The ocean absorbs ~25% of anthropogenic CO₂ emissions annually, mediated by physio-chemical and biological processes. While physical drivers of oceanic CO₂ uptake are relatively well characterized, biological contributions remain poorly constrained. Advancing our understanding of biological controls is essential for monitoring and predicting climate-driven changes in the ocean carbon cycle. Therefore, we first propose a new global ocean ecological biome delineation based on ocean colour remote sensing. Next, we examine CO₂ dynamics in a temperate shelf sea at high spatial and temporal resolution. Ecological biomes, i.e. regions of coherent biological and biogeochemical structure, have proven essential for carbon cycle studies. Yet, existing classifications rely heavily on physical variables (e.g. sea surface temperature, SST) with limited biological representation. We present a biologically-informed segmentation at 0.25° resolution based on 26 years of satellite ocean color data (ESA Ocean Colour Climate Change Initiative, OC-CCI), spanning the open and coastal ocean. Our biomes capture key surface ocean ecosystem features, including primary productivity, particulate organic carbon, and phytoplankton community structure. Their relevance for carbon cycle research is further demonstrated through biome-scale estimates of biological modulation of the seawater partial pressure of CO₂ (pCO₂). Secondly, we examine pCO₂ dynamics within selected biome regions in the North Sea over the past decade at unprecedented (1km daily) resolution using in-situ pCO2 observations (Surface Ocean CO2 Atlas, SOCAT) and satellite observations of i.a. ocean colour (OC-CCI) and SST. By applying regionally-optimized retrieval algorithms, we estimate key biogeochemical drivers of pCO₂, including chlorophyll-a, suspended particulate matter and particulate organic carbon. We identify distinct biogeochemical regions shaped by primary productivity, riverine inputs, and sediment dynamics, with varying impacts on pCO2 dynamics, from locally enhancing the CO2 uptake to degassing CO2. This study provides new insights into coastal carbon dynamics applicable to coastal regions globally.

Coastal regions are highly dynamic environments, where physical, chemical, and biological interactions regulate carbon exchange between the ocean and the atmosphere. Extreme warming events, such as marine heatwaves (MHWs), can strongly disrupt these fluxes, yet their fine-scale, short-term impacts on the full carbonate system remain poorly understood. The Belgian Part of the North Sea (BPNS), with its extensive in-situ observations of key carbonate system variables, provides an ideal setting to study these effects. In this study, we combine in-situ carbonate system observations from the Surface Ocean CO₂ Atlas (SOCAT) and the Integrated Carbon Observation System (ICOS) with machine learning to reconstruct daily, 1 km-resolution maps of sea surface partial pressure of CO₂ (pCO₂, 2000–2024) and dissolved inorganic carbon (DIC, 2017–2024). Using CO2SYS, we derive pH and total alkalinity, completing the full carbonate system and enabling high-resolution assessment of air–sea CO₂ fluxes (FCO₂) during MHWs. From 2000–2024, over 100 MHW events were detected in the BPNS using Hobday et al.’s (2016) criteria, with a 90% threshold, minimum five-day duration, up to two-day gaps, and the 1983–2012 SST climatology (daily ESA SST CCI and C3S data at 0.05°, interpolated to 1 km). On average, MHWs lasted two weeks, reached ~0.29 °C above the 90th percentile, and affected ~19% of the region. FCO₂ anomalies during MHWs, expressed as a percentage relative to the climatological FCO₂, represent deviations in CO₂ flux: positive anomalies indicate increased outgassing or reduced uptake, negative anomalies enhanced uptake or reduced outgassing. Preliminary analyses show average flux anomalies of 15 ± 13% (5%-trimmed mean), with short (

Marine heatwaves are periods of anomalous sea surface temperatures (SST) sustained for long periods. These heatwaves have wide ranging impacts on marine ecosystems and biodiversity but can also alter the marine carbonate system and air-sea CO2 exchange. The regional responses of the carbonate system and air-sea CO2 exchange are likely to be different and vary during and after the marine heatwave. Within this work, we used a satellite observation–based approach to detect heatwaves in the SST records, alongside a well-characterised observational carbonate system dataset (OceanSODA ETHZ), to examine changes in the marine carbonate system before, during, and after five documented marine heatwaves: (1) Southern Ocean (2016), (2) Northeast Pacific (“The Blob”; 2015), (3) Western Australia (2011), (4) South Pacific (2016) and (5) Equatorial Indian Ocean (2016). In all heatwaves, a significant reduction of dissolved inorganic carbon (DIC) was observed during the event with DIC anomalies increasing in magnitude beforehand and declining afterwards. The magnitude of these anomalies differed among the four heatwaves. Variations in DIC and SST appeared to drive anomalies in the fugacity of CO2 (fCO2 (sw)) and pH, due to their influence on carbonate chemistry. The impact of these heatwaves on the air-sea exchange of CO2 varied during the heatwaves lifetime and the region and was driven by a combination of the carbonate system state, thermodynamics, and meteorological condition. We identified that the strongest anomalies in the carbonate system, and air-sea CO2 exchange did not always coincide with the heatwave period but could occur prior to or after the heatwave. These results therefore support a need to consider the temporal sequence of any compounding events and their feedbacks.

We present new estimates of primary production and net community production for the European Arctic. Primary production estimates were computed using a regional algorithm that showed higher accuracy in the Greenland Sea compared to previous studies. This improvement was achieved by integrating multiple sources of local data collected during expeditions of the Institute of Oceanology of the Polish Academy of Sciences (2015–2022), as well as campaigns of the Norwegian Polar Institute. The algorithm accounts for the local vertical distribution of chlorophyll and local particulate absorption spectrum, which significantly enhanced algorithm performance. Using this approach, we generated a time series of phytoplankton seasonal cycles for 1998–2022, revealing a more prolonged bloom period than previously reported. Our calculations indicate that total phytoplankton production is 11–150% higher than earlier estimates, implying stronger CO₂ uptake in this sector of the Arctic Ocean. The higher values primarily result from including the subsurface chlorophyll maximum, which is underrepresented in satellite observations and often omitted in models. Moreover, the use of level two satellite products extended coverage into high-latitude regions, yielding estimates in areas that level three products previously reported as zero. In parallel we present the latest advances of this work by estimating net community production from BGC-Argo float observations (2012-2024), which provides an independent constraint on regional carbon fluxes.

