Abrupt increase in east Indonesian rainfall from flooding of the Sunda Shelf ~9500years ago

Abrupt increase in east Indonesian rainfall from flooding of the Sunda Shelf ~9500years ago
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Abrupt increase in east Indonesian rainfall from  󿬂 ooding of the Sunda Shelf  w 9500 years ago Michael L. Grif  󿬁 ths a , b , * , Russell N. Drysdale c , Michael K. Gagan d , Jian-xin Zhao e , John C. Hellstrom f  , Linda K. Ayliffe d , Wahyoe S. Hantoro g a Department of Earth System Science, University of California, Irvine, CA 92697-3100, USA b Department of Environmental Science, William Paterson University, Wayne, NJ 07470, USA c Department of Resource Management and Geography, Melbourne School of Land and Environment, The University of Melbourne, Parkville, VIC 2010, Australia d Research School of Earth Sciences, The Australian National University, Canberra, ACT 0200, Australia e Centre for Microscopy and Microanalysis, The University of Queensland, Brisbane, QLD 4072, Australia f  School of Earth Sciences, The University of Melbourne, Parkville, VIC 3010, Australia g Research and Development Center for Geotechnology, Indonesian Institute of Sciences, Bandung 40135, Indonesia a r t i c l e i n f o  Article history: Received 29 February 2012Received in revised form18 June 2012Accepted 7 July 2012Available online xxx Keywords: SpeleothemIndonesiaSunda Shelf Monsoon a b s t r a c t We present a precisely dated, multi-proxy stalagmite record from Liang Luar Cave, Flores (southeastIndonesia) that reveals a rapid increase in Indonesian monsoon rainfall at  w 9.5 ka. A  “ ramp- 󿬁 tting ” method for detecting statistically signi 󿬁 cant in 󿬂 ections in a time-series was applied to the stalagmite d 18 O, Mg/Ca, and Sr/Ca pro 󿬁 les to quantify the precise timing and magnitude of an abrupt increase inmonsoon strength over a period of   w 350 years. Previously published lake-level records from themonsoon-affected Australian interior show a sudden intensi 󿬁 cation of the Australian monsoon at w 14 ka. However, our records indicate that monsoon intensi 󿬁 cation in Flores occured w 4 e 5 kyr later.The timing of the monsoon shift in Flores is synchronous with the rapid expansion of rainforest innortheast Australia and regional freshening of the southern Makassar Strait which, under present-dayconditions, is sensitive to monsoon variability. The freshening of southern Makassar was coeval withan abrupt w 1.5   C cooling in the upper thermocline of the Timor Sea  w 9.5 ka, indicative of reducedsurface heat transport by the Indonesian Through 󿬂 ow (ITF) when the Java Sea opened during postglacialsea-level rise. This suggests that the abrupt increase in monsoon rainfall on Flores was not due toa change in the ITF  e  because a decrease in rainfall would be expected to accompany cooler local seasurface temperatures (SSTs)  e  but rather by the sudden increase in ocean surface area and/or temper-ature in the monsoon source region as the Sunda Shelf   󿬂 ooded during deglaciation. We propose that itwas the abrupt intensi 󿬁 cation of the monsoon through the late deglaciation that maintained thesubsequent structure of the ITF following the  󿬂 ooding of the Sunda Shelf at w 9.5 ka.Published by Elsevier Ltd. 1. Introduction Palaeoclimate records and climate model experiments haveshownthatthedominantmechanismcontrollingtropicalmonsoonsystems is orbital-scale variations in summer insolation(Ruddiman, 2006). However, in the region affected by theAustralian e Indonesian summer monsoon (AISM), there is stillsome disagreement. Evidence from lake-level records in northernAustralia suggest that AISM activity was reduced during the LastGlacial Maximum (LGM), and became re-established during theLateglacial-to-Holocene transition (Wyrwoll and Miller, 2001;Magee et al., 2004; Miller et al., 2005). These changes have been attributed to boreal winter insolation and its in 󿬂 uence on the AISMthrough the cross-equatorial  󿬂 ow of air from the semi-permanentSiberian