The sequestration efficiency of the deep ocean: Fast computations using transport matrices built from ACCESS-ESM1.5 archivesBenoît Pasquier,Richard J. Matear,Matthew A. Chamberlain,Tilo Ziehn,David K. Hutchinson,François W. Primeau,Yi Liu,Ann Bardin COSIMA Working Group MeetingApril 10, 2025Online
What Sets Neodymium Distributions in the Ocean and How Does This Impact its Utility as a Paleoceanographic Proxy?Sophia K. V. Hines,Paige M. Wise,Benoît Pasquier,Seth G. John AGU Fall MeetingDecember 9, 2024Washington, DC, USA
Neodymium (Nd) isotope ratios have emerged as a promising paleoceanographic proxy for deep ocean circulation; however, their utility relies on a clear understanding of what determines Nd isotope distributions in the ocean today. If inputs to the surface ocean dominate, then Nd isotope values in the ocean interior can be used to reconstruct water mass mixing in the past. If benthic sources dominate, then the isotopic composition of a water mass can be modified along its flow path, complicating the interpretation of past ocean reconstructions. We use a new tracer-enabled transport matrix model, the Global Neodymium Ocean Model (GNOM), to investigate the controls on the marine Nd cycle. Parameters that set Nd sources to the ocean and cycling within the ocean are not prescribed, but instead simultaneously optimized given realistic ranges based on experiments and observations. Due to its computational efficiency, we can use this model to explore a large range of parameter space and determine how each source influences observed isotope distributions. To test the robustness of the model, given the large number of free parameters, we run model optimizations systematically omitting sources from sediments, rivers, and dust. We also test the importance of sedimentary fluxes at different depths. We find that sediments, rivers, and dust are the most important sources of Nd to the ocean, but that shallow sediments contribute more than deep sediments.Thus, rather than a simple ‘top-down’ or ‘bottom-up’ view of sedimentary Nd cycling, we find that a variety of sources at various depths contribute to Nd cycling in the oceans.
Shallow vs Deep: Identifying the magnitude and locations of marine Nd sources with GNOM Paige M. Wise,Benoît Pasquier,Seth G. John,Sophia K. V. Hines Goldschmidt ConferenceAugust 18, 2024Chicago, IL, USA
Neodymium (Nd) isotope ratios serve as a valuable paleo-proxyies aiding in unraveling the complexities of ocean circulation's impact on carbon drawdown. Nd isotopes (εNd) have the potential to elucidate the ocean's configuration during significant climate transitions, wherein the deep ocean acted as a major carbon reservoir. Despite their potential, questions persist regarding Nd cycling in the ocean. To address this, we utilize the Global Neodymium Ocean Model (GNOM), a tracer-enabled transport matrix model, offering rapid computational insights. GNOM identifies rivers, dust, and sediments as primary Nd sources and prefers a "Top-Down" sediment release mechanism, a topic of recent debate. Notably, our model experiments challenge this paradigm, revealing a nuanced interplay between surface and deep ocean sources. We explore the impacts of restricting specific sources, demonstrating the pivotal role of shallow sediments in shaping Nd distributions. Enhanced sediment reactivity, particularly in extreme εNd ranges, significantly influences benthic fluxes. Moreover, our findings suggest different and distinct controls on the Atlantic εNd profile and the Pacific εNd profile, with implications for paleo-tracing methodologies. By elucidating the "Top-down vs Bottom-up" debate, our study underscores the importance of sedimentary processes in modulating marine Nd cycling and underscores the necessity for cautious interpretation of εNd records, particularly in the Pacific Ocean. Through GNOM modeling, we advance understanding of Nd isotope dynamics, shedding light on crucial drivers of oceanic εNd variability.
The Ocean's Carbon and Oxygen Cycles in Future Steady-State Climate ScenariosBenoît Pasquier,Mark Holzer,Matthew A. Chamberlain,Richard J. Matear,Nathaniel L. Bindoff,François W. Primeau Ocean Sciences MeetingFebruary 20, 2024New Orleans, LA,
What would be the state of the marine carbon and oxygen cycles if the ocean's physical and thermodynamic state was frozen in time? We answer this question by embedding a simple data-constrained model of the nutrient, carbon, and oxygen cycles in steady circulations that correspond to perpetual 2090s conditions as simulated for the RCP4.5 and RCP8.5 scenarios. Focusing on steady-state changes from preindustrial conditions allows us to capture the response of the system on all timescales, not just on the sub-centennial timescales of typical transient simulations. We find that biological production experiences only modest declines because the reduced nutrient supply by a more sluggish future circulation is counteracted by warming-stimulated growth. Organic-matter export declines by 15–25 % due to reductions in both biological production and export ratios, the latter driven by warming-accelerated shallow respiration and reduced subduction of dissolved organic matter. The future biological pump becomes stronger in the sense that it cycles a 30–70 % larger regenerated carbon inventory accumulated over dramatically longer sequestration times, while pump efficiency is reduced in the sense that preformed DIC is shunted away from biological utilization to outgassing. Near-surface paths of preformed DIC become more important in the future as weakened ventilation isolates the deep ocean. Thus, while regenerated DIC cycling becomes slower, preformed DIC cycling speeds up in the future for inventory changes of similar magnitude. Intense global deoxygenation (25–65 % less oxygen globally) is driven primarily by the circulation through (i) increased residence times and total oxygen utilization despite reduced respiration rates, underlining reexposure times as the key link between regenerated carbon and oxygen inventories, and (ii) the decrease of the fraction of the ocean interior that is ventilated from high latitudes, which partially shuts down the corresponding cold and oxygen-rich water supply to the deep ocean.
