Conference Agenda

The carbonate pump, an integral component of the biological carbon pump, plays a pivotal role in regulating the global carbon cycle by facilitating the production, sinking, and sequestration of particulate inorganic carbon (PIC), while also modulating the export and transfer of particulate organic carbon (POC). However, progress in understanding these complex dynamics remains limited by the scarcity of PIC observations from surface to depth.

Here we present two autonomous optical sensor prototypes designed to measure PIC concentrations in seawater. Both exploit the birefringence of calcium carbonate, the primary constituent of PIC, by detecting near-forward depolarized scattering with either linear or circular polarizers. Laboratory experiments confirmed that both designs were sensitive to variations in the concentration of PIC derived from cultured coccolithophores, a major calcifying plankton group, across an oceanic concentration range spanning more than three orders of magnitude. The linear prototype demonstrated higher sensitivity, while the circular prototype yielded stronger optical signals and improved alignment stability. and the circular prototype offering greater mechanical stability. We also observed differences in mass-specific depolarization between species, reflecting variations in coccolith morphology.

PIC sensor prototypes were deployed on several research cruises, operating in underway flow-through mode at a sampling rate of 1 Hz. Sensor signals correlated very well with PIC concentration obtained from discrete water samples, demonstrating the capability for autonomous, high-resolution sensing of PIC in surface waters. These sensors provide much-needed in situ observations to improve satellite-based PIC retrieval algorithms and to advance our understanding of the biological processes driving the carbonate pump.

Griet Neukermans^1,2, Clémence Goyens¹, Alexandre Castagna¹, Qiming Sun¹, Andrea van Langen Roson^1,2, Nils Haentjes³, Emmanuel Boss³, Thanos Gkritzalis², and Peter Landschützer².

¹ Marine Optics and Remote Sensing Group (MarSens), Ghent University, Ghent, Belgium

² Flanders Marine Institute (VLIZ), Ostend, Belgium

³ School of Marine Sciences, University of Maine, Maine, USA

Integrating measurements of carbonate chemistry, marine reflectance, and bio-optical properties is essential to capture and understand the coupled physical–biogeochemical processes driving CO₂ dynamics and to link in situ observations with satellite remote sensing. This is particularly so in coastal shelf seas, comprising optically-complex waters with strong spatial and temporal variations in biological activity and carbonate chemistry.

We build on the observational capacity of RV Simon Stevin, a Flemish ICOS (Integrated Carbon Observation System) Ocean Station operating in the North Sea, equipped with state-of-the-art sensors for continuous measurement of carbonate system parameters on pumped surface water, including the partial pressure of CO₂ (pCO₂). We expanded the vessel’s underway system with a flow-through Autonomous uNderway near-real-Time HYperspectraL Optical Properties PackagE (ANTHYLOPE), measuring hyperspectral backscattering (Sequoia hyperBB), attenuation, and absorption (Seabird AC-S), single wavelenght backscattering, fluorescence of CDOM and Chlorophyll-a (RBR Tridente), UV fluorometry (Seapoint SUVF), particle size distribution (Sequoia LISST-200X), red light attenuation (LISST-Tau) and Particulate Inorganic Carbon (prototype optical sensor, developed in collaboration with Sequoia Scientific), complemented by a thermosalinograph (Seabird TSG). Lastly, an autonomous hyperspectral radiometry system for the measurement of above-water reflectance (R_rs), (IMO DALEC) was mounted on a pole on the bow of the vessel.

Our integrated monitoring system was first put in operation in May 2024 and has been tested and improved during several measurement campaigns. Here, we present the data processing and quality control pipelines and discuss the challenges associated with the operation of the ANTHYLOPE and DALEC systems. We present preliminary results on the mulitple uses of the integrated dataset. First, we show that the characteristics of the particle assemblage (particle concentration, composition, and size) can be retrieved from inherent optical properties. Next, we show the improved retrieval of biogeochemical variables by leveraging the hyperspectral nature of the signals. We also test and (re-) calibrate commonly used remote sensing algorithms for the retrieval of SPM, POC, and Chlorphyll-a from R_rs. Lastly, we investigate the spatio-temporal dynamics of pCO₂ and examine its physical, chemical, and biological drivers.

