Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Ocean mover’s distance: using optimal transport for analysing oceanographic data
Remote sensing observations from satellites and global biogeochemical models have combined to revolutionize the study of ocean biogeochemical cycling, but comparing the two data streams to each other and across time remains challenging due to the strong spatial-temporal structuring of the ocean. Here, we show that the Wasserstein distance provides a powerful metric for harnessing these structured datasets for better marine ecosystem and climate predictions. The Wasserstein distance complements commonly used point-wise difference methods such as the root-mean-squared error, by quantifying differences in terms of spatial displacement in addition to magnitude. As a test case, we consider chlorophyll (a key indicator of phytoplankton biomass) in the northeast Pacific Ocean, obtained from model simulations, in situ measurements, and satellite observations. We focus on two main applications: (i) comparing model predictions with satellite observations, and (ii) temporal evolution of chlorophyll both seasonally and over longer time frames. The Wasserstein distance successfully isolates temporal and depth variability and quantifies shifts in biogeochemical province boundaries. It also exposes relevant temporal trends in satellite chlorophyll consistent with climate change predictions. Our study shows that optimal transport vectors underlying the Wasserstein distance provide a novel visualization tool for testing models and better understanding temporal dynamics in the ocean.  more » « less
Award ID(s):
1748166
PAR ID:
10379164
Author(s) / Creator(s):
; ; ; ; ; ; ; ; ; ;
Date Published:
Journal Name:
Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
Volume:
478
Issue:
2262
ISSN:
1364-5021
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; (, Global Biogeochemical Cycles)
    Abstract Nitrate is an essential nutrient for phytoplankton growth and is a primary component of ocean carbon cycling. In this study, we developed a neural network constrained by the high spatial and temporal coverage of BGC‐Argo floats to predict nitrate in a consistent way throughout space and time in the Southern Ocean, a key area for ocean carbon uptake and controlling global ocean nutrient distributions. After correcting for physical and sampling biases using the Biogeochemical Southern Ocean State Estimate model, we show that annual net community production (ANCP), originally calculated from seasonal nitrate drawdown, reveals the greatest production around the 45–55°S meridional band, and an average basin‐wide ANCP of 3.91 ± 0.13 PgC y−1with a significant increase of 0.67% y−1from 2004 to 2022. We also highlight that using the common nitrate seasonal drawdown method to derive ANCP might underestimate the true carbon export at depth by about one third. Our findings align with previous studies, which indicate an increase in surface satellite chlorophyll‐aand model export fluxes. Our results demonstrate the potential of leveraging machine learning constrained by BGC‐Argo observations to study long‐term changes of biogeochemical processes in the ocean. 
    more » « less
  2. ; ; ; ; (, Remote Sensing)
    Harmful algal blooms (HABs) are increasing in frequency in the Chesapeake Bay, posing risks to marine ecosystems, water quality, and public health. Chlorophyll-a (Chl-a) is a widely used indicator of algal biomass, and satellite observations such as Sentinel-3 Ocean and Land Color Instrument (OLCI) enable large-scale monitoring of bloom dynamics. However, cloud cover and atmospheric interference frequently introduce missing pixels in daily satellite products, reducing temporal continuity and limiting monitoring reliability. Satellite-derived chlorophyll-a (Chl-a) data exhibit substantial missingness, with daily pixel gaps ranging from approximately 52.30% to 100% (mean ≈ 88.95%). This study evaluates spatial interpolation, EOF-based, supervised machine-learning, deep-learning, and convolutional autoencoder approaches for reconstructing missing Chl-a values. Sentinel-3 OLCI Chl-a data from 2023–2024 were used for model training, while data from 2025 served as a temporally independent test set to avoid spatiotemporal leakage. To simulate cloud-induced data gaps, artificial missingness scenarios ranging from 50% to 90% were applied for the Inverse Distance Weighting (IDW) and Data Interpolating Empirical Orthogonal Functions (DINEOF) baseline approaches, while machine-learning, deep-learning, and convolutional autoencoder models were evaluated using real satellite-derived missing observations. The evaluated models include IDW, DINEOF, K-Nearest Neighbors (KNN), Random Forest (RF), Extra Trees (ET), XGBoost, a Long Short-Term Memory (LSTM) network, and a Temporal Data Interpolating Convolutional Autoencoder (Temporal DINCAE). Model performance was assessed using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), prediction bias, and the coefficient of determination (R2). Results indicate that tree-based ensemble models outperform spatial interpolation and EOF-based methods, with XGBoost achieving the best overall performance (R2 ≈ 0.86; RMSE ≈ 9.61 mg m−3). The LSTM model achieved lower prediction errors (RMSE ≈ 5.87 mg m−3; MAE ≈ 2.16 mg m−3), highlighting the benefit of incorporating temporal dependencies, although with slightly reduced variance capture. The convolutional autoencoder-based Temporal DINCAE model achieved strong reconstruction performance (R2 ≈ 0.84; RMSE ≈ 11.15 mg m−3). Uncertainty quantification shows that Extra Trees tends to underestimate uncertainty with narrower prediction intervals, whereas XGBoost provides better-calibrated but wider intervals. 
