

only been reported by Zhu et al.(2000a),who investigated L33at

Weinan (Fig.13).The GMB occurs over a 60-cm stratigraphic interval

and has a N –S –N pattern with VGPs located principally on American

and east Asian longitudes.Reversed polarity directions are recorded

after a phase with VGPs that cluster in the Atlantic Ocean near southern

Africa.The total duration of the directional change for the GM transition

at Weinan is estimated at about 9.4kyr. 5.Geomagnetic excursions in Chinese loess Geomagnetic excursions are short-lived episodes with intermediate or opposite polarity that occur on timescales of a few thousand

years Fig.13.Detailed paleomagnetic record of the Gauss –Matuyama transition from the

Weinan section (Zhu et al.,2000a

).Fig.12.Detailed paleomagnetic records of the upper and lower Olduvai transitions from the Baoji section (Yang et al.,2008).Sections A and B are separated by about 300m.153

Q.Liu et al./Earth-Science Reviews 150(2015)139–167

(e.g.,Laj and Channell,2007;Roberts,2008).During excursions,the geomagnetic?eld deviates from the usual time-averaged geocentric axial dipole con?guration by more than40–45°(Gubbins,1999). Seven validated(Mono Lake,Laschamp,Blake,Iceland Basin,Pringle Falls,Big Lost,Stage17)and six“possible”Brunhes Chron excursions (Norwegian–Greenland Sea,Calabrian Ridge0,Calabrian Ridge2, West Eifel4,West Eifel2,West Eifel1)have been identi?ed and widely validated(Laj and Channell,2007;Roberts,2008).Seven Matuyama Chron excursions(Brunhes precursor,Kamikatsura,Punaruu,Bjorn, Gardar,Gilsa,Pre-Olduvai)and?ve subchrons(Santa Rosa,Jaramillo, Cobb Mountain,Olduvai,Réunion)have been identi?ed and validated according to Laj and Channell(2007)and Roberts(2008).In the follow-ing sections,we discuss the three most frequently studied geomagnetic excursions from Chinese loess sequences(i.e.,the Mono Lake,Laschamp and Blake excursions)and summarize studies that discuss other possi-ble excursions.

5.1.Mono Lake and Laschamp excursions

The MLE was?rst documented by Denham and Cox(1971)from the Wilson Creek Formation at Mono Lake,California,and was subsequently investigated by Liddicoat and Coe(1979).The age of the MLE was esti-mated as33ka,with its duration estimated at approximately2kyr (Benson et al.,2003;Channell,2006;Laj and Channell,2007).The LE was the?rst geomagnetic excursion reported in the literature (Bonhommet and Babkine,1967;Bonhommet and Zahringer,1969) from lava?ows of the Puy de Dome region of the Massif Central, France.The age of the LE is estimated as41ka B.P.with a duration of ap-proximately2kyr(Laj and Channell,2007).

Both the MLE and LE have been reported in Chinese loess sequences (Zhu et al.,1999,2000b,2006b,2007;Pan et al.,2002).Detailed paleo-magnetic investigations of three parallel sets of oriented samples from L1(Malan loess)at the Weinan section,Shaanxi Province,revealed two distinct anomalous directional intervals with low normalized rem-anence intensities(Fig.14a).Based on twelve14C and TL ages and the magnetic susceptibility timescale of Kukla et al.(1988),the age of the two excursions was determined to lie between46.8and37.4ka for the LE and between27.1and26ka for the MLE,respectively.These two excursions are also recorded at the Lingtai section(Zhu et al., 2000b)(Fig.14b).This further indicates that geomagnetic excursions are recorded in the Chinese loess.Nevertheless,as argued above for po-larity transitions,high-?delity recording of excursional?eld morphol-ogies is not expected.Detailed paleomagnetic studies of the stratigraphic interval from S1to S0at Luochuan,Xifeng,and Yichuan (Zhu et al.,2007)(Fig.14c–e)revealed no anomalous paleomagnetic di-rections at Yichuan(Fig.14e),with abnormal directions only observed for2specimens at a stratigraphic position of about7.8m at Xifeng (Fig.14d).At Luochuan,only a low-amplitude paleomagnetic

anomaly

Fig.16.Magnetostratigraphy for the Weinan section reported by Pan et al.(2002).(a)Low-?eld magnetic susceptibility,(b)ChRM declination and(c)inclination,(d)virtual geomagnetic pole(VGP)latitude,(e)magnetic polarity zonation derived from the paleomagnetic record,and(f)geomagnetic polarity timescale(GPTS)from Cande and Kent(1995).Ages for geomag-netic excursions in the Brunhes and Matuyama chrons are from Laj and Channell(2007)and Roberts(2008).Red lines in(d)represent possible geomagnetic excursions or polarity tran-sition intervals.(For interpretation of the references to color in this?gure legend,the reader is referred to the web version of this article.)

