Although quantitative isotope data from speleothems has been used
to evaluate isotope-enabled model simulations, currently no consensus exists
regarding the most appropriate methodology through which to achieve this. A
number of modelling groups will be running isotope-enabled palaeoclimate
simulations in the framework of the Coupled Model Intercomparison Project
Phase 6, so it is timely to evaluate different approaches to using the
speleothem data for data–model comparisons. Here, we illustrate this using
456 globally distributed speleothem
Earth system models (ESMs) are routinely used to project the consequences of
current and future anthropogenic forcing of climate, and the impacts of
these projected changes on environmental services (e.g. Christensen et al.,
2013; Collins et al., 2013; Kirtman et al., 2013; Field, 2014). ESMs are
routinely evaluated using modern and historical climate data. However, the
range of climate variability experienced during the period for which we have
reliable historic climate observations is small, much smaller than the
amplitude of changes projected for the 21st century. Radically
different climate states in the geologic past provide an opportunity to test
the performance of ESMs in response to very large changes in forcing,
changes that in some cases are as large as the expected change in forcing at
the end of the 21st century (Braconnot et al., 2012). The use of
“out-of-sample” testing (Schmidt et al., 2014) is now part of the
evaluation procedure of the Coupled Model Intercomparison Project (CMIP).
Several palaeoclimate simulations are being run by the Palaeoclimate
Modelling Intercomparison Project (PMIP) as part of the sixth phase of CMIP
(CMIP6-PMIP4), including simulations of the Last Millennium (LM, 850–1850
CE,
Although these CMIP6-PMIP4 time periods were selected because they represent a range of different climate states, the choice also reflects the fact that global syntheses of palaeoenvironmental and palaeoclimate observations exist across them, thereby providing the opportunity for model benchmarking (Kageyama et al., 2017). However, both the geographic coverage and temporal coverage of the different types of data are uneven. Ice core records are confined to polar and high-altitude regions and provide regionally to globally integrated signals of forcings and climatic responses. Marine records provide a relatively comprehensive coverage of the ocean state for the LGM, but low rates of sedimentation mean they are less informative about the more recent past (Hessler et al., 2014). Lake records provide qualitative information of terrestrial hydroclimate, but the most comprehensive source of quantitative climate information over the continents is based on statistical calibration of pollen records (see for example Bartlein et al., 2011). However, pollen preservation requires the long-term accumulation of sediments under anoxic conditions and is consequently limited in semi-arid, arid and highly dynamic wet regions such as in the tropics.
Oxygen isotope records (
As with other model evaluation studies, much of the diagnosis of isotope-enabled ESMs has focused on modern day conditions (e.g. Joussaume et al., 1984; Hoffmann et al., 1998, 2000; Jouzel et al., 2000; Noone and Simmonds, 2002; Schmidt et al., 2007; Roche, 2013; Xi, 2014; Risi et al., 2016; Hu et al., 2018). However, isotope-enabled models have also been used in a palaeoclimate context (e.g. Schmidt et al., 2007; LeGrande and Schmidt, 2008, 2009; Langebroek et al., 2011; Caley and Roche, 2013; Caley et al., 2014; Jasechko et al., 2015; Werner et al., 2016; Zhu et al., 2017). The evaluation of these simulations has often focused on isotope records from polar ice cores and from marine environments. Where use has been made of speleothem records, the comparison has generally been based on a relatively small number of the available records. Furthermore, all of the comparisons make use of an empirically derived correction for the temperature-dependent calcite–water oxygen isotope fractionation at the time of speleothem formation that is based on synthetic carbonates (Kim and O'Neil, 1997). This fractionation is generally poorly constrained (McDermott, 2004; Fairchild and Baker, 2012), does not account for any kinetic fractionation at the time of deposition and is not suitable for aragonite samples. Thus, using a single standard correction and not screening records for mineralogy introduces uncertainty into the data–model comparisons.
SISAL (Speleothem Isotopes Synthesis and Analysis), an international working
group under the auspices of the Past Global Changes (PAGES) project
(
In this paper, we examine a number of issues that need to be addressed in
order to use speleothem data, specifically data from the SISAL database, for
model evaluation in the palaeoclimate context and make recommendations about
robust approaches that should be used for model evaluation in CMIP6-PMIP4.