    more » « less
  3. ; ; ; ; ; ; (, Journal of Geophysical Research: Oceans)
    Abstract Ocean chlorophyll time series exhibit temporal variability on a range of timescales due to environmental change, ecological interactions, dispersal, and other factors. The differences in chlorophyll temporal variability observed at stationary locations (Eulerian perspective) or following water parcels (Lagrangian perspective) are poorly understood. Here we contrasted the temporal variability of ocean chlorophyll in these two observational perspectives, using global drifter trajectories and satellite chlorophyll to generate matched pairs of Eulerian‐Lagrangian time series. We found that for most ocean locations, chlorophyll variances measured in Eulerian and Lagrangian perspectives are not statistically different. In high latitude areas, the two perspectives may capture similar variability due to the large spatial scale of chlorophyll patches. In localized regions of the ocean, however, chlorophyll variability measured in these two perspectives may significantly differ. For example, in some western boundary currents, temporal chlorophyll variability in the Lagrangian perspective was greater than in the Eulerian perspective. In these cases, the observing platform travels rapidly across strong environmental gradients and constrained by the shelf topography, potentially leading to greater Lagrangian variability in chlorophyll. In contrast, we found that Eulerian chlorophyll variability exceeded Lagrangian variability in some key upwelling zones and boundary current extensions. In these cases, variability in the nutrient supply may generate intermittent chlorophyll anomalies in the Eulerian perspective, while the Lagrangian perspective sees the transport of such anomalies off‐shore. These findings aid with the interpretation of chlorophyll time series from different sampling methodologies, inform observational network design, and guide validation of marine ecosystem models. 
    more » « less
  4. ; ; ; ; ; (, Global Biogeochemical Cycles)
    Abstract Southern Ocean (SO) phytoplankton chlorophyll is highly variable on sub‐seasonal time scales. Although the SO is the windiest ocean basin globally, it is not conclusively understood how storms impact SO phytoplankton dynamics. Much of our existing knowledge stems from satellites, but biases due to data gaps from cloud cover and low solar angles remain unquantified. Here, we use ocean–sea‐ice simulations with the Community Earth System Model to quantify the climatological 1997–2018 imprint of storms on chlorophyll and phytoplankton dynamics in the ice‐free SO. Additionally, by comparing the full‐field model output to synthetic satellite observations, we quantify sampling biases in satellite‐derived estimates. We find that both the sign and the magnitude of the average surface chlorophyll imprint vary substantially across storms but last for at least 4 days after the storm passing. Based on our analysis, more than one third of the storms explain the majority of local non‐seasonal chlorophyll variability, but satellite‐derived storm imprints are often too large in magnitude. On the day of the storm passing, changes in vertical mixing predominantly cause surface chlorophyll anomalies, and reduced light availability due to enhanced cloud cover outweighs the enhanced nutrient availability due to entrainment. Interestingly, storms imprint differently on total net primary production than on surface chlorophyll, demonstrating the difficulty to derive carbon‐cycle impacts from a surface‐chlorophyll assessment. With SO future storm activity projected to increase, complementing satellite observations with other observing technologies, for example, profiling floats, is necessary to better constrain how storms impact biological carbon cycling in the SO. 
    more » « less
  5. ; ; ; ; (, Biogeosciences)
    Abstract. In the vast Pacific Ocean, remote islands and atolls induce mesoscale and sub-mesoscale processes that significantly impact the surrounding oligotrophic ocean, collectively referred to as the island mass effect (IME). These processes include nutrient upwelling and phytoplankton biomass enhancement around islands, creating spatial and temporal heterogeneity in biogeochemical properties. Previous algorithms developed for detecting the IME using satellite data are based on monthly or longer averages of satellite-derived chlorophyll a concentrations. As such, they tend to underestimate the true extent of this phenomenon because they do not take into account sub-mesoscale and short-term temporal variations and because of the sensitivity of the detection algorithm to single-pixel variability. Here we present a new approach that enhances satellite data recovery by merging products from multiple sensors and applying the POLYMER atmospheric correction. By integrating modeled surface currents with higher-temporal-resolution satellite observations, we dynamically track chlorophyll a enhancements associated with the IME and the advection of detached patches and filaments over distances exceeding 1000 km from their source. Our findings, applied to four island groups in the South Pacific, suggest that the ecological influence of the IME on the oligotrophic ocean is much larger than previously recognized. This work provides a foundation for improved mechanistic understanding of the IME and suggests broader implications for ocean ecology in subtropical regions. The approach developed here could also be applied in studies on biological responses to other mesoscale and sub-mesoscale processes in other parts of the world's oceans. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Text Encoded Conversion
  • XML
  • Save / Share this Record
  • Send to Email