high-pressure system. In contrast, some palaeoclimatemodel simulations have shown that increased local summer inso-lation can lead to stronger monsoon activity over northernAustralia (Chappell and Syktus, 1996). These divergent views mayindicate that theAISM does notbehavein a simple linear fashion tovariations in either Northern or Southern Hemisphere insolation.Rather, it may be controlled by the complex interplay betweeninsolation and other forcing mechanisms such as sea-surfacetemperature (Liu et al., 2003), sea level (Grif  󿬁 ths et al., 2009), andland-cover change (Miller et al., 2005). The Indo-Paci 󿬁 c Warm Pool (IPWP), in particular, plays animportant role in the global climate system through a number of  *  Corresponding author. Department of Earth System Science, University of Cal-ifornia, Irvine, CA 92697-3100, USA. E-mail address:  m.grif  󿬁 (M.L. Grif  󿬁 ths). Contents lists available at SciVerse ScienceDirect Quaternary Science Reviews journal homepage: 0277-3791/$  e  see front matter Published by Elsevier Ltd. Quaternary Science Reviews xxx (2012) 1 e 7 Please cite this article inpress as: Grif  󿬁 ths, M.L., et al., Abrupt increase in east Indonesian rainfall from 󿬂 oodingof the Sunda Shelf  w 9500 yearsago, Quaternary Science Reviews (2012),  dynamical mechanisms, including the Intertropical ConvergenceZone (ITCZ) and the related AISM, El Niño/Southern Oscillation(ENSO), Madden Julian Oscillation (MJO) and Indian Ocean Dipole(IOD).TheintenseupliftofwatervapourovertheIPWPprovidesthelargestsourceofheatandmoisturetotheglobalclimatesystem.Thestrengthandlocationofthisconvectionishighlyvariableandcloselytied to spatial gradients in sea-surface temperatures. A signi 󿬁 cantcomponent of the IPWP, and also the Earth ’ s thermohaline circula-tion, is the Indonesian Through 󿬂 ow (ITF), which is responsible fortransporting a large amount of water [10 e 15 Sverdrup (Sv), where1Sv ¼ 1millioncubicmeterspersecond]andheat( w 0.5PW,where1PW ¼ 10 15 W)fromthewesternPaci 󿬁 cWarmPool(WPWP)to12  Sin the Indian Ocean (Vranes et al., 2002). Under present-day condi-tions,theITFtransportsrelatively freshandcool watertotheIndianOcean thermocline, with w 80% of the total amount being exportedthrough the Makassar Strait (Fig. 1) (Fieux et al., 1994; Schneider, 1998; Gordon et al., 2003). During the AISM (Dec e Mar) northwestwinds push low-salinity South China Sea (SCS) water southeastacross the Sunda Shelf (through the Java Sea) and into thesouthern Makassar Strait (Fig. 1), setting up a northward pressuregradient in the surface layer. This seasonal body of relatively freshwater restricts the  󿬂 ow of warmer surface water from movingsouthward into the Indian Ocean during the austral summer,a patternwhichisreversed in theboreal summerwhentheeasterlytrades eliminate the obstructing pressure gradient by shifting moresaline water from the Banda Sea into the Makassar Strait (Linsleyet al., 2010).Hence, the shortage of highly resolved, precisely dated palae-oclimate records from the AISM domain means that a coherentexplanationforitsvariabilityintimeandspaceremainselusive.Weaim to reduce this knowledge gap by employing statistical analysisof geochemical tracers extracted from a previously publishedstalagmite record from Liang Luar Cave, western Flores, Indonesia(Grif  󿬁 ths et al., 2009, 2010a, 2010b, 2012). Speci 󿬁 cally, we utilizea statistical  “ ramp- 󿬁 tting ”  method (Mudelsee, 2000) to rigorouslycharacterize abrupt shifts in the speleothem record in order tobetterassess the evolution of the monsoon system through the latedeglaciation. 