Sea Surface Height Anomaly (SLA) is a signature of the mesoscale dynamics of the upper ocean. Sea surface temperature (SST) is driven by these dynamics and can be used to improve the spatial interpolation of SLA fields. In this study, we focused on the temporal evolution of SLA fields. We explored the capacity of deep learning (DL) methods to predict short-term SLA fields using SST fields. We used simulated daily SLA and SST data from the Mercator Global Analysis and Forecasting System, with a resolution of (1/12)° in the North Atlantic Ocean (26.5–44.42°N, −64.25–41.83°E), covering the period from 1993 to 2019. Using a slightly modified image-to-image convolutional DL architecture, we demonstrated that SST is a relevant variable for controlling the SLA prediction. With a learning process inspired by the teaching-forcing method, we managed to improve the SLA forecast at 5 days by using the SST fields as additional information. We obtained predictions of 12 cm (20 cm) error of SLA evolution for scales smaller than mesoscales and at time scales of 5 days (20 days) respectively. Moreover, the information provided by the SST allows us to limit the SLA error to 16 cm at 20 days when learning the trajectory.
Understanding Phytoplankton community dynamics in response to environmental shifts is crucial for assessing the impact of climate change on marine biology. To this end, satellite observations offer a dataset spanning two decades, capturing diverse sea surface parameters, including temperature, ocean color, and surface height. Notably, ocean color data is processed to derive sea surface chlorophyll-a concentration, widely acknowledged as a reliable proxy for phytoplankton biomass. Lately, advances in ocean color observation allow us to describe the phytoplankton community structure in terms of groups (broad functional or taxonomic groups) or size classes. Although these advances provide more detailed information on phytoplankton diversity and structure, these datasets suffer from spatial and temporal coverage limitations due to strict quality control in the presence of atmospheric aerosols, clouds, sea ice, etc... As a result, studies examining phytoplankton trends over the past two decades and future projections rely on incomplete chlorophyll-a and ocean color data. Therefore this compromises the identification of consistent trends within phytoplankton datasets.In this study, we address this issue using a deep-learning approach. Our method constructs an attention network that learns from the available satellite dataset of Chla and phytoplankton size classes images (weekly and one-degree-degraded spatial resolution) while assimilating information from gap-free sea surface physics data originating from satellite observations and assimilated numerical models). The primary objective is to estimate the phytoplankton dataset based on the knowledge of physical factors, while filling the gaps within this datasetThe trained deep-learning model allows us to discern patterns and correlations between chlorophyll concentration and the phytoplankton size classes on one hand, and the physics-based data on the other hand. From a phytoplankton weekly database spanning from 1997 to 2020, with 50% missing pixels, our approach demonstrates promising results in replicating chlorophyll concentration and accurately inferring phytoplankton size classes.The methodology shows the potential of deep-learning for robust ecological applications but mainly lays the groundwork for future trend studies on phytoplankton communities.
Phytoplankton abundance and community structure are crucial to the natural carbon cycle and the sustainability of marine ecosystems. However, their long-term natural variability and their response to anthropogenic climate change remain poorly understood. Since 1997, numerous algorithms have been developed to estimate these two parameters using ocean colour remote sensing. These tools have enabled 25 years of continuous, global-scale observations, providing invaluable insights into the variability of phytoplankton. However, analysing recent trends using these datasets presents significant challenges. Temporal inconsistencies and discontinuities in satellite time series - resulting from sensor transitions or decommissioning - introduce biases that complicate trend detection. Here, we present an inter-comparison framework designed to address these limitations by analysing multiple ocean colour remote sensing products. We implemented a robust statistical approach to minimize the influence of temporal discontinuities on trend detection. Results reveal both regions with robust trends across products and areas of significant disagreement, with the latter being largely prevalent. This study provides novel insights into recent phytoplankton dynamics while highlighting current limitations in our capacity to monitor these changes. Our findings emphasize the importance of multi-product analyses for reliable trend assessment in ocean colour remote sensing.