Optimal parameters for the ocean's nutrient, Californiarbon, and oxygen cycles compensate for circulation biases but replumb the biological pumpBenoît Pasquier,Mark Holzer,Matthew A. Chamberlain,Richard J. Matear,Nathaniel L. Bindoff,François W. Primeau AMOS National ConferenceFebruary 5, 2024Canberra, CT, Australia
Modeling the ocean's carbon and oxygen cycles accurately is challenging because of uncertainties in both biogeochemistry and ocean circulation. One solution to reduce the mismatch between observed and modelled biogeochemical tracer fields is biogeochemical parameter optimization. I will present recent work where we show that parameter optimization may compensate for biases of the embedding circulation model by altering the inner workings of the biological pump, with implications for estimating the future response of the ocean system to environmental change. To this end we embedded a mechanistic model of the ocean's coupled nutrient, carbon, and oxygen cycles into two circulation models: OCIM2, which is data-assimilated, and ACCESS-M, which is built from the ACCESS1.3 climate model. We find that parameter optimization reduces mismatch with tracer observations and reproduces the global biological pump strength and regenerated inventories for both circulations, but ACCESS-M export production optimizes to twice that of OCIM2 to compensate for ACCESS-M having lower sequestration efficiencies driven by less efficient particle transfer and shorter residence times. Idealized simulations forcing complete Southern Ocean nutrient utilization show that the response of the optimized system is strongly sensitive to the embedding circulation. In ACCESS-M, Southern Ocean nutrient and carbon trapping is partially short circuited by unrealistically deep mixed layers. For both circulations, intense Southern Ocean production deoxygenates Southern-Ocean-sourced deep waters, muting the imprint of circulation biases on oxygen. Our findings highlight that the biological pump's plumbing needs careful assessment to predict the biogeochemical response to ecological changes, even when optimally matching observations.
Quantifying the Dominant Fluxes of the Modern Marine Neodymium Cycle through Model OptimizationPaige M. Wise,Lowell Stott,Seth G. John,Benoît Pasquier AGU Fall MeetingDecember 15, 2023San Francisco, California, USA
Neodymium (Nd) isotopes serve as a valuable proxy for water mass mixing due to their biologically inert nature and distinct isotopic signatures (εNd). This makes εNd a significant tracer for reconstructing past ocean circulation in paleoceanography. However, uncertainties persist in understanding the dominant fluxes and sources/sinks of the global ocean Nd cycle, limiting its full potential as a paleo-proxy. Field data alone has been insufficient in constraining the marine Nd budget, prompting the utilization of models like the Global Neodymium Ocean Model (GNOM). GNOM stands out for its computational efficiency, embedded data-assimilation circulation scheme, and diagnostic capabilities. Built on a transport-matrix model framework, GNOM enables simultaneous optimization of sources/sinks and internal cycling parameters within prescribed bounds to match observational data. Continuous improvements, such as discretization of river concentration parameters and refined dust parameter ranges, are ongoing since its initial release. Accurately modeling modern ocean Nd cycling with GNOM will enhance the implementation of εNd as a water mass mixing paleo-proxy. In future versions of GNOM, we plan to incorporate paleo-circulation matrices, including the Last Glacial Maximum, to further improve our understanding of past ocean circulation dynamics. This will contribute to more robust paleoceanographic reconstructions using εNd as a key tool in unraveling the Earth's climatic history.
Global distribution of nickel sources and sinks from a diagnostic modelSeth G. John,Hengdi Liang,Benoît Pasquier,Mark Holzer,Sam Silva Goldschmidt ConferenceJuly 9, 2023Lyon, France doi: 10.7185/gold2023.20635
Nickel is a micronutrient for phytoplankton in the oceans. While it generally has a ‘nutrient-type’ profile, there remain uncertainties about the biological and abiotic processes responsible for cycling Ni. Here we apply a diagnostic modeling technique to evaluate the 3D spatial distribution of Ni sources and sinks in the global ocean. In contrast to a prognostic model, where specific biogeochemical processes are encoded, the diagnostic approach couples global OCIM circulation and a global Ni climatology in order to determine the location of Ni sources and sinks, without invoking explicit biogeochemistry. First we predicted a global Ni climatology using various machine learning techniques, choosing simple multiple linear regression as the optimal approach. Ni fluxes were then diagnosed with a nutrient-restoring model. We find that Ni uptake in the surface oceans is spatially co-located with P uptake and not with Si uptake, suggesting that Ni is not incorporated into diatom frustules despite the global similarity between Si and Ni distributions. We also find that Ni/P uptake ratios increase as Ni is depleted, providing evidence against a pool of non-bioavailable Ni in the surface ocean. Ni regeneration is deeper than P but shallower than Si, suggesting a decoupling between regeneration of biogenic Ni and P from sinking particles.