2. Study site and methods The stalagmite used in this study (LR06-B1) was collected fromLiang Luar, a w 1.7 km-long cave (8  32 0 N, 120  26 0 E; located at anelevation of   w 550 m a.s.l.) developed in late-Miocene reefalcarbonates on the Indonesian island of Flores (Fig. 1). The localregion has a mean annual temperature (MAT) of   w 25   C andreceives an average rainfall of 1200 mm annually. Most of the caverecharge ( w 69%) is delivered by the AISM between November andMarch, when the lower tropospheric winds shift from easterly(austral winter dry season) to north-westerly (austral summer wetseason). The stalagmite, which measured 1.25 m in length, wasactive at the time of collection. It formed w 800 m from the caveentrance in a large chamber with a high humidity (close to 100%).Oxygen isotope ( d 18 O) values were determined from calcitepowders, drilled at 1 mm increments, using a GV InstrumentsGV2003 continuous- 󿬂 ow isotope ratio mass spectrometer. Resultsare expressed as the deviation in per mil ( & ) between the sampleand the VPDB standard using the delta notation. The analyticaluncertainty (1 s ) of the standard material (a Carrara Marble calledNew1) was 0.08 & for  d 18 O. Subsamples were analysed for Mg, Srand Ca on a Varian Liberty 4000 inductively coupled plasmaatomic emission spectrometer (ICP-AES) [see Grif  󿬁 ths et al.(2010b) for a full description of the methods]. The  d 18 O andtrace elements were tied to a previously published  230 Th/ 234 Udepth-age model (Grif  󿬁 ths et al., 2009) constructed in two-stagesusing a Bayesian-Monte Carlo approach (Drysdale et al., 2005)from a total of 33 U-series dates. Principal Components Analysis(PCA) was performed on the  d 18 O, Mg/Ca, and Sr/Ca time-series toextract the dominant signal embedded within the three proxytime-series. 3. Results The most striking feature common to all three stalagmiteproxy time-series is the sharp decrease in values that began w 9.5 ka (Fig. 2). The close coupling between the three variables atthis time provides strong evidence for a single environmentalcontrol. Previous work on stalagmite LR06-B1, supported bya replicate stalagmite record from the same cave (LR06-B3), hasshown the  d 18 O to largely re 󿬂 ect variations in the intensity and/oramount of AISM rainfall (Grif  󿬁 ths et al., 2009), with lower  d 18 Ovalues corresponding to higher rainfall amounts. This pattern hassince been validated by isotope-enabled climate model simula-tions (Lewis et al., 2010). The two abrupt decreases in stalagmite d 18 O of  w 0.5 & centred at w 10.8 and w 9.5 ka therefore suggestincreases in rainfall on western Flores at this time. Noteworthy,however, is that some of the observed decreasing trend in  d 18 Othrough the early Holocene (Fig. 2a) may be explained by theconcomitant increase in local SSTs (Stott et al., 2004) and thedecrease in  d 18 O of the ocean ( “ ice-volume effect ” ) (Siddall et al.,2003), as previously noted in Grif  󿬁 ths et al. (2009, 2010a). However, the two step-wise shifts in speleothem  d 18 O at  w 10.8and w 9.5 ka occur over a period of  w 100 years (Fig. 2b), which istoo rapid to be attributed to changes in SSTs and ice-volume;hence, these other effects are likely to be negligible with Fig.1.  Location of Flores (red  󿬁 lled box) and other sites discussed in the text. Colouredcircles represent published marine sediment records (MD41: Rosenthal et al., 2003;MD81: Stott et al., 2004; MD90: Steinke et al., 2006; MD65: Levi et al., 2007; MD78: Xu et al., 2008; MD62: Renssen et al., 2009; MD76, 70GGC, 13GGC: Linsley et al., 2010). Site numbers in and around Australia denote the following: (1) Lake Eyre; (2) Lynch ’ sCrater, Queensland; (3) Lake Euramoo, Queensland; (4) Chillagoe Caves, Queensland;(5) ODP site 820; (6) Lake Gregory, Western Australia; (7) Fitzroy Crossing, WesternAustralia. Grey shading shows the approximate position of the shoreline at 10 kacalculated from reconstructed gridded sea-level data ( The pink box shows the region used to calculate the change in% of exposed land through the Holocene, which is shown in Fig. 4b. The solid blue lineshows the average summer (DJF) 96-h back trajectory moisture pathway for the period1960 e 2000 using the HYSPLIT model (Draxler and Rolph, 2003). Trajectories were started at an elevation of 1000 m (approximate cloud base height in the tropics). M.L. Grif   󿬁 ths et al. / Quaternary Science Reviews xxx (2012) 1 e 7  2 Please citethis