156Q.Liu et al./Earth-Science Reviews150(2015)139–167

associated with the LE is observed in the stratigraphic interval at 5–5.7m(Fig.14c).Recording of the LE in the central-southern CLP sug-gests that loess was deposited continuously at the studied locations at timescales of the duration of the LE(~2kyr)during the last glacial (Zhu et al.,2007).Lack of recording of the LE and MLE in some sections, however,suggests that deposition was not continuous at some locations over these time intervals.

5.2.Blake excursion

The Blake excursion(~120ka B.P.)was?rst de?ned by Smith and Foster(1969)from a paleomagnetic study of four deep-sea sediment cores from the Blake Outer Ridge,North Atlantic Ocean,following obser-vations from other marine sediment cores(see summary in Laj and Channell(2007)).Although early paleomagnetic studies reported the absence of the Blake event during the Brunhes Chron in the central CLP(e.g.,Liu et al.,1988;Zheng et al.,1992),anomalous geomagnetic di-rections have been observed in the upper Lanzhou section in the west-ern CLP(Burbank and Li,1985;Derbyshire et al.,1987;Rolph et al.,1989).Subsequently,a directional anomaly that was assigned to the Blake event was reported from the top of the Jingyuan section,also from the western CLP(Yue et al.,1991).The?rst detailed study of the Blake event in the Chinese loess was that of Zhu et al.(1994b)at the Xi-ning section,western CLP(Fig.15a).Over700oriented specimens from a5.7-m stratigraphic interval of S1were thermally demagnetized to ob-tain a magnetic stratigraphy.A56-cm-thick interval with anomalous paleomagnetic directions was observed,which is made up of3thin in-tervals of reversed polarity separated by two short normal polarity in-tervals.The duration of the Blake event was estimated to be5.3±0.6kyr based on correlation between Chinese loess and the marine ox-ygen isotope record.The duration of each of the six recorded polarity changes was suggested to be a few hundred years(Zhu et al.,1994b). Age estimates for the Blake excursion at the Xining section range from 117.1±1.2ka to111.8±1.0ka.

Another study of the Blake excursion was carried out at the Jiuzhoutai section,near Lanzhou city,Gansu Province(Fang et al., 1997).More than400oriented specimens were sampled from a4-m in-terval in the lower part of S1.Two thin intervals with reversed

polarity

Fig.17.Magnetostratigraphy for the Luochuan section from Liu et al.(2010).(a)Low-?eld magnetic susceptibility,(b)ChRM declination and(c)inclination,(d)VGP latitude,(e)magnetic polarity zonation from the paleomagnetic data,and(f)geomagnetic polarity timescale(GPTS)from Cande and Kent(1995).Ages for geomagnetic excursions in the Brunhes and Matuyama chrons are from Laj and Channell(2007)and Roberts(2008).Colored lines in(d)represent intervals with anomalous paleomagnetic directions.(For interpretation of the ref-erences to color in this?gure legend,the reader is referred to the web version of this article.)

157

Q.Liu et al./Earth-Science Reviews150(2015)139–167

paleomagnetic directions separated by a short normal polarity interval were observed,across a total stratigraphic thickness of 42.5cm (Fig.15b).At the Jiuzhoutai section,the Blake excursion was argued to have started at 119.9ka and ended at ca 114.5ka B.P.with a duration of 5.5kyr,which is consistent with the record from the Xining section (Zhu et al.,1994b ).Durations for the lower and upper reversed polarity intervals and the intermediate short normal polarity interval within the event are estimated as 2.58,1.61,and 1.31kyr,respectively.

In the central CLP,detailed analysis of S1was carried out at the Huanxian (Zheng et al.,1995)and Lingtai sections (Zhu et al.,2000b ),Gansu Province,to study the Blake excursion.For the Huanxian section,oriented block samples from S1and the upper part of L2were continu-ously sampled over a total stratigraphic thickness of about 4.6m (Fig.15c).The normalized remanence record for Huanxian (the ratio of the NRM to the anhysteretic remanent magnetization (NRM/ARM))undergoes a distinct decay in the lower part of the sampled section,which was interpreted to suggest the position of the Blake excursion.However,the paleomagnetic record from Huanxian was unstable,which makes it dif ?cult to accept this evidence as indicating the pres-ence of the Blake excursion.At the Lingtai section,only one specimen (at a stratigraphic level of ~9.3m)recorded abnormal paleomagnetic directions for the studied interval of S1(Fig.15d).