PCO2: A simple biogeochemistry model embedded in a simple ocean circulation model in matrix form Benoît Pasquier UNSW Ocean Research CarnivalFebruary 22, 2023Sydney, NSW, Australia
The Biogeochemical Balance that Controls Oceanic Nickel Cycling in the Modern and Past OceansSeth G. John,Rachel Kelly,Xiaopeng Bian,Shun-Chung Yang,Feixue Fu,Magdalene I. Smith,Nathan Lanning,Hengdi Liang,Benoît Pasquier,Emily Seelen,Mark Holzer,Tim M. Conway,Jessica N. Fitzsimmons,David Hutchins Goldschmidt ConferenceJuly 10, 2022Honolulu, Hawaii, USA doi: 10.46427/gold2022.12708
Nickel is an important element in the oceans, utilized by enzymes involved in important marine processes including nitrogen fixation, uptake of fixed nitrogen, and methanogenesis. However, important questions remain about the processes which control the distribution of Ni in modern ocean, and the processes which control the global ocean Ni inventory on geological timescales. We have combined the analysis of Ni in field samples, in vitro lab culturing experiments, and modeling, in order to present a novel view of global Ni biogeochemical cycling. Two key features of the global Ni distribution are a deep concentration maximum reminiscent of Si, and surface Ni concentrations which never drop much below 2 nM. Common explanations for these features include a presumed presence of Ni within diatom silicate frustules, and the presence of ~2 nM non-bioavailable Ni in the surface oceans. However, neither supposition is consistent with our new data, which show very low Ni in diatom frustules, and the ability of phytoplankton to deplete Ni below 2 nM when provided sufficient macronutrients. Instead, we use the AWESOME OCIM modeling framework to explore alternative explanations for Ni distribution which are consistent with our data. We propose a new view of the Ni cycle in which surface ocean oligotrophic gyre Ni concentrations are set by high latitude nutrient uptake, with Ni being depleted slightly more slowly than macronutrients N and P, leaving 2 nM residual Ni after macronutrient depletion. We attribute the deeper depth maximum of Ni to reversible scavenging. This new view suggests that surface ocean Ni concentrations are highly sensitive both to Ni uptake in upwelling regions and to scavenging intensity. Both of these processes are expected to have varied in the past ocean, possibly leading to large and rapid fluctuations in surface ocean Ni concentrations, placing important ocean processes on a knife's edge between Ni feast and famine.
Modeling Marine Ecosystems At Multiple Scales Using JuliaGaël Forget,Benoît Pasquier,Zhen Wu JuliaConJuly 28, 2021Online
Life in the oceans is strongly connected to our climate. In this workshop, you will learn to use packages from the JuliaOcean and JuliaClimate organizations that provide a foundation for studying marine ecosystems across a wide range of scales. We will first run agent-based models to explore individual microbes and processes that drive species interactions. On the other end of the model hierarchy, we will simulate planetary-scale transports that control ocean biogeography and climate change.
Global Controls on the Distribution of Nickel in the OceansSeth G. John,Shun-Chung Yang,Xiaopeng Bian,Rachel Kelly,David Hutchins,Feixue Fu,Hengdi Liang,Benoît Pasquier,Emily Seelen Goldschmidt ConferenceJune 22, 2020Online doi: 10.46427/gold2020.1217
Nickel appears to be an important micronutrient for phytoplankton living in the modern oceans, based on the known use of Ni as a cofactor for biologically important enzymes such as superoxide dismutase and urease. It also has a ‘nutrient-type’ global distribution, whereby Ni is depleted in the surface ocean and increases between the Atlantic and Pacific in deep waters. Yet, many questions remain about the biological cycling of Ni in the oceans. Particularly, there is uncertainty about the mechanisms by which Ni is maintained at concentrations which never fall much below 2 nM in oligotrophic gyres, and whether Ni could have been a limiting nutrient in the past oceans. Using a combination of experiments, observations, and modeling, we propose a new comprehensive model of Ni and Ni isotope (δ60Ni) cycling in the oceans. We find that two models are broadly consistent with oligotrophic gyre Ni concentrations, either there is a pool of ~2 nM Ni which is not biologically available, or the biological uptake of Ni from upwelling waters until macronutrients are depleted, leaving about 2 nM residual Ni. However, targeted modeling studies of Ni and δ60Ni distributions in the North Pacific, as well as experiments on Ni bioavailability, show no evidence of an inert Ni pool, thus favoring the second model. The amount of Ni present in oligotrophic gyres is therefore highly dependent on the relative uptake rates of Ni and macronutrients in upwelling regions, represented by the parameter β. Observed β for modern Southern Ocean diatoms is consistent with the model-predicted β. However, a survey of β values reported for cultured and natural phytoplankton shows great variability in this parameter, depending on species and growth conditions. Thus, oligotrophic gyre Ni concentrations may be highly dependent on the exact species composition and growth conditions in the Southern Ocean and other upwelling regions. This suggests that Ni concentration variability in the past surface ocean may be much greater than previously considered, and that Ni ‘famines’ in the past ocean may have been frequent.