article inpress as: Grif  󿬁 ths,M.L., et al., Abrupt increase in east Indonesianrainfall from 󿬂 oodingof the SundaShelf  w 9500 yearsago, Quaternary Science Reviews (2012),  respect to the much larger rainfall  “ amount-effect ”  over theseshort intervals.The Sr/Ca and Mg/Ca of stalagmite LR06-B1 co-vary toa statistically signi 󿬁 cant degree during the Holocene (Grif  󿬁 thset al., 2010b), suggesting their co-variability is strongly in 󿬂 u-enced by a common environmental control, or set of controls.Strong and systematic Mg/Ca and Sr/Ca co-variations in speleo-thems have been attributed to changes in seepage water chemistryduring drier periods (Fairchild and Treble, 2009). Thus, intervals of  decreased rainfall recharge (i.e. drier periods) result in dewateringof fractures in the karst hostrock, which enhances ventilation andtriggers CO 2  degassing  ‘ upstream ’  of the stalagmite. This degassingpromotes  ‘ prior calcite precipitation ’  along the hostrock fracturesand on the stalactite tips, resulting in higher Mg/Ca and Sr/Caratios of the meteoric water due to the preferential loss of the Ca 2 þ ion because of the partition coef  󿬁 cients ( D Mg  and  D Sr ) being lessthan unity (Fairchild et al., 2000; Fairchild and Treble, 2009). The converse occurs during wetter phases. Further support for this PCPmechanism (and hence cave hydrology) controlling the Mg/Ca andSr/Ca ratios was recently provided by coupled  14 C and  d 13 Cmeasurements from the same specimen (Grif  󿬁 ths et al., 2012). The effects of cave temperature can be ruled out for explaining theobserved shifts in Mg/Ca at  w 9.5 ka because the assumedtemperature variability through this period (Stott et al., 2004;Grif  󿬁 ths et al., 2010a) is too small to account for the observedchanges in Mg/Ca (Gascoyne, 1983; Huang and Fairchild, 2001). Similarly, the speleothem growth rate can be discarded asa possible mechanism for the  w 9.5 ka shift in Sr/Ca. Whilst thepartitioning of Sr into calcite is sensitive to growth rate (Huangand Fairchild, 2001), the observed shift in Sr/Ca through theearly Holocene would require the growth rate to have varied byorders of magnitude (Tesoriero and Pankow, 1996; Huang and Fairchild, 2001), which is not the case for this specimen(Grif  󿬁 ths et al., 2009); the growth rate varies by no more thana factor of three over the Holocene. It is worth pointing out,however, that some of the shorter-term discrepancies in thedecadal- to centennial scale trends (between the different proxies)may be explained by these additional effects.A strong relationship is also observed between the stalagmite d 18 O and the Sr/Ca and Mg/Ca ratios (Fig. 2). This positive co-variation provides additional evidence that the Sr/Ca and Mg/Catrends re 󿬂 ect effective water excess (de 󿬁 cit) due to increases(decreases) in rainfall. Slight discrepancies between the Mg/Ca andSr/Ca ratios and the  d 18 O pro 󿬁 les likely re 󿬂 ect the dependence of  d 18 O values on factors additional to rainfall amount (e.g. moisturetrajectories, Cruz et al., 2007), whereas Mg and Sr variations areprimarily controlled by hydrological processes unaffected bygeological sources.Further support for hydrological control of the three proxies isthe strong relationship between the  󿬁 rst principal component(PCA1) of the LR06-B1 multi-proxy record, which explains 67% of the variance and is interpreted to re 󿬂 ect the dominant rainfallsignal among the three proxy time series, and initial uraniumisotope activity ratios ([ 234 U/ 238 U] I ) from the same specimen(Fig. 3b). Speci 󿬁 cally, the [ 234 U/ 238 U] I  proxy, interpreted to re 󿬂 ectchanges in palaeohydrology (e.g. Hellstrom and McCulloch, 2000;Grif  󿬁 ths et al., 2010b), displays an abrupt transition to lower values(i.e. wetter conditions) also at w 9.5 ka (Fig. 3b).The hydrologically controlled changes in the LR06-B1 proxies at w 9.5 ka point to a major increase in monsoon strength. However,such a climate shift can only be regarded as signi 󿬁 cant if thedifference in the