Characterizing the temporal uncertainty in palaeoclimate records is
crucial for analysing past climate change, correlating climate
events between records, assessing climate periodicities,
identifying potential triggers and evaluating climate model
simulations. The first global compilation of speleothem isotope
records by the SISAL (Speleothem Isotope Synthesis and Analysis)
working group showed that age model uncertainties are not
systematically reported in the published literature, and these are only
available for a limited number of records (ca. 15 %,
Speleothems are a rich terrestrial palaeoclimate archive that forms
from infiltrating rainwater after it percolates through the soil,
epikarst and carbonate bedrock. In particular, stable oxygen and
carbon isotope (
The Speleothem Isotope Synthesis and Analyses (SISAL) working group is an international effort under the auspices of Past Global Changes (PAGES) to compile speleothem isotopic records globally for the analysis of past climates (Comas-Bru and Harrison, 2019). The first version of the SISAL database (Atsawawaranunt et al., 2018a, b) contained 381 speleothem records from 174 cave sites and has been used for analysing regional climate changes (Braun et al., 2019a; Burstyn et al., 2019; Comas-Bru and Harrison, 2019; Deininger et al., 2019; Kaushal et al., 2018; Kern et al., 2019; Lechleitner et al., 2018; Oster et al., 2019; Zhang et al., 2019). The potential for using the SISAL database to evaluate climate models was explored using an updated version of the database (SISALv1b; Atsawawaranunt et al., 2019) that contains 455 speleothem records from 211 sites (Comas-Bru et al., 2019).
Summary of the dating information on which the original age–depth
models are based
Cave sites included in the version 1, 1b and 2 of the SISAL database on the World Karst Aquifer Map (WOKAM; Goldscheider et al., 2020).
SISAL is continuing to expand the global database by including new
records (Comas-Bru et al., 2020a). Although most of the records in
SISALv2 (79.7 %; Fig. 1a) have been dated using the generally very
precise, absolute radiometric
We attempted to construct age–depth models for 533 entities in an automated mode. For eight records, this automated construction failed for all methods. For these records we provide manually constructed chronologies where no age model previously existed and added a note in the database with details on the construction procedure. Age models for 21 records were successfully computed but later dropped in the screening process due to inconsistent information or incompatibility for an automated routine. In total, we provide additional chronologies for 512 speleothem records in SISALv2.
The SISAL chronology provides alternative age–depth models for SISAL
records that are not composites (i.e. time series based on more than
one speleothem record), that have not been superseded in the database
by a newer entity and which are purely
To allow a comprehensive cross-examination of uncertainties, seven
age–depth modelling techniques were implemented here across all
selected records. Due to the high number of records (
Major challenges arise through hiatuses (growth interruptions) and age
reversals. We developed a workflow to deal with records with known
hiatuses that allowed the construction of age–depth models for
20 % of the records with one or more hiatuses (Roesch and Rehfeld,
2019; details below for each age–depth modelling technique). Regarding
the age reversals, we distinguish between tractable reversals (with
overlapping confidence intervals) and non-tractable reversals (i.e.
where the 2-sigma dating uncertainties do not overlap) following the
definition of Breitenbach et al. (2012). Details such as the hiatus
treatment and outlier age modification are recorded in a log file
created when running the age models. We followed the original author's
choices regarding date usage. If an age was marked as “not used” or
“usage unknown”, we did not consider this in the construction of the
new chronologies except in OxCal, where dates with “usage unknown”
were considered.
Linear interpolation ( Linear regression Bchron Bacon OxCal COPRA (co StalAge
The structure of the SISAL database version 2. Fields and tables marked with (*) refer to new information added to SISALv1b; see Tables 1 and 2 for details. The colours refer to the format of that field: Enum, Int, Varchar, Double or Decimal. More information on the list of predefined menus can be found in Atsawawaranunt et al. (2018a).
The data are stored in a relational database (MySQL), which consists
of 15 linked tables:
Details of the
Changes made to the dating table to accommodate the new age models. These changes are marked with (*) in Fig. 3.
Changes were also made to the dating table (
Changes made to tables other than the
The dating and the sample tables were modified to accommodate the inclusion of new entities in the database. Specifically, the predefined option lists were expanded, options that had never been used were removed, and some typographical errors in the field names were corrected; these changes are listed in Table 3.
The quality control procedure for individual records newly
incorporated in the SISALv2 database is based on the steps described
in Atsawawaranunt et al. (2018a). We have updated the Python database
scripts to provide a more thorough quality assessment of individual
records. Additional checks of the dating table resulted in
modifications in the
Summary of the modifications applied to records already in version 1 (Atsawawaranunt et al., 2018b) and version 1b (Atsawawaranunt et al., 2019) of the SISAL database. Mistakes in previous versions of the database were identified as outlined in the Supplement and through analysing the data for the SISAL publications.
Continued.
Visual summary of quality control of the automated SISAL
chronology construction. The evaluation of the age–depth models for each
method (
Illustration of the impact of the age model choice on
reconstructed speleothem chronology illustrated by the KNI-51-H speleothem
record (entity_id 342; Denniston et al., 2013b). Panel
Information on new speleothem records (entities) added to the SISAL_v2 database from SISALv1b (Comas-Bru et al., 2019). There may be multiple entities from a single cave, here identified as the site. Latitude (Lat) and Longitude (Long) are given in decimal degrees north and east, respectively.
Continued.
Continued.
Continued.
Continued.
Scatterplot of average uncertainties in the
Analyses of the data included in SISALv1 (Braun et al., 2019a; Burstyn
et al., 2019; Deininger et al., 2019; Kaushal et al., 2018; Kern
et al., 2019; Lechleitner et al., 2018; Oster et al., 2019; Zhang
et al., 2019) and SISALv1b (Comas-Bru et al., 2019) revealed a number
of errors in specific records that have now been corrected. These
revisions include, for example, updates in mineralogies
(
We used an automated approach to age–depth modelling in R because of
the large number of records. Roesch and Rehfeld (2019) have described
the basic workflow concept and tested it using all of the
age-modelling approaches used here except OxCal. The basic workflow
involves step-by-step inspection and formatting of the data for the
different methods, and the use of predefined parameter choices is specific to each method. Each age-modelling method is called
sequentially. An error message is recorded in the log file if a
particular age-modelling method fails, and the algorithm then
progresses to the next method. If output is produced for a particular
age-modelling method, these age models are checked for monotonicity.
Finally, the output standardization routine writes out, for each
entity and age-modelling approach, the median age model, the ensembles
(if applicable) and information of which hiatuses and dates were used
in the construction of the age models. These outputs are then added to
the
The general approach for the OxCal age models was similar, and
step-by-step details and scripts are provided at
An overview of the evaluation results for the age–depth models
constructed in automated mode is given in Fig. 4. Three nested
criteria are used to evaluate them. Firstly, chronologies with
reversals (Check 1) are automatically rejected (score
The original age–depth models for every entity are available in SISALv2. However, given the lack of age uncertainties for most of the records, we recommend considering the SISAL chronologies with their respective 95 % confidence intervals whenever possible. No single age–depth modelling approach is successful for all entities, and we therefore recommend that all the methods for a specific entity are used together in visual and/or statistical comparisons. Depending on methodological choices, age–depth models compatible with the dating evidence can result in considerable temporal differences for transitions (Fig. 5). For analyses relying on the temporal alignment of records (e.g. cross-correlation), age–depth model uncertainties should be considered using the ensemble of compatible age–depth models as described in, for example, Mudelsee et al. (2012), Rehfeld and Kurths (2014) and Hu et al. (2017).
Global and regional temporal coverage of entities in the SISALv2.
Percentage of entities uploaded to the different versions of the
SISAL database with respect to the number of records identified by the SISAL
working group as of November 2019. The number of identified records includes
potentially superseded speleothem records. Regions are defined as: Oceania
(
The database is available in SQL and CSV format from
SISALv2 contains 353 976
The published age–depth models of all speleothems are accessible in
the
This second version of the SISAL database has an improved spatial coverage compared to SISALv1 (Atsawawaranunt et al., 2018b) and SISALv1b (Fig. 3; Atsawawaranunt et al., 2019). SISALv2 contains most published records from Oceania (80.2 %), Africa (73.7 %) and South America (77.6 %), but improvements are still possible in regions like the Middle East (42.3 %) and Asia (64.8 %; Table 6).
