Chang, Inseong, Young-Ho Kim, Young-Gyu Park, Hyunkeun Jin, Gyundo Pak, Andrew C Ross, and Robert Hallberg, January 2026: Assessing vertical coordinate system performance in the Regional Modular Ocean Model 6 configuration for Northwest Pacific. Geoscientific Model Development, 19(1), DOI:10.5194/gmd-19-187-2026187-216. Abstract
The Northwest Pacific (NWP) has a complex ocean circulation system and is among the regions most affected by climate change. To facilitate rapid responses to marine incidents and effectively address climate variability impacts, the Korea Institute of Ocean Science and Technology (KIOST) developed the Korea Operational Oceanographic System–Ocean Predictability Experiment for Marine Environment (KOOS-OPEM), a high-resolution regional ocean prediction system based on Modular Ocean Model version 5 (MOM5). In this study, the base model of KOOS-OPEM was upgraded to MOM6 to enhance its regional ocean modeling capabilities. A key advancement of MOM6 is its flexible vertical coordinate system enabled by a Lagrangian remapping system. Taking advantage of this feature, we evaluated the impact of vertical coordinate choices on model performance by comparing the HYBRID (z∗-isopycnal) and ZSTAR (z∗) configurations. Model outputs from the 2003–2012 period were assessed against multiple observational datasets and reanalysis products to determine their ability to reproduce key oceanographic features. The results indicated that HYBRID better preserved stratification and reduced spurious diapycnal mixing, significantly improving the representation of North Pacific Intermediate Water (NPIW). In contrast, ZSTAR exhibited excessive diapycnal mixing, resulting in a thicker isopycnal layer associated with NPIW and a salinity bias of approximately 0.2 psu. An idealized age tracer experiment further confirmed that ZSTAR facilitates excessive downward diffusion of younger surface waters, eroding the minimum salinity layer of the NPIW. In tidal simulations, HYBRID outperformed ZSTAR in reproducing M2 tidal amplitudes in the Yellow Sea, where stratification plays a key role. Conversely, ZSTAR underestimated these amplitudes due to its limitations in representing stratification. Despite its advantages, HYBRID underperformed in high-latitude regions, exhibiting larger temperature and salinity biases between 100 and 600 m depth, with temperature biases reaching approximately −1 °C. This discrepancy arose because HYBRID maintained fewer active layers in weakly stratified regions, reducing vertical resolution and leading to errors in water mass representation. To mitigate these issues and improve HYBRID's performance in high-latitude regions, adjustments to target density profiles are necessary. In addition, both configurations showed limitations in simulating winter SST, largely due to insufficient vertical resolution in the surface mixed layer. To address these issues, adopting a finer surface layer resolution (e.g., 1 m instead of 2 m) will further enhance the model's representation of mixed-layer processes.
Regional ocean models enable the generation of computationally affordable and regionally tailored ensembles of near-term forecasts and long-term projections of sufficient resolution to serve marine resource management. Climate change, however, has created marine resource challenges, such as shifting stock distributions, that cut across domestic and international management boundaries and have pushed regional modeling efforts toward “coastwide” approaches. Here, we present and evaluate a multidecadal hindcast with a Northeast Pacific regional implementation of the Modular Ocean Model, version 6, with sea ice and biogeochemistry that extends from the Chukchi Sea to the Baja California Peninsula at 10 km horizontal resolution (MOM6-COBALT-NEP10k, or NEP10k). This domain includes an Arctic-adjacent system with a broad, shallow shelf seasonally covered by sea ice (the eastern Bering Sea), a sub-Arctic system with upwelling in the Alaska Gyre and predominant downwelling winds and large freshwater forcing along the coast (the Gulf of Alaska), and a temperate, eastern boundary upwelling ecosystem (the California Current Ecosystem). The coastwide model was able to recreate seasonal and cross-ecosystem contrasts in numerous ecosystem-critical properties including temperature, salinity, inorganic nutrients, oxygen, carbonate saturation states, and chlorophyll. Spatial consistency between modeled quantities and observations generally extended to plankton ecosystems, though small to moderate biases were also apparent. Fidelity with observed zooplankton biomass, for