mean values either side of the in 󿬂 ection point islargerthanthestandarderrorof themeans.Tostatisticallyquantifythetimingandmagnitudeof thechangesintheproxies,weapplieda ramp- 󿬁 tting function to the  d 18 O, Mg/Ca, and Sr/Ca time series.This method detects the presence and timing of a linear changefromonesteadystatetoanother(Mudelsee,2000).Themeanproxy values prior to and after the transition in each geochemical pro 󿬁 lewere determined by weighted least-squares regression overa 1500-year long interval (Fig. 2b). The timing of the beginning andend points of the ramp were calculated by a brute-force searchusing a prescribed standard deviation of the data series (Mudelsee,2000). The uncertainty of the ramp timing, expressed as a standarderror, was calculated using 400 bootstrap simulations.From Fig. 2b, it is evident that the onset of the shifts in the Mg/Ca, Sr/Ca and  d 18 O pro 󿬁 les occurs at about the same time. There aredifferences, however, in the duration of the shifts: the Mg/Ca hasaduration of473years whilethe Sr/Caand  d 18 O recordshavemuchshorter durations of 199 and 109 years, respectively. As outlinedabove, these slight discrepancies in the trends are likely due tootherfactorsinadditiontoabove-cavehydrology(e.g.temperature,growth rate, isotopic composition of the ocean source) havingdifferent effects on the behaviour of the various proxies in thespeleothem,aswellasthedifferentsensitivitythresholdsthatneedto be crossed before each proxy responds to a given level of hydrological forcing. a 9.54 ka9.49 ka 9.54 ka 473 yrs199 yrs109 yrs b    S  r   /   C  a   (  x   1   0   0   0   ) -7-6-5-4         δ    1   8    O   (   ‰    V   P   D   B   )   M  g   /   C  a   (  x   1   0   0   0   ) 0 2 4 6 8 10    S  r   /   C  a   (  x   1   0   0   0   ) -6.0         δ    1   8    O   (   ‰    V   P   D   B   ) -5.5-5.0-4.5    M  g   /   C  a   (  x   1   0   0   0   ) 1020308.8 9.0 9.2 9.4 9.6 9.8 10.015202530 Fig. 2.  Multi-proxy geochemical records for stalagmite LR06-B1 over the period12.65 e 0 ka. (a) Time series of stalagmite  d 18 O, Sr/Ca, and Mg/Ca. (b) Ramp 󿬁 t regressionmodel (Mudelsee, 2000) applied to the  d 18 O, Sr/Ca, and Mg/Ca time series over theperiod 10.15 e 8.65 ka. Grey shading highlights the duration of the shift in mean valuesfor all three proxies. M.L. Grif   󿬁 ths et al. / Quaternary Science Reviews xxx (2012) 1 e 7   3 Please cite this article inpress as: Grif  󿬁 ths, M.L.,et al., Abrupt increase in eastIndonesianrainfall from 󿬂 oodingof the SundaShelf  w 9500 yearsago, Quaternary Science Reviews (2012),  4. Discussion and conclusions 4.1. Comparison with other records The timing of the  w 9.5 ka monsoon intensi 󿬁 cation in Flores,inferred from the LR06-B1 AISM multi-proxy record, is in goodagreement with tropical pollen records from northeast Australia.For example, comparison of the LR06-B1 PCA1 with the  󿬁 rst prin-cipal component of rainforest pollen taxa at Lake Euramoo in theAtherton Tablelands (Haberle, 2005), which explains 63% of the varianceandis interpretedto re 󿬂 ectanincrease ineffectiverainfall(Fig. 3c), reveals that the  󿬁 nal transition to rainforest in north-eastern Australia occurred around the same time (within datinguncertainty) as the sharp increase in monsoon rainfall on Flores(Fig.3);asimilarpatternwasfoundinpollenrecordsfromODP820(Moss and Kershaw, 2007), which is located offshore of Townsville in northeast Queensland (Fig. 1). A speleothem  d 18 O record of Australian monsoon variability from Chillagoe Caves (Turney et al.,2006)(Fig.1),alsoindicatesanabruptincreaseinrainfallduringthe early Holocene, consistent (within dating uncertainty) with Flores.Incontrast,pollenrecordsfromnearbyLynch ’ sCrater(Turneyetal.,2004) suggest that rainforest was established somewhat earlier at w 10.9 ka. Also, lake-level and  󿬂 uvial records from Lake Eyre, LakeGregory, and Fitzroy Crossing (Fig. 1) in the north Australian inte-rior (Wyrwoll and Miller, 2001; Magee et al., 2004) suggest monsooninitiationasearlyas14ka,morethan4 