Co-environing the ocean and climate: The Argo program

Introduction

For most of history, people have had little knowledge of processes unfolding beneath the surface of the ocean. While scientists have gradually gathered more information from the depths over the last century, until recently deep-sea data points remained sparse and painstakingly collected. While this is still the case for some parts of the ocean, the first two decades of the twenty-first century have seen a steep increase in the amount of subsurface ocean data that are collected and made available for scientists and others to use. To a large extent, this change is due to the development of autonomous sub-surface sensors that have freed many ocean observations from a dependence on expensive ships and allowed for continuous sampling of vast areas (Wong et al., 2020).

With an extensive fleet of robotic sensors, the Argo program is at the center of this new strategy for ocean data collection. Argo floats are neutrally buoyant instruments that drift in the upper 2000 meters of the ocean, recording data on currents, temperature and salinity. The program is an international collaboration, with around 30 countries contributing one or more floats. The US is responsible for more than half the fleet, with many other countries contributing between several hundred and only a few floats each. In the past two decades, the program has produced extensive data for oceanographic, climate and Earth System sciences, informing fundamental insights on ocean currents, the mixing of water masses, ocean heat content and interactions between the ocean and the atmosphere. These data are crucial for climate studies, including predictive climate change models.

This article outlines the historical context of the ideas and practices that inform the Argo program and considers some of the implications of the program for contemporary interpretations of ocean science and sustainability. We describe the developments in ocean and climate science and governance that led to the proposal of the Argo program in the 1990s and help explain why, once proposed, the program was quickly implemented. We then look closely at the goals and motivations set out in the Argo proposal and follow these principles in subsequent reviews and updates, including recent expansions of the program. We focus our analysis on the question of how the technologies of the floats, the data they record, and the aims and expectations formulated around the program have ‘environed’ the global ocean and informed an increasingly climate-centered view of the marine realm of the planet. In this context, we also highlight how the Argo program connects to, and is part of, an increasing focus on datafication and programming as forms of Earth System governance. We suggest that specifically, Argo has co-environed the ocean with climate change in ways that have helped create new views of the ocean's historicity and its relationship to human temporalities. This historicity extends to human pasts as well as ocean futures through its ties to “cultures of prediction” associated with climate sciences and policy. Considering these cultures of prediction further, we argue that Argo data and ocean science and governance more broadly are increasingly intertwined with predictive aims, and that this has some important implications in relation to development of the so-called ‘blue economy’ with further references to notions of control, active management, economic development and climate change mitigation.

The concept of environing helps us tie the scientific progress and practices associated with the Argo program to the idea of the environment as nature that has been transformed through human agencies, including material as well as knowledge-based processes. We show that while the global ocean has been studied for centuries with gradually accumulating knowledge, Argo represents the inclusion of (parts of) the world ocean in a global environmental conception in a new way (Argo is not alone in this process, for accounts of preceding and related historical developments see e.g., Camprubi and Robinson, 2016; Helmreich, 2009; Höhler, 2021; Lidström et al., 2022; Rozwadowski, 2018). Our analysis builds on other approaches that address representation, visualization and quantification of scientific knowledge and their imaginaries (eg. Lynch and Wolgar, 1990; Coopmans et al., 2014; Vertesi and Ribes, 2019). The terms ‘environing technologies’ and ‘environing media’ help us situate our analysis within the framework of environmental history as the study of the intersection between scientific observation, knowledge representation, material transformation, policy formation and cultural imaginaries that result in the formation of what is widely understood in the present as ‘the environment.’ Within this framework, we unpack Argo's role in changing the nature of human-ocean relations over the past decades. Our analysis helps to operationalize the idea of environing by applying it to the wide-ranging and complex yet interconnected set of Argo practices, moving from the physical floats and sensor technologies to cultural perceptions and policy practices, each with their own set of historical and contextual situatedness. We show how the terms of environing, environing technologies and environing media highlight different elements of these processes. We also contribute to developing the concept through exploring its interlinkages with neighboring analyses of increasing datafication, digitalization and programming as integral to Earth System science and governance.

The upper 2000 meters of the water column that the original, or “Core”, Argo program measures represent about half of the global ocean. Deep Argo floats, as well as floats that can measure additional biogeochemical parameters, known as BGC-Argo, are under development as part of a process to extend Core Argo into these ocean dimensions. There are also Polar Argo floats that can navigate the Arctic and Southern Oceans. This article focuses primarily on the Core Argo fleet. Our analysis draws on the scientific discourse surrounding the program, produced by Argo scientists, as well as on synthesizing ocean and climate science reports and frameworks for international ocean and climate governance. We have visited the Argo lab at Scripps Institution of Oceanography (SIO), a central node in the creation and early development of the program as well as in the network's present operations and expansions, including scientific as well as technological development and manufacturing. Due to the important role of SIO in the Argo program to date and our focus on this lab, together with the dominance of US Argo floats compared to other national fleets, there is a certain focus on SIO and the US context in our study. Complementary studies of other nodes in the Argo network would add important perspectives to our analysis.