Phytoplankton form the base of marine food webs and play a central role in the ocean carbon cycle. Satellite observations of ocean color provide global information on phytoplankton pigments: Chlorophyll-a proxy for the total phytoplankton biomass and other accessory pigments which provide diagnostic markers for key functional phytoplankton groups and size classes; but these records remain relatively short and are affected by noise. By contrast, satellite-derived physical variables such as sea surface temperature and circulation have longer, more stable observational records, motivating efforts to relate biological variability to physical drivers and raising the possibility of reconstructing part of the biological variability from physical observations alone. However, it remains unclear what aspects of phytoplankton pigments temporal variability such approaches can robustly reproduce, particularly beyond the dominant seasonal cycle. The general problem addressed here is to assess the ability and limitations of neural networks to reconstruct seasonal to interannual variability of satellite-derived surface pigments at the global scale from physical data only. In this study, we show that neural networks can successfully reproduce the seasonal variability of phytoplankton pigments across the global ocean using surface physical drivers, achieving performances close to those of climatological estimates. In oceanic regions with strong interannual variability, such as the Tropical Pacific, part of the interannual signal can also be recovered (up to 50% of ENSO-related variability). However, outside these regions, interannual variability is often poorly reconstructed due to the dominance of seasonal signals, measurement noise, and the absence of clear physical signatures associated with biological changes. These results demonstrate that neural-network-based reconstructions capture physically driven components of phytoplankton variability but remain limited in their ability to infer weak or decoupled biological signals beyond seasonal variability.
Understanding the dynamics of phytoplankton communities in response to physicalenvironmental changes is essential for evaluating the impact of climate change on marineecosystems. Satellite observations provide a rich dataset spanning over two decades,capturing physical sea surface parameters such as temperature, salinity, and sea surfaceheight, alongside biological insights such as ocean color. Ocean color data, in particular, isprocessed to estimate sea surface chlorophyll-a concentrations — a widely recognized proxyfor phytoplankton biomass. Recent advancements in ocean color observation have furtherenabled the characterization of phytoplankton community structure in terms of functionalgroups or size classes.However, linking satellite-derived physical parameters to biological indicators remainschallenging due to spatial and temporal variability.Can physical data reliably predict patterns in ocean color, such as chlorophyll-aconcentrations and phytoplankton community structures, and potentially assess theirvariations? This study addresses this question through a deep-learning approach, utilizingan attention-based autoencoder model to learn relationships between physical variables andocean color data, including chlorophyll-a concentrations and phytoplankton size classes atweekly and 1° spatial resolution.Our trained deep-learning model effectively captures patterns and correlations betweenphysical parameters, chlorophyll concentrations, and phytoplankton size classes. It enablesdetailed exploration of how physical factors influence biological variability across differenttemporal scales. Utilizing a phytoplankton database spanning 1997–2023, this approachdemonstrates promising results in replicating chlorophyll concentrations, inferringphytoplankton size classes, and shedding light on the potential links between physical andbiological data.This study highlights the potential of machine learning for ecological research, contributing tomore accurate trend analyses. Understanding phytoplankton variability is critical for marineecosystem management, given their role in global carbon cycling. This methodologyunderscores the value of deep-learning to anticipate phytoplankton dynamics underchanging environmental conditions.
High-resolution satellite observations are essential for studying fine-scale ocean processes. Yet, present satellite sea surface salinity (SSS) products remain too coarse to resolve many fine-scale structures. We investigate denoising diffusion models as a generative framework for SSS downscaling in a controlled proof-of-concept experiment based on GLORYS reanalysis fields. A multichannel diffusion prior is trained on 1/12° SSS, sea surface temperature (SST), and sea surface height (SSH) fields, and is then conditioned at inference time on a synthetically degraded coarse SSS observation (1/3°) together with high-resolution (1/12°) auxiliary SST and/or SSH. Conditioning is performed through pseudo-inverse guidance, which steers the generated samples toward states that are compatible with the coarse observation while remaining within the learned GLORYS-consistent multivariate distribution. We also test a gradient-enhancement procedure designed to increase contrast during inference. Experiments in the Gulf Stream region compare models conditioned on SST only, SSH only, and both variables. Validation over the year 2020 uses root-mean-square error (RMSE), structural similarity (SSIM), gradient distributions, and temporal Fourier spectra. In the present GLORYS configuration, conditioning on SST substantially improves accuracy relative to SSH alone; combining SST and SSH yields further gains, comparable to a strong convolutional baseline under RMSE/SSIM, while additionally providing an ensemble of plausible reconstructions. The gradient-enhanced sampler increases structural contrast but can risk amplifying part of the variability, illustrating a trade-off between pixel-wise accuracy and structural realism. Overall, the results support guided diffusion as a promising framework for SSS downscaling and uncertainty-aware reconstruction, while showing that transfer to real satellite SSS products will require product-aware observation operators, uncertainty weighting, and independent in-situ validation.