Nikolay P. Nezlin

Chlorophyll-a (Chl-a) concentration is a key indicator of water quality and phytoplankton biomass, used to monitor algal blooms in coastal and offshore waters. Traditional Chl-a prediction methods involve a two-step process: atmospheric correction to convert top-of-the-atmosphere (TOA) data to water-leaving radiance, followed by a machine learning or band-ratio algorithm. However, atmospheric correction is challenging in optically complex waters, reducing prediction accuracy. This study introduces a novel single-step approach that directly predicts Chl-a from TOA data using machine learning, bypassing the need for atmospheric correction. An attention-based data-driven atmospheric compensation model is also applied, incorporating contextual features such as sea surface pressure and sun geometry to enhance predictions. The experiments used real-world satellite data collected over the Chesapeake Bay and the Florida coast. The experimental results in the Florida regions demonstrate that using a multilayer perceptron (MLP) with TOA data achieves a logarithmic mean absolute error (Log MAE) of 1.830, outperforming both the Log MAE of 1.843 obtained with Level-2 reflectance data and Log MAE of 4.087 from the Chlorophyll 3-band algorithm (OC3V). The data-driven atmospheric compensation model improves this further, achieving a Log MAE of 1.745. A similar trend is observed in the Chesapeake Bay region data experiments. A key advantage of TOA data is that the model can predict Chl-a in regions previously excluded in the Level-2 reflectance data due to Level-2 quality checks, thus providing broader coverage.

Monitoring chlorophyll-a concentration (Chl-a) is essential for assessing aquatic ecosystem health, yet its retrieval using remote sensing remains challenging in turbid coastal waters because of the intricate optical characteristics of these environments. Elevated levels of colored (chromophoric) dissolved organic matter (CDOM) and suspended sediments (aka total suspended solids, TSS) interfere with satellite-based Chl-a estimates, necessitating alternative approaches. One potential solution is machine learning, indirectly including non-Chl-a signals into the models. In this research, we develop machine learning models to predict Chl-a concentrations in the Chesapeake Bay, one of the largest estuaries on North America’s East Coast. Our approach leverages the Extra-Trees (ET) algorithm, a tree-based ensemble method that offers predictive accuracy comparable to that of other ensemble models, while significantly improving computational efficiency. Using the entire ocean color datasets acquired by the satellite sensors MODIS-Aqua (>20 years) and VIIRS-SNPP (>10 years), we generated long-term Chl-a estimates covering the entire Chesapeake Bay area. The models achieve a multiplicative absolute error of approximately 1.40, demonstrating reliable performance. The predicted spatiotemporal Chl-a patterns align with known ecological processes in the Chesapeake Bay, particularly those influenced by riverine inputs and seasonal variability. This research emphasizes the potential of machine learning to enhance satellite-based water quality monitoring in optically complex coastal waters, providing valuable insights for ecosystem management and conservation.

Phytoplankton growth in estuaries is regulated by a complex combination of physical factors with freshwater discharge usually playing a dominating role controlling nutrient and light availability. The role of other factors, including upwelling-generating winds, is still unclear because most estuaries are too small for upwelling to emerge. In this study, we used remotely sensed proxies of phytoplankton biomass and concentration of suspended mineral particles to compare the effect of river discharge with the effect of upwelling events associated with persistent along-channel southerly winds in the Chesapeake Bay, a large estuary where upwelling and its effects on biogeochemical dynamics have been previously reported. The surface chlorophyll-a concentrations (Chl-a) were estimated from Visible Infrared Imaging Radiometer Suite (VIIRS) satellite data using the Generalized Stacked-Constraints Model (GSCM) corrected for seasonal effects by comparing remotely sensed and field-measured data. Light limitation of phytoplankton growth was assessed from the concentration of suspended mineral particles estimated from the remotely sensed backscattering at blue (443 nm) wavelength bbp(443). The nine-year time series (2012–2020) of Chl-a and bbp(443) confirmed that a primary factor regulating phytoplankton growth in this nearshore eutrophic area is discharge from the Susquehanna River, and presumably the nutrients it delivers, with a time lag up to four months. Persistent southerly wind events (2–3 days with wind speed >4 m/s) affected the water column stratification in the central part of the bay but did not result in significant increases in remotely sensed Chl-a. Analysis of model simulations of selected upwelling-favorable wind events revealed that strong southerly winds resulted in well-defined lateral (East–West) responses but were insufficient to deliver high-nutrient water to the surface layer to support phytoplankton bloom. We conclude that, in the Chesapeake Bay, which is a large, eutrophic estuary, wind-driven upwelling of deep water plays a limited role in driving phytoplankton growth under most conditions compared with river discharge. Integr Environ Assess Manag 2022;18:921–938. © 2022 SETAC KEY POINTS River discharge is a primary factor regulating phytoplankton growth in the Chesapeake Bay. Upwelling-generating wind events were insufficient to support phytoplankton blooms. Generalized Stacked-Constraints Model (GSCM) is a useful method for processing ocean color satellite imagery in the nearshore areas.