Satellite-derived particulate backscattering (b_bp) provides key insights into large-scale ocean biology and biogeochemistry, serving as a proxy for phytoplankton biomass and particulate organic carbon. These data underpin estimates of carbon stocks, fluxes, and productivity in coupled physical–biogeochemical models. However, the paucity of in situ multi-band b_bp measurements limits robust uncertainty assessment of satellite products (Brewin et al., 2023). To address these observational gaps, Surface Velocity Programme (SVP) drifting buoys, historically used to validate SST and SSS, have been equipped with bio-optical and oxygen sensors, defining the Biogeochemical-SVP drifters (BGC-SVP). The high sampling frequency combined with a Lagrangian approach enables it to overpass numerous pixels in a single day, thus providing large in situ datasets for validation activities of b_bp. In the context of ESA INSPIRE project, here, we present results for the first comparison of satellite and in-situ b_bp using observations from BGC-SVP Lagrangian drifters which provide b_bp at 470 and 532 nm. To this aim, we tested different algorithms (e.g. QAA, GSM) and satellite sensors (e.g., PACE/OCI, Sentinel-3/OLCI) to quantitatively evaluate the performance of the retrievals. The in-situ data derived entirely from the first BGC-SVP drifter array deployed in the Mediterranean Sea during the ITINERIS’EYES cruise performed in July 2025. In the next future, a coordinated array of BGC-SVP drifters, BGC-Argo floats, and other autonomous platforms working in synergy could provide the required surface and subsurface data at the temporal, spatial, and spectral (e.g., multi- or hyperspectral resolution) scales of interest to future satellite missions.

Particulate organic carbon is central to oceanic carbon export and biogeochemical cycling, yet robust global observations of its mass concentration (POC) remain challenging due to limitations in remote sensing and in situ techniques. The expanding BGC-Argo float array can support integration of surface POC estimates from satellite remote-sensing reflectance R_rs with subsurface observations, but consistent algorithms are essential to avoid platform-induced biases. This study evaluates the performance of a multivariable POC algorithm, referred to as Model-B (Koestner et al., 2024), that uses the particulate backscattering coefficient b_bp and concentration of chlorophyll-a (Chla) as inputs, and is applicable to both BGC-Argo float and satellite observations with a 2-step approach. Three matchup datasets are explored: in situ R_rs and in situ POC (N = 509), satellite R_rs and in situ POC (N = 223), and satellite R_rs and BGC-Argob_bp(700) and Chla (N = 4448). For estimating POC from R_rs, the Model-B input of b_bp(700) is derived using QAA-v5 using either MODIS-Aqua or in situ R_rs, and the Chla input is estimated with the OCI algorithm. Independently, POC is also derived from Model-B using vertically-resolved b_bp(700) and Chla from scattering and fluorescence sensors on BGC-Argo floats. Initial results show that the multivariable Model-B performs comparably across both in situ and satellite matchup datasets to the R_rs-based hybrid POC algorithm developed for global satellite applications (Stramski et al., 2022). Some positive biases for Model-B at low POC occur which are likely driven by uncertainties in b_bp(700) and Chla inputs. For the MODIS–BGC-Argo matchups, Model-B estimates from BGC-Argo floats in the surface layer show promising consistency with satellite hybrid POC algorithm results, particularly when satellite-derived Chla is used instead of float-based fluorescence estimates of Chla. Further evaluation is ongoing to assess regional and temporal agreement, refine BGC-Argo Chla fluorescence corrections, and explore merging satellite and BGC-Argo data into a 3-D POC product.

The coastal ocean plays a critical role in the ocean carbon cycle, yet our current understanding of the different pools and fluxes of organic carbon is limited by strong local dynamics, which requires high spatial and temporal resolution of observations covering relatively large areas across the shelf seas. Satellite remote sensing of carbon pools and fluxes at high spatial resolution and daily frequency can cover this gap in observations, but new algorithms need to be developed and validated.