The ocean’s biological gravitational pump (BGP) –a set of food-web processes that generate organic particles that gravitationally sink from the surface to the deep ocean– contributes to locking away atmospheric CO2. Despite its importance for the carbon cycle and climate, the BGP remains poorly constrained by observations owing to the ocean’s vastness, strong spatiotemporal variability, and the high cost of particle measurements. Moreover, current biogeochemical models used in climate simulations lack a process-based, mechanistic representation of the complex, food-web interactions driving the BGP, instead reducing them to a few globally uniform parameters. As a result, their capacity to capture environmental responses and realistically project future changes in the BGP is limited. We present a novel mechanistic model, the Stochastic Lagrangian Aggregate Model of Sinking particles, version 2 (SLAMS-2.0), which explicitly simulates and tracks the formation, interactions and transformations of large numbers of biologically-produced particles within the BGP. The model is forced by satellite and hydrographic climatologies of surface ocean carbon and depth-resolved biogeochemical variables, and validated against multi-tracer particle flux observations, particle number concentrations, and particle size distributions from six contrasting time-series sites. Unlike existing biogeochemical models, SLAMS-2.0 produces fundamental BGP characteristics –such as the transfer efficiency of particulate organic carbon flux– as emergent properties rather than fixed parameterisations. Here, we will outline the architecture of SLAMS-2.0, present preliminary results from a global-ocean simulation, and discuss its potential for improving understanding of the BGP in the today’s climate and its response to future change.

Shelf seas are important for carbon sequestration and carbon cycle, but shelf sea observations for carbon pools are often sparse, or highly uncertain. Alternative can be provided by reanalyses, but these are often expensive to run. We propose to use an ensemble of neural networks (i.e. deep ensemble) to learn from a coupled physics-biogeochemistry model the relationship between the directly observable variables and variety of carbon pools (detritus, DOC, zooplankton, heterotrophic bacteria and DIC). We demonstrate for North-West European Shelf (NWES) sea environment, that when the deep ensemble trained on a model free run simulation is applied to the NWES reanalysis, it is capable to reproduce the reanalysis outputs for carbon pools and additionally provide uncertainty information. We focus on explainability of the results and discuss potential use of the deep ensembles for future climate what-if scenarios. We suggest that model-informed machine learning presents a viable alternative to expensive reanalyses, or existing satellite algorithms, so it could complement observations wherever they are missing and/or highly uncertain.

Understanding ocean carbon cycling and its sensitivity to climate change requires integrating diverse observations to resolve the roles of multiple carbon pumps, including the biological and microbial carbon pumps (BCP and MCP, respectively). While BCP exports organic carbon from surface waters to the deep ocean via sinking particles (biological gravitational pump) and actively mediated transport driven by physical and biological processes (physical and migration pumps), the MCP transforms labile dissolved organic carbon into refractory forms, contributing to long-term carbon storage. Despite their shared roles in regulating carbon fluxes, the coupling between these two pumps remains poorly understood, particularly in most oligotrophic areas, the subtropical gyres. This study is conducted under the European Space Agency’s Ocean Carbon pillar, within the framework of the “Satellite-based observations of Carbon in the Ocean: Pools, fluxes and Exchanges” (SCOPE) project, which aims to improve observation-based estimates of carbon pools and fluxes and to support the development of satellite products for ocean carbon cycling. Focusing on the core North Atlantic Subtropical Gyre (NASTG), we investigate the coupling between BCP and MCP by integrating satellite Ocean Color observations, in situ BioGeoChemical-Argo (BGC-Argo) float profiles, and 4D observation-based reconstructions. To this aim, we compare BCP efficiency derived from satellite-based export production (EP) and primary production (PP)—SCOPE products—with instantaneous particulate organic carbon (POC) fluxes from BGC-Argo profiles. We also assess the contributions of different BCP pathways, —particularly the physical injection pump, to their coupling with the MCP. We find a significant correlation between the downward export of particles from the BCP in the productive layer and the intensity of the MCP, with a discernible half-month time lag between the two processes. This synergic approach helps to “connect the puzzle” of carbon export and transformation in one of the ocean’s most nutrient-poor regions, offering new insights into the biogeochemical functioning of the NASTG .

The biodiverse and rapidly changing Greater Cape Floristic Region (GCFR) of southern Africa is outlined by coastal bays that receive dissolved organic matter (DOM) from rivers draining complex catchments composed of natural, agricultural, and urban land classes. As part of NASA's BioSCape field campaign (October-November 2023), we characterized the optical properties of three GCFR coastal bays (St. Helena, Walker, and Algoa) and four inland systems (Rietvlei wetland, Zeekoevlei lake, Theewaterskloof dam, and Klein River Estuary) in relation to carbon cycling. Measurements of the optical properties of colored DOM (CDOM), including absorption at 300 nm (ag300), spectral slope (S275-295), and fluorescence, highlighted the bio-optical complexity associated with terrestrial influences from contrasting watersheds, urban disturbances, biological activity, and rapid transformations along this dynamic coastline. CDOM in the coastal bays was characterized by an order of magnitude lower ag300 (0.5 – 2.8 m-1) and considerably higher S275-295 (0.020 – 0.031 nm-1) compared to upstream waters (25 – 71 m-1 and 0.012 – 0.019 nm-1, respectively), suggesting intense biological production and/or photochemical degradation. Coastal DOM was mostly (>75%) composed of protein-like fluorescent products, indicative of bacterial utilization, while inland DOM was mostly composed of humic and highly aromatic materials. Satellite retrievals, using OLCI and MSI imagery, captured the disproportionate yet ephemeral impact of extreme events – intense riverine discharge following extensive periods of drought – on coastal CDOM plumes, and revealed that seasonal hydrological cycles, catchment biodiversity, and human activity are the primary drivers of biogeochemical variability along this globally significant, coastal biodiversity hotspot.