The Argo program

The Argo program is a global array of about 4000 autonomous profiling floats that roam the ocean. Once deployed, the floats sink to a depth of one thousand meters where they drift with the current for ten days. They then sink another thousand meters before rising to the surface whilst recording a temperature and salinity profile. When they surface, they transmit their data via satellite to regional Argo Data Assembly Centers (DACs), where the data undergo a standardized real-time data quality control process before being sent to Argo Global Data Assembly Centers (GDACs), one located in Brest in France, and one in Monterey in California. From the GDACs, the raw data are made available for immediate purposes, such as weather forecasting, within 12 hours. The data are then further processed and eventually published as high-quality data for scientific research, usually within a year. The program was operationalized between 1999 and 2007, when it reached full implementation.

Neutrally buoyant floats were invented by British oceanographer John Swallow in the mid-twentieth century to research ocean circulation (Gould, 2005; Swallow, 1955). Initially, the floats were tracked by ships that determined their location and movements and eventually recovered them. The dependence on ships was broken in the 1980s when floats were designed that were able to transmit their data to satellites and then sink again to repeat the cycle. This decreased the floats’ operating costs and increased their operating lifetime. A parallel development expanded the use of the floats from passive drifting devices to active recorders of additional data points through added sensors. As the technology advanced, it also became more robust. By the 1990s, it had become sufficiently reliable and – importantly – cost-effective to allow for large-scale application.

The 1990s also saw the expansion of so-called “satellite oceanography”, which emphasized the need for complimentary subsurface data (Conway, 2006; Höhler, 2021). In 1992, the US National Aeronautics and Space Administration (NASA) and the French Centre national d'études spatiales (CNES) launched a satellite ocean topography experiment, or TOPEX/Poseidon. The second satellite, launched in 2001, was named Jason, the leader of the Argonauts in Greek mythology. The Argo program was named to signal its partnership with the satellite program.

The most immediate scientific predecessor to the Argo program was the World Ocean Circulation Experiment (WOCE), a ship-based project that traversed the ocean between 1990 and 1998 collecting subsurface data points for a one-time ‘snapshot’ of ocean conditions. Jessica Lehman has argued that while “WOCE was an ambitious project that drew on extensive international collaboration and emerging technologies that continue to play a significant role in how the global environment is known and governed”, the program also encountered “productive limits” by discovering that ocean circulation was much more variable and complex than previously thought, including many small-scale, nonlinear processes in addition to larger-scale circulation patterns, suggesting that the value of the one-time ‘snapshot’ produced by WOCE would quickly diminish as time went by (Lehman, 2021). The productive limits encountered by WOCE, together with the use of profiling floats within the project, inspired plans for a different approach where a much larger number of autonomous floats would replace individual ship-based measurements to provide a continuous flow of data, creating a dynamic instead of static representation of different water masses able to reflect their movements and changes over time.

Building on these three developments – the float technology, the increasing recording of surface data by satellites and the realization of a highly dynamic ocean – inspired plans for the Argo program (for a longer account of this history, see Lidström et al., 2022). Dean Roemmich, oceanographer at Scripps Institution of Oceanography in California and lead author of the initial Argo proposal, explains that “the global oceanographic community became aware of floats thanks to WOCE, so a lot of people thought of their potential use” (pers. comm.). In 1997, Roemmich authored a document titled The Argo Prospectus together with colleagues to be used to gauge interest in developing autonomous profiling floats for a large-scale program (this document has since been lost, information provided by Roemmich). The prospectus was presented at the US National Oceanographic and Atmospheric Administration (NOAA), where it was received favorably. The scientists involved in formulating the idea proceeded to contact their scientific counterparts in nations seen as crucial for making the program an international collaboration, while individuals from NOAA did the same on the bureaucratic side, so that “there was a somewhat coordinated effort on two levels”. The strategy was successful, and also beyond the US, according to Roemmich, “the idea was received very positively” (pers. comm.).