example, was limited to first-order seasonal and cross-system contrasts. Temporally, simulated monthly surface and bottom temperature anomalies in coastal regions (<500 m deep) closely matched estimates from data-assimilative ocean reanalyses. Performance, however, was reduced in some nearshore regions coarsely resolved by the model's 10 km resolution grid and for point measurements. The time series of satellite-based chlorophyll anomaly estimates proved more difficult to match than temperature. System-specific ecosystem indicators were also assessed. In the eastern Bering Sea, NEP10k robustly matched observed variations, including recent large declines, in the area of the summer bottom water “cold pool” (<2 °C), which exerts a profound influence on eastern Bering Sea fisheries. In the Gulf of Alaska, the simulation captured patterns of sea surface height variability and variations in thermal, oxygen, and acidification risk associated with local modes of interannual to decadal climate variability. In the California Current Ecosystem, the simulation robustly captured variations in upwelling indices and coastal water masses, though discrepancies in the latter were evident in the Southern California Bight. Enhanced model resolution may reduce such discrepancies, but any benefits must be carefully weighed against computational costs given the intended use of this system for ensemble predictions and projections. Meanwhile, the demonstrated NEP10k skill level herein, particularly in recreating cross-ecosystem contrasts and the time variation of ecosystem indicators over multiple decades, suggests considerable immediate utility for coastwide retrospective and predictive applications.
Liao, Enhui, Laure Resplandy, Fan Yang, Yangyang Zhao, Sam J Ditkovsky, Manon Malsang, Jenna Pearson, Andrew C Ross, Robert Hallberg, and Charles A Stock, September 2025: A high-resolution physical-biogeochemical model for marine resource applications in the Northern Indian Ocean (MOM6-COBALT-IND12 v1.0). Geoscientific Model Development, 18(18), DOI:10.5194/gmd-18-6553-2025. Abstract
We introduce and evaluate the regional ocean model MOM6-COBALT-IND12 version 1 coupling the MOM6 ocean dynamics model to the Carbon, Ocean Biogeochemistry and Lower Trophics (COBALT) biogeochemical model at a horizontal resolution of °. The model covers the northern Indian Ocean (from 8.6° S to the northern continental boundaries), central to the livelihoods and economies of countries that comprise about one-third of the world’s population. We demonstrate that the model effectively captures the key physical and biogeochemical basin-scale features related to seasonal monsoon reversal, interannual Indian Ocean Dipole and multi-decadal variability, as well as intraseasonal and fine-scale variability (e.g., eddies and planetary waves), which are all essential for accurately simulating patterns of coastal upwelling, primary productivity, temperature, salinity, and oxygen levels. Well represented features include the timing and amplitude of the monsoonal blooms triggered by summer coastal upwelling and winter mixing, the strong contrast between the high evaporation/high salinity Arabian Sea and high precipitation/high runoff/low salinity Bay of Bengal, the seasonality of the Great Whirl gyre and coastal Kelvin upwelling/downwelling waves, as well as the physical and biogeochemical patterns associated with intraseasonal and interannual variability. Quantitatively, the model exhibits relatively small biases, as reflected by root mean square error (RMSE) values in key variables: sea surface temperature (0.25–0.3 °C), mixed layer depth (7–8.09 m), sea level anomaly (0.02 m), sea surface salinity (0.53–0.71 psu), vertical chlorophyll (0.03–0.3 mg m−3), subsurface temperature (0.33 °C), and subsurface salinity (0.07 psu). A major model bias (16 µmol kg−1 of oxygen) is the larger oxygen minimum zone simulated in the Bay of Bengal, a common challenge of ocean and Earth system models in this region. This bias was partly mitigated by improving the representation of the export and burial of organic detritus to the deep ocean (e.g., sinking speed, riverine lithogenic material inputs that protect organic material and burial fraction), and water-column denitrification (e.g., nitrate-based respiration at higher oxygen levels) using observational constraints. These results indicate that the regional MOM6-COBALT-IND12 v1.0 model is well suited for physical and biogeochemical studies on timescales ranging from weeks to decades, in addition to supporting marine resource applications and management in the northern Indian Ocean.