In the ESA project “Carbon from earth Observation between Ocean and Land (COOL)”, we aim to estimate carbon pools and fluxes for which algorithms are relatively mature, and hence we can have some certainty in their application in the coastal ocean at the global scale in the near future. These include Particulate Organic Carbon (POC), Particulate Inorganic Carbon (PIC), Dissolved Organic Carbon (DOC) and Primary Production (PP). Using data from Sentinel-3 OLCI at 300 m resolution and Sentinel-2 MSI at 60 m, we aim to produce an internally consistent coastal ocean carbon satellite dataset in selected European coastal regions. We make use of in situ datasets for evaluation and validation of these satellite products and investigate the newly produced Earth Observation datasets in the Baltic Sea and in the upwelling areas of the western Gallican coast, northwest Spain.

Satellite observations of particulate backscattering (bbp) have greatly enhanced our understanding of ocean biology and biogeochemistry on large scales, serving as proxies for phytoplankton biomass or particulate organic carbon. bbp is essential for estimating organic carbon stocks and fluxes and ocean productivity, which is subsequently incorporated into coupled physical-biogeochemical models. However, the paucity of in situ multi-band bbp data hinders efforts to quantify uncertainties in satellite bbp and its derived products. To address these gaps, Surface Velocity Programme (SVP) drifting buoys have been equipped with bio-otical and oxygen sensors, defining the Biogeochemical (BGC)-SVP drifters. The high sampling frequency combined with a Lagrangian approach enables it to overpass numerous pixels in a single day, thus providing large in situ datasets for validation activities, not achievable by other in situ platforms. Here, we present a novel Observing System (OS) to validate satellite bbp products, integrating remote sensing, Lagrangian modeling, and in situ data, for the Mediterranean Sea. Main innovations are: integration of Lagrangian simulations using a sub-grid kinematic model applied to ocean currents datasets, and the constraing of simulated trajectories with gapped satellite bbp data to assess variability and identify optimal deployment sites and time for BGCSVP drifters, maximizing match-up opportunities. Different criteria are established as the beachtime time, the total potential and bin-specific matchups. Preliminary results suggest the Ionian Sea as the best site to reduce drifter beaching but also to capture low-mid bbp values over the entire year. Higher bbp values could be captured during winter and spring in the northwestern Mediterraean Sea. The development of an OS is a foundational step from research to sustained operations. The OS framework here developed can be extended to global ocean and has potential applications for validating other ocean color variables across ongoing (e.g., Sentinel-3/OLCI, PACE/OCI) and future satellite missions (e.g., ESA CHIME).

The particulate backscattering coefficient (bbp) is an indicator of phytoplankton biomass, particulate organic carbon, and particle size distribution in the ocean. It serves as input for modeling net marine primary production and net community production. Since bbp can be estimated through satellite imagery, it plays a fundamental role in quantification primary production on a global scale and evaluating its spatial patterns. Therefore, accurate satellite-based bbp is required to constrain coupled physical and biogeochemical models, thereby improving climate projections. To date, most of the European Space Agency (ESA) Ocean Science Cluster-funded projects that utilize bbp have relied on global operative products (i.e., ESA OC-CCI). However, these products lack associated uncertainty compared to in situ measurements, limiting out understanding of their impact on ocean productivity and organic carbon export.

The ESA INSPIRE project aims to address this gap by developing an advanced Observing System (GOS) specifically tailored for validating satellite bbp products, integrating remote sensing, Lagrangian modelling and in situ data. This involves using a new generation of Surface Velocity Programme (SVP) drifting buoys equipped with bio-optical and oxygen sensor, named Biogeochemical (BGC)-SVP drifters. Designed for extended deployment periods, they offer a promising solution for collecting data in challenging marine environments by the combination of the Lagrangian approach and a high sampling frequency. This project seeks optimize drifter deployment locations to maximises the number of in situ observations usable for match-up activities. Lastly, satellite bbp products will be validated with in situ measurements collected using BGC-SVP drifters deployed both the global ocean with a particular focus in the Mediterranean Sea. The drifters launched in the Mediterranean Sea were acquired through the ITINERIS (Italian Integrated Environmental Research Infrastructures System) project.

The present study and the use of BGC-SVP drifters could be impactful in relation to the next generation of altimetry (e.g., NASA SWOT), hyperspectral ocean color satellite missions (e.g., NASA PACE, NASA GLIMR, ESA Sentinel Next Generation, and ESA CHIME), and future lidar mission (e.g., ASI CALIGOLA) for the detection of ocean processes from fine to larges scales both in space and time.