The Argo program was officially proposed in 1998 in a report titled On the design and implementation of Argo: a global array of profiling floats, co-authored by a group of eleven scientists that called themselves the Argo Science Team (AST, later the Argo Steering Team) representing the US, South Africa, Canada, France, Germany, Japan and Australia, and including one woman and ten men. The proposal lays out a detailed plan, including aims for the envisioned number and geographical coverage of floats, together with a tentative time plan and best practices for data management. The main intention was to provide technological infrastructure needed to improve studies of climate variability and prediction. The authors of the proposal summarized the program as “though ambitious, both doable and worth doing” (AST, 1998).

The ocean and climate change

The proposal for Argo was the result of a growing recognition of the importance of the ocean for climate studies. Considerations of significant interconnections between the ocean and climate became prominent with the advent of Earth System science in the 1980s. Until then, the ocean and climate systems had mostly been thought of as separate, though an understanding of interactions between them had emerged earlier in the twentieth century, notably in the fields of paleoceanography and paleoclimatology. A 1959 landmark volume edited by future IPCC founder Bert Bolin titled The Atmosphere and the Sea in Motion very much prefigured the work that would be done under Bolin's IPCC leadership from 1988 to 1997 (Bolin, 1959). One chapter on “Changes in the carbon dioxide content of the atmosphere and sea due to fossil fuel combustion” investigated the buffering effect of the ocean of atmospheric CO₂ from burning fossil fuels and started to demonstrate that not only had a significant increase occurred over the past century, but much less of the CO₂ was absorbed by the ocean than was previously thought. In this sense, an instrumentalization of the ocean in the service of climate and atmosphere was already present in these early scientific efforts.

Early General Circulation Models (GCM) disregarded the known interaction between climate and ocean for the sake of simplicity, but in the 1970s stronger computers and more data allowed for the development of coupled GCMs of ocean, atmosphere and ice that predicted the global and regional impacts of increasing carbon dioxide levels (Benson, 2020). The coupling still served primarily to understand the changing climate rather than the ocean itself. In addition, the role of all living beings on Earth known as the biosphere following the proposal of Russian geochemist Vladimir Vernadsky from the 1920s (Rispoli, 2022) was still left out of the equation, in spite of growing insight that this sphere had the capacity to regulate carbon through its ability to sequester it in biomass growth. This started to change in the 1980s. The specific notion of an interconnected Earth System came from a rather small group of scientists from the NASA advisory board, who formed the Earth System science committee (ESSC) in the mid-1980s to develop and promote the study of the whole planet as a single system (Barton, 2020; Edwards, 2010). Scientifically, their effort grew out of the insight that the spheres of the Earth were in constant interaction and that understanding global environmental change presupposed a holistic understanding accounting for biogeochemical processes. Arguably as important as the discovery that human activities could impact the climate was the insight that the atmosphere and the ocean had been in a constant process of interrelated change over billions of years (eg. Hines et al., 2024). In 1986, the ESSC published an influential report that contained an engineer-like diagram depicting the interactions between the spheres of the Earth that has become known by the name of the committee's chairman as the Bretherton diagram (NRC, 1986). The diagram, which includes human activities as a separate box to convey its potential forcing in the system, conveyed by means of arrows and boxes that ocean dynamics and atmospheric physics are interlinked, as is global moisture and marine biogeochemistry in a system of feedback loops inspired by cybernetics, notably in the work of James Lovelock and his Gaia Hypothesis. The Bretherton diagram provided a conceptual basis and visual tool for the new field of Earth System science developed by the ESSC, which would use satellites and computer modelling to study the planet as an integrated system with interconnections between the land, air, water, and biota. Around the same time, a report from an early meeting organized by NASA to establish what would become the International Geosphere-Biosphere Programme (IGBP) declared that “the Earth is changing even as we try to understand it in ways that involve the interplay of land and sea, of oceans, air and biosphere” (NRC, 1988: v). The report insisted that a unified perspective of the planet was necessary.

In this process, the IPCC emerged as the key actor for the stabilization of the epistemic strategy of predictive global climate modelling as the default mode of operations for environmental policy (Borie et al., 2021). As the IPCC institutionalized climate modelling, global mean surface temperature, which relied on a specific form of ocean sensing, appeared as the variable of interest to understand (and govern) anthropogenic climate change, as opposed to other indicators like precipitation, radiative forcing or ocean heat content. The IPCC took form at the same time that notions of sustainability were starting to be promoted as a response to the increasing evidence of anthropogenic impacts on the environment, including climate, initially with the Brundtland commission report in 1987, which famously stated that the world needed to change to a trajectory of sustainable development that would allow present generations to meet their needs, without compromising those of future generations (WCED, 1987). In sketching a roadmap to what would become a hugely influential concept for global environmental governance, the report stated that “the accumulation of knowledge and the development of technology can enhance the carrying capacity of the resource base” (Brundtland, 1987: 42). There were thus high hopes tied to technological development for environmental purposes to facilitate the type of economic extractivist structures that had built Western prosperity in the postwar period to continue while decreasing environmental harm. This included increasing recognition of a changing climate and its ties to the ocean, as a need to limit anthropogenic impacts was being formulated.