The 2016 El Niño is the strongest warm ENSO phase of the 21st century recorded to date, leading to the development of severe temperature anomalies - marine heatwaves (MHWs) - across the equatorial Pacific Ocean. Here we analyze biogeochemical model output and machine learning (ML)-based reconstructions of Argo oxygen and backscattering measurements to demonstrate the impact of the 2016 El Niño in weakening oceanic carbon export - the transfer of biogenic carbon from the surface ocean to depth. We show that equatorial Pacific anomalies in modeled carbon flux, and ML-based optical particle backscatter and ecosystem respiration, display interannual oscillations linked to ENSO cycles, with an acute reduction in export (- 50 %), respiration (- 30 %) and backscatter (- 20% ), during the 2016 El Niño. This plunge in export production is attributed to a large decrease in chlorophyll biomass observed from space, and modeled ecological changes in phytoplankton community composition.

The Southern Ocean plays a vital role in global CO2 uptake, but the magnitude and even the sign of the flux remain uncertain. Physical mechanisms for carbon uptake are emphasized, with the role of biology in surface ocean carbon uptake potentially underestimated in coastal high latitude systems, and the influence of phytoplankton phenology is underexplored. This study focuses on the West Antarctic Peninsula, a case study region experiencing rapid climate change, to examine shifts in seasonal surface ocean carbon dioxide uptake. We used 20 years of in situ air‐sea CO2 flux data from research vessels and satellite‐derived Chlorophyll‐a data from OC-CCI as a proxy for phytoplankton biomass. OC-CCI was used for the Chlorophyll-a record because the OC-CCI atmospheric correction approach (POLYMER) results in more representative spatial distributions of coastal phytoplankton biomass than other approaches due to adjacency effects along bright icy coastlines in polar regions. We observed that the seasonal cycles of both air‐sea CO2 flux and Chlorophyll‐a concentration intensified poleward. The amplitude of the seasonal cycle of the non‐thermal component of surface ocean pCO2 increased with increasing latitude, while the amplitude of the thermal component remained relatively stable. These results suggest that pronounced biological uptake occurs over the shelf in austral summer despite reduced CO2 solubility in warmer waters, which typically limits carbon uptake through physical processes. Chlorophyll-a concentrations and air-sea CO2 fluxes were tightly coupled across all years, especially when phytoplankton biomass was high. These results suggest that an ocean-color-based air-sea CO2 flux algorithm may be developed to estimate surface ocean carbon uptake from space.

A major challenge in understanding the oceanic carbon cycle is estimating the sinking flux of organic carbon exiting the sunlit surface ocean, i.e., carbon export. Existing algorithms derive carbon export from satellite ocean color, but often display poor accuracy. One reason is that they neglect offsets created temporally by the lag between production and export, which combined with horizontal advection can result in a spatial offset of hundreds of kilometers in dynamic regions such as Eastern Boundary Upwelling Systems. Here we use a Lagrangian “growth-advection” (GA) satellite-derived product, where plankton succession and export are mapped onto surface oceanic circulation following coastal upwelling, thus explicitly representing these offsets. We show that the GA product succeeds in representing export measured off the California coast, despite relying exclusively on satellite winds and currents (no ocean color). In situ export is best represented by a combination of GA export, proportional to modeled zooplankton, and export derived from ocean color, related to local phytoplankton. Both products also correlate with a long-term time series of abyssal carbon flux. However, their spatial and temporal patterns are very different, underscoring the need to better constrain, and take into account, zooplankton contribution to export and its offset from primary production. These results provide insights on export spatiotemporal patterns and a path toward improving satellite-derived carbon export in the California Current and beyond.

Phytoplankton play a central role in the Earth System. Through the production of organic carbon, phytoplankton drive major processes in the ocean carbon cycle and form the basis of almost all life in the ocean. Phytoplankton have high biodiversity and this is complemented by their functional diversity, which recognises that different types of phytoplankton play varied roles in marine ecosystems and in biogeochemical cycles of the ocean. It is therefore fitting that the Global Climate Observing System (GCOS) has included phytoplankton in the ocean biosphere Essential Climate Variable (ECV), together with zooplankton. With many of the satellite-retrieval algorithms related to phytoplankton maturing over time, we are now in a position to produce such phytoplankton products to the ensemble of the European Space Agency (ESA) Climate Change Initiative (CCI). Here we present an overview of the ‘PHYTOplankton biomass and diversity Climate Change Initiative’ (PHYTO-CCI) project that aims to develop satellite-based data products for two ECVs identified by the GCOS: phytoplankton carbon biomass and pigment diversity. In the PHYTO-CCI project, satellite retrieval algorithms will be compared and combined using optical water classification to produce ECV products with associated uncertainty estimates. These products will be validated using both in situ and model data, followed by a comprehensive scientific assessment. The value of the new ocean biosphere ECVs will be demonstrated through their application in climate research and their relevance for supporting marine ecosystem services. The phytoplankton biomass and diversity ECVs are critical for understanding the structure and function of marine ecosystems, their role in the Earth System, and how they may be affected by global warming and other human-driven impacts.

A thorough understanding of the global carbon pools and cycle is essential to understand and predict the effects of climate change. Coastal waters play a key role in the global carbon cycle but remain poorly understood due to their optical complexity, high spatial variability, and sensitivity to climate change. Satellite remote sensing data can provide high spatial and temporal resolution for carbon monitoring at local, regional, and global scales. However, existing sensors are not optimised for dynamic coastal zones. Sentinel-2 MSI (S2) offers high spatial resolution, while Sentinel-3 OLCI (S3) provides better spectral band configuration, temporal resolution, and radiometric sensitivity, though its spatial resolution may be insufficient for highly heterogeneous coastal waters. In the ESA CoastalCarbonMapper project, we tested the applicability of both S3 and S2 for mapping carbon fractions—Total Organic Carbon (TOC), Dissolved Organic Carbon (DOC), Particulate Organic Carbon (POC), and Dissolved Inorganic Carbon (DIC)—in coastal waters. We aimed to develop and validate algorithms using in situ data and S2 and S3 imagery, addressing: (1) What are the optical proxies for different carbon fractions in coastal waters? (2) Which algorithms are most suitable for coastal carbon mapping? Bio-optical and physical water parameters were measured directly in the field, and water samples were collected to analyse carbon fractions and optically active water constituents in the laboratory. Measurements at the test sites were taken four times during the ice-free season of 2023–2024. Based on the collected data, potential optical proxies were identified, and retrieval algorithms for carbon fractions were developed and validated. The study represents a step forward in the remote sensing of coastal waters and Earth observation science. If adopted, the proposed carbon fraction products could allow for significant progress in different fields, from research to monitoring and policy making.