At least two possibilities have been proposed to explain paleomag-netic records of the Blake excursion from western CLP loess sections compared to the central CLP.First,at the Jiuzhoutai section,the Blake event is recorded just below the upper boundary between S1SS3(MIS5e)and L2(MIS6),within a bioturbated soil horizon (Ah)with high organic carbon content and loose structure.In the central CLP,these Ah paleosol horizons may have been bioturbated and eroded dur-ing the sharp climatic change just after the time of paleosol formation (Fang et al.,1997).However,the dry climate and high dust accumula-tion rate in the western CLP could have minimized such disturbances (Fang et al.,1997).Second,although the NRM acquisition mecha-nism in Chinese loess remains unclear in detail,it is widely accepted that the ChRM recorded in Chinese loess could be a combined DRM and CRM (e.g.,Zhu et al.,1994a;Spassov et al.,2003a;Zhao and Roberts,2010;Liu and Zhang,2013).Generally,CRMs become more important through neoformation of pedogenic ferrimagnetic minerals during warm and humid periods (Zhu et al.,2000b ).In the central and eastern CLP,paleoclimate was relatively warmer and more humid than in western and northern parts.Owing to enhanced pedogenesis and weathering in relatively humid areas,geomagnetic recording via long-term secondary remanence acquisi-tion could have replaced any primary DRM to modify or smooth the paleomagnetic signal (Zheng et al.,1995).However,it is not known to what extent these effects contributed to variable recording of the Blake excursion,as illustrated in Fig.15.

Paleomagnetic anomalies that have been referred to as the Blake ex-cursion have also been identi ?ed in S1from the Weinan section on the southern margin of the CLP (Pan et al.,2002)(Fig.16),and from the Luochuan section in the CLP hinterland (Liu et al.,2010)(Fig.17).Uncertainty remains concerning the extent to which the directional anomaly faithfully records the Blake excursion,especially under the in-?uence of pedogenesis,and the mechanisms responsible for variable re-cording between different parts of the CLP (Zheng et al.,1995;Zhu et al.,2000b ).Further work is needed to clarify these

issues.

Fig.18.Estimations of relative paleointensity for the L1unit from the Lingtai section by Pan et al.(2001).(a)Best-?t slopes of the linear portions between ARM and NRM estimated using the pseudo-Thellier approach of Tauxe et al.(1995).Normalized (b)NRM 300/IRM 1.2,(c)NRM 300/χ,and (d)NRM 300/ARM 300.NRM 300and ARM 300are after thermal treatment at 300°C.SIRM means an IRM gained in a ?eld of 1.2T.NRM =natural remanent magnetization;ARM =anhysterestic remanent magnetization;IRM =isothermal remanent magnetization.

158Q.Liu et al./Earth-Science Reviews 150(2015)139–167

5.3.Other possible excursions

The best resolved magnetic stratigraphy from Chinese loess is from Weinan (Pan et al.,2002)(Fig.16).Oriented samples were continuously collected from S0to S13(97m in thickness).Geomagnetic reversals or excursions recovered from S0to S13were named CLP1to CLP10(Fig.16).CLP1,2,and 3were correlated to the Mono Lake,Laschamp,and Blake excursions,respectively,and CLP6corresponds to the MBB (Pan et al.,2002).CLP8and 9correspond to the upper and lower bound-aries of the Jaramillo subchron,respectively (Pan et al.,2002).

Pan et al.(2002)suggested that CLP7(within unit L9)corresponds to the Santa Rosa fe1f8ab82f60ddccda38a0efpared to other loess units,L9contains a marker bed that is referred to as the upper silt layer and is character-ized by a coarser texture,greater friability and a markedly higher pro-portion of coarse silt and ?ne sand.Remagnetizations in L9have been reported at the Sanmenxia (Wang et al.,2005)and Luochuan sections