The ocean is one of the largest sinks of anthropogenic CO₂ emissions, yet quantification of its carbon storage and distribution remains uncertain. Satellite measurements of ocean colour, sea surface temperature, salinity, altimetry and atmospheric CO₂ offer global unique coverage. However, effective and transparent integration of these diverse datasets persists as a complex problem. This study introduces an XAI prototype model that aims to predict air-sea CO₂ fluxes, using combined Earth observation and in-situ data. Over the North Atlantic, a regional test case was used where SOCAT in situ pCO₂ measurements were combined with multi-sensor satellite data from Sentinel-3 (OLCI chlorophyll-a, SLSTR SST, altimetry SSH) and ASCAT winds. Machine learning models (XGBoost) were used to predict surface pCO₂ and air-sea fluxes were computed through traditional bulk formulations.

For interpretability, SHAP values were utilized to quantify the relative contribution of environmental drivers to flux estimates. The findings show that sea surface temperature anomaly dictated variability for the 2019 marine heatwave and chlorophyll-a contributed significantly during seasonal bloom events. Outputs were then framed into a Carbon Knowledge Graph (CKG), linking flux estimates to their drivers and uncertainties. Additionally, this approach highlights the potential for scalable applications to other ocean basins in support of global climate assessments. The framework shows how explainable machine learning and knowledge organization can provide open, policy-relevant monitoring of the ocean carbon cycle.

Dissolved organic carbon (DOC) plays a crucial role in ecological and biogeochemical processes. Many models have been developed to estimate DOC in coastal waters at a local scale using ocean-colour remote-sensing data. However, there is currently no global algorithm capable of addressing the variability and complexity of DOC dynamics in coastal waters. In the Satellite-based observations of Carbon in the Ocean: Pools, fluxes and Exchange (SCOPE) project, funded by the European Space Agency (ESA), we aim to address this gap by developing a global DOC satellite retrieval algorithm for coastal waters by using daily, 4-km resolution data from the European Space Agency (ESA) Ocean Colour Climate Change Initiative (OC-CCI) from 1997 to 2023, combined with sea surface salinity (GLORYS12v1) and temperature (ESA SST-CCI, version 3.0). For model development, we matched these satellite datasets with in situ DOC concentration data from the CoastDOM v1 database. Multiple statistical methods, including multiple linear regression (MLR), random forest regression (RF), and extreme gradient boosting (XGBoost), were tested, with the best performance achieved by a RF model using sea surface salinity and temperature, the remote sensing reflectance at 560 nm and total absorption at 412 nm. Although the developed algorithm showed high performance, the relatively coarse resolution of OC-CCI poses challenges, as it may fail to resolve sharp DOC gradients in dynamic coastal zones such as river plumes and estuaries, potentially reducing accuracy in those areas. Still, OC-CCI offers climate-quality data for a longer period of time compared to individual ocean-colour sensors. Expanding in situ observations, especially in underrepresented areas, will further enhance model accuracy and applicability. This work contributes to a better understanding of carbon dynamics in coastal ecosystems and provides a robust tool for future satellite-based assessments of DOC in global coastal waters.

The research project investigates the complex interplay between underwater light dynamics and carbon cycling in the ecologically and economically significant coastal waters of the Sundarbans. These coastal ecosystems, characterized by persistently high turbidity due to sediment discharge from the Ganges-Brahmaputra-Meghna river system, pose unique challenges for environmental monitoring and resource management. The concentration of Total Suspended Solids (TSS), often exceeding 210 mg/L, significantly limits the availability of Photosynthetically Active Radiation (PAR), which is critical for phytoplankton growth—the foundation of aquatic food webs. In this study, we employ direct, high-accuracy in-situ measurements and laboratory analyses to accurately characterize the underwater light environment and carbon dynamics. We have conducted field studies during peak monsoon conditions to capture maximum turbidity effects and assess vertical profiles of PAR, turbidity, and chlorophyll-a (Chl-a) across multiple sampling stations, notably around Dublar Char and Kuakata. The data collected evaluates how variations in light characteristics and carbon pools influence ecosystem productivity indicators that are vital for fisheries, aquaculture, and overall water quality.