Approaching the climate-ocean relation from a different angle, Naomi Oreskes has shown how oceanography has been shaped by its relevance for military purposes, to the effect that after the end of the cold war, the field was pressed to find a new sense of purpose, as well as source of funding, a crucial aspect for a field of expensive research (Oreskes, 2021). By then, climate change was emerging as the most evident social concern to which oceanographers could tie their scientific questions. Already in 1965, opposing the close ties between ocean science and the US Navy, oceanographer William von Arx had stated: “If, because of operational expense, it is to be the eternal fate of oceanography that its pursuit must be tied to some compelling cause, it would seem that the problems of weather and climate would provide its most appropriate justification” (cited from Oreskes, 2021: 397). Two decades later, in the 1980s, oceanographers suggested that measuring the ocean's temperature could be a way to measure the temperature of the planet as a whole. The first attempt to do this was initiated by scientists at SIO, who proposed using acoustic tomography to measure the speed of sound in the ocean and from that derive its temperature (Oreskes, 2021: 395–468). However, after initial testing for what became known as the Acoustic Tomography of Ocean Climate (ATOC) project was conducted in 1991, concerns grew that continuous sound waves transmitted through more or less the entire ocean would pose risks to marine animals, especially whales and other cetaceans. As the project proposal came under increasing scrutiny, its theoretical foundations also came into doubt. From high ambitions of definitively detecting global warming in the ocean, ATOC scientists lowered their aspirations to claims that ocean acoustics together with satellite altimetry and other data could help monitor ocean conditions and test climate models. After a short operating period from 1995 to 1998, concerns about the impact of the sound signal on marine life remained, and the program was discontinued in 2000, after two decades of controversy. The Argo program was introduced just as ATOC came to an end, by scientists at the same institution (but not the same group of scientists) who instead of sound waves proposed the use of neutrally-buoyant floats to collect climate-relevant ocean data. In contrast to ATOC, the Argo program was quickly accepted and implemented, with initial float deployments taking place within a year of the proposal.

Co-environing the ocean and climate

While it was known before Argo that the vast volumes of water in the ocean play a central role in the planetary climate system, there was little data or knowledge on the extent or details of the ocean's regulatory function and interaction with the atmosphere. At the time Argo was proposed, it was widely recognized that extended systematic observations were essential to monitoring the climate and that these observations needed to include the global ocean: “developing a fundamental understanding of ocean processes” was recognized as “prerequisite to establishing a deeper and more predictive description of climate” (Siedler, 2013: 2). The development of neutrally buoyant autonomous floats in the 1990s and their use in WOCE made a global ocean observation system feasible (Gould et al., 2013). The data on ocean temperature, salinity and velocity eventually provided by Argo floats enable increased understanding of ocean circulation, which underpins heat distribution and carbon sequestration, and thus climate regulation (Roemmich et al., 2019). The aim and ability of Argo to contribute observational data to inform not just ocean but importantly climate science was foregrounded in the initial program proposal:

A plan for a new network of autonomous profiling floats is described with the potential to greatly enhance the present level of upper ocean temperature and salinity measurement. Such enhancements are urgently needed to sustain improved understanding of climate variability and ocean variability over a range of space and time scales and to underpin a range of operational oceanographic applications. (AST, 1998: ii)

A 2009 review repeated the claim that “Argo's most valuable contribution will be its observations of climate-related ocean variability on seasonal to decadal time scales and beyond” (Roemmich et al., 2009: 38–39). In 2019 the future aims of Argo were summarized as delivering “operational reanalysis and forecasting capability, and assessment of the state and variability of the climate system with respect to physical, biogeochemical, and ecosystems parameters” (Roemmich et al., 2019).