We focus particularly on interpretation issues that could be overlooked in
using speleothem records and we show the strengths and limitations of
different comparison techniques. We use the MH and LGM time periods, partly
because the
Section 2 introduces the data and the methods used in this study. Section 2.1 introduces the isotope-enabled model simulations for the modern
(1958–2013), the
ECHAM5-wiso (Werner et al., 2011; Werner, 2019) is the isotope-enabled version of the ECHAM5 Atmosphere Global Circulation Model (Roeckner et al., 2003, 2006; Hagemann et al., 2006). The water cycle in ECHAM5 contains formulations for evapotranspiration of terrestrial water, evaporation of ocean water, and the formation of large-scale and convective clouds. Vapour, liquid and frozen water are transported independently within the atmospheric advection scheme. The stable water isotope module in ECHAM5 computes the isotopic signal of different water masses through the entire water cycle, including in precipitation and soil water.
ECHAM5-wiso was run for 1958–2013 using an implicit nudging technique to
constrain simulated fields of surface pressure, temperature, divergence and
vorticity to the corresponding ERA-40 and ERA-Interim reanalysis fields
(Butzin et al., 2014). The
At best, the speleothem isotopic signal will be an average of the
precipitation
We use two sources of modern isotope data for assessment purposes: (i)
The GNIP database (IAEA/WMO, 2018) provides raw monthly
The OIPC dataset provides a gridded long-term (1960–2017) global record of
modern
We use an updated SISAL database (SISALv1b: Atsawawaranunt et al., 2019),
which provides revised versions of 45 records from SISALv1 and includes 60 new records (Table 1). SISALv1b has isotope records from 455 speleothems
from 211 cave sites distributed worldwide. Because the isotopic
fractionation between water and
List of speleothem records that have been added to SISALv1 (Atsawawaranunt et al., 2018a, b) to produce SISALv1b (Atsawawaranunt et al., 2019) sorted alphabetically by site name. Elevation is in metres above sea level (m a.s.l.), latitude in degrees north and longitude in degrees east.
Recent data suggest that many calcite speleothems are precipitated out of
isotopic equilibrium with waters (Daëron et al., 2019). Therefore, we
have converted speleothem calcite data to their drip-water equivalent using an
empirical speleothem-based fractionation factor that accounts for any
kinetic fractionation that may arise in the precipitation of calcite
speleothems in caves (Tremaine et al., 2011):
We use the fractionation factor from Grossman and Ku (1986) as formulated in
Lachniet (2015) to convert aragonite speleothems to their drip-water
equivalent:
We use the V-PDB to V-SMOW conversion from Coplen et al. (1983) as in Sharp (2007):
We have used mean annual surface air temperature from CRU-TS4.01 (Harris et
al., 2014) for the OIPC comparison and ECHAM5-wiso simulated mean annual
temperature for the SISAL-model comparison as a surrogate for modern and
past cave air temperature (Moore and Sullivan, 1997). There are
uncertainties in this conversion because several factors are unknown, e.g.