Integrating satellite remote sensing data provides a broader spatial context and allows validation against our field observations, thereby enhancing the reliability of the results. This research aligns with national developmental goals by addressing critical environmental challenges and promoting sustainable resource management practices. It actively contributes to global Sustainable Development Goals (SDGs) related to climate action, marine resource sustainability, and food security. By bridging the data gap in this optically complex region, our study produces actionable insights that guide policy development and climate resilience initiatives crucial for the livelihoods of coastal communities.

Keywords: Light dynamics, Carbon cycling, Aquatic productivity, Remote sensing, Coastal management.

Accurate estimation of phytoplankton functional types (PFTs) from space remains challenging in optically complex waters, where global algorithms often fail to capture regional bio-optical variability. We present advances in regionalized algorithm development and validation for two contrasting yet similarly challenging basins: the Baltic Sea and the Black Sea. In the Baltic Sea, we refined empirical diagnostic pigment (DP)-based approaches using High-Performance Liquid Chromatography (HPLC) datasets from multiple sub-basins, enabling improved estimation of phytoplankton size classes (PSCs) and key functional groups such as cryptophytes, green algae, and dinoflagellates. Nanoplankton dominated basin-wide (~46% of chlorophyll a), while picoplankton prevailed offshore and microplankton peaked in nearshore regions. In the Black Sea, we combined hierarchical clustering, principal component analysis, and network-based community detection to derive region-specific coefficients for PFT–pigment relationships from HPLC measurements at 690 stations collected across 12 bio-optical campaigns. Applying these algorithms to multi-decadal satellite chlorophyll-a datasets (1998–2024), we reconstructed spatial and temporal patterns of PFTs, with microplankton dominating nutrient-rich coastal zones (70–80% of chlorophyll a), nanoplankton showing broad distribution (~30–40%), and picoplankton prevailing offshore (>60%). Both regionalized models significantly reduced errors compared to global approaches, particularly for cryptophytes, haptophytes, and prochlorophytes, and showed consistency with microscopy-based observations. These results demonstrate the potential of tailored algorithms to enhance the monitoring of phytoplankton community structure in coastal seas, thereby advancing our capacity to assess ecosystem dynamics and their role in regional carbon cycling.

The Yangtze River Estuary in China – one of the most turbid estuaries worldwide - is a biogeochemical transformer of autochthonous and allochthonous dissolved organic carbon (DOC) that shapes coastal ecosystem functioning and bio-diversity. Here, we used almost one decade of satellite data from Sentinel-3/OLCI, to examine, for the first time, spatial patterns, seasonal cycles, and interannual variabilities of DOC across this important and complex ecosystem from space. Four atmospheric correction (AC) approaches (C2RCC, ACOLITE, MUMM, and POLYMER) were first evaluated for OLCI using a rich field radiometric data collected across the estuary over the past ten years. We found that ACOLITE was the best-performing AC method, with mean absolute percent difference of 15%. An ocean color DOC algorithm was then developed using a machine-learning (random forest) approach for the Yangtze River Estuary based on a comprehensive bio-optical data collected in this system, and applied to generate a long-term DOC data record from OLCI (2016-2025). Higher DOC concentrations were consistently observed at the estuary mouth, strongly influenced by river discharge, while sharp gradients and distinct DOC plumes were captured on the inner East China Sea continental shelf, consistent with freshwater riverine export and monsoon. This study offered the first comprehensive observation of DOC dynamics across the estuary from the space and an analysis of the key factors driving biogeochemical variability at seasonal and interannual scales.

The central Iceland Basin (CIB) is a key region of the subpolar North Atlantic, where phytoplankton blooms drive substantial fluxes of carbon and nutrients, playing a critical role in regional biogeochemical cycles. In this study, we tuned a spectrally-resolved model of marine primary production using in situ datasets from the CIB. Measurements of phytoplankton pigments, light absorption coefficients, and photosynthesis–irradiance (P–E) parameters were obtained from several cruises of the CIB that covered various phases of the seasonal cycle, characterised by different levels of resource limitation. Matchups between in situ chlorophyll-a measurements and Level-3 satellite ocean colour data products reveal sensor-specific differences. Profiles of Chlorophyll-a fluorescence from CTD casts and BGC-Argo floats, corrected for non-photochemical quenching, were used to characterise the shape and magnitude of the biomass profile. The chlorophyll profile parameters and phytoplankton absorption measurements were incorporated into the radiative transfer model, and modelled profiles of photosynthetically active radiation (PAR) and downwelling irradiance at 412 and 490 nm were compared with measured irradiance profiles obtained from ship-based deployments and floats. Using the regionally-tuned primary production model, profiles of instantaneous production, integrated over the day, were computed. We discuss the implications of our findings for satellite-based estimates of primary production.