Despite occasional recognition of its importance, the ocean has been underrepresented in both climate science and climate governance. Allison and Bassett (2015) note that “The ‘out of sight, out of mind’ nature of ocean change is reflected in the ocean's lack of visibility in global climate change policy debates, including the annual Conference of the Parties to the UN Framework Convention on Climate Change” (778, see also Galland et al., 2012; Gallo et al., 2017; Levin and Le Bris, 2015). In the early GCMs of the 1970s and 1980s, treatment of the ocean was rudimentary, leading to fundamental critiques of the models (Heymann et al., 2017). As the models improved in the 1990s, they still lacked substantial observational data for the ocean, and its representation continued to be heavily simplified (Weart 2010). A shift is seen in the IPCC's Fifth Assessment Report (AR5, IPCC, 2013), which highlights a new availability of relevant ocean observations, showing a general lack of observations prior to the year 2000. Riser et al. (2016) point out that “It is apparent from this report that the recent expansion of the ocean observing system clearly sets AR5 apart from its predecessor 4, with most of this change ascribed to the existence of the Argo array” (146). In 2019, the IPCC published a Special Report on the Ocean and Cryosphere in a Changing Climate (SROCCC), which notes that while “In situ ocean subsurface temperature and salinity observations have increased in spatial and temporal coverage since the middle of the nineteenth century,” observations on a scale approaching the global had not been possible before Argo: “near global coverage (60°S-60°N) of the upper 2000 m has been achieved since 2007 due to the international Argo network” (IPCC, 2019: 99). The Argo program is similarly included as a milestone for climate observations in the most recent IPCC report, which states that “After 2006 direct [ocean heat content] estimates for the upper 2000 m layer benefit from the near-global ARGO array with its superior coverage over 60°S–60°N” (IPCC, 2021: 174, 349). Reflecting on these contributions and the scale of the data produced by the Argo program, Argo scientist Sarah Purkey notes that “the way we [the scientific community] see the ocean [today] is really through Argo” (pers. comm.).

The idea of seeing the ocean through a specific technology is not in itself new – as Melody Jue among others has pointed out, “the (deep) ocean emerges as an object of knowledge only through chains of mediation and remote sensing” (Jue, 2020: 3). This is true for past views of the ocean as much as for present ones; only the technologies and the visions they enable have changed, and increasing coverage and influence over the view of the ocean comes with new power relations. Here we understand these processes as acts of ‘environing,’ a concept proposed by environmental historians Paul Warde and Sverker Sörlin in 2009 as a way of understanding how what is commonly referred to as ‘the environment’ comes into being. Briefly defined, environing is the process by which ‘nature’ or ‘wilderness’. i.e., that which lies beyond the reach of human influence and control, becomes ‘environment’, understood as nature impacted, monitored, shaped, known, or otherwise part of human societies. The deep and global ocean has remained outside this realm of the human for longer than most other parts of the planet (Ramirez-Llodra et al., 2011). The Argo program, we argue, is playing a central role in incorporating more of the ocean into the measured, monitored and modelled ‘environment’, and into human environmental imaginations (Buell, 1995).

Environing is not a neutral or mechanistic process and the outcomes are multi-faceted, including physical impacts on the environment but also changes that result from knowledge and value formations, where certain perspectives take precedence over others. The scale and kind of information that Argo provides environ the ocean in particular ways, through the affordances and limitations of the physical floats as well as the strategies and practices, such as what parts of the ocean to prioritize, what sensors to include, and how to manage the data, agreed on by the Argo community. More specifically, we want to highlight here, the Argo program co-environs the ocean with the planetary climate and climate change, by bringing into view a specifically climate-relevant ocean. The link between Argo and climate is fundamental; as phrased by one group of Argo scientists, the focus of the program is on “climate-relevant variability on seasonal to decadal timescales, multi-decadal climate change, improved initialization of coupled ocean–atmosphere climate models and constraining ocean analysis and forecasting systems” (Riser et al., 2016: 145). This means, in the specific case of Argo, providing data that reflect geophysical dimensions (temperature and circulation) of the upper 2000 meters of the water column (the depth range of the original and still majority of Argo floats) and the kind of continuous and vast data flows that meet the needs of predictive climate change models hegemonic in climate sciences. This view may occlude other potential ways of relating to and understanding the ocean beyond the technocratic paradigm of which the Argo program forms part.

Making the ocean historical

Describing a process rather than a state of being, environing is an inherently historical and historicizing concept. It incorporates the ‘natural’ environment into the societal and understands socio-natural relationships as historically situated, evolving, mutual, and driven by historical agencies. Viewing Argo as an “environing technology” (Sörlin and Wormbs, 2018) and more broadly as “environing media” (Wickberg and Gärdebo, 2022), helps to show how the program is part of a “history of environing, as the study of how all human activity demarcates and generates an environment, an outside that haunts the space in which people choose to act” (Warde, 2009: 73). Identifying the co-environing of the ocean and climate that Argo performs contextualizes the program in relation to different environing processes that societies have undertaken that can be studied historically in an expanded notion of the environmental aligned with current Anthropocene conditions.