cave temperature and
We compare the modern temporal variability in the SISAL records with
ECHAM5-wiso by extracting simulated
Data–model comparisons are generally made by comparing (1) anomalies between
a palaeoclimate simulation and a control period with (2) data anomalies with
respect to a modern baseline. There is no agreed standard defining the
interval used as a modern baseline for palaeoclimate reconstructions. Some
studies have used modern observational datasets which cover a specific and
limited period of time and some use the late 20th century as a
reference. We investigate the appropriate choice of modern baseline for the
speleothem records by comparing the interval centred on 1850 CE with
alternative intervals covering the late 20th century, specifically
1961–1990 and 1850–1990 CE, and we assess the impact of these choices on
both mean
We use the published age–depth models for each speleothem record. There is
no information about the temporal uncertainties on individual isotope
samples for most of the records in SISALv1b. This precludes a general
assessment of the impact of temporal uncertainties on data–model
comparisons. Nevertheless, we assess these impacts for the LGM for two
records (entity BT-2 from Botuverá cave: Cruz et al., 2005; and entity
SSC01 from Gunung-buda cave: Partin et al., 2007) for which new age–depth
models have been prepared using COPRA (Breitenbach et al., 2012). We created
1000-member ensembles of the age–depth relationship using the original
author's choice of radiometric dates and
To explore the use of absolute isotope data for model evaluation, we
extracted absolute data for two transects illustrating key features of the
geographic isotope patterns during the modern, MH and LGM periods. Each
transect follows the great circle line between two locations. The span of
each regional transect varies to maximize the number of SISAL records
included. We extracted model outputs for the same transects at
1.12
The presence or absence of speleothems in the temperate zone has long been interpreted as a direct indication of an interstadial or stadial climate state (Gordon et al., 1989; Kashiwaya et al., 1991; Baker et al., 1993), while in dry regions speleothem growth indicates a pluvial climate (Vaks et al., 2006) and in episodically cold regions responds to the absence of permafrost (Atkinson et al., 1978; Vaks et al., 2013). Speleothem distribution through time approximates an exponential curve in many regions around the world (e.g. Ayliffe et al., 1998; Jo et al., 2014; Scroxton et al., 2016). This relationship suggests that the natural attrition of stalagmites is independent of the age of the specimens and approximately constant through time, despite potential complications from erosion, climatic changes and sampling bias. The underlying exponential curve can, therefore, be thought of as a prediction of the number of expected stalagmites given the existing population. Intervals when climate conditions were more or less favourable to speleothem growth can then be identified from changes in the population size by subtracting this underlying exponential curve (Scroxton et al., 2016). We apply this approach at a global level to the unscreened SISAL data by counting the number of individual caves with stalagmite growth during every 1000-year period from 500 kyr BP to the present. Growth was indicated by a stable isotope sample at any point in each 1000-year bin, giving 3866 data points distributed in 500 bins. We use cave numbers, rather than the number of individual speleothems, to minimize the risk of over-sampled caves influencing the results. Random resampling (100 000) of the 3866 data points was used to derive 95 % and 5 % confidence intervals. The number of speleothems cannot be reliably predicted by a continuous distribution when numbers are low, so we do not consider intervals prior to 266 kyr BP – the most recent interval with less than four records.
There are many regions of the world where the absence of carbonate lithologies means that there will never be speleothem records (Fig. 1a). Nevertheless, SISALv1b represents a substantial improvement in spatial coverage compared to SISALv1, particularly for Australasia and Central and North America (Fig. 1a, Table 1), and the sampling for regions such as Europe and China is quite dense. Thus, SISALv1b provides a sufficient coverage to allow the data to be used for model evaluation. The temporal distribution of records is uneven, with only ca. 40 at 21 kyr increasing to > 100 records at 6 kyr and > 110 for the last 1000 yr (Fig. 1b). A pronounced regional bias exists towards Europe during the Holocene. Regional coverage is relatively even during the LGM, except for Africa, which is under-represented throughout (< 4 % of total). Nevertheless, there is enough coverage to facilitate data–model comparisons for the MH and LGM for most regions of the world.
Spatio-temporal distribution of SISALv1b database.
The global occurrence of speleothems through time approximates an exponential distribution (Fig. 2a). Anomalously high numbers of speleothems are found in the last 12 kyr, between 128 and 112 kyr BP and during interglacials MIS 1 and 5e (and the early glacial MIS 5d). There are fewer speleothems than expected between 73 and 63 kyr BP and during MIS 2 (Fig. 2b). These deviations could arise from sampling biases, but it is unlikely that such biases would lead to differences between the tropics and temperate regions. Differences between curves constructed for both tropical and temperate regions (Fig. 2c) suggest that, at least for the last 130 ka, deviations from expected stalagmite growth in the extra-tropics correspond to variability on glacial and interglacial scales. Thus, the speleothem data indicate similar climatic sensitivity, even at a global level, to that demonstrated for sub-continental and regional scales by earlier authors, despite their use of much smaller numbers and far less precise age data than in the SISAL dataset.
Distribution of the number of single caves with speleothem growth
through time.