The Arctic Ocean and polar seas are undergoing rapid environmental change driven by global warming. Harsh climatic conditions and the logistical challenges of working in these regions have long limited direct in situ observations, leaving key biological processes poorly understood. Yet, the dynamics of carbon export are central to quantifying the ocean’s role in regulating climate. Here, we study these biological mechanisms using high-temporal resolution in situ measurements from 178 biogeochemical Argo (BGC-Argo) floats, one IAOOS (Ice Atmosphere Arctic Ocean Observing System) and 9 Ice-Tethered Profilers (ITPs) deployed across ice-free and sea ice–covered in the Arctic ocean and polar seas. We used data from autonomous platforms deployed in the central Arctic Ocean, Canadian polar and subpolar waters, the Greenland Sea, and the Norwegian Sea to better understand the phenology and mechanisms of phytoplankton blooms, the biological gravitational pump, the mixed-layer pump, and the eddy subduction pump and annual net community production. Results revealed a clear latitudinal gradient in the efficiency of the different carbon pumps, as well as distinct differences in bloom timing and magnitude between ice-covered and ice-free regions. In this research, we will test the hypothesis that there is a strong relationship emerges between bloom magnitude and carbon export to the deep ocean. Future work will require higher-spatial-resolution approaches to link phytoplankton functional types with their specific contributions to the biological carbon pump, ultimately improving predictions of carbon export in a rapidly changing Arctic.

The Northern Humboldt Current System (NHCS) is one of the world’s most productive upwelling ecosystems, yet knowledge of its phytoplankton composition and variability remains limited. To address this gap, we applied the PHYSAT methodology, originally developed by Alvain et al. (2008) to classify phytoplankton groups from satellite ocean-color data, marking its first use in an Eastern Boundary Upwelling Ecosystem. This work was supported by the long-term in situ phytoplankton monitoring program of the Peruvian Sea Institute (IMARPE). Analysis of SeaWiFS and MODIS data (2003–2010) identified five major phytoplankton groups: diatoms, nano-eukaryotes, Synechococcus spp., Prochlorococcus spp., and coccolithophorids. Methodological adaptations included additional quality-control filtering and the adjustment of reflectance anomaly ranges to capture monthly variability, with a focus on diatoms. Results revealed strong seasonal and spatial patterns. Diatoms dominated coastal waters year-round, peaking in austral summer and declining in winter, while nano-eukaryotes showed the opposite pattern. Offshore oligotrophic regions were characterized mainly by Synechococcus spp., and to a lesser extent Prochlorococcus spp. These trends were consistent across both 9 km and 1° spatial resolutions. This study provides the first regional baseline of phytoplankton distribution in the NHCS, offering new perspectives on mesoscale variability and ecosystem responses to El Niño–Southern Oscillation (ENSO) events.

The North Sea and Baltic Sea are two highly productive, interconnected marginal seas in northern Europe that play a vital role in regional carbon cycling. Both are strongly influenced by inputs of terrestrial carbon but differ fundamentally in character. The Baltic Sea is a wind-driven, brackish water system that is almost completely enclosed by land and has residence times on the order of decades. In contrast, the North Sea is a tidally-driven, marine system, on the edge of the North Atlantic with residence times on the order of months. Episodic deep inflows of salty, oxygenated North Sea water penetrate the deep basins of the Baltic Sea, providing temporary oxygen supply to otherwise persistent hypoxic zones, while brackish, surface Baltic Sea water drains into parts of the North Sea carrying with it a net export of carbon. Both systems have densely populated and intensively used coastlines, exposed to climate change and ever-increasing anthropogenic pressures. To understand the net carbon uptake behaviour of the coupled system, and how this might change in response to perturbations in atmospheric and river forcing, we use a coupled hydrodynamic–biogeochemical model, together with an extensive observational dataset, to investigate dissolved inorganic (DIC) and organic carbon (DOC) pools over the past 25 years. We quantify large-scale carbon budgets, assess the turnover times of pelagic and benthic pools, and explore the drivers that shape long-term changes in carbon inventories. Results show that the North Sea is strongly influenced by Atlantic exchange and functions as a short-memory system, while the Baltic Sea retains perturbations for decades due to restricted circulation. Atmospheric CO₂ and alkalinity inputs emerge as dominant drivers of change, while eutrophication control and warming modulate seasonal and interannual variability. These insights are critical for understanding regional carbon sequestration potential and the response of coastal seas to climate change.

Understanding the carbon cycle is fundamental to climate change research, as marine ecosystems play a crucial role in regulating carbon storage. Particulate Organic Carbon (POC), the organic carbon in sinking plankton and detritus, is a key component of this cycle, transferring carbon from the ocean surface to the deep sea through biological processes. While Earth System Models (ESMs) are essential for predicting carbon cycle changes, recent studies highlight persisting uncertainties in marine carbon export, which pose major challenges for climate projections. A major source of uncertainty in ESMs lies in their inconsistent ability to represent phytoplankton size classes (PSCs) and their distinct biogeochemical impacts. Diatoms, the dominant contributor to the large phytoplankton (or microhytoplankton) biomass, are projected to decline as the ocean warms and stratifies. Consequently, regional and global declines in net primary production and POC export are closely tied to diatom occurrence in current ESMs, reflecting the prevailing paradigm. Here, we address PSC-related biases in the PISCESv2.0–NEMO4.0.4 model by implementing a restoring technique that constrains surface phytoplankton biomass using ESA satellite observations. The method distinguishes two functional groups—small phytoplankton (pico- and nanophytoplankton) and large phytoplankton (diatoms)—and enables us to quantify the impact of PSC biases on the global biological carbon pump during the satellite era. By identifying PSC bias patterns, their potential underlying mechanisms, and biogeochemical impacts, our approach provides new insights into the biological pump’s response to climate change.