Metals, Models, and Isotopes: Insights into the Biogeochemical Cycling of Trace Nutrients in the OceanSeth G. John,Tim M. Conway,Tom Weber,Timothy DeVries,Alessandro Tagliabue,Hengdi Liang,Benoît Pasquier Goldschmidt ConferenceJune 21, 2020Online doi: 10.46427/gold2020.1216
The past few years has seen revolutions in apparently dissimilar fields, one in the availability of ocean trace-metal concentration and isotope data available through the GEOTRACES program, and another in techniques to model steady-state global tracer distributions using OCIM techiques. Yet, at the intersection between the two, recent efforts have leveraged new data and new modeling techniques in order to better understand the global biogeochemical cycling of trace- metal nutrients in the oceans. OCIM type models are particularly well suited to exploring trace-metal biogeochemical cycling. New tools such as the AWESOME OCIM allow such models to be run on a laptop, using code specifically written for non-expert modelers. Steady-state tracer distributions can be solved in just a few seconds, allowing for thousands of model runs as models are ‘tuned’ to match observations. In this way, various hypotheses about metal isotope cycling can be tested. Hypotheses can be rejected if there is no version of the model consistent with observations, while plausible models can be used to probe the specific conditions (parameter values) under which biogeochemical processes might occur. OCIM models of several trace-metal nutrients hilight the power of this approach. Fe concentrations and δ56Fe can be used in OCIM models to fingerprint Fe sources, and quantify biological cycling. Zn and δ66Zn models constrain both the rates of biological Zn uptake and the pathways by which Zn is transferred into the deep ocean. Cd and δ114Cd models can be used to elucidate the mechanisms behind the global similarity between Cd and P distributions. Ni and δ60Ni hilight the particular biological controls which allow excess Ni to persist globally in oligotrophic gyres, and hint at the possibility of past-ocean Ni limitation. And Cu models, along with the lack of δ65Cu variability, can be used to hilight previously unconsidered sources of Cu to the oceans.
Julia users and tools for oceanographyGaël Forget,Benoît Pasquier,Alexander Barth,Milan Klöwer,Ali Ramadan,Gregory L. Wagner,Constantinou Navid Ocean Sciences MeetingFebruary 17, 2020San Diego, California, USA
There has been a visible uptick in oceanography and climate applications of the Julia language since it reached the v1.0 milestone last year. The growth and appeal for this language were recently highlighted by Nature magazine. It seems very timely to offer this rapidly growing community of open source developers and users an opportunity to meet in person, advertise their recent efforts, and engage with the oceanographic community at large. A tentative agenda for this workshop would include: a brief general presentation of Julia, a survey of existing efforts, a hackathon-type session with both experienced and new users interacting, and open-ended discussion time.
AIBECS.jl: the ideal tool for marine biogeochemistry modelling Benoît Pasquier,François Primeau Ocean Sciences MeetingFebruary 16, 2020San Diego, California, USA
Progress in global marine biogeochemistry increasingly relies on models. In general, converting conceptual mechanisms into useful numerical simulations comes with large computational costs and significant amounts of time developing code. A recently developed alternative is the AWESOME OCIM (A Working Environment for Simulating Ocean Movement and Elemental cycling in an Ocean Circulation Inverse Model, or AO), which drastically reduces development time with its graphical user interface (GUI). Building on the OCIM's steady-state matrix formulation of the ocean circulation, the AO allows computationally efficient simulations of single tracers controlled by linear mechanisms. Here, we present the AIBECS (for Algebraic Implicit Biogeochemical Elemental Cycling System), an open-source, user-friendly, fast, and modular Julia package, which aims to extend the AO's capabilities to nonlinear, multi-tracer systems, embedded in other circulations than just the OCIM. Instead of a GUI, the AIBECS provides an application programming interface (API) to create global steady-state marine biogeochemistry models in a few lines of code, allowing oceanographers to focus on research instead of wasting precious time reinventing the wheel. Because of its ease of use, the AIBECS is perfect for teaching and exploratory research. Yet, it is designed for cutting-edge research owing to its advanced algorithms and diagnostic capabilities. Under the hood, the AIBECS runs with state-of-the-art nonlinear-system solvers and auto-differentiation algorithms. Combined with highly-efficient parameter optimization or uncertainty analysis tools, it aspires to be at the forefront of global biogeochemistry modeling. In short, the AIBECS is the ideal tool for exploring global marine biogeochemical cycles.