Environing media and environing technologies highlight the feedback loop between knowing and changing the environment historically, as different forms of datafication and digitalization come to afford shifting environmental epistemologies that in turn warrant alternating responses and material interventions into the observed object of the environment (Wickberg, 2023). In this way, the analytical juxtaposition of media, technology and environment becomes an impasse to study the varying expressions of an ever-changing human-Earth relationship beyond the long-standing separation of nature and culture, which much recent scholarship has tried to overcome. Together, these concepts propose a complementary route of understanding the ontological and epistemological construction of environments critically and historically by focusing on how the global environment itself, as an epistemic object, is fundamentally studied, conceived of, and mediated. Chains of technical, scientific and cultural mediation can hence be traced from the data gathering in distributed sensor networks, through the processing of this data into epistemic objects and concepts, to the policy, politics and public perception of a given phenomenon such as climate change.

In this theoretical context, we can see how the Argo program, as a specific technology, media and cultural practice, is environing the ocean and contributing to making it part of the environment defined as nature that is known, studied, measured, monitored, shaped and governed by societies. While the ocean has been studied for centuries, and marine scientific knowledge gradually assembled, Argo signifies a considerable shift in the extent to which the global ocean can be considered ‘environed’ through the interdependence between Argo data and climate models. As Paul Edwards has pointed out in A Vast Machine (2010), without models, there are no data. Edwards wrote this in order to clarify misconceptions about the nature of climate models, which were current among skeptics at the time, and explained that while models are simulations, the “real data” that skeptics referred to can only become legible through models. Turning this insight around, we can see how the vast data flows produced by the Argo program become legible and used through climate models, which affords a historically specific conception of the world ocean marked by the defining challenge of anthropogenic climate change. Understanding this process as environing leans on a Foucauldian understanding of historically situated discursive orders which regulate the knowable and sayable at a given moment in time, which reveals, denaturalizes and uncovers how we come to know and understand the ocean through the dynamic interplay between technologies and discourses through chains of mediation. This recursive feedback loop unfolds as more data is fed into models revealing an increasingly urgent state of the climate system from a human perspective.

By co-environing the ocean with climate change, Argo is eroding the ocean's traditional appearance as timeless and ahistorical (Lidström 2023) and instead enabling views of the ocean as a historical object with relevance on human timescales. Argo scientists Riser et al. (2016) describe how the ocean emerges as a historical object through Argo data:

Given that they provide a comprehensive baseline of today's ocean state, the Argo observations have been particularly useful in examining ocean changes on timescales of decades and longer. A stunning example has used contemporary Argo observations in conjunction with data from the HMS Challenger expedition, carried out in the second half of the nineteenth century. The study reveals a warming of the ocean over the past 135 years of nearly 0.6°C near the sea surface, tapering to near zero at depths close to 1000 m [...]. Over the upper 900 m of much of the ocean there is an average increase in temperature of over 0.3°C over the 135-year study interval. This work, further supported by analogous results in climate models, underscores the changing nature of ocean properties and the need to sustain global observing systems over long periods and, further, that recent changes in ocean temperature probably predate the sparse global-scale ocean data-sets of the past half century. (148)

As co-environed with climate change, and interrelated with Anthropocene ideas and implications, the ocean is moreover being incorporated into the “cultures of prediction” that dominate climate sciences. While predictability has gained prominence as a goal and ordering principle in a range of environmental sciences and management areas over the twentieth century, it is especially pronounced in climate research. As Martin-Nielsen (2017) and others have shown, this was not inevitable but the result of “fundamental decisions about which types of knowledge are important, which epistemic standards are used to judge that knowledge, and which applications of that knowledge are regarded as useful and socially relevant” (100) made in the 1960s and 1970s, which to a large extent phased out other ways of knowing and researching climate change. Heymann et al. (2017) show how these “cultures of prediction” developed in the second half of the twentieth century as a result of successive decisions, and that as a consequence, within climate science and governance “predictive claims and the understandings, values and norms they shape and carry have become self-evident and normalized ways of experiencing the world” (5, 103). Argo data align with and contribute to this development, as seen in the primary role assigned to prediction and forecasting in recent proposals for expanding the program under the new name OneArgo (combining the original “Core” Argo program with “Deep” and “Biogeochemical” Argo floats that can reach the full depth of the ocean and measure additional parameters):