The first-order spatial patterns shown by the SISAL speleothem records
during the modern period (1960–2017;
Comparison of SISAL data with observational and simulated
Comparison of the SISAL records with
Simulated inter-annual variability is less than that shown in the GNIP data (Fig. 4). Although there are missing values for the GNIP station data, we have
also removed these intervals from the simulations, so incomplete sampling is
unlikely to explain the difference between the observed and simulated
inter-annual variability. Our results are consistent with the general
tendency of climate models to underestimate the sensitivity of extreme
precipitation to temperature variability or trends (Flato et al., 2014).
ECHAM5 is known to underestimate inter-annual variability in regions where
precipitation is dominantly convective (i.e. the tropics), as well as in
summer in extra-tropical regions (e.g. in southern Europe) because
convective precipitation operates on small spatial scales and has a large
random component, even for a given large-scale atmospheric state (Eden et
al., 2012). The inter-annual variability of the modern speleothem records is
lower than both the simulated and the GNIP data, reflecting the impact of
karst and in-cave processes that effectively act as a low-pass filter on the
signal recorded during speleothem growth (Baker et al., 2013). Thus,
smoothing the simulated
Modern global inter-annual
The selection of a modern or pre-industrial base period is a first step in
reconstructing speleothem
Number of SISALv1b speleothem records available for key time
periods. Mid-Holocene (MH):
A relatively good agreement exists between the sign of the simulated and
observed
ECHAM5-wiso weighted
The simulated changes in
An alternative approach to examine the realism of simulated changes is to
compare the LGM and MH periods directly, which improves the number of
records for which anomalies can be calculated (Fig. 5c;
Age uncertainties inherent to the speleothem samples representing the LGM
could partially explain the LGM data–model mismatches. A global assessment
of the impact of time-window width on the MH and LGM anomalies shows that
reducing the window width from
LGM period definitions and their impact on SISAL
The number of sites available in SISALv1b means that quantitative data–model
comparisons using the traditional anomaly approach are limited in scope.
Approaches based on comparing trends in absolute
The first-order spatial gradient in observed
Latitudinal isotopic transect for Asia during the
Longitudinal isotopic transect for Europe during the
Our analyses illustrate a number of possible approaches for using speleothem
isotope data for model evaluation. The discontinuous nature of most
speleothem records means that the number of sites available for conventional
anomaly-mode comparisons is potentially limited. To some extent this is
mitigated by the fact that differences between the modern and pre-industrial
isotope values are small, permitting the calculation of anomalies using a
longer baseline interval (1850–1990 CE). The use of smaller intervals of
time in calculating MH or LGM anomalies (Figs. S5 and 6) does not have a
significant impact either on the mean values or the number of records
provided the interval is
Screening of published speleothem isotope data is essential to produce meaningful data–model comparisons. The SISAL database facilitates screening for mineralogy, which has a substantial effect on isotope values because of differences in water-carbonate fractionation factors for aragonite or calcite that are more pronounced at lower temperatures (Fig. S1).
Based on the limited number of records available at the LGM, speleothem age uncertainties have only a limited impact on mean isotope values, and propagation of such uncertainties as well as any model uncertainties would nevertheless substantially improve the robustness of data–model comparisons.
Based on our analyses, we therefore recommend that model evaluation using
speleothem records should do the following:
filter speleothem records with respect to their mineralogy and use the
appropriate equilibrium fractionation factor: Tremaine et al. (2011) for
converting isotope data from either calcite or
aragonite-corrected-to-calcite samples to their drip water equivalent, and
Grossman and Ku (1986) as reformulated by Lachniet (2015) for converting
isotope data from aragonite samples; use the interval between 1850 and 1990 as the reference period for
speleothem isotope records; use speleothem isotope data averaged for the intervals use speleothem isotope data averaged for the interval use absolute values only to assess data–model first order spatial patterns; focus on multi-decadal to millennial timescales if using transient
simulations for data–model comparisons.
Speleothem records show the same first-order spatial patterns as are available
in the Global Network of Isotopes in Precipitation (GNIP) data and are
therefore a good reflection of the
Using the traditional anomaly approach to data–model comparisons, there is
consistency between the sign of observed and simulated changes in both the
MH and the LGM. However, the ECHAM5-wiso model underestimates the changes in
Only a limited number of speleothem records are continuous over long periods of time and the need to convert these to anomalies with respect to modern times is a drawback. The limited number of records covering the LGM make the comparisons for this period particularly challenging. Nevertheless, continued expansion of the SISAL database will increase its usefulness for model evaluation in future. Furthermore, we have shown that alternative approaches using absolute values could help examine spatial trends and diagnose systematic offsets.