F-1 algorithm: Efficient differentiation through large steady-state problems Benoît Pasquier,François W. Primeau UNSW Applied Maths SeminarAugust 7, 2019Sydney, NSW, Australia
Introducing a Julia package to efficiently differentiate an objective function defined implicitly by the solution of a large PDE system. Steady-state systems of nonlinear partial differential equations (PDEs) are common in engineering and the biogeosciences. These systems are typically controlled by parameters that can be estimated efficiently using second-order optimization algorithms. However, computing the gradient vector and Hessian matrix of a given objective function defined implicitly by the solution of large PDE systems is seldom economical. A fast and easy-to-use algorithm is introduced for computing the gradient and Hessian of an objective function implicitly constrained by a steady-state PDE system. The algorithm, which is based on the use of hyperdual numbers, is called the F-1 algorithm, because it requires only one factorization of the constraint-equation Jacobian. Careful examination of the relationships that arise from differentiating the PDE system reveal analytical shortcuts that the F-1 algorithm leverages. Benchmarks of the F-1 algorithm against five numerical differentiation schemes are shown in the context of optimizing a global steady-state model of the marine phosphorus cycle that depends explicitly on m=6 parameters. In this context, the F-1 algorithm computes the Hessian 16 to 100 times faster than other algorithms, allowing for the entire optimization procedure to be performed 4 to 26 times faster. This is because other algorithms require O(m) to O(m2) factorizations, which suggests even greater speedups for larger problems. A live demonstration of using the F-1 algorithm, which is implemented as a Julia package, is given.
Introducing AIBECS.jl, a Julia package for creating global marine biogeochemistry modelsBenoît Pasquier,François W. Primeau,J. Keith Moore CCRC SeminarsAugust 7, 2019Sydney, NSW, Australia
Running standard global marine biogeochemistry models comes with large computational costs and with a steep learning curve. The recently developed AWESOME OCIM (for A Working Environment for Simulating Ocean Movement and Elemental cycling in an Ocean Circulation Inverse Model) is an attractive alternative because it offers better user experience thanks to a MATLAB Graphical User Interface (GUI), and faster computations thanks to the OCIM's steady-state matrix formulation of the ocean circulation. Benoît will introduce and give a live demonstration of the AIBECS (for Algebraic Implicit Biogeochemical Elemental Cycling System), an open-source, user-friendly, fast, and modular Julia package, which aims to provide a solution to the limitations of the AWESOME OCIM. The AIBECS provides an Application Programming Interface (API) to generate global steady-state marine biogeochemistry models in just a few lines of code. Under the hood, the AIBECS comes with state-of-the-art nonlinear-system solvers, auto-differentiation algorithms, and was designed with parameter optimization/estimation and uncertainty analysis in mind. The AIBECS allows researchers to focus on the science rather than spending time reinventing the wheel for differentiating convoluted systems, for solving nonlinear problems, or for leveraging cryptic linear-algebra shortcuts. Because of its ease of use, the AIBECS is ideal for teaching and exploratory research. The AIBECS is also ideal for cutting-edge research owing to its open-source design, its modularity, its advanced algorithms, and its novel-diagnostic capabilities.
Southern Ocean silicic-acid leakage: Sensitivity to diatom physiology and its Si isotope signature analyzed in an inverse model Mark Holzer,Benoît Pasquier,Tim DeVries,Mark Brzezinski AMOS National ConferenceJune 13, 2019Darwin, NT, Australia
We quantify the response of the global nutrient cycles and phytoplankton community structure to Southern Ocean iron fertilization using an inverse model of the coupled Fe-P-Si cycles that contains a mechanistic representation of phytoplankton abundance and nutrient colimitation. We find that different parameterizations of the iron dependence of the Si:P uptake ratio allow equally good fits to the observational nutrient climatology but produce very different responses to iron fertilization. The different Si:P parameterizations considered are also consistent with the available observational data of diatom physiology but the steady-state response to iron fertilization depends sensitively on the rate with which the Si:P uptake ratio decreases with increasing dissolved-iron concentration (DFe). For a sufficiently rapid decrease of Si:P, iron fertilization leads to silicic acid becoming partially untrapped from the Southern Ocean and leaking to low latitudes with an accompanying shift in floral composition. If Si:P decreases more slowly with increasing DFe, iron fertilization leads to strengthened Southern Ocean silicon trapping. For all cases, the global response of the biological phosphorus and silicon pumps, and its implications for carbon uptake, are dominated by the Southern Ocean. The Si-isotope signature of the opal flux to the sediments distinguishes between different forms of the Si:P uptake ratio, with leakage producing isotopically lighter sediments similar to the observational record for the last glacial maximum.