e 5kyrearlierthanindicated by the LR06-B1 multi-proxy record. Discrepanciesbetween these records may be due to: (1) different response timesof lake versus cave systems; (2) dating uncertainties; (3) preser-vation biases; and (4) different regional rainfall patterns underdifferent orbital and sea level con 󿬁 gurations. For example, usingthe fully coupled NCAR Community Climate Model (CCSM3),Wyrwoll et al. (2012) showed pronounced regional variability inrainfall over the AISM domain during different orbital geometries.Also, DiNezio et al. (2011) demonstrated, using multi-modelensembles, marked variability over the AISM domain during theLast Glacial Maximum (LGM) when global eustatic sea level was w 120 m lower than present. Hence, the monsoon signal preservedin our record maybe more representative of the proximal moisturesource, rather than an intensi 󿬁 cation of the AISM  per se , thereforegiving rise to a serendipitous relationship with the northeastQueensland pollen and speleothem records. Another possibility isthat the late deglacial ampli 󿬁 cation of the monsoon in this regionwas a two-phase process, and as such, our speleothem record onlycaptured the latter phase, owing to its shorter length of record. Forexample,takingintoaccountthelowerdatingresolutionofthelakerecords, it is possible that the higher lake levels during the earlydeglaciation could represent an earlier wet period around the timeof the Heinrich Stadial 1 ( w 14.5 e 17.5 ka), as is re 󿬂 ected in spe-leothem records from Liang Luar (Linda Ayliffe, personal commu-nication) and Chillagoe Caves (Turney et al., 2006). The duration of the stepwise transition in stalagmite LR06-B1PCA1 occurs over a period of  w 350 years (Fig. 3) and highlights thepotential role for the (Indonesian) monsoon system to abruptlyswitch non-linearly from one steady state to another. Furtherevidence to support this pattern is provided by a compilation of  Globigerinoides ruber   sea-surface  d 18 O ( d 18 O sw ) records from theIPWP (Linsley et al., 2010). Fig. 3d shows the difference in  d 18 O sw between 400-year, non-overlapping binned averages of SouthernMakassar marine cores (70GGC, MD62, 13GGC and MD65) andWPWP cores (MD41, MD81, MD76). The difference in  d 18 O sw  isinterpreted to re 󿬂 ect changes in the salinity gradient between thetwo regions, whereby a larger difference indicates a less salinesouthern Makassar Strait. From Fig. 3d it is evident that the relativedifference in  d 18 O sw  between the two regions increased from w 0.3 &  prior to  w 9.5 ka to  w 0.7 &  after  w 9.5 ka, re 󿬂 ectinga regional freshening of the southern Makassar Strait at this time.Linsley et al. (2010) suggested that this freshening was initiatedwhen rising post-glacial sea level reached  30 m and the  󿬂 oodingoftheSundaShelfreachedalevelwheretheKarimataStrait(i.e.theshallowest point)  󿬂 ooded, reconnecting the South China Sea (SCS)with the Makassar Strait via the Java Sea (Bard et al., 1996;Hanebuth et al., 2000; Siddall et al., 2003; Sathiamurthy and Voris, 2006) (Fig. 1). Hence, low-salinity waters that accumulated in the SCS because of its semi-enclosed basin before w 9.5 ka could nowreach the southern Makassar Strait. While the reconnection of theSCS with the Makassar Strait likely initiated the larger  d 18 O sw difference between the Makassar Strait and the areas to the northandeast, the increase in AISMmoisture (evident in our speleothemrecord) likely maintained the lower salinity of the surface seawaterfollowing the stabilisation of eustatic sea level just after 9.5 ka, assuggested by Linsley et al. (2010). 4.2. Mechanisms TrendsinthestalagmiteLR06-B1PCA1 representchanges intheamountofmonsoonrainfallattheLiangLuarsite.Comparisonwithother regional and global climate archives (Fig. 4) allows anassessment of possible AISM controls through the end of the lastdeglaciation. Previously we demonstrated that the deglacial rise in abcd Fig. 3.  