Mismatches between simulations and observations can reflect the issues with
the experimental design, problems with the model or uncertainties in the
observations (Harrison et al., 2015). The failure to include changes in
atmospheric dust loading, for example, has been put forward as an
explanation of data–model mismatches in both the MH and the LGM (e.g.
Hopcroft et al., 2015; Messori et al., 2019). Missing processes and
feedbacks, such as climate-induced vegetation or land-surface changes, could
also contribute to mismatches (e.g. Yoshimori et al., 2009; Swann et al.,
2014). Uncertainties caused by the specific structure of the model or
assigned model parameter values could also contribute to data–model
mismatches (Qian et al., 2016). Ultimately, there needs to be an assessment
of the contribution of all these factors to data–model mismatches, but here
we have only focused on potential uncertainties associated with the
speleothem data. Our initial analyses suggest age uncertainty contributes
little to the uncertainties in the estimates of LGM speleothem isotope
values. However, it is still important to propagate dating uncertainties for
data–model comparison. Site-specific controls may have a much larger effect
on the
Comparisons with speleothem data can be seen as a complement to model evaluation using other types of palaeoenvironmental data and palaeoclimatic reconstructions (see for example MARGO Project Members, 2009; Harrison et al., 2014). They are particularly useful because they provide insights into how well state-of-the-art models reproduce the hydrological cycle and atmospheric circulation patterns. The ability to reproduce past observations provides additional confidence in the ability of climate models to simulate large climate changes, such as those expected by the end of the 21st century (Braconnot et al., 2012; Schmidt et al., 2014). However, mismatches between model simulations and palaeo-observations are also useful because they can help to pinpoint issues that may need to be addressed in developing improved models or in better experimental protocols (Kageyama et al., 2018), providing that these mismatches do not arise because of misunderstanding or misinterpretation of the observations themselves. By providing a protocol for using speleothem data for data–model comparisons that accounts for uncertainties in the observations, we anticipate that at least such causes of data–model mismatches will be minimized.
The SISAL (Speleothem Isotopes Synthesis and AnaLysis Working Group)
database version 1b is publicly available through the University of Reading
repository at
The supplement related to this article is available online at:
SISAL working group members who coordinated data gathering for the SISAL database are listed here in alphabetical order: Syed Masood Ahmad (Department of Geography, Faculty of Natural Sciences, Jamia Millia Islamia, New Delhi 110025, India), Yassine Ait Brahim (Institute of Global Environmental Change, Xi'an Jiaotong University, Xi'an, Shaanxi, China), Sahar Amirnezhad Mozhdehi (School of Earth Sciences, University College Dublin, Belfield, Dublin 4, Ireland), Monica Arienzo (Division of Hydrologic Sciences, Desert Research Institute, 2215 Raggio Parkway, 89512 Reno, NV, USA), Kamolphat Atsawawaranunt (School of Archaeology, Geography & Environmental Sciences, Reading University, Whiteknights, Reading, RG6 6AH, UK), Andy Baker (School of Biological, Earth and Environmental Sciences, University of New South Wales, Kensington 2052, Australia), Kerstin Braun (Institute of Human Origins, Arizona State University, P.O. Box 874101, 85287 Tempe, Arizona, USA), Sebastian Breitenbach (Sediment & Isotope Geology, Institute of Geology, Mineralogy & Geophysics, Ruhr-Universität Bochum, Universitätsstr. 