The number of past and future regenerations of iron in the ocean and its intrinsic fertilization efficiencyBenoît Pasquier,Mark Holzer MIT Follows Lab Group MeetingJanuary 2, 2019Boston, MA, USA
Developing a new, open-source, user-friendly, fast, modular, global marine biogeochemistry model (in Julia)Benoît Pasquier MIT EAPS Sack-lunch seminarJanuary 1, 2019Boston, MA, USA
One of the most successful tools made available to oceanographers recently is the Ocean Data View (ODV) software, used "for the analysis and visualization of oceanographic and meteorological data sets." Part of the success of ODV comes from its simplicity, which makes it attractive to oceanographers. Current global marine biogeochemistry models, on the other hand, lack this appeal and generally come with a steep learning curve. So steep, in fact, that very few non-modelers ever use these models on their own and computational tasks are delegated to collaborating model experts. This problematic barrier to entry in modeling, which hinders the progress of oceanography, has recently been lowered with the release of the AWESOME OCIM (A Working Environment for Simulating Ocean Movement and Elemental cycling in an Ocean Circulation Inverse Model, or AO), a software with a graphical user interface (GUI) for modeling a single biogeochemical tracer, using — under the hood — linear algebra. Here I will present some work in progress for developing a new marine biogeochemistry model that builds on the ideas behind the AO and provides users with an open-source, user-friendly, fast, and modular software. Among other goals, this model aims to allow users to represent the coupling of multiple tracers (such as dissolved iron (DFe) and macronutrients) that can be embedded in different models of the ocean's circulation (e.g., swap in the OCIM transport with an average of the MITgcm circulation). The model also addresses a number of features that are currently absent in the AO, like the ability to represent non-linear mechanisms (such as the scavenging of DFe by sinking particles). Additionally, a powerful feature of the OCIM that is not leveraged by the AO is its computational speed, which allows for objective parameter optimization and sensitivity analysis. Most of the model's capabilities stem from the language it is developed in, the Julia language, which is a free, open-source, general-purpose, and fast alternative to MATLAB (the language in which the AO is written).
Offline parameter optimization for global marine biogeochemical modelsBenoît Pasquier UCI Primeau Lab Group MeetingFebruary 18, 2018Irvine, California, USA
Inverse-model estimates of the ocean's coupled phosphorus, silicon, and iron cycles.Benoît Pasquier,Mark Holzer Ocean Sciences MeetingFebruary 12, 2018Portland, Oregon, USA
A key question about the iron cycle is how much export is supported by the different iron sources (aeolian, sedimentary, and hydrothermal). Previous studies shut down a given source to quantify its importance, but this approach cannot quantify the source's contribution to export in the unperturbed state because of the system's nonlinearities. Moreover, such studies used forward models that were not objectively data constrained. Here, we estimate the ocean's steady-state coupled phosphorus, silicon, and iron cycles using a new inverse model embedded in a data-assimilated circulation. The model features the redissolution of scavenged iron, a subgrid topography parameterization, and three phytoplankton functional classes. Phytoplankton concentrations are represented implicitly in the formulation of nutrient uptake. Efficient numerics allow us to optimize biogeochemical parameters against observations. Because iron sources are uncertain, largely due to poorly constrained scavenging, we generate a family of state estimates for a wide range of source strengths. All states have similar fidelity to the observations. From our state estimates, we quantify the relative contribution of each iron source to export production. This is done non-invasively by labelling each iron type with a suitable passive tracer. We find that the phosphorus and opal exports are well constrained at 8.1±0.3 TmolP yr-1 and 171.±3. TmolSi yr-1. The exports supported by each iron type have well constrained patterns. Sedimentary iron supports export primarily in shelf and upwelling regions, while hydrothermal iron supports export mostly in the Southern Ocean. Per source-injected molecule, aeolian iron supports 3.1±0.8 times more phosphorus export and 2.0±0.5 times more opal export than the other iron types. Conversely, per injected molecule, sedimentary and hydrothermal iron support 2.3±0.6 and 4.±2. times less phosphorus export, and 1.9±0.5 and 2.±1. times less opal export than the other iron types.