Uncertainties in the ocean carbon budget, whether estimated from observations or models, reflect an incomplete understanding of carbon cycle processes. In the frame of the ESA-funded SCOPE project, we address these uncertainties by assimilating satellite-derived phytoplankton carbon (PhyC) into the ocean biogeochemical component of an Earth System Model. Two kinds of global simulations are performed: a control run with free-evolving biogeochemistry and a second bunch of experiments that include assimilation of PhyC observations. Both simulations apply physical data constraints to ensure consistent ocean circulation. The assimilation of PhyC improves the representation of surface biological activity and carbon fluxes, especially in regions with limited in situ observations. Validation against independent datasets shows reduced uncertainties in key biogeochemical variables. These results highlight the value of satellite-derived ocean biogeochemical observations for improving model representation of marine carbon processes and reducing uncertainty in ocean carbon budgets.

Satellite ocean color data provide a comprehensive, daily to weekly scale view of phytoplankton biomass dynamics in the global ocean, which is essential for resolving the seasonal cycle of the biological carbon pump. However, capturing the underlying dynamics requires knowledge of subsurface chlorophyll-a profiles, often demanding complex model constraints. Here we present a novel approach to project surface chlorophyll-a data to subsurface profiles using a machine learning framework trained with Biogeochemical (BGC) Argo float observations. The method assumes a Markov process in the seasonal sequence of chlorophyll-a profiles measured by BGC Argo floats and predicts subsurface variability through a Hidden Markov Model (HMM). We tested the system in the Norwegian Sea, where clusters of BGC Argo profiles have been available since 2013. The HMM’s observable vector includes satellite chlorophyll-a, sea surface temperature, ERA5 downward shortwave radiation, and mixed-layer depth from core Argo floats. The Root Mean Square Error (RMSE) of the reconstructed chlorophyll-a profiles varies with depth and season, with the highest errors (50–100 m) at the base of the mixed layer during the spring bloom. Recently, the system was extended to retrieve particulate organic carbon (POC) and dissolved inorganic carbon (DIC) profiles, providing a more consistent representation of the seasonal cycle of the biological carbon pump in the Norwegian Sea.

The temperature dependence of phytoplankton photosynthesis and growth has been widely incorporated into satellite algorithms of primary production and global biogeochemical models. The culture study of Eppley (1972) showed that marine microalgae achieved their maximum growth rates at temperatures close to those at which the cells were initially collected, underscoring the importance of temperature as a factor governing the ecophysiology of phytoplankton cells. The parameters of the photosynthesis-irradiance (P-E) response curves account for variability in the two primary determinants governing carbon fixation in the natural environment: the amount of biomass present and light availability. Thus, the P-E parameters serve as valuable indicators of photosynthetic efficiency. Temperature is believed to impact the photosynthetic characteristics of phytoplankton cells through its effect on enzyme kinetics. Temperature also has a role in setting the density structure of the surface lit layer, thereby influencing the secondary determinants of algal growth, such as the supply of nutrients and light history. These factors, in turn, govern the seasonal succession of phytoplankton taxa. Using a multi-year dataset of P-E response curves, we investigate how photosynthetic performance of natural phytoplankton assemblages varies across a wide range of temperatures in the western North Atlantic. We will also explore how information on phytoplankton community structure may serve as a useful indicator of photosynthetic efficiency, based on the premise that the forcing variables governing phytoplankton diversity also regulate the secondary determinants known to limit rates of carbon fixation.

The gradual change in meteorological regimes associated with Climate Change is rapidly modifying land ecosystems. In several parts of the world, vegetated landscapes are becoming drier and hotter, hence more prone to burn. These changing trends in global fire activity influence the global carbon budget, sometimes in non-obvious ways. Large wildfires have the potential to perturb ocean biogeochemical balance through aerosols deposition. The deposition of wildfire' aerosols has been associated with the exceptional occurrence of phytoplankton fertilisation events, harmful algal blooms, changes in phytoplankton community composition and phenology, and anomalous bacterial activity. Understanding these relationships is critical to better constrain the net effect of wildfires in the global carbon budget. However, the casual links between wildfire activity and marine biogeochemical responses are complex and they require interdisciplinary efforts as they depend on the chemical composition at source, the aerosol's lifetime and the biogeochemical state of the receptor waters. The ESA-funded project PYROMAR puts the focus on this increasingly relevant component of the global carbon cycle. The project brings together experts on ocean colour and aerosols satellite data, fire dynamics, sea-ice, atmospheric chemistry and ocean biogeochemistry with three objectives: (1) build an inventory of ocean biogeochemical responses to wildfire aerosols, (2) understand the limiting factors controlling the response, and (3) identify long-term trends in the coupling between fire-prone and marine ecosystems. PYROMAR will target these objectives in three regional sites: the Arctic Ocean, the Californian upwelling and the South Atlantic. This poster will present the different tasks and tools deployed in PYROMAR's strategy as well as its complementarity with ESA-EO clusters and on-going projects.

A fundamental building block in marine primary production models is the set of model parameters essential to compute underwater light penetration and photosynthesis. A variety of models are available for satellite-based computations of primary production, which may be classified as available-light models, absorbed-light models, and growth-rate models. Regardless of the choice of the model, a set of four parameters would enable the implementation of any of them, and allow interchange of modes in a consistent manner (Sathyendranath et al. 2009). They are the initial slope of the photosynthesis-irradiance curve, the light-saturation parameter, the specific absorption of phytoplankton, and the carbon-to-chlorophyll ratio. In models where primary production is computed for multiple functional types, these parameters have to be assigned for each of them. Often, the parameters are assigned based on culture experiments. Though desirable, if only for comparison with laboratory experiments, it is not easy to obtain field data on these parameters, since, typically, phytoplankton rarely occur as single-species populations in the field. A notable exception is the work of Uitz et al. (2008, 2010), in which they estimated photosynthetic parameters from field data for three size classes, and then computed size-class-specific primary production using satellite data. In this work, we have used a large dataset of photosynthesis-irradiance parameters from various parts of the global ocean, combined with HPLC data on pigment composition, to identify samples dominated by a single phytoplankton type. The data, segregated by dominant phytoplankton type, are then analysed to estimate mean photosynthesis parameters and other bio-optical properties, including spectral specific absorption coefficient for each phytoplankton type. The results will be presented and compared with other published results.