Chlorophyll underpins ocean productivity yet simulating chlorophyll across biomes, seasons and depths remains challenging for earth system models. Inconsistencies are often attributed to misrepresentation of the myriad nutrient supply, growth and loss processes that govern phytoplankton biomass. They may also arise, however, from unresolved or misspecified photoacclimation or photoadaptation responses. A series of global ocean ecosystem simulations were conducted to assess these latter sensitivities: alternative photoacclimation schemes implicitly modulated investments in light harvesting versus photodamage avoidance and other cellular functions. Photoadaptation experiments probed the impact of adding low- and high-light adapted phytoplankton ecotypes. Results showed that photoacclimation and photoadaptation alternatives generate chlorophyll differences exceeding a factor of 2 in some regions and seasons. In stratified waters, photoadaptation and acclimation to light levels over mixing depths consistent with the timescale of photoadaptation (days) benefitted model performance. In regions and seasons with deep mixed layers, surface-skewed photoacclimation yielded improved fidelity across satellite chlorophyll products. Large photoacclimation-driven differences in chlorophyll concentration had small impacts on primary productivity and carbon export, unlike those arising from changes in the nutrient supply. Improved photoacclimation and photoadaption constraints are thus needed to reduce ambiguities in the drivers of chlorophyll change and their biogeochemical implications.
Using tide gauge (TG) observations, we identify pronounced multidecadal fluctuations in sea level along the US Northeast Coast (USNEC) superimposed on a long-term increasing trend. This multidecadal sea level variability, largely arising from fluctuations in the buoyancy-driven Atlantic meridional overturning circulation (AMOC), substantially modulates the frequency of flood occurrences along the USNEC and serves as a source of multiyear predictability. Using an initialized dynamical downscaling decadal prediction system with a 1/12° ocean resolution, we demonstrate that flood frequency along the USNEC can be predicted on multiyear to decadal timescales. The long-term increasing trend in flood frequency, mainly driven by increasing greenhouse gases and associated radiative forcing changes, can be predicted a decade ahead. Furthermore, detrended flood frequency along the USNEC exhibits prediction skill for up to 3 years, as verified by TG observation. This multiyear prediction skill is achieved using prediction models that are initialized from our best estimate of observed AMOC.
Balwada, Dhruv, Ryan Abernathey, Shantanu Acharya, Alistair Adcroft, Judith Brener, V Balaji, Mohamed Aziz Bhouri, Joan Bruna, Mitchell Bushuk, Will Chapman, Alex Connolly, Julie Deshayes, Carlos Fernandez-Granda, Pierre Gentine, Anastasiia Gorbunova, William Gregory, Arthur Guillaumin, Shubham Gupta, Marika M Holland, J Emmanuel Johnsson, Julien Le Sommer, Ziwei Li, Nora Loose, Feiyu Lu, Paul A O'Gorman, Pavel Perezhogin, Brandon G Reichl, Andrew C Ross, Aakash Sane, Sara Shamekh, Tarun Verma, Janni Yuval, Lorenzo Zampieri, Cheng Zhang, and Laure Zanna, December 2024: Learning machine learning with Lorenz-96. Journal of Open Source Education, 7(82), DOI:10.21105/jose.00241. Abstract
Machine learning (ML) is a rapidly growing field that is starting to touch all aspects of our lives, and science is not immune to this. In fact, recent
work in the field of scientific ML, i.e. combining ML and with conventional scientific problems, is leading to new breakthroughs in notoriously hard problems, which might have seemed too distant till a few years ago. One such age-old problem is that of turbulence closures in fluid flows. This closure or parameterization problem is particularly relevant for environmental fluids, which span a large range of scales from the size of the planet down to millimeters, and remains a big challenge in the way of improving forecasts of weather and projections of climate.