150, IA E5-179, 44801 Bochum, Germany), Yuval Burstyn (Geological Survey of Israel, 32 Yesha'yahu Leibowitz, 9371234, Jerusalem, Israel; Institute of Earth Sciences, Hebrew University of Jerusalem, Edmond J. Safra campus, Givat Ram, 91904 Jerusalem, Israel), Sakonvan Chawchai (MESA Research unit, Department of Geology, Faculty of Sciences, Chulalongkorn University, 254 Phayathai Rd, Pathum Wan, 10330 Bangkok, Thailand), Andrea Columbu (Department of Biological, Geological and Environmental Sciences, Via Zamboni 67, 40126, Bologna, Italy), Michael Deininger (Institute of Geosciences, Johannes Gutenberg University Mainz, Johann-Joachim-Becher-Weg 21, 55128 Mainz, Germany), Attila Demény (Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences, Hungarian Academy of Sciences, Budaörsi út 45, 1112 Budapest, Hungary), Bronwyn Dixon (School of Archaeology, Geography & Environmental Sciences, Reading University, Whiteknights, Reading, RG6 6AH, UK; School of Geography, University of Melbourne, Melbourne, 3010, Australia), István Gábor Hatvani (Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences, Hungarian Academy of Sciences, Budaörsi út 45, 1112 Budapest, Hungary), Jun Hu (Department of Earth Sciences, University of Southern California, 3651 Trousdale Parkway, 90089 Los Angeles, California, USA), Nikita Kaushal (Department of Earth Sciences, University of Oxford, South Parks Road, Oxford, OX1 3AN, UK), Zoltán Kern (Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences, Hungarian Academy of Sciences, Budaörsi út 45, 1112 Budapest, Hungary), Inga Labuhn (Institute of Geography, University of Bremen, Celsiusstr. 2, 28359 Bremen, Germany), Matthew S. Lachniet (Dept. of Geoscience, University of Nevada Las Vegas, P.O. Box 4022, 89154 Las Vegas, NV, USA), Franziska A. Lechleitner (Department of Earth Sciences, University of Oxford, South Parks Road, OX1 3AN Oxford, UK), Andrew Lorrey (National Institute of Water & Atmospheric Research, Climate Atmosphere and Hazards Centre, 41 Market Place, Viaduct Precinct, Auckland, New Zealand), Monika Markowska (University of Tübingen, Hölderlinstr. 12, 72074 Tübingen, Germany), Carole Nehme (IDEES UMR 6266 CNRS, Geography department, University of Rouen Normandie, Mont Saint Aignan, France), Valdir F. Novello (Instituto de Geociências, Universidade de São Paulo, São Paulo, Brazil), Jessica Oster (Department of Earth and Environmental Sciences, Vanderbilt University, Nashville, TN, 37206, USA), Carlos Pérez-Mejías (Department of Geoenvironmental Processes and Global Change, Pyrenean Institute of Ecology (IPE-CSIC), Avda. Montañana 1005, 50059 Zaragoza, Spain, now at: Institute of Global Environmental Change, Xi’an Jiaotong University, Xi’an, Shaanxi, China), Robyn Pickering (Department of Geological Sciences, South Africa and Human Evolution Research Institute, South Africa), Natasha Sekhon (Department of Geological Sciences, Jackson School of Geosciences, University of Texas, Austin, TX, 78712, USA), Xianfeng Wang (Earth Observatory of Singapore, Nanyang Technological University, 636798, Singapore), Sophie Warken (Institute of Environmental Physics, Ruprecht-Karls-Universität Heidelberg, Im Neuenheimer Feld 229, 69120 Heidelberg, Germany).