(Jin and Liu,2011a ).Wang et al.(2005)attributed the anomalous mag-netization in L9at Sanmenxia to remagnetization caused by realign-ment of magnetic grains.They suggested that magnetic particles in L9were physically reoriented by in ?ltrating rainwater during formation of S8in the succeeding interglacial period,and were ?nally reoriented by the geomagnetic ?eld during the Brunhes Chron.A VRM that overprinted ChRM carriers (mainly large PSD and MD magnetite)and randomization of a DRM caused by inef ?cient magnetic particle align-ment by the geomagnetic ?eld during deposition were later proposed to explain the remagnetization of L9at the Luochuan section (Jin and Liu,2011a ).In contrast,two normal polarity zones in L9at the Linghui section in the Baoji area (Yang et al.,2004)and at the Songjiadian sec-tion in the Sanmenxia area (Wang et al.,2010)have been interpreted to represent the Kamikatsura and Santa Rosa excursions,respectively.Therefore,it remains unclear whether CLP7at Weinan is an excursion or a remagnetization.

CLP10occurs within L13and was correlated to the Punaruu excur-sion by Pan et al.(2002),although anomalous paleomagnetic directions in L13have been correlated previously to the Cobb Mountain subchron (Guo et al.,1998a ).Two additional episodes have been reported from the Weinan section,CLP4and CLP5(Fig.16e).Episode CLP4occurs in the lower part of L3and has an inferred mid-depth age of 280ka.Pan et al.(2002)correlated this event to the Fram Strait/CR0excursion,following Langereis et al.(1997).Episode CLP5,which occurs in the upper part of L5,has an age of 422ka,which correlates to an un-named excursion reported by Langereis et al.(1997).The CR0and the un-named excursions documented from core KC-01B from the Calabri-an Ridge,Ionian Sea,by Langereis et al.(1997)lack strong independent support and are de ?ned by single samples in each case.These two anomalies were de ?ned as “possible ”excursions by Roberts (2008)and need further veri ?cation.The origin of the CLP4and CLP5paleo-magnetic anomalies in the Weinan section,thus,remain unclear.

Liu et al.(2010)presented a high-resolution magnetic stratigraphy from the Luochuan section (Fig.17)based on 1284samples with a sam-pling interval of 10cm.Beside the Blake excursion in S1,and the anom-aly in L9that has been proposed as a remagnetization (Jin and Liu,2011a ),several paleomagnetic anomalies are observed,including those at the top of L5,and within S5,L10,L14,and the top of L15(marked by “?”in Fig.17e),respectively.Whether the anomalies discussed above provide reliable records of geomagnetic variations needs further investigation.The paleomagnetic recording ?delity asso-ciated with such anomalous intervals could be tested by measuring sev-eral parallel sets of specimens across a given horizon,which would also enable more robust assessment of the exact stratigraphic interval spanned by an inferred excursion (e.g.,Zhu et al.,1993;Jin and Liu,2010).We suggest that when only one set of loess samples is used for studies of high-frequency geomagnetic processes,results should be treated with suspicion.Additionally,a more precise loess timescale is needed to constrain the age of identi ?ed anomalies to enable correla-tion with excursion records in marine sediments and/or radiometrically dated lava ?ows.Timescales for Chinese loess sequences commonly em-ploy a large lock-in depth for geomagnetic reversal boundaries (e.g.,Zhou and Shackleton,1999;Heslop et al.,2000)when performing astronomical tuning,which results in inaccurate age control for tuning (e.g.,Ding et al.,2002;Sun et al.,2006).Accurate determination of lock-in depths for geomagnetic reversal boundaries is,therefore,also needed to reduce age uncertainties in loess studies.6.RPI records from Chinese loess

Determination of RPI from sediments is important for developing an understanding of geomagnetic ?eld evolution and for use as an inde-pendent timescale (Guyodo and Valet,1999;Valet et al.,2005;Roberts et al.,2013).The basic assumption for RPI studies is that the NRM intensity is a function of both the alignment of magnetic grains with the magnetizing (geomagnetic)?eld and the concentration

of

Fig.19.Normalized remanence records for lower L8and upper S8at Luochuan (Jin and Liu,2010).(a)NRM 300/SIRM 300,(b)NRM 300/ARM 300,and (c)NRM 300/χfor depths from 230to 30cm.The bold (red)lines represent a three-point running average through the data.NRM 300,ARM 300,and SIRM 300are after thermal treatment at 300°C.The shaded interval corresponds to the Matuyama –Brunhes transitional interval based on paleomagnetic directions.

159

Q.Liu et al./Earth-Science Reviews 150(2015)139–167

本文来源：https://www.bwwdw.com/article/b28q.html