Links between the Flores palaeomonsoon record and pollen rainforest taxafrom Lake Euramoo, AthertonTablelands, and marine sediment records from the IPWP.Comparison of (a) the  󿬁 rst multi-proxy principal component (PCA1) of the stalagmiteLR06-B1  d 18 O, Sr/Ca, Mg/Ca records, (b) the hydrologically-sensitive initial [ 234 U/ 238 U]for LR06-B1, and (c) PCA1 from rainforest pollen taxa from Lake Euramoo, AthertonTablelands (Haberle, 2005; Donders et al., 2007). In (a), the thick black line shows a 10- point running mean. (d) Difference in marine  d 18 O sw  of binned averages (400 year non-overlapping) between the southern Makassar sediment cores (MD70, MD62, 13GGC,MD65) and the western Paci 󿬁 c cores (MD41, MD81, MD76). The sharp increase in thedifference in  d 18 O sw  between the two regions at w 9.5 ka suggests a freshening of thesouthern Makassar Strait at this time (Linsley et al., 2010). The grey shading highlightsthe timing of the Younger Dryas, which is characterized by increased AISM rainfallcoeval with an expansion of rainforest in northeast Australia. Positions and analyticaluncertainties of the radiometric dates through the sharp transition in the stalagmite(black) and pollen (grey) records are shown along the bottom. M.L. Grif   󿬁 ths et al. / Quaternary Science Reviews xxx (2012) 1 e 7  4 Please citethis article inpress as: Grif  󿬁 ths,M.L., et al., Abrupt increase in east Indonesianrainfall from 󿬂 oodingof the SundaShelf  w 9500 yearsago, Quaternary Science Reviews (2012),  sea level had a dominant in 󿬂 uence on AISM moisture at Liang Luarbased on the  d 18 O record presented in this study and anotheradjacent stalagmite record (Grif  󿬁 ths et al., 2009).In addition to providing strong evidence for a rapid inundationof the Sunda Shelf at w 9.5 ka, the near-synchronous freshening of the Makassar Strait (Linsley et al., 2010, Fig. 4c) with a  w 1.5   Cdecrease in Timor Sea thermocline temperatures (Xu et al., 2008,Fig. 4d) suggests that freshening of the surface-ocean water insouthernMakassarincreasedthenorthwardpressuregradient,andthus inhibited the  󿬂 ow of warmer surface-layer water into theIndian Ocean. Hence, ITF heat transport was likely enhanced,relative totoday, prior to9.5 ka (Linsleyet al., 2010). Given this, it is unlikely that the abrupt reduction in ITF heat transport at w 9.5 kahad any impact on AISM strength in Flores, because the oppositesignal would be expected (i.e. weaker summer monsoon) if it wasin 󿬂 uential.OurpreferredexplanationforthelatedeglacialincreaseinAISMrainfall is related tothe 󿬂 oodingof theKarimata Straitas theSundaShelf was inundated during rapid deglacial sea-level rise. Lewiset al. (2010, 2011) used the NASA GISS ModelE equipped with vapour source distribution (VSD) tracers to show that wet season(DJF) precipitation sources for southern Indonesia are relativelylocal ( w 340 km mean vapour transport distance), with the meanlocus located west of Flores. Hence, assuming that monsoon sourcemoisture and trajectories were similar to present during the earlyHolocene (Fig. 1), it is possible that, prior to the  󿬂 ooding of theKarimata Strait, the reduced ocean-surface area over the sourceregion either: (1) limited the availability of moisture fuelling themonsoon; and/or (2) greatly affected atmospheric circulationpatterns over east Indonesia. Evidence for the latter scenario hasbeenprovidedbyanensembleofGCMexperimentsduringtheLGM(DiNezio et al., 2011). Results of the simulations suggest that when eustatic sea level was lower, there was a weakening of theascending branch of the Walker circulation over Indonesia, whichwas associated with a reduction in convection driven by enhancedland cooling over exposed areas.Another,somewhatcomplementary,hypothesisisthatsea-levelrise plus Northern Hemisphere insolation-induced warming of theSCS resulted in the intrusion of relatively warmer SCS surfacewaters (Steinke et al., 2006, Fig. 4e) into the Sunda area, as the Karimata Strait was  󿬂 ooded at  w 9.5 ka, thus resulting in higherevaporation over the monsoon source region. Feedback mecha-nisms associated with SST variability have a dominant in 󿬂 uence onAISM intensity over northern Australia during the Holocene (Liuet al., 2003; Wyrwoll et al., 2007). In a climate model simulation for11ka,Liuetal.