The temporal distribution of records for the past 2000 years is good,
with 181 speleothems covering at least one-third of this period and 84
records covering the entire last 2000 years (
This updated SISALv2 database now not only provides the basis for comparing a large number of speleothem-based environmental reconstructions on a regional to a global scale but also allows for comprehensive analyses of stable-isotope records on various timescales, from multi-decadal to orbital.
The supplement related to this article is available online at:
The following SISAL working group members contributed with either data or age-modelling advice to SISALv2: James Apaéstegui (Instituto Geofísico del Perú, Lima, Peru), Lisa M. Baldini (School of Health and Life Sciences, Teesside University, Middlesbrough, UK), Shraddha Band (Geoscience Department, National Taiwan University, No. 1, Sect. 4, Roosevelt Road, Taipei 106, Taiwan), Maarten Blaauw (School of Natural and Built Environment, Queen's University Belfast, UK), Ronny Boch (Institute of Applied Geosciences, Graz University of Technology, Rechbauerstraße 12, 8010 Graz, Austria), Andrea Borsato (School of Environmental and Life Sciences, University of Newcastle, Challaghan 2308, NSW, Australia), Alexander Budsky (Institute for Geosciences, Johannes Gutenberg University Mainz, Johann-Joachim-Becher-Weg 21, 55128 Mainz, Germany), Maria Gracia Bustamante Rosell (Department of Geology and Environmental Science, University of Pittsburgh, USA), Sakonvan Chawchai (Department of Geology, Faculty of Science, Chulalongkorn University, Bangkok 10330, Thailand), Silviu Constantin (Emil Racovita Institute of Speleology, Bucharest, Romania, and Centro Nacional de Investigación sobre la Evolución Humana, CENIEH, Burgos, Spain), Rhawn Denniston (Department of Geology, Cornell College, Mount Vernon, IA 52314, USA), Virgil Dragusin (Emil Racovita Institute of Speleology, 010986, Strada Frumoasă 31, Bucharest, Romania), Russell Drysdale (School of Geography, University of Melbourne, Melbourne, Australia), Oana Dumitru (Karst Research Group, School of Geosciences, University of South Florida, 4202 E. Fowler Ave., NES 107, Tampa, FL 33620, USA), Amy Frappier (Department of Geosciences, Skidmore College, Saratoga Springs, New York, USA), Naveen Gandhi (Indian Institute of Tropical Meteorology, Homi Bhabha Road, Pashan, Pune-411008, India), Pawan Gautam (Centre for Earth, Ocean and Atmospheric Sciences, University of Hyderabad, India; now at Geological Survey of India, Northern Region, India), Li Hanying (Institute of Global Environmental Change, Xi'an Jiaotong University, China), Ilaria Isola (Istituto Nazionale di Geofisica e Vulcanologia, Pisa, Italy), Xiuyang Jiang (College of Geography Science, Fujian Normal University, Fuzhou 350007, China), Zhao Jingyao (Institute of Global Environmental Change, Xi'an Jiaotong University, China), Kathleen Johnson (Dept. of Earth System Science, University of California, Irvine, 3200 Croul Hall, Irvine, CA 92697 USA), Vanessa Johnston (Research Centre of the Slovenian Academy of Sciences and Arts ZRC SAZU, Novi trg 2, Ljubljana, Slovenia), Gayatri Kathayat (Institute of Global Environmental Change, Xi'an Jiaotong University, China), Jennifer Klose (Institut für Geowissenschaften, Johannes Gutenberg University Mainz, Germany), Claire Krause (Geoscience Australia, Canberra, Australian Capital Territory, 2601, Australia), Matthew Lachniet (Department of Geoscience, University of Nevada Las Vegas, Las Vegas, NV 89154, USA), Amzad Laskar (Geosciences Division, Physical Research Laboratory, Navrangpura, Ahmedabad 380009, India), Stein-Erik Lauritzen (University of Bergen, Earth science, Norway), Nina Lončar (University of Zadar, Department of Geography, Trg Kneza Višeslava 9, 23000, Zadar, Croatia), Gina Moseley (Institute of Geology, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria), Allu C. Narayana (Centre for Earth, Ocean and Atmospheric Sciences, University of Hyderabad, India), Bogdan P. Onac (University of South Florida, School of Geosciences, 4202 E Fowler Ave, Tampa, FL 33620, USA and Emil