The efficiency of different iron sources in supporting the ocean's global biological pumpBenoît Pasquier,Mark Holzer UCI Half-baked seminar, Department of Earth System ScienceNovember 15, 2017Irvine, California, USA
Response of the biological pump to perturbations in the iron supply: Global teleconnections diagnosed using an inverse model of the coupled phosphorus-silicon-iron nutrient cycles Benoît Pasquier,Mark Holzer AMOS National ConferenceFebruary 7, 2017Canberra, Australia
We construct a mechanistic inverse model of the ocean's coupled phosphorus, silicon, and iron cycles and analyze the response of the biological pump to perturbations in the iron supply. The nutrient cycles are embedded in a data-assimilated steady global ocean circulation. Biological nutrient uptake is parameterized in terms of nutrient, light, and temperature limitations on growth for three functional classes of phytoplankton. A matrix formulation of the discretized nutrient equations permits efficient numerical solutions that allow optimization of key biogeochemical parameters by minimizing the misfit between modelled and observed concentrations. We perturb the iron supply for a variety of scenarios and systematically quantify the teleconnections in nutrient utilization across the global ocean ecosystem. Specifically, Green-function techniques are used to quantify the transport pathways and timescales with which the perturbations in the nutrient fields are propagated, thus mediating the teleconnections. We find that carbon and opal export can have opposite responses to changes in the iron supply. For example, a globally uniform reduction in the aeolian iron input increases opal export outside of the Southern Ocean but decreases carbon export there. A path-density transport diagnostic applied to the nutrients shows that enhanced iron limitation can untrap silicon from the Southern Ocean and increase opal export outside of the Southern Ocean. However, enhanced iron limitation also leads to non-diatom phytoplankton exporting less carbon in the tropics and to an increase in the biomass fraction of diatoms, which increases the Si:C export ratio. In addition, we investigate the amplitudes of the iron perturbations necessary to eliminate iron limitation and the resulting changes in the patterns of macronutrient limitation.
Exploring iron control on global productivity: "FePSi", an inverse model of the ocean's coupled phosphate, silicon and iron cycles Benoît Pasquier,Mark Holzer Postgrad ConferenceJune 8, 2016Sydney, NSW, Australia
"FePSi" is the first data-constrained mechanistic inverse model coupling the iron (Fe), phosphorus (P), and silicon (Si) oceanic cycles. The nutrient cycling is embedded ina data-assimilated steady global circulation. Biological nutrient uptake is parameterized in terms of nutrient, light, and temperature limitations on growth for 3 classes of phytoplankton that are not transported explicitly. A sparse matrix formulation of the discretized nutrient tracer equations allows for efficient numerical solutions using Newton's method, which facilitates the objective optimization of the key biogeochemical parameters. The optimization minimizes the misfit between modeled and observed nutrients and chlorophyll fields. We explore the nonlinear, counterintuitive and asymmetric responses of the biological pump and nutrient cycles to changes in the aeolian iron supply for a variety of scenarios. Specifically, Green-function techniques are employed to quantify in detail the pathways and timescales with which those perturbations are propagated throughout the world oceans, determining the global teleconnections that mediate the response of the global ocean ecosystem.
Iron Source Attribution and the Age of Dissolved Iron in the OceanMark Holzer,Marina Frants,Benoît Pasquier Ocean Sciences MeetingFebruary 23, 2016New Orleans, Louisiana, USA
We quantify the contributions of the aeolian, sediment, and hydrothermal iron sources to the concentration of dissolved iron (dFe) and calculate the mean age of these contributions since their injection into the ocean. Our calculations use estimates of the global iron cycle from a simple inverse model constrained by a data-assimilated global circulation and by available dFe measurements, including the GEOTRACES Intermediate Data Product. We compare our rigorously calculated source contributions with commonly employed source anomalies, i.e., the differences between solutions with and without the source in question. We find that such source anomalies strongly underestimate the true contribution from any given source (by as much as a factor of two for the hydrothermal source) because of the nonlinearity of the iron scavenging. Aeolian iron is the largest contributor in the Southern Ocean euphotic zone, where its mean age reveals that this iron is supplied from depth on average a few hundred years after deposition from the atmosphere. Hydrothermal and sedimentary iron each contribute roughly 20% to the total dFe concentration in the Southern Ocean euphotic zone. Hydrothermal iron tends to have the oldest surface ages except where hydrothermal sources occur near the surface where the scavenging rate is high. The systematics of the source contributions, iron ages, and iron-age distributions are quantified across a family of solutions with a range of aeolian source strengths, all of which are consistent with currently available dFe observations.
Iron control on global productivity: an efficient inverse model of the ocean's coupled phosphate, silicon, and iron cycles Benoît Pasquier,Mark Holzer Ocean Sciences MeetingFebruary 22, 2016New Orleans, Louisiana, USA
We construct a data-constrained mechanistic inverse model of the ocean's coupled phosphorus and iron cycles. The nutrient cycling is embedded in a data-assimilated steady global circulation. Biological nutrient uptake is parameterized in terms of nutrient, light, and temperature limitations on growth for two classes of phytoplankton that are not transported explicitly. A matrix formulation of the discretized nutrient tracer equations allows for efficient numerical solutions, which facilitates the objective optimization of the key biogeochemical parameters. The optimization minimizes the misfit between the modelled and observed nutrient fields of the current climate. We systematically assess the nonlinear response of the biological pump to changes in the aeolian iron supply for a variety of scenarios. Specifically, Green-function techniques are employed to quantify in detail the pathways and timescales with which those perturbations are propagated throughout the world oceans, determining the global teleconnections that mediate the response of the global ocean ecosystem. We confirm previous findings from idealized studies that increased iron fertilization decreases biological production in the subtropical gyres and we quantify the counterintuitive and asymmetric response of global productivity to increases and decreases in the aeolian iron supply.