In support of efforts to protect eelgrass beds, halt biodiversity loss, and promote recovery, the SwedCoast-BlueCarb project applies satellite Earth Observation (EO) to Swedish coastal waters. Funded by the Swedish and UK Space Agencies, the project combines EO and in situ data to assess the impacts of climate-change mitigation in contrasting test areas: the CDOM-dominated Baltic Sea around Kalmar, and the Swedish west coast that's strongly influenced by Baltic outflow. Initial activities established collaborations with academic partners and monitoring programmes. The EO processing has focused on generating consistent datasets with atmospheric correction methods tested, and a modelling approach developed to retrieve both water optical properties and submerged vegetation from surface reflectance. Copernicus Sentinel-2 imagery (20 m) is used to map eelgrass (Zostera marina) and bladderwrack (Fucus vesiculosus) extent, together with uncertainty estimates, where vegetation occurs within detectable depths. In addition, laboratory analyses support the optical modelling by characterising the absorption and reflectance spectra of submerged vegetation. Now that the initial modelling approach is working, the ongoing work utilises the benefits of a machine learning model to accelerate the modelling process, allowing for faster processing and, consequently, systematic monitoring across large spatial and temporal scales. Sentinel-3 data (300 m) provide complementary information on water optical status, a key factor in determining light availability and ecosystem health. In parallel, commercial WorldView-2 imagery (2 m) is being evaluated to demonstrate the potential of very high-resolution mapping. Ultimately, an automated processing chain will deliver EO-based products openly through a GIS-style portal. These products will enable local authorities and conservation groups to track the condition of eelgrass, identify restoration priorities, and assess the effectiveness of management measures. By quantifying vegetation extent and water optical properties, the project supports blue-carbon conservation goals and strengthens the evidence base for climate-change mitigation in coastal ecosystems.

Seagrass ecosystems store a disproportionately large amount of the ocean’s total carbon, but the synergistic impact of climate and anthropogenic interactions has led to severe habitat loss and carbon storage capability of the ecosystem. High dynamicity of the seagrass ecosystem and associated blue carbon stock necessitates timely monitoring of blue carbon stock to determine variability in the source dynamics and budget of these vulnerable ecosystems. Conventional field measurements of seagrass biomass and sediment organic carbon are laborious and economically non-viable, necessitating the need to develop regional algorithms for monitoring seagrass biomass and organic carbon using satellite data. This study assessed the spatio-temporal variability of seagrass biomass and sediment organic carbon stock along the Palk Bay, South-east coast of India, through a combination of in-situ surveys and satellite-based modelling for the period January 2022 -December 2023. Two permanent monitoring sites—Chinnapalam and Mandapam were selected following pilot surveys, with additional sampling conducted from Thondi to Thangachimadam to validate satellite-derived estimates. Field data on seagrass biomass, sediment, and water quality were collected and analysed, while Sentinel-2 imagery (2015–2023) was processed to map annual and seasonal variability in seagrass cover. NDVI- based seagrass above ground biomass was also obtained from Sentinel data, and validated with field observations. Results revealed significant seasonal variability in total seagrass biomass, with higher values during the wet season (778.29 ± 227.06 g dwt m⁻²) compared to the dry season, whereas the sediment organic carbon showed higher concentrations in the dry season (1.03 ± 0.23%) compared to the wet season (0.72 ± 0.06%). Among species, Cymodocea serrulata showed the highest biomass, while Halophila ovalis and Halodule pinifolia had comparatively lower values. Above-ground (AGB) and below-ground biomass (BGB) also exhibited significant species-level and seasonal variability, contributing to seasonal differences in total organic carbon stock. Empirical models linking NDVI with in-situ AGB (R² = 0.80) enabled satellite-based estimation of biomass and carbon stock, based on AGB-carbon stock relationship for the Palk Bay. Validation with independent field data showed strong agreement with seagrass AGB (R² = 0.67) and carbon stock (R2 = 0.72). This integrated field–satellite approach provides a robust framework for mapping blue carbon resources at regional scales, reducing economic and human efforts. The outputs were then incorporated in a WebGIS application, that offers valuable decision-support tools for identifying priority areas for seagrass management and restoration to enhance climate mitigation potential.

The ongoing rise in ocean acidification (OA) presents a significant threat to coral reefecosystems, particularly affecting the vulnerability of coral species to environmental stressors.This research focuses on quantifying the impact of OA on coral reefs surrounding Saint MartinIsland, Bangladesh—an area recognized for its coral biodiversity. Through systematic ecologicalsurveys, water quality analyses, and predictive modeling using machine learning (ML), we aimto illustrate the changing conditions of coral growth rates and resilience linked to OA over time.Coral growth is negatively impacted by OA due to reduced skeletal density, making theseecosystems more susceptible to bioerosion and storm damage, thereby impairing their structuralintegrity and resilience.Historical data shows that live coral cover surrounding Saint Martin Island has decreaseddramatically, from 20–25% in the 1980s to below 10% today. This decline has been exacerbatedby both global climate change and localized stressors, such as coastal pollution and overfishing.Our analysis will specifically explore how these stressors interact with OA to further decreasecoral resilience, supporting findings that highlight the severe implications of combined threatsfaced by coral ecosystems. Implementation of modern ML techniques allows us to predict futurerisks and develop proactive management strategies for coral conservation and restoration,focusing on promoting resilient coral species and effectively managing local impacts.Furthermore, the findings of this research will be pivotal in guiding stakeholders towardsadaptive solutions, such as the establishment of locally managed marine protected areas and therestoration of protective habitats like mangroves and seagrass beds, which can help mitigate theeffects of acidification. This comprehensive understanding is crucial for sustaining not only thecoral reefs themselves but also the myriad of ecosystem services they provide, vital for localcommunities reliant on fisheries and tourism. This research aims to fortify the resilience of SaintMartin Island’s coral reefs against the backdrop of ongoing climate change, ensuring thelongevity of their ecological and economic contributions.