As a critical component of the oceanic carbon cycle, phytoplankton carbon should be monitored operationally. However, the reliability of ocean colour algorithms developed so far is yet to be fully established across different oceanic conditions. As part of the ESA-funded project ‘Satellite-based observations of Carbon in the Ocean: Pools, fluxes and Exchanges’ (SCOPE), we conducted a comprehensive intercomparison of representative algorithms developed previously for computing phytoplankton carbon from satellite ocean colour. Four categories of phytoplankton-carbon retrieval algorithms were considered based on their structure and suitability for global application: (1) particle backscattering-based empirical relationship, (2) particle backscattering-based allometric semi-analytical algorithm, (3) absorption-based allometric semi-analytical algorithm, and (4) photoacclimation-based algorithm. We compiled a large in situ database of phytoplankton carbon consisting of the available flow cytometry-based measurements and directly measured phytoplankton carbon to assess the performance of the algorithms. In situ matched-up algorithm outputs were generated using ESA’s Ocean Colour CCI v6 data archive. Results of this extensive intercomparison will be presented to assess the performance and consistency of the candidate algorithms, and the commonalities and major differences will be highlighted. The suitability of applying the candidate algorithms, either individually or in combination, for routinely estimating phytoplankton carbon in the global ocean will be discussed. Our results will contribute to enhanced understanding of the oceanic carbon cycle and the estimation of the ocean carbon budget using satellite remote sensing.

Particulate Inorganic Carbon (PIC) is an important component of the marine carbon cycle, due to its dual role in carbon sequestration and release. In the pelagic environment, PIC is produced by calcifying plankton, particularly coccolithophores. Mapping of coccolithophore bloom areas and monitoring PIC concentration have been performed from space using multispectral ocean colour satellite since the SeaWiFS mission.

Here we provide a first evaluation of the geostationary Meteosat Third Generation (MTG) Flexible Combined Imager (FCI) data to map areas where PIC is the dominant source of scattering and to retrieve PIC concentration. In situ data were collected during a field campaign in the North Sea in June 2025, when several water types were sampled, including a coccolithophore bloom (PIC concentration between 10 and 80 mg m^-3). PIC was measured via ICP-OES from discrete water samples, and continuously during ship transit with a prototype LISST-PIC optical sensor. Previously published waveband difference algorithms were adapted to the FCI wavebands.

The classification method confidently detected areas with PIC above 40 mg m^-3 (3.3 mmol m^-3), while correctly excluding areas with a high non-coccolithophore particle load (e.g. the East Anglian plume). The PIC retrieval algorithm showed a Mean Absorlute Percentage Deviation (MAPD) of 40 % against discrete samples (N = 7) and 55 % against the continuous PIC measurements (N = 84), with most of the deviation arising from estimation bias (overestimation). While these results were obtained from a single campaign and a single coccocolithophore bloom, they suggest that the FCI can be used to detect areas where PIC is the main scatterer and for PIC quantification, warranting additional work to improve algorithm design. The high temporal frequency of FCI can potentially provide better coverage of areas with frequent cloud cover, as well as resolve tidal variations in coastal environments where coccolithophores are known to occur (e.g. English channel).

In situ bio-optical datasets are essential for the assessment of the uncertainties of satellite ocean colour measurements and derived products. This is especially critical in coastal waters, where land adjacency effects, complex atmospheric aerosol mixtures, high loads of optically active components in particular high concentration of chromophoric dissolved organic matter and bottom reflectance effects contaminate the signal that reaches the satellite.

The Tara Europa expedition, the ocean component of the Traversing European Coastlines (TREC) program carried a comprehensive sampling of coastal ecosystems all along the European coast in 2023 and 2024. The Tara Europa expedition offered the unique opportunity of an oceanographic survey from a unique platform, using the same set of protocols, instruments, and sample analysis, collocated with a rich biological dataset describing the microbiologic diversity in detail.