SISAL working members who submitted data to the SISAL database are listed here in alphabetical order: Tim Atkinson (Departments of Earth Sciences & Geography, University College London, WC1E 6BT, UK), Avner Ayalon (Geological Survey of Israel, 32 Yesha'yahu Leibowitz, 9371234 , Jerusalem), James Baldini (Department of Earth Sciences, Durham University, DH1 3LE, UK), Miryam Bar-Matthews (Geological Survey of Israel, 32 Yesha'yahu Leibowitz, , 9371234 , Jerusalem), Juan Pablo Bernal (Centro de Geociencias, Universidad Nacional Autónoma de México, Campus UNAM Juriquilla, Querétaro 76230, Querétaro, Mexico), Ronny Boch (Graz University of Technology, Institute of Applied Geosciences, Rechbauerstrasse 12, 8010 Graz, Austria), Andrea Borsato (School of Environmental and Life Science, University of Newcastle, 2308 NSW, Australia), Meighan Boyd (Department of Earth Sciences, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK), Chris Brierley (Department of Geography, University College London, WC1E 6BT, UK), Yanjun Cai (State Key Lab of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an 710061, China), Stacy Carolin (Institute of Geology, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria), Hai Cheng (Institute of Global Environmental Change, Xi'an Jiaotong University, China), Silviu Constantin (Emil Racovita Institute of Speleology, str. Frumoasa 31, Bucharest, Romania), Isabelle Couchoud (EDYTEM, UMR 5204 CNRS, Université Savoie Mont Blanc, Université Grenoble Alpes, 73370 Le Bourget du Lac, France), Francisco Cruz (Instituto de Geociências, Universidade de São Paulo, São Paulo, Brazil), Rhawn Denniston (Department of Geology, Cornell College, Mount Vernon, IA, 52314, USA), Virgil Drăguşin (Emil Racovita Institute of Speleology, Str. Frumoasa 31, Bucharest, Romania), Wuhui Duan (Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, 100029, China), Vasile Ersek (Department of Geography and Environmental Sciences, Northumbria University, Newcastle upon Tyne, UK), Martin Finné (Department of Archaeology and Ancient History, Uppsala University, Sweden), Dominik Fleitmann (Department of Archaeology, School of Archaeology, Geography and Environmental Science, Whiteknights, University of Reading RG6 6AB, UK), Jens Fohlmeister (Institute for Earth and Environmental Sciences, University of Potsdam, Karl-Liebknecht-Str. 24–25, 14476 Potsdam, Germany), Amy Frappier (Department of Geosciences, Skidmore College, Saratoga Springs, NY 12866, USA), Dominique Genty (Laboratoire des Science du Climat et de l'Environment, CNRS, L'Orme des Merisiers, 91191 Gif-sure-Yvette Cedex, France), Steffen Holzkämper (Department of Physical Geography, Stockholm University, 106 91 Stockholm, Sweden), Philip Hopley (Department of Earth and Planetary Sciences, Birkbeck, University of London, Malet St, London, WC1E 7HX, UK), Vanessa Johnston (Karst Research Institute, Research Centre of the Slovenian Academy of Sciences and Arts, Titov trg 2, 6230, Postojna, Slovenia), Gayatri Kathayat (Institute of Global Environmental Change, Xi'an Jiaotong University, China), Duncan Keenan-Jones (School of Historical and Philosophical Inquiry, University of Queensland, St Lucia QLD 4072, Australia), Gabriella Koltai (Institute of Geology, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria), Ting-Yong Li (Chongqing Key Laboratory of Karst Environment, School of Geographical Sciences, Southwest University, Chongqing 400715, China; Field Scientific Observation & Research Base of Karst Eco-environments at Nanchuan in Chongqing, Ministry of Nature Resources of China, Chongqing 408435, China), Mahjoor Ahmad Lone (High-Precision Mass Spectrometry and Environment Change Laboratory (HISPEC), Department of Geosciences, National Taiwan University, Taipei 10617, Taiwan, China. Research Center for Future Earth, National Taiwan University, Taipei 10617, Taiwan, China), Marc Luetscher (Swiss Institute for Speleology and Karst Studies (SISKA), Rue de la Serre 68, CH-2301 La Chaux-de-Fonds, Switzerland), Dave Mattey (Department of Earth Sciences, Royal Holloway University of London, Egham, Surrey, TW20 0EX, UK), Ana Moreno (Dpto. de Procesos Geoambientales y Cambio Global, Instituto Pirenaico de Ecología-CSIC, Zaragoza, Spain), Gina Moseley (Institute of Geology, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria), David Psomiadis (Imprint Analytics GmbH, Werner von Siemens Str. 1, A7343 Neutal, Austria), Jiaoyang Ruan (Guangdong Provincial Key Lab of Geodynamics and Geohazards, School of Earth Sciences and Engineering, Sun Yat-sen University, Guangzhou 510275, China), Denis Scholz (Institute for Geosciences, University of Mainz, Johann-Joachim-Becher-Weg 21, 55128 Mainz, Germany), Lijuan Sha (Institute of Global Environmental Change, Xi'an Jiaotong University, Xi'an, Shaanxi, China), Andrew Christopher Smith (NERC Isotope Geoscience Facility, British Geological Survey, Nottingham, UK), Nicolás Strikis (Departamento de Geoquímica, Universidade Federal Fluminense, Niterói, Brazil), Pauline Treble (ANSTO, Lucas Heights NSW, Australia), Ezgi Ünal-İmer (Department of Geological Engineering, Middle East Technical University, Ankara, Turkey), Anton Vaks (Geological Survey of Israel, 32 Yesha'yahu Leibowitz, 9371234, Jerusalem), Stef Vansteenberge (Analytical, Environmental & Geo-Chemistry, Department of Chemistry, Vrije Universiteit Brussel, Belgium), Ny Riavo G. Voarintsoa (Institute of Earth Sciences, The Hebrew University in Jerusalem, Israel), Corinne Wong (Environmental Science Institute, The University of Texas at Austin, 2275 Speedway, Austin TX 78712, USA), Barbara Wortham (Department of Earth and Planetary Science, University of California, Davis, USA), Jennifer Wurtzel (Research School of Earth Sciences, Australian National University, Canberra, ACT, Australia/ARC Centre of Excellence for Climate System Science, Australian National University, Canberra, ACT, Australia), Haiwei Zhang (Institute of Global Environmental Change, Xi'an Jiaotong University, China).