Racoviță Institute of Speleology, Cluj-Napoca, Romania), Jacek Pawlak (Institute of Geological Sciences, Polish Academy of Sciences, 00-818, Twarda 51/55, Warsaw, Poland), Christopher Bronk Ramsey (Research Laboratory for Archaeology and the History of Art, Oxford University, Oxford, UK), Isabel Rivera-Collazo (Department of Anthropology and the Scripps Institution of Oceanography, UC San Diego, USA), Carlos Rossi (Dept. Petrología y Geoquímica, Facultad de Ciencias Geologicas, Universidad Complutense, Madrid, Spain), Peter J. Rowe (School of Environmental Sciences, University of East Anglia, NR4 7TJ, Norwich Research Park, Norwich, UK), Nicolás M. Stríkis (Department of Geochemistry, Universidade Federal Fluminense, Niterói, Brazil), Liangcheng Tan (State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an 710075, China), Sophie Verheyden (Politique scientifique fédérale belge BELSPO, Bvd. Simon Bolivar 30, 1000 Brussels, Belgium), Hubert Vonhof (Max Planck Institute for Chemistry, Mainz, Germany), Michael Weber (Johannes Gutenberg University Mainz, Germany), Kathleen Wendt (Institute of Geology, University of Innsbruck, Austria), Paul Wilcox (Institute of Geology, University of Innsbruck, Austria), Amos Winter (Dept. of Earth and Environmental Systems, Indiana State University, USA), Jiangying Wu (School of Geography, Nanjing Normal University, Nanjing, China), Peter Wynn (Lancaster Environment Centre, University of Lancaster, Lancaster, LA1 4YQ, UK) and Madhusudan G. Yadava (Geosciences Division, Physical Research Laboratory, Navrangpura, Ahmedabad 380009, India).
LCB is the coordinator of the SISAL working group. LCB, SPH and KR designed
the new version of the database. KR coordinated the construction of the new
age–depth models except OxCal. All age–depth models except OxCal were run by
CR and KR. LCB coordinated the construction of the OxCal age–depth models,
which were run by SAM and LCB. LCB implemented the changes in the v2 of the
database with the assistance of KA. SMA, YAB, AB, YB, MB, AC, MD, AD, BD,
IGH, JH, NK, ZK, FAL, AL, BM, VFN, JO, CPM, NSc, NSi, BMW, SW and HZ
coordinated the regional data collection and the age model screening. SFMB,
MB and DS provided support for COPRA, Bacon and StalAge, respectively. JF
assisted in the quality control procedure of the SISAL database. Figures 1,
4 and 5 were created by CR and KR. Figures 2, 3 and 6 were created by LCB.
All authors listed as “SISAL working group members” provided data for this
version of the database and
The authors declare that they have no conflict of interest.
This study was undertaken by SISAL (Speleothem Isotopes Synthesis and Analysis), a working group of the Past Global Changes (PAGES) project, which in turn received support from the Swiss Academy of Sciences and the Chinese Academy of Sciences. We thank SISAL members who contributed their published data to the database and provided additional information when necessary. We thank all experts who engaged in the age–depth model evaluation. The authors would like to acknowledge Avner Ayalon, Jordi López, Bahadur Singh Kotlia and Dennis Rupprecht.
The design and creation of v2 of the database were supported by funding to Sandy P. Harrison from the ERC-funded project GC2.0 (Global Change 2.0: Unlocking the past for a clearer future; grant no. 694481) and the Geological Survey Ireland Short Call 2017 (Developing a toolkit for model evaluation using speleothem isotope data; grant no. 2017-SC-056) award to Laia Comas-Bru. Sandy P. Harrison and Laia Comas-Bru received additional support from the ERC-funded project GC2.0 and from the JPI-Belmont project “PAlaeo-Constraints on Monsoon Evolution and Dynamics (PACMEDY)” through the UK Natural Environmental Research Council (NERC). Laia Comas-Bru and Belen Martrat received support from the CSIC scientific international collaboration programme I-LINKA20102 IBCC-lo2k. Kira Rehfeld and Denis Scholz acknowledge support by the Deutsche Forschungsgemeinschaft (DFG; codes RE3994/2-1 and SCHO 1274/11-1).
This paper was edited by Thomas Blunier and reviewed by Oliver Bothe and Judson W. Partin.