The capability to anticipate the exceptionally rapid warming of the Northwest Atlantic Shelf and its evolution over the next decade could enable effective mitigation for coastal communities and marine resources. However, global climate models have struggled to accurately predict this warming due to limited resolution; and past regional downscaling efforts focused on multi-decadal projections, neglecting predictive skill associated with internal variability. We address these gaps with a high resolution (1/12°) ensemble of dynamically downscaled decadal predictions. The downscaled simulations accurately predicted past oceanic variability at scales relevant to marine resource management, with skill typically exceeding global coarse-resolution predictions. Over the long term, warming of the Shelf is projected to continue; however, we forecast a temporary warming pause in the next decade. This predicted pause is attributed to internal variability associated with a transient, moderate strengthening of the Atlantic meridional overturning circulation and a southward shift of the Gulf Stream.
Using a 1/12° regional model of the Northwest Atlantic Ocean (MOM6-NWA12), we downscale an ensemble of retrospective seasonal forecasts from a 1° global forecast model. To evaluate whether downscaling improved the forecast skill for surface temperature and salinity and bottom temperature, the global and downscaled forecasts are compared with each other and with a reference forecast of persistence using anomaly correlation. Both sets of forecasts are also evaluated on the basis of mean bias and ensemble spread. We find that downscaling significantly improved the forecast skill for monthly sea surface temperature anomalies in the Northeast US Large Marine Ecosystem, a region that global models have historically struggled to skillfully predict. The downscaled sea surface temperature (SST) predictions for this region were also more skillful than the persistence baseline across most initialization months and lead times. Although some of the SST prediction skill in this region stems from the recent rapid warming trend, prediction skill above persistence is generally maintained after removing the contribution of the trend, and patterns of skill suggestive of predictable processes are also preserved. While downscaling mainly improved the SST anomaly prediction skill in the Northeast US region, it improved bottom temperature and sea surface salinity anomaly skill across many of the marine ecosystems along the North American east coast. Although improvements in anomaly prediction via downscaling were ubiquitous, the effects of downscaling on prediction bias were mixed. Downscaling generally reduced the mean surface salinity biases found in the global model, particularly in regions with sharp salinity gradients (the Northern Gulf of Mexico and the Northeast US). In some cases, however, downscaling amplified the surface and bottom temperature biases found in the global predictions. We discuss several processes that are better resolved in the regional model and contribute to the improved skill, including the autumn reemergence of temperature anomalies and advection of water masses by coastal currents. Overall, the results show that a downscaled high-resolution model can produce improved seasonal forecast skill by representing fine-scale processes that drive predictability.
Using a recently developed 1/12th degree regional ocean model, we establish a link between U.S. East Coast sea level variability and offshore upper ocean heat content change. This link manifests as a cross-shore mass redistribution driven by an offshore thermosteric sea level response to subsurface warming or cooling. Approximately 50% of simulated monthly to interannual coastal sea level variance south of Cape Hatteras can be statistically accounted for by this mechanism, realized as a function of regional ocean hypsometry, gyre scale warming, and the depth dependence of density change. This response to offshore warming explains the nonstationarity of U.S. East Coast sea level covariance, a specifically observed and modeled behavior after ~ 2010. Since approximately 2010, elevated rates of sea level rise south of Cape Hatteras can be partly explained as the result of shoreward mass redistribution due to offshore subsurface warming within the North Atlantic subtropical gyre. These results reveal a mechanism that connects local coastal sea level to a broader region and identifies the influence of regional heat content changes on coastal sea level. This analysis presents a framework for identifying new regions that may be susceptible to enhanced sea level rise due to ocean warming and helps bridge the gap between quantifying large scale change and anticipating local coastal impacts that can make flooding and storm surge more acutely damaging.
Amaya, Dillon J., Michael G Jacox, Melanie R Fewings, Vincent S Saba, Malte F Stuecker, Ryan R Rykaczewski, Andrew C Ross, and Charles A Stock, et al., April 2023: Marine heatwaves need clear definitions so coastal communities can adapt. Nature, 616, DOI:10.1038/d41586-023-00924-2.