Distinguishing between local, anthropogenic nutrient inputs and large‐scale climatic forcing as drivers of coastal phytoplankton biomass is critical to developing effective nutrient management strategies. Here we assess the relative importance of these two drivers by comparing trends in chlorophyll‐a between shallow coastal (0.1–16.5 km) and deep offshore (17–700 km) areas, hypothesizing that coastal regions influenced by anthropogenic nutrient inputs may have different spatial and temporal patterns in chlorophyll‐a concentration from offshore regions where coastal inputs are less influential. Quarterly conductivity‐temperature‐depth (CTD) fluorescence measurements collected from three southern California continental shelf regions since 1998 were compared to chlorophyll‐a data from the more offshore California Cooperative Fisheries Investigations (CalCOFI) program. The trends in the coastal zone were similar to those offshore, with a gradual increase of chlorophyll‐a biomass and shallowing of its maximum layer since the beginning of observations, followed by chlorophyll‐a declining and deepening from 2010 to present. An exception was the northern coastal part of SCB, where chlorophyll‐a continued increasing after 2010. The long‐term increase in chlorophyll‐a prior to 2010 was correlated with increased nitrate concentrations in deep waters, while the recent decline was associated with deepening of the upper mixed layer, both linked to the low‐frequency climatic cycles of the Pacific Decadal Oscillation and North Pacific Gyre Oscillation. These large‐scale factors affecting the physical structure of the water column may also influence the delivery of nutrients from deep ocean outfalls to the euphotic zone, making it difficult to distinguish the effects of anthropogenic inputs on chlorophyll along the coast.

The effect of upwelling events, storm water discharge, and local circulation on phytoplankton blooms in the central California and Southern California Bight (SCB) coastal zones was analyzed using 10+ years (1997–2007) of remotely sensed surface chlorophyll concentration (CHL, derived from SeaWiFS ocean color), sea surface temperature, and modeled freshwater discharge. Analysis of variability and factors associated with phytoplankton blooms was conducted using the offshore extension of zones of CHL > 5 mg m⁻³; this method excludes terrestrial interference that complicates the use of ocean color to investigate phytoplankton blooms in coastal waters. In the SCB, blooms were most frequent in spring and associated with the spring transition to an upwelling regime. Along the Central Coast, blooms persisted from spring to autumn during seasonal intensification of upwelling. Offshore CHL extensions showed a significant positive trend during 1997–2007, with maxima in 2000–2001 and 2005–2006 that coincided with higher than normal frequency of upwelling events. Upwelling was found to be a major factor driving phytoplankton blooms, although the standard upwelling index derived from large‐scale atmospheric circulation was decoupled from the frequencies of both upwelling events and phytoplankton blooms. Areas of longer residence time associated with natural boundaries between coastal ocean regions had more extensive and persistent blooms. The influence of storm water discharge on offshore CHL extension appeared to be limited to areas in close proximity to major river mouths. These “hot spots” were also co‐located with ocean outfalls of Publicly Owned Treatment Works (POTW) discharge and, in some cases, longer residence time of coastal waters.

Satellite-derived (TOPEX/Poseidon, Jason-1, Jason-2, SeaWIFS (Sea-Viewing Wide Field-of-View Sensor) and MODIS (Moderate Resolution Imaging Spectroradiometer) Aqua) weekly and monthly products for sea surface height, sea surface temperature (SST) and chlorophyll-a (CHL) concentration were used to analyse the trends of physical–biological coupling. In the variability of chlorophyll, no interannual trends were found for the Sea of Oman (1997–2008). However, the variation of chlorophyll within the annual cycle has increased. A similar tendency was evaluated for the variability of the energy of mesoscale eddies. The median level of kinetic energy and the coefficient of variation of this energy within the annual cycle both increased from 1997 through 2008.