The plumbing of the global biological pumpBenoît Pasquier,Mark Holzer AMOS National ConferenceJuly 16, 2015Brisbane, QLD, Australia
We quantify the timescales and pathways that set the efficiency of the biological pump (the fraction of the phosphate inventory that is regenerated). We use a data-constrained phosphorus-cycling model embedded in a steady data-assimilated ocean circulation to quantify the pump's leaks of preformed phosphate, its sources of regenerated phosphate, and the pathways with which the combined biogenic particle transport and the water circulation teleconnect different regions of the global euphotic zone. These pathways are quantified by a path density, which is the concentration of phosphate that was last utilized in a region A and that will reemerge into the euphotic zone of a region B, partitioned according to the A–to–B transit-time. Suitable integrals of this path density, computed efficiently by direct matrix inversions, yield the phosphate mass in transit, its flow rate, and its residence time in the aphotic zone. We find that a pump efficiency of (39 ± 2)% has dominant contributions from the Eastern Equatorial Pacific (25 ± 1)%, from the Southern Ocean (SO) (21 ± 1)%, and from the Eastern Equatorial Atlantic (EEqA) (12 ± 1)%. The pump’s 61% leak originates predominantly in the SO (75%) and in the SubPolar North Atlantic (17%). While the SO euphotic zone is a large leak of preformed phosphate, it is also the major receptor of phosphate reemerging from depth: The SO euphotic zone is the destination of (62 ± 6)% of the regenerated inventory and of (69 ± 5)% of the preformed inventory. The mean interior residence time of regenerated phosphate reemerging in the SO depends on where it was last utilized: 69 ± 1 years if last utilized in the SO and 500 ± 20 years if last utilized outside the SO. The transit-time distribution of the mass of regenerated phosphate last taken up in the EEqA and reemerging in the SO euphotic zone is bimodal, pointing to two distinct pathways which are quantified using the phosphate path density.
An efficient inverse model of the ocean's coupled nutrient cyclesBenoît Pasquier,Mark Holzer Postgrad ConferenceJune 11, 2015Sydney, NSW, Australia
We construct a data-constrained model of the ocean's coupled macronutrient and micronutrient cycles. The model focuses initially on phosphate and dissolved iron. The nutrient cycling is embedded in a data-assimilated steady ocean circulation. Biological nutrient uptake is parameterized in terms of nutrient and physical limitations on plankton growth, without the need of tracers for the concentration of phytoplankton. The uptake parameterization is formulated using a novel, versatile functional form that is able to capture different plankton classes, both in terms of size and species. A matrix formulation of the discretized partial differential equations allows for very efficient solutions and facilitates the objective optimization of key model parameters by minimizing the mismatch with the observed global nutrient climatology. This approach matches observed phosphate and iron concentration with RMS errors of less than 10%. In the near future, the model will allow us to quantify the timescales and pathways with which perturbations in the iron supply are communicated throughout the world ocean's ecosystem. Including the ocean's silicon cycle will elucidate the role of diatoms in the biological pump and the sensitivity of elemental ratios to iron perturbations.
Plumbing of the biological pumpBenoît Pasquier,Mark Holzer Postgrad ConferenceSeptember 1, 2014Sydney, NSW, Australia
Nutrient transport and biological teleconnections of Ocean surface regions are diagnosed in a data-assimilated circulation model coupled to a jointly optimized simple phosphorus cycling model. These teleconnections paint the plumbing of the biological pump: phytoplankton takes up phosphate (PO4) in the euphotic layer, pumps it as dissolved organic phosphorus (DOP) through the whole water column, until it is remineralized in the ocean depths and ultimately transported to the surface where it is available again for uptake. The biological pump efficiency (39%) is redefined and computed. Ocean surface regions where biological production takes its origin are defined, out of which major contributors as well as biological leaks are determined: The high latitude ocean regions provide a large quantity of phosphate to other oceans, but contribute very lightly themselves to the biological pump. While the easternmost part of the tropical oceans produce large quantities of utilized phosphorus which is pumped for long average transit times. Using Green functions and adjoint techniques, flow rates, masses in transit, and timescales of biologically utilized nutrients between surface regions of interest are computed: the easternmost part of tropical ocean basins provide the majority of the biology, which reemerges predominantly in the high latitudes. Time-dependent path densities of major teleconnections are then computed, which paint a very detailed and quantitative picture of the phosphorus cycle within the ocean interior: paths associated with long transit times (>1000 yrs) spread through the entire ocean, but are concentrated in the deep Northernmost Pacific, while paths associated with short transit times (<500 yrs) are more localised, and restricted mostly to the surface currents.