The coastal ocean is a region of socio-economic and ecological importance. Yet, this system is under immense pressure from global climate change and other anthropogenic hazards, which threaten ecosystem services and increase the vulnerability of the growing coastal population and infrastructure. Whilst the coastal ocean is under pressure, it can also be part of the solution to manage and adapt to changes, with the phytoplankton ecosystem as an example of this. Primary production by phytoplankton plays an important role in the global carbon cycle through the conversion of inorganic carbon in the water to organic carbon via photosynthesis. This process is not only important for global climate regulation, but also essential for supporting all coastal ecosystems and the services they provide. In this study, we explored changes in phytoplankton primary production in the global coastal ocean from 1998-2022 within Longhurst's ecological provinces. We address three key questions: (1) In which coastal provinces does primary production undergo significant changes? (2) What are the underlying causes of these changes? And (3) Is the aggregation of data into large areas (such as the ecological provinces) suitable for investigating the underlying causes of any observed change in the global coastal ocean? To address these questions, we have undertaken trend analysis of primary production model outputs based on the Ocean-Color Climate Change Initiative (OC-CCI) data, and applied a linear regression analysis using a seasonal decomposition of the time series and autoregressive integrated moving average (ARIMA) method. We further explored the impact of sea surface temperature (from SST CCI), phytoplankton chlorophyll-a concentration from OC-CCI (while recognising that chlorophyll-a concentration was an input to the primary production calculations, and that, therefore the two are not independent) and upwelling (Bakun Index) on primary production. We also studied the impacts of missing data (with a systematic bias) and of data aggregation methods at individual pixel level on apparent or artificial trends. Results showed that there were five (out of 23) coastal provinces with statistically significant increasing or decreasing linear trends in primary productivity, and that all regions with significant trends are also upwelling regions. Through this work we contribute to developing an understanding of the changes experienced in the global coastal ocean, which is essential knowledge for management of coastal challenges and pressures globally. This is a contribution to the UKRI NERC funded FOCUS Project (NE/X006271/1), the ESA SCOPE Project and the Simons Foundation CBIOMES Project.

Monitoring marine carbon pools and fluxes is critical for understanding the ocean's role in the global carbon cycle. Satellite ocean color data provide a unique tool for assessing key biogeochemical parameters, but their accuracy relies on robust regional algorithms. Since 2002, the Ocean Optics Laboratory at the Shirshov Institute of Oceanology has developed an electronic Atlas for the Russian seas based on satellite data and validated regional algorithms (http://optics.ocean.ru). The Atlas offers enhanced accuracy for bio-optical characteristics, which are fundamental proxies for quantifying carbon-related processes, primarily chlorophyll-a and coccolithophore concentrations. This study leverages over than two decades (1998–2024) of satellite observations processed with these regional algorithms to analyze interannual variability and trends in key parameters linked to the ocean carbon cycle in the Barents, Kara, Laptev, White, Baltic, Black, and Caspian seas. We present analyzed trends in these parameters, which provide valuable insights for future carbon cycle studies. Notably, a significant positive trend in coccolithophore concentration in the northeastern Black Sea confirms the intensification of blooms, which has implications for calcium carbonate production and export fluxes. Similarly, a positive trend in chlorophyll-a concentration in the northern Barents Sea, likely associated with regional sea ice decline, points to increased primary production and potential carbon drawdown. Such long-term observations are crucial for validating and improving climate models that represent high-latitude and marginal seas.

Marine net primary production is a cornerstone of the global carbon cycle and a foundation of marine ecosystems. As one of the largest carbon fluxes on the planet, it sustains marine biodiversity, underpins global ocean ecosystems and supports critical ecosystem services. Climate change is disrupting marine primary production in complex and poorly understood ways, with major implications for the carbon cycle, food security and climate feedbacks. Yet there remains little consensus on either the direction or magnitude of projected change. To address this uncertainty, we analyse remote sensing net primary production trends using six different algorithms, and benchmark them against fifteen divergent model projections. Our results suggest that future declines in production are more likely than current models predict, and that even the best-performing models still underestimate the magnitude of ongoing declines. However, large uncertainties remain, as trend estimates and model rankings depend strongly on the choice of remote sensing algorithm. To address this knowledge gap, we applied a subset of these algorithms to biogeochemical-Argo measurements. Although this offers less spatial and temporal coverage than satellites, it provides critical information at depth. Most notably, the disagreement in trend direction across ocean biomes disappears in this framework, and estimated changes are often much larger than those inferred from remote sensing alone. Together, these findings highlight both the promise and the limitations of current approaches to quantifying ocean productivity and its role in the carbon cycle. They underscore the urgent need for an integrated strategy that brings together satellite observations, autonomous platforms and biogeochemical models. Such integration is essential not only to constrain projections, but also to generate new mechanistic understanding of the drivers of ocean productivity change, knowledge critical for improving models and informing climate policy.