Gomez, Fabien A., Sang-Ki Lee, Charles A Stock, Andrew C Ross, Laure Resplandy, Samantha A Siedlecki, Filippos Tagklis, and Joseph E Salisbury, June 2023: RC4USCoast: a river chemistry dataset for regional ocean model applications in the US East Coast, Gulf of Mexico, and US West Coast. Earth System Science Data, 15(5), DOI:10.5194/essd-15-2223-20232223-2234. Abstract
A historical dataset of river chemistry and discharge is presented for 140 monitoring sites along the US East Coast, the Gulf of Mexico, and the US West Coast from 1950 to 2022. The dataset, referred to here as River Chemistry for the U.S. Coast (RC4USCoast), is mostly derived from the Water Quality Database of the US Geological Survey (USGS) but also includes river discharge from the USGS's Surface-Water Monthly Statistics for the Nation and the U.S. Army Corps of Engineers. RC4USCoast provides monthly time series as well as long-term averaged monthly climatological patterns for 21 variables including alkalinity and dissolved inorganic carbon concentration. It is mainly intended as a data product for regional ocean biogeochemical models and carbonate chemistry studies in the US coastal regions. Here we present the method to derive RC4USCoast and briefly describe the rivers' carbonate chemistry patterns. The dataset is publicly available at https://doi.org/10.25921/9jfw-ph50 (Gomez et al., 2022).
We present the development and evaluation of MOM6-COBALT-NWA12 version 1.0, a 1/12∘ model of ocean dynamics and biogeochemistry in the northwest Atlantic Ocean. This model is built using the new regional capabilities in the MOM6 ocean model and is coupled with the Carbon, Ocean Biogeochemistry and Lower Trophics (COBALT) biogeochemical model and Sea Ice Simulator version-2 (SIS2) sea ice model. Our goal was to develop a model to provide information to support living-marine-resource applications across management time horizons from seasons to decades. To do this, we struck a balance between a broad, coastwide domain to simulate basin-scale variability and capture cross-boundary issues expected under climate change; a high enough spatial resolution to accurately simulate features like the Gulf Stream separation and advection of water masses through finer-scale coastal features; and the computational economy required to run the long simulations of multiple ensemble members that are needed to quantify prediction uncertainties and produce actionable information. We assess whether MOM6-COBALT-NWA12 is capable of supporting the intended applications by evaluating the model with three categories of metrics: basin-wide indicators of the model's performance, indicators of coastal ecosystem variability and the regional ocean features that drive it, and model run times and computational efficiency. Overall, both the basin-wide and the regional ecosystem-relevant indicators are simulated well by the model. Where notable model biases and errors are present in both types of indicator, they are mainly consistent with the challenges of accurately simulating the Gulf Stream separation, path, and variability: for example, the coastal ocean and shelf north of Cape Hatteras are too warm and salty and have minor biogeochemical biases. During model development, we identified a few model parameters that exerted a notable influence on the model solution, including the horizontal viscosity, mixed-layer restratification, and tidal self-attraction and loading, which we discuss briefly. The computational performance of the model is adequate to support running numerous long simulations, even with the inclusion of coupled biogeochemistry with 40 additional tracers. Overall, these results show that this first version of a regional MOM6 model for the northwest Atlantic Ocean is capable of efficiently and accurately simulating historical basin-wide and regional mean conditions and variability, laying the groundwork for future studies to analyze this variability in detail, develop and improve parameterizations and model components to better capture local ocean features, and develop predictions and projections of future conditions to support living-marine-resource applications across timescales.
