δ18O to temperature converter

Introduction

This tool converts carbonate δ18O data to temperatures. It automates the process of applying a number of correction methods which have been developed in the literature, as well as providing convenient access to a range of published species- and scenario-specific calibrations. It also automates two routine but technically complex tasks: performing paleocoordinate rotations and reconciling age datums between required datasets.

If you use output from this tool in publications, please cite the following publication:

The tool output also includes a short summary of the methods used for each run, with citations. This information may be used in publications using this tool, in whole, in part, or in adaptation, subject to the journal's standards. (At a minimum you should cite the references listed in order to give credit to the authors of the underlying datasets and methods used.)

Documentation

The δ18O paleothermometer is based on the observed relationship between δ18O and temperature in inorganic and biogenic carbonates (for a review, see Sharp, 2017). δ18O data are typically converted to temperatures using an empirical calibration such as that of Kim & O'Neil (1997):

T = 16.1 - 4.64 (δ18Oc - δ18Ow - 0.27) + 0.09 (δ18Oc - δ18Ow - 0.27)2

where T is temperature (°C), δ18Oc is the oxygen isotope composition of the carbonate (‰ VPBD), and δ18Ow is the oxygen isotope composition of the water in which it was precipitated (‰ VSMOW). Much of the difficulty in converting δ18Oc to temperature arises from the requirement of knowing δ18Ow, which may vary on both global and local scales. This tool implements multiple published methods of estimating δ18Ow, as well as correcting for carbonate-ion effects on δ18Oc (see below).

NOTE: While this tool can perform a variety of corrections, you are still responsible for pre-screening your data for diagenetic alteration or other external biases. For example, use of δ18O data from foraminifera must consider factors such as diagenetic recrystallization, depth habitat, shell size, and the presence of gametogenic calcite. For a review, see Pearson (2012).

Calibrations

A variety of calibrations are provided. The choice of calibration is up to the user; it is generally best practice to identify those listed calibrations which are most specific to your use-case (e.g., species being measured) and then review the original papers to verify that your use of the calibration is appropriate. Full citations for each calibration are provided on the results page. The tool will warn you if your δ18O data are outside the valid range for the calibration, but may not warn about other special considerations.

In the absence of a calibration specific to your use-case, some common calibrations in the literature are:

Site age

Site ages are required by several correction methods. A single age may be specified for the entire dataset, or sample-specific ages may be given as an age column in the datasheet (see the example at the top of the converter page). All ages are given in millions of years ago.

The Timescale option specifies which age datums to use when pulling data from internal datasets (such as global seawater δ18O reconstructions). To avoid inconsistencies, you should select the same timescale used for your own data's ages. (Internal datasets are converted between timescales by linear interpolation between magnetochron boundary ages, following the method used in the classic OSDN timescale tool; the timescale of your own data will not be modified.)

Site location

Site locations are required by several correction methods. A single latitude/longitude may be specified for the entire dataset, or sample-specific latitudes/longitudes may be given as lat and long columns in the datasheet (see the example at the top of the converter page). All latitudes/longitudes are decimal, i.e., 16° 30' E = 16.5, 16° 30' W = –16.5, 32° 45' N = 32.75, 32° 45' S = –32.75.

The tool may optionally perform paleocoordinate rotations to estimate site locations when the carbonates were precipitated, using given age information. Paleocoordinate rotations are performed using the GPlates web API, with ages rounded to the nearest 100 ka to reduce the number of API calls. If you wish to manually specify paleocoordinates (e.g., based on a different plate reconstruction), give the appropriate coordinates in the lat and long fields and specify none for the paleocoordinate adjustment. (Note that performing paleocoordinate rotations can cause the tool to run significantly slower, as it must wait for GPlates to respond.)

Global water δ18O ('ice volume' effect)

Because polar precipitation is enriched in 16O, δ18Ow varies globally with polar ice volume / sea level (for a review, see Sharp, 2017). You may specify δ18Ow manually or select one of several included water δ18Ow datasets to automatically estimate it from each sample's age field. These datasets are typically constructed by assuming that the benthic δ18O record reflects a combination of temperature and ice volume, and then subtracting out an independent record of temperature (e.g., Mg/Ca-based bottom-water temperatures) or ice volume (e.g., a geologic record of sea level such as NJSL) to determine the residual δ18Ow. Each of these records has different limitations, so you should review the original papers to verify that your use-case is appropriate. Full citations for each dataset are provided on the results page. The tool will warn you if your ages are outside the valid range for the dataset, but may not warn about other special considerations.