(2003)showedthat,despiteinsolationbeinglow,AISMprecipitationwasenhancedduetothepresenceofwarmSSTsoff northwest Australia. Their results indicate that monsoon vari-ance over the northwest region of the continent could not beexplained without feedbacks associated with changes in SST.Comparison of our AISM record with southern Makassar Strait SSTs(Linsley et al., 2010, Fig. 4e) indicates no signi 󿬁 cant shift intemperature through the w 9.5 ka interval, suggesting that this wasnot the primary mechanism for the abrupt monsoon shift.However,  󿬂 ooding of the Sunda Shelf coincided with the warmestSSTs of the Holocene. Hence, it is possible that: (1) there wasacritical thresholdin local SSTs that was surpassed, which initiatedthe AISM; and/or (2) the response of the monsoon system todeglacial warming of local SSTs was delayed.Model experiments (e.g. Chappell and Syktus, 1996) have alsoshown that Southern Hemisphere summer insolation is the domi-nant external forcing mechanism governing AISM variability.Whilst this is likely the case over orbital time scales (Linda Ayliffe,personal communication), we  󿬁 nd little evidence for insolationbeing solely responsible for the abrupt rise in Indonesian rainfall at w 9.5 ka. The sharp transition in the AISM from one steady state toanother occurs over a period of  w 350 years (Fig. 3b), and demandsan explanation consistent with a rapid, rather than gradual, changein environmental boundary conditions. A contrasting view is thenotion that the AISM is not controlled by Southern Hemispheresummer insolation  per se , but by the strength of the East Asianwinter monsoon (EAWM) (Magee et al., 2004; Miller et al., 2005). Strong coupling of the AISM (Grif  󿬁 ths et al., 2009) and EAWM(Yancheva et al., 2007) was observed during the Younger Dryascooling. However, there is insuf  󿬁 cient evidence to vindicate theEAWM as being a dominant AISM forcing during the earlyHolocene.The results of this study lead to us to three primary conclu-sions: (1) there was a signi 󿬁 cant shift in east Indonesian rainfall at w 9.5 ka that was coincident with the  󿬂 ooding of the Sunda Shelf;(2) the late deglacial rainfall increase was probably not the resultof an abrupt change in ITF heat transport, but rather due toa larger/warmer monsoon source region (stemming from thebroad-scale  󿬂 ooding of the Sunda Shelf) and/or increasedconvection as exposed land areas became inundated; and (3) abcde Fig. 4.  Comparison of the Flores palaeomonsoon record with other proxy data over theperiod 13 e 0 ka. (a) The stalagmite LR06-B1 PCA1 record. Black line shows a 10-pointrunning mean. (b) Relative sea-level reconstruction (black line) based on marinesediments from the Red Sea (Siddall et al., 2003), and % land exposed over the SundaShelf (grey line and circles) calculated from the reconstructed gridded sea leveldatabase ( The region used tocalculate the % land exposed is shown by the pink box in Fig. 1. (c) Same as (d) inprevious  󿬁 gure. (d) MD78 (Timor Sea)  Pulleniatina obliquiloculata  Mg/Ca upper ther-mocline temperatures (Xu et al., 2008). Anomalies are calculated as departures from modern SSTs. (e) Mg/Ca  Globigerinoides ruber   SST records from the SCS marine coreMD90 (black circles; Steinke et al., 2006) and average southern Makassar Strait marinecores 13GGC, 70GGC, MD62, MD65 (grey circles; Linsleyet al., 2010). Anomalies for thesouthern Makassar Strait record are departures from the last 2000 years whileanomalies for the SCS are departures from modern. M.L. Grif   󿬁 ths et al. / Quaternary Science Reviews xxx (2012) 1 e 7   5 Please cite this article inpress as: Grif  󿬁 ths, M.L.,et al., Abrupt increase in eastIndonesianrainfall from 󿬂 oodingof the SundaShelf  w 9500 yearsago, Quaternary Science Reviews (2012),
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks