Oceanic Carbon Cycling Response to Global Temperature Changes - CYCLOCARB

Summary – Introduction – Specific objectives approach and methods – Originality, innovative and interdisciplinary aspect – References – Figures and data

Specific objectives approach and methods

In the CYCLOCARB project, we will ultimately develop statistical models akin to Semi-Empirical Models ³⁵ (SEMs) of the carbon cycle. Though this approach does not formally include specific biogeochemical and physical processes, SEMs provide some constrains on the integrated carbon and climate system. To develop such new statistical model we will develop a unique extensive quality controlled paleoceanographic database covering the last glacial cycle and modern times. It will include proxies for temperature, ice volume, and carbon cycling and partitioning within the climatic system on a global scale. Advanced numerical methods (Convolutional Neural Networks (CNN)³⁶ and Monte Carlo approach) will be used to analyze this extensive dataset and document and quantify crucial aspect of modern and past carbon cycling and climate. By constraining global carbon cycling variations in the past and present and their link to the Earth’s temperature and climate, CYCLOCARB will yield some fundamental constrains to assess the role of the global ocean in past and ongoing climatic changes. Milestones and deliverable associated to each Working Packages are detailed in section 3.1.

WP1: Building a multiproxy database (month 1- month 13)

A database for the most commonly measured marine proxies (SST proxies; benthic and planktonic foraminifers stable isotopes (δ¹³C and δ¹⁸O)), and both organic and inorganic carbon content of the sediment) will be developed; partly relying on a database I already built during my postdocs (organic and inorganic carbon) and the ones developed at the host institution. Records will be gathered from online dataset repositories, and formatted into an appropriate numerical format. Mostly, Pangaea and NOAA repositories will be mined, following routines I already established for other studies ^5,11,37-39. A particular attention will be paid on the format and programing languages that will be used, taking into account ongoing initiatives on this topic (in particular OC3 (http://pastglobalchanges.org/science/wg/oc3), currently developing a database for surface and deglacial δ¹³C and Climate12K in which I’m implied, but also climateproxiesfinder, and SENSETROP (http://www.sensetrop.org/)). When available, raw data, and extended metadata will be gathered, including sample nature (species), analyzing protocol and laboratory, and calibration used. The SST proxies will be systematically included in the database, completing ongoing SENSETROP/COMPARE research project at CEREGE. I will rely on existing database structure (LinkedEarth) and will systematically include reconstruction based on Mg/Ca, alkenones, assemblages of foraminifers and dynocysts, as well as on TEX86 for the entire globe. Chronological constrains will also be gathered and used to revisit the age model of the records. ¹⁴C measurements will be collected and recalibrated using newly published IntCal2020 calibration curve. Model output for modern and past reservoir ages will be used to correct for circulation and carbon cycling variations impact on reservoir ages ⁴⁰. Homogenized age models will be created for the most useful archives using revisited ¹⁴C ages and regional benthic δ¹⁸O stacks, providing a common consistent timeframe adapted to the studied period (the last glacial cycle). Additionally, datasets for modern/preindustrial/surface sediment will be included in the database, in order to train CNN, and facilitate global evaluation and understanding of the modern processes, and the reconstruction of past conditions.

Deliverable: A website (D1.1, month 5), and the database (D1.2, month 12)

Milestones: Decision on database structure/selection criteria (M1.1, month 5), and on age model workflow (M1.2, month 11).

WP2: Variations of net carbon flux in the system over the last 150 ka and contribution of the biological pump to atmospheric pCO₂ changes (m11-m19)

I will perform statistical analyses (EOFs) on the selected δ¹³C records in order to define consistent 3 dimensional domains. I will include planktonic species δ¹³C in these analyses to infer surface and subsurface conditions. Artificial Neural Network trained on modern/surface sediment datasets will be tested to calculate 3 dimensional variations of global δ¹³C over times, as well as organic and inorganic carbon burial in modern and past environment, providing a mean δ¹³C reconstruction with incertitude envelop for the entire system. If this method is ineffective, I will use transfers function. For now, the literature only provides estimation of the global δ¹³C for the last deglaciation (Fig. 1) ^8,34. By assembling δ¹³C of benthic and planktonic foraminifers, and carbonate and organic carbon content in marine sediment within a single database, the quantification of the whole climatic system δ¹³C variation across a glacial cycle, and the changes in carbon budget and distribution in the system will be drastically improved.

Deliverable: submission of a scientific article (D2 month 19).

Milestones: Decision on reconstruction method used, and on focus periods (M2 month 15)

WP3: Global temperature variations over the last glacial cycle (m19-m26)

An extensive up-to-date database of SSTs will pave the way for scientific advances in both model/proxy, and proxy types comparison. I will perform a detailed evaluation of the available records by comparing proxies’ types, and I will use the SST compilation from low latitudes sites to evaluate the climate sensitivity to pCO2 changes as initially planned in the COMPARE project ²⁵ in collaboration with scientists from the host institution. Ultimately, I will reconstruct proxy specific global SST over the last 150 ka, allowing the study of spatial and temporal evolution of the surface temperature response to multiple forcing. This aspect will benefit from my recent participation to an initiative for global temperature reconstruction over the Holocene (Climate12K). High latitude records will allow me to evaluate the impact of ice cap variability on the global climate, and to study the structure of the latitudinal temperature gradient and its evolution through times, a crucial driver of global climatic conditions ⁴¹. Moreover, the inclusion of δ¹⁸O measurements will provide valuable information on relative changes in oceanic circulation and temperature, as well as on continental ice volume. Such a multiproxy database, will facilitate comparison of the SST proxies on a global scale, and should increase our understanding the proxies themselves.

Deliverable: submission of a scientific article (D3 month 27).

Milestones: Decision on reconstruction method used, and on focus periods (M3 month 23)

WP4: Interactions between carbon budget, carbon distribution and global climate (m27-m34)

The combined study of climate/temperature and carbon cycling proxies, and their relative timing and geographical structure will help quantifying the relative importance of processes associated to biological pump variability and carbon budget variations. I will develop empirical models of the integrated carbon cycling and climate using statistical relationships. I will evaluate the links between oceanic δ¹³C distribution, temperature, and the changes in the pCO₂ using data constrained semi empirical model, and compare them to previously published studies based on coupled climate model ^17,42. Collaboration has already been initiated with PhD students from Bern University to design specific experiments for this study (A. Jeltsch-Thömmes & G. Battaglia, F. Joos ³⁹). Sensitivity experiments are performed with Bern3D, testing the impact of changes in oceanic circulation, nutrient availability and remineralization depth on δ¹³C of DIC and on inorganic and organic carbon burial, with open and closed system configuration. I will use the inferred relationships between Δδ¹³C and ΔpCO₂ with the actual data-based reconstructions to quantitatively evaluate the net impact of the biological pump on the atmospheric pCO₂ and global temperature over the last glacial cycle.

I will also perform cross proxy analyses, taking advantage of the several proxy types included in a single database, and minimizing the age model uncertainties. Studying the timing of the changes in both the carbon cycling and SST will also be strongly facilitated with the CYCLOCARB multiproxy database. I will compare the reconstructions to the geographic and temporal signature of different processes explored using numerical simulations in terms of PCO₂, carbon budget and distribution, and temperature.

Deliverable: submission of a scientific article (D4 month 34).

Milestones: Decision on reconstruction method used, and on focus periods (M4 month 31)

WP5: Administration of the project, Dissemination, Mentoring and Job and Funding applications (over the entire duration of the project): see details in section 1.2, point 5 to 8, and section 3.1.