Within the ESA-funded Hyperspectral Bio-Optical Observations Sailing on Tara (HyperBOOST) project, PML, CNR, LOV and UMaine extended the variables collected during the TREC integrated sampling by including bio-optical measurements relevant to present and future satellite ocean colour missions. This provided a comprehensive dataset encompassing in-situ hyperspectral radiometry, bio-optical properties, optically active components, biogeochemical and biodiversity relevant data for optically complex waters. Continuous inline optical measurements were combined with laboratory analyses of surface water samples collected at more than 200 stations along the European coasts.

This dataset will provide the opportunity to explore the complexity of Dissolved and Particulate Organic matter optical properties across the land-sea interface, opening the possibility to improve DOC and POC quantification from satellite imagery.

The diffuse export and biogeochemical processing of dissolved organic carbon (DOC) along coastlines remain under-quantified controls on coastal water quality and the ocean carbon cycle. Coastal DOC processes are difficult to quantify, because the coastal zone features steep gradients in DOC concentration and organic matter composition, dynamic biogeochemical and physical processes, and complex waters containing diverse sources in-water constituents. Hyperspectral imagery can facilitate the quantification of coastal DOC dynamics at large scales by separating multiple in-water constituents and characterizing subtle changes in ocean color to map biogeochemical and physical properties. To fully utilize a new generation of existing and planned hyperspectral satellite imagers, new regionally transferable algorithms are needed to retrieve coastal DOC concentration and composition at continental to global scales. To address this need, we developed hyperspectral algorithms for retrieving both DOC concentration as well as the quality and composition of colored dissolved organic matter (CDOM) in coastal waters characterized by varying water quality and ecological conditions. An in-situ dataset of hyperspectral radiometry, CDOM, DOC, and ancillary parameters was compiled from research cruises on the Atlantic, Gulf, Pacific, and Arctic coasts of the United States for algorithm development and validation. Algorithms utilizing hyperspectral ocean color, overall, outperformed existing multispectral approaches for retrieving coastal DOC and allowed for the retrieval of novel indicators of organic matter source and composition based on CDOM spectral shape. These algorithms were applied to hyperspectral satellite imagery of contrasting urbanized and natural coastal estuaries to demonstrate their utility and transferability for quantifying coastal DOC processes under diverse environmental conditions and anthropogenic impacts.

This study explores oceanic carbon fluxes from space by focusing on Net Community Production (NCP), the balance between primary production and community respiration. NCP estimates the if a region is consuming or producing carbon, which on long time scales can be translated to an estimate of carbon export flux. The project's goal is to develop a novel algorithm that estimates NCP directly from hyperspectral Rrs, without requiring ancillary data and capable of capturing both positive and negative NCP values in optically complex waters. To build the algorithm we are collecting in situ NCP and Rrs data in the South Atlantic Bight (SAB) region, which offers broad coastal-to-offshore gradients in both NCP and optical properties. NCP is measured using the Pressure of In Situ Gases Instrument (PIGI) flow-through system, which estimates NCP from dissolved oxygen and nitrogen gas concentrations, offering continuous measurements and a cost-effective alternative to traditional Oxygen/Argon methods. Concurrent hyperspectral remote sensing reflectance (Rrs) data is collected using the Solar Tracking Radiometry Platform (So-Rad), which provides high-resolution radiometric observations. Both instruments operate autonomously aboard the R/V Savannah, which primarily operates in the SAB. Here, we present initial results from the first field season (100 days in 2025), including early progress on neural network-based model development. Looking ahead, a second algorithm based on MODIS bands is planned to assess long-term trends in SAB NCP over the past 25 years using satellite data archives.

Ocean carbon research faces a persistent challenge: high-quality in-situ measurements are extremely sparse and expensive to collect, yet these data are essential for understanding marine primary production and its role in global climate processes. Current satellite-based approaches struggle with limited validation data, a problem recognised by the IPCC as a major constraint on ocean carbon cycle understanding.

We demonstrate how GeoFoundation models pre-trained on abundant unlabelled Sentinel-3 satellite data can dramatically improve performance when only small amounts of in-situ measurements are available. Our approach first pretrains a model using 512,000 Sentinel-3 tiles spanning global ocean regions to learn generalizable features, then fine-tunes on sparse oceanographic measurements.

The benefits for data-limited applications are substantial. Our primary production model achieved meaningful performance using only 103 in-situ measurements—representing just 6% of the pixels in a single satellite image. When training data was reduced to just 19 observations, the foundation model maintained strong performance whilst conventional approaches failed. This demonstrates the approach's potential to extract maximum value from existing sparse datasets and opportunistically collected measurements.

Beyond improved statistical performance, the model captures realistic spatial patterns over large oceanic regions where no training data exists. Large-scale inference reveals detailed coastal productivity structures that conventional physical models typically under-predict, suggesting the approach has learnt meaningful oceanographic relationships.

We also evaluated the approach for chlorophyll-a concentration estimation using 274 global in-situ measurements. The GeoFoundation model substantially outperformed existing methods, achieving lower RMSE compared to decision tree approaches and the operational Sentinel-3 OLCI Level-2 neural network product. Crucially, when applied to large-scale inference over the North Sea, the foundation model produced spatial patterns with higher Structural Similarity Index Measure (SSIM) (0.88) to the operational product compared to models trained from scratch (0.82), and the decision tree (0.68), demonstrating improved ability to capture realistic oceanographic features.

The implications for operational ocean carbon monitoring are significant. This methodology could enhance existing observation networks by maximising the value of each expensive ship-based measurement and support carbon cycle research in data-poor regions.

An important prerequisite for understanding the role and response of ocean ecosystems to rising atmospheric CO₂ levels and global warming is accurate and well-characterized regional, basin and global scale measurements of oceanic Net Primary Productivity (NPP). Currently global estimates of NPP from satellite data used in global ocean carbon cycling and climate studies, continue to suffer from uncertainties because of their dependence on: 1) satellite derived fields of phytoplankton biomass that are constructed using algorithms incapable of providing product accuracies over regional and global scales, 2) limited estimates of phytoplankton photosynthetic quantum yields (ϕ) that are currently obtained primarily aboard research vessels, and 3) inadequate methods for scaling local in-situ ϕ measurements to regional and basin scales. Here we have utilized the Absorption based Model (AbPM), that exploits the inherent optical absorption properties of phytoplankton derived from remotely sensed reflectance data rather than phytoplankton biomass as an input, and a novel bio-optical classification scheme called Bio-Optical Measurement and Evaluation System (BIOMES) for scaling limited in-situ estimates of ϕ to obtain global maps of NPP. We have used the global collection of in-situ NPP datasets to assess the performance of AbPM derived estimates of NPP against those obtained using more widely used biomass-based models.

The particle size distribution (PSD) is a key property related to the structure and function of marine ecosystems, as well as biogeochemical and optical properties. A recently developed PSD ocean color algorithm retrieves the parameters of an assumed power-law PSD (slope ranging from 2.5 to 6.0, and scaling parameter, N_o) using a 2-component particle bio-optical model, applied to mostly phytoplankton-sized particles and smaller. The PSD can be used to retrieve size-partitioned phytoplankton carbon. Here, we extrapolate the PSD to larger particles to retrieve global mesozooplankton carbon biomass and diel vertical migration (DVM) as part of a larger project aimed at assessing zooplankton contribution to vertical carbon transport via biogenic hydrodynamic transport. A significant finding is that retrieval results are realistic when the PSD slope is fixed at 4.0, rather than variable spatially and temporally within the 2.5 to 6.0 range as in the original PSD retrievals. This fixed slope of 4.0 is consistent with the so-called Sheldon hypothesis; the significance of this finding is discussed, also drawing on recent developments indicating a very conserved marine ecosystem size spectrum. Furthermore, retrievals are realistic when the tuned scaling parameter (N_o) is used, but the tuning was previously designed for POC and phytoplankton carbon retrievals, and it is independent of the mesozooplankton retrievals. The novel algorithm has been applied to the latest monthly OC-CCI v6.0 merged ocean color data set; these data and code have been published. Basic characteristics of this data set are presented and discussed. The PSD-based mesozooplankton carbon biomass and DVM retrievals are validated against two data sets independent of ocean color PSD: a lidar-based DVM retrieval, and the MAREDAT in-situ zooplankton biomass data set. Results of that validation are presented and implications of the time lag between phytoplankton and zooplankton biomass are discussed.