Responses of the Earth system to climate change may not be gradual, but abrupt in the form of tipping points. A new ESA project ‘TIME’ on Tipping points and abrupt changes In Marine Ecosystems, is designed to investigate eight vulnerable elements of the marine ecosystems for evidence of loss of resilience or for signs of abrupt changes, based on satellite data. For one of these elements: coccolithophores, a type of phytoplankton covered with highly reflective calcium plates, is unique in its ability to be studied globally even using coarse visible satellite data. Previous studies have shown that blooms of coccolithophores are starting to appear in new areas such as sub-polar waters, and are becoming more prominent at mid-latitudes. In TIME we have extended our dataset to generate a consistent 45-year time series of AVHRR data for analysis of coccolithophores, which will be used to investigate the drivers of change using AI tools. In this presentation, we will present preliminary results of this unique dataset. pim@pml.ac.uk

The strong control that carbon dioxide (CO2) emissions have over Earth's climate requires their accurate quantification and study. The ocean annually absorbs more than a quarter of all CO2 emissions and observation-based estimates of this ocean carbon uptake (sink) have now become a key component within annual global carbon budget assessments used to guide policy. And these ocean observations form one of only two, key observational pillars and constraints within annual carbon assessments, and their uncertainties directly impact the closure of the total budget. These assessments are fairly unique in the area of Essential Climate Variables (ECVs), as data synergy approaches are fundamentally required in their generation, whereas most other ECVs require only single parameters or measurements. These ocean assessments rely heavily on multiple satellite, in situ and re-analysis datasets, but uncertainties and errors within these datasets are often unknown or poorly constrained with unknown consequences. To address these issues, the Ocean Carbon for Climate (OC4C) project are now developing the first ocean carbon assessment that focusses on using climate data records. The first version (for 1980 to 2024) uses five climate data records, and this has enabled the regional and global uncertainties to be comprehensively characterised. This presentation will present the new dataset along with example case studies to illustrate some of the many advances. This will include i) an example erroneous signal that appears when climate records are not used, ii) how inclusion of biological signals can reduce the uncertainties, iii) the power of community led experiments to investigate underlying signals, iv) how an enhanced ocean budget can be used to interrogate the global carbon budget and the land component within this, and v) how the dataset has already been used for the 2025 Planetary Health Check assessment – an effort that charts humanity’s impact on the Earth System to guide government policy.

Rapid Arctic warming and permafrost thaw are mobilizing large pools of terrestrial organic carbon (OC) into rivers, deltas, and coastal seas. Quantifying these land–ocean fluxes is essential for constraining regional and global carbon budgets, predicting carbon cycle feedbacks, and assessing impacts on vulnerable Arctic shelf ecosystems. Yet major uncertainties remain: most satellite retrievals were designed for large water bodies, while Arctic rivers and nearshore zones are characterized by narrow channels, strong salinity and optical gradients, low sun angles, and adjacency effects that challenge algorithm performance. Here, we assess the potential of optical remote sensing to trace organic carbon across Arctic land–ocean compartments. Building on high-frequency river monitoring and multi-platform campaigns in the Mackenzie–Beaufort region, we evaluate Sentinel-2 MSI and Sentinel-3 OLCI algorithms, including atmospheric correction approaches (Acolite, Polymer, C2RCC) and inherent optical property retrieval schemes (band ratios, semi-analytical, neural networks). Results show that while absolute reflectances differ across atmospheric corrections, consistent spectral shapes and newly derived bio-optical relationships enable robust retrieval of dissolved and particulate OC across riverine, deltaic, and shelf waters. Remote sensing captures the strong seasonality of Arctic rivers, including spring freshet and rain-driven pulses, and reveals the lateral spread and variability of river plumes across the shelf. Our findings demonstrate that remote sensing can capture both the seasonal dynamics of OC export and the spatial heterogeneity of carbon pathways across compartments and salinity gradients. Comparisons with in situ data reveal limitations of ocean colour algorithms, guiding their applicability in Arctic environments. Satellite monitoring offers overlooked potential in Arctic rivers and coastal transition zones, providing synoptic coverage that complements sparse and costly field data. Through regional validation, performance assessment, and application of the most reliable combinations of atmospheric correction and IOP retrieval algorithms, we demonstrate how carbon remote sensing from space can reduce uncertainties in Arctic land–ocean flux estimates.

Quantifying the ocean carbon budget and understanding how it is responding to anthropogenic forcing is a major goal in climate research. It is widely accepted that the ocean has absorbed around a quarter of CO2 emissions released anthropogenically, and that the ocean uptake of carbon has increased in proportion to increasing CO2 emissions. Yet, our understanding of the pools of carbon in the ocean, the processes that modulate them, and how they interact with the land and atmosphere, is not satisfactory enough to make confident predictions of how the ocean carbon budget is changing. Improving our understanding requires a holistic and integrated approach to ocean carbon cycle research, with monitoring systems capable of filling the gaps in our understanding. Satellite observations can play a major role in this. The ESA-funded ‘Satellite-based observations of Carbon in the Ocean: Pools, fluxes and Exchanges’ (SCOPE) project aims to provide the best possible characterisation of the ocean carbon budget from satellite observations and further the understanding of its variability in space and time. Here, we present the development of an internally consistent dataset of the carbon pools, fluxes and exchanges that are observable from space, including dissolved inorganic and organic carbon, particulate inorganic and organic carbon, phytoplankton carbon, primary and export production, and air-sea CO2 and land-sea exchange. This satellite-based ocean carbon dataset is harmonised in space and time, based on climate-quality input data from the ESA’s Climate Change Initiative and fully error-characterised. This allows us, for the first time, to address both the physico-chemical and biological processes that drive the ocean carbon cycle in a consistent manner. We use the newly developed satellite-based ocean carbon dataset to analyse trends in each component of the ocean carbon cycle since the start of the ocean-colour data record in 1997. This will provide insight into how satellite observations can aid in the assessment of the ocean carbon budget in a climate context and provide useful information to evaluate and improve climate models.

Satellite-derived models of Primary Production (PP) is a critical tool to constrain the global carbon system and to better understand the mechanism of marine ecosystems with with unprecedented spatial resolution and coverage. While arguably one of the most significant developments in biological oceanography over the last 30 years, there is still questions about what the models predict (the output lies somewhere between Net and Gross PP) or how to validate the models with in situ observations. PP is furthermore only a part of the cycling of carbon in the surface ocean. To address these challenges, we have developed purely data-driven models to better constrain upper-ocean fluxes including export production (EP). We use decision-tree models to predict primary and export production from satellite-derived properties. The approach is expanded by including depth as one input feature, allowing for depth-resolved predictions. Our approach allows us to generate purely data-driven global models with high skill showing that depth resolved PP and EP estimates are feasible. We are able to predict well constrained vertical relationships for PP and EP closely analogous to earlier analytical work. The models can in conjunction also estimate community respiration, remineralization attenuation (the Martin relationship), and export efficiency.