OneArgo will enable biogeochemical and ecosystem forecasting and new long-term climate predictions for which the deep ocean is a key component. Driving forward a revolution in our understanding of marine ecosystems and the poorly-measured polar and deep oceans, OneArgo will be instrumental to assess sea level change, ocean carbon fluxes, acidification and deoxygenation. […] OneArgo is a strategic and cost-effective investment which will provide decision-makers, in both government and industry, with the critical knowledge needed to navigate the present and future environmental challenges, and safeguard both the ocean and human wellbeing for generations to come. (Thierry et al., 2025)

A predicted ocean

As we have outlined, at the time Argo was proposed, global climate modelling had been stabilized as a dominant epistemic strategy for researching climate change; the program was developed to fit into a well-established framework and purpose related to climate prediction, where critical boundary work had already been done in the decades leading up to the turn of the century to make predictive models the preferred and authoritative method and alternative approaches had been marginalized (Heymann et al., 2017). Several social science scholars (eg. Sheila Jasanoff, Mike Hulme, Silke Beck) and others have critiqued the epistemic effects of this framework for climate science and policy, in which numerical calculations of future changes are placed at the start of a causal chain in which complex processes and relationships are reduced to climate change mitigation through socioeconomically viable pathways in the integrated assessment models (IAMs) that inform the reports from the IPCC. This “climate reductionism” is visible in the prominence given to global mean surface temperature as the icon of global change and consolidated in documents like the 2015 Paris agreement to limit global warming to 1.5 or 2 degrees C above pre-industrial levels (Hulme, 2023). The influential IPCC mitigation scenarios rely on largely untested negative emission technologies and can thereby convey a view that the limitation of warming is possible, while ignoring the staggering human and environmental impacts the scaling up of these technologies will have: “Climate is represented as a complex, multifaceted system which nonetheless tends towards equilibrium through the mechanistic resolution of multiple processes, and an outcome that can be captured in a single measure – whether this be global temperature, net global carbon emissions, or global economic output” (Borie et al., 2021).

As the ocean is brought into this context, it is transformed from being understood as a static backdrop into a human environment, and thereby into a governable object (Lövbrand and Stripple, 2011). The shift from scientific observations to research explicitly intended for policy advise constitutes a particular form of environing, characterized by a “mission drift, from collecting to also evaluating and advising” (Sörlin and Wormbs, 2018: 113). The move between these realms can appear seamless, and they may overlap, but scientific aims can also be mistaken for policy goals, and vice versa, and difficulties associated with separating observations from normative interpretations can have important consequences for both science and policy (eg. Hulme, 2013; Howe, 2014). Aims of predictability lend themselves to this kind of drift, as they move easily between scientific research and environmental governance. Likewise, they move easily between climate and ocean policy, and as the ocean has figured increasingly in the former, aims of prediction have spilled into the latter. For example, one of seven desired outcomes identified for the ongoing UN Decade of Ocean Science for Sustainable Development (2021–2030) is “A predicted ocean”, incorporating aims of observing, understanding, and predicting:

The vast volume of the ocean is neither adequately mapped nor observed, nor is it fully understood. Exploration and understanding of key elements of the changing ocean, including its physical, chemical and biological components and interactions with the atmosphere and cryosphere, are essential, particularly under a changing climate. Such knowledge is required from the land-sea interface along the world's coasts to the open ocean and from the surface to the deep ocean seabed. It needs to include past, current and future ocean conditions. More relevant and integrated understanding and accurate prediction of ocean ecosystems and their responses and interactions will underpin the implementation of ocean management that is dynamic and adaptive to a changing environment and changing uses of the ocean (UNESCO-IOC, 2021: 18).

As part of the process of large-scale quantified knowledge production there has been a successive move away from direct and experiential observations of particular environmental phenomena to ever more mediated forms of knowledge that are increasingly automated, to the extent that we may speak of a mediated planet, understood as an interconnected, increasingly detailed and continuously updated data-driven view of the Earth and the biosphere (Wickberg et al., 2024). Jennifer Gabrys (2017) calls this process “becoming environmental of computation” and points to how the multifaceted processes of sensing environments are produced through exchanges of energy, materialities, relations and milieus. This is not just the process of the planet as an environmental object of observation or study, but rather of novel ways of perceiving, understanding and experiencing environments contingent on data and large-scale computation since the postwar era, which have intensified over the past decades and to which Argo adds a marine dimension. The effect, following Gabrys, is that the marine environment can be seen as programmable for particular functions and concretized across technologies, peoples, practices, and non-human entities as techno-geographies. Understanding this programmability of the ocean as part of a longer historical process, it appears as a logical consequence of the cybernetic imaginaries of command and control of the 1960s described above, where the Earth appears similar to a computer that can deliver desired outcomes once the system is mastered. The Argo program helps to inscribe the ocean within this rationale, supported by the IPCC and the UNFCCC.