The default option is currently the Rohling et al. (2021) CENOGRID-based record, which was constructed by subtracting a high-resolution, astronomically-tuned benthic δ18O stack (Westerhold et al. 2020) from a multi-proxy sea-level reconstruction (Rohling et al. 2021). This record is appropriate for general marine samples back to 40 Ma and is probably the highest-resolution record of this interval available at time of publication.

Local water δ18O ('salinity' effect)

In addition to global variation, δ18Ow also varies spatially with evaporation and local hydrography. This variation broadly covaries with salinity, so it is often referred to as the 'salinity effect' (for a review, see Sharp, 2017). This effect has traditionally been difficult to account for; this tool provides several methods from the literature, which are described in the sections below.

Both global and local corrections are applied simultaneously, with the local correction dataset being normalized relative to 0‰ to reflect local deviations from global δ18Ow. That is,

δ18Ow = δ18Ow (global) + δ18Ow (local)

Both a global and local correction are therefore required to fully determine δ18Ow for each sample. (The exception is the modern ocean, where mean global δ18Ow is close to 0‰, so only a local correction is required.)

For methods which rely on geographic datasets (such as LeGrande & Schmidt 2006), you may specify the area to average over. This is useful to get a better sense of the regional mean δ18Ow when the exact paleocoordinates or local hydrography may not be known. Three methods are provided:

If no valid δ18Ow values are available within the specified range (such as when the paleocoordinates fall on land in the dataset), this calculation will return NaN (Not a Number) and the temperature conversion for that sample will fail.

Modern/glacial data

For samples from intervals with climate and paleogeography similar to the modern, it may appropriate to simply use modern local δ18Ow. A subset of the LeGrande & Schmidt 2006 modern interpolated δ18Ow maps are provided for this purpose. Select the water depth most appropriate for your sample. A surface dataset for the Last Glacial Maximum are also provided (Tierney et al. 2020).

Latitudinal methods

δ18Ow broadly covaries with latitude, so a reasonable approximation of δ18Ow can be produced from paleolatitude alone. This tool provides two such approximation:

None of these methods perform well at high northern latitudes, where the more complex basin geography can cause large local variations in δ18Ow (see Zachos et al. 1994, Gaskell et al. 2022).

Spatial method

Gaskell et al. (2022) introduced a method for estimating δ18Ow from isotope-enabled GCM simulations, allowing explicit consideration of paleogeography and climate state. For details on this method, see the original publication; briefly, surface δ18Ow values are taken from a suite of GCM runs with variable pCO2, representing different possible climate states for a given paleogeography. For improved precision, δ18Ow is interpolated between these runs using natural cubic splines, using bottom-water temperature to define where each sample falls on these splines. This method therefore requires a bottom-water temperature record in addition to sample latitudes/longitudes.

This method is valid globally, subject to the appropriateness of the paleogeography in the available datasets.

Carbonate system effects

The δ18O of many carbonates covaries negatively with water [CO32-] and/or pH (see Spero et al. 1997 and others), potentially biasing temperature reconstructions. This tool can automatically adjust δ18Oc for changes in seawater carbonate chemistry based on empirical measurements of the slope of this relationship and a record of seawater [CO32-].

Note that the [CO32-] records presently included have a relatively coarse resolution, so they are not appropriate for simulating changes in seawater carbonate chemistry over (e.g.) interglacial timescales. Spatial and depth variation in [CO32-] is also not considered. As an alternative, you may specify [CO32-] manually.

Demonstration dataset

As a demonstration, consider a dataset of Trilobatus sacculifer δ18O data from ODP 999A in the Caribbean (de Schnepper et al., 2013). A .csv file of this dataset is provided here. Use the following options:

The resulting temperatures should look like this file.

Contact and updates

Inquiries may be directed to Daniel E. Gaskell, the current maintainer of this tool. Source code is available on GitHub.

References cited