LCB is the coordinator of the SISAL working group. LCB and SPH designed the study. LCB and SPH wrote the first draft of the manuscript with contributions from MW, NS, KR and CVP. LCB did the analyses and created Figs. 1, 3–5, 7, 8 and S1–6. MW provided the ECHAM5-wiso model simulations and helped with model analyses. NS did the analyses on speleothem growth over time and created Fig. 2. KR did the analysis on the LGM uncertainties and created Fig. 6. All authors contributed to the last version of this paper. The authors listed in the “SISAL working group” team contributed to this study by coordinating data gathering, by database construction or with speleothem data submitted to the SISAL database. SB created the COPRA age–depth models used in this study. TA and DG contributed original unpublished data to the SISAL database. AB, BW, JB, MA, MSL, SB, TA, VJ and ZK helped edit the paper.
The authors declare that they have no conflict of interest.
SISAL (Speleothem Isotopes Synthesis and Analysis) is a working group of the Past Global Changes (PAGES) programme. We thank PAGES for their support for this activity. We thank the World Karst Aquifer Mapping project (WOKAM) team for providing us with the karst map presented in Fig. 1a. The authors would like to thank the following data contributors: Petra Bajo, Dominique Blamart, Russell Drysdale, Frank McDermott and Jean Riotte. Laia Comas-Bru and Sandy P. Harrison acknowledge support from the ERC-funded project GC2.0 (Global Change 2.0: Unlocking the past for a clearer future, grant no. 694481). Sandy P. Harrison also acknowledges support from the JPI-Belmont project “PAleao-Constraints on Monsoon Evolution and Dynamics (PACMEDY)” through the UK Natural Environmental Research Council (NERC). Laia Comas-Bru also acknowledges support from the Geological Survey Ireland (Short Call 2017; grant number 2017-SC-056) and the Royal Irish Academy's Charlemont Scholar award 2018. Cristina Veiga-Pires acknowledges funding from the Portuguese Science Foundation (FCT) through the CIMA research centre project UID/MAR/00350/2013. Kira Rehfeld was supported by Deutsche Forschungsgemeinschaft (DFG) grant no. RE3994/2-1.
Financial support for SISAL activities that have lead to this research has been provided by the Past Global Changes (PAGES) programme; the European Geosciences Union (grant no. W2017/413); the Irish Centre for Research in Applied Geosciences (iCRAG); the European Association of Geochemistry (Early Career Ambassadors program 2017); the Quaternary Research Association UK; the Navarino Environmental Observatory, Stockholm University; University College Dublin (grant no. SF1428), Savillex (UK); John Cantle; Ibn Zohr University, Morocco; the University of Reading; the European Research Council (grant no. 694481); the Natural Environment Research Council (JPI-Belmont project “PAleao-Constraints on Monsoon Evolution and Dynamics (PACMEDY)”); the Geological Survey Ireland (grant no. 2017-SC-056); the Royal Irish Academy (Charlemont Scholar award 2018); the Portuguese Science Foundation (grant no. UID/MAR/00350/2013); and the Deutsche Forschungsgemeinschaft (grant no. RE3994/2-1).
This paper was edited by Zhengtang Guo and reviewed by three anonymous referees.