Ultimately, anthropogenic climate change is, as its name suggests, an interference with the Earth System caused by humans, and the epistemology of systems thinking applied to the ocean may make it appear as though intervening with the ocean is easier than curbing greenhouse gas emissions. The idea of predictability, entrenched in climate science and policy, functions as an environing technology that effectuates “a process of translation – from observation, data and knowledge to environmental usability” (Sörlin and Wormbs, 2018: 115). Argo data informs improved understanding of ocean circulation, which in turn impacts sequestration and distribution of atmospheric carbon. Such data lay the grounds for an “unprecedented opportunity to map the detailed structure of the global ocean temperature and salinity fields, at both surface and subsurface levels, and both globally and locally,” allowing for the development of “climate indicators” that were “previously impossible” (Riser et al., 2016: 146). Such data flows and the detailed and predictive views they allow are a precondition not only for understanding air-sea interaction in relation to anthropogenic climate change, but also for recent proposals to increase the ocean's ability to sequester CO₂ through marine carbon dioxide removal (mCDR), including strategies such as iron fertilization, artificially-induced phytoplankton blooms, sinking of organic matter, and direct injection of liquid CO₂, controversial forms of geoengineering that aim to reverse or slow the effects of anthropogenic climate change. Claims of predictability are central in this discussion, as the deemed safety of deploying any such technologies would depend on the ability to predict the outcome and judge associated risks. In this way, the increasingly detailed view of the ocean offered by Argo is both informing the epistemology of climate change and potentially allowing for a new level and form of intentional anthropogenic impact in the form of ocean-based climate interventions.

Conclusion: Interpretations of ocean sustainability

The Argo program has successfully datafied parts of the world ocean through a continuous and vast flow of physical data. After two decades of operation, Argo scientists could state that the collective efforts of “a multi-national team of dedicated scientists, engineers, and data experts […] have allowed Argo to revolutionize the way large-scale oceanographic data are collected, disseminated, and analyzed” (Wong et al., 2020: 19). The importance of the program, as well as its limitations, are seen in the separation made by the IPCC of the parts of the ocean that are accounted for by Argo, and the parts that Argo floats leave mostly unsampled. The distinction is seen in the levels of certainty assigned to observations of different depths of the ocean; the IPCC's summary of scientific understanding of ocean warming states that an increase in ocean heat content is “very likely in the 700–2000 m layer, with high confidence since 2006”, while it is only “likely” that “the OHC below 2000 m has increased since 1992 for the ocean below 2000 meters” (IPCC, 2021: 350). This difference has profound implications as the level of certainty dictates what makes it into the influential Summary for Policy Makers, the most widely read part of the IPCC reports, which mostly leaves out the deep ocean (Levin, 2021, Pillar et al., 2024). While extensions of the Argo program will help to further datafy these remote and deeper layers, the ocean is vast, and some parts will remain resistant to datafication. This resistance and its implications are essential to consider. There is a risk that data-resistant dimensions of the ocean, such as deep-sea life, are marginalized in governance priorities if such data is scarce or lacking at the same time that certain amounts or kinds of data are seen as prerequisite for governance actions (Bridges and Howell, 2025).

Increasingly datafied and modelled insights about the state of the ocean also play a role in projecting the marine environment as a space for anthropogenic purposes, ranging from increasing carbon sequestration to a new resource frontier for the so called ‘blue economy’. Such promises and the narratives they apply place the ocean within a neoliberal discourse according to which sustainable development is synonymous with increased economic exploitation, and in which terms such as “blue humanities” align academic research with the rationale for ‘blue’ economies or ‘blue’ growth (Deloughrey, 2023).

Looking back at Argo's first decade of operation, Argo scientists concluded that “Argo has achieved more than anyone imagined it would ten years ago,” but added that “the hardest work lies ahead – sustaining the program, broadening its applications and user base, and ensuring that its global observations benefit people in all nations” (Roemmich et al., 2009: 43). The pursuit of these aims will continue to co-environ the ocean and climate in ways that include continued datafication paired with associated challenges of turning data into knowledge and making benefits and usage of data equitable. It will be important to balance data-based applications, such as predictive climate models, with research and governance directed at basic understanding of the ocean in its own right and not only as a component of the climate system. How the new data flows from the deep ocean are made legible through models and other tools to inform our understanding of the ocean is determined by a variety of scientific, technological, social and political factors in complex ways.

Highlights