Ross, Andrew C., and Charles A Stock, October 2022: Probabilistic extreme SST and marine heatwave forecasts in Chesapeake Bay: A forecast model, skill assessment, and potential value. Frontiers in Marine Science, 9:896961, DOI:10.3389/fmars.2022.896961. Abstract
We test whether skillful 35-day probabilistic forecasts of estuarine sea surface temperature (SST) are possible and whether these forecasts could potentially be used to reduce the economic damages associated with extreme SST events. Using an ensemble of 35-day retrospective forecasts of atmospheric temperature and a simple model that predicts daily mean SST from past SST and forecast atmospheric temperature, we create an equivalent ensemble of retrospective SST forecasts. We compare these SST forecasts with reference forecasts of climatology and damped persistence and find that the SST forecasts are skillful for up to two weeks in the summer. Then, we post-process the forecasts using nonhomogeneous Gaussian regression and assess whether the resulting calibrated probabilistic forecasts are more accurate than the probability implied by the raw model ensemble. Finally, we use an idealized framework to assess whether these probabilistic forecasts can valuably inform decisions to take protective action to mitigate the effects of extreme temperatures and heatwaves. We find that the probabilistic forecasts provide value relative to a naive climatological forecast for 1-2 weeks of lead time, and the value is particularly high in cases where the cost of protection is small relative to the preventable losses suffered when a heatwave occurs. In most cases, the calibrated probabilistic forecasts are also more valuable than deterministic forecasts based on the ensemble mean and naive probabilistic forecasts based on damped persistence. Probabilistic SST forecasts could provide substantial value if applied to adaptively manage the rapid impacts of extreme SSTs, including managing the risks of catch-and-release mortality in fish and Vibrio bacteria in oysters.
Efforts to manage living marine resources (LMRs) under climate change need projections of future ocean conditions, yet most global climate models (GCMs) poorly represent critical coastal habitats. GCM utility for LMR applications will increase with higher spatial resolution but obstacles including computational and data storage costs, obstinate regional biases, and formulations prioritizing global robustness over regional skill will persist. Downscaling can help address GCM limitations, but significant improvements are needed to robustly support LMR science and management. We synthesize past ocean downscaling efforts to suggest a protocol to achieve this goal. The protocol emphasizes LMR-driven design to ensure delivery of decision-relevant information. It prioritizes ensembles of downscaled projections spanning the range of ocean futures with durations long enough to capture climate change signals. This demands judicious resolution refinement, with pragmatic consideration for LMR-essential ocean features superseding theoretical investigation. Statistical downscaling can complement dynamical approaches in building these ensembles. Inconsistent use of bias correction indicates a need for objective best practices. Application of the suggested protocol should yield regional ocean projections that, with effective dissemination and translation to decision-relevant analytics, can robustly support LMR science and management under climate change.
Ross, Andrew C., R G Najjar, and Ming Li, January 2021: A Metamodel-Based Analysis of the Sensitivity and Uncertainty of the Response of Chesapeake Bay Salinity and Circulation to Projected Climate Change. Estuaries and Coasts, 44, DOI:10.1007/s12237-020-00761-w. Abstract
Numerical models are often used to simulate estuarine physics and water quality under scenarios of future climate conditions. However, representing the wide range of uncertainty about future climate often requires an infeasible number of computationally expensive model simulations. Here, we develop and test a computationally inexpensive statistical model, or metamodel, as a surrogate for numerical model simulations. We show that a metamodel fit using only 12 numerical model simulations of Chesapeake Bay can accurately predict the early summer mean salinity, stratification, and circulation simulated by the numerical model given the input sea level, winter–spring streamflow, and tidal amplitude along the shelf. We then use this metamodel to simulate summer salinity and circulation under sampled probability distributions of projected future mean sea level, streamflow, and tidal amplitudes. The simulations from the metamodel show that future salinity, stratification, and circulation are all likely to be higher than present-day averages. We also use the metamodel to quantify how uncertainty about the model inputs transfers to uncertainty in the output and find that the model projections of salinity and stratification are highly sensitive to uncertainty about future tidal amplitudes along the shelf. This study shows that metamodels are a promising approach for robustly estimating the impacts of future climate change on estuaries.
Most present forecast systems for estuaries predict conditions for only a few days into the future. However, there are many reasons to expect that skillful estuarine forecasts are possible for longer time periods, including increasingly skillful extended atmospheric forecasts, the potential for lasting impacts of atmospheric forcing on estuarine conditions, and the predictability of tidal cycles. In this study, we test whether skillful estuarine forecasts are possible for up to 35 days into the future by combining an estuarine model of Chesapeake Bay with 35‐day atmospheric forecasts from an operational weather model. When compared with both a hindcast simulation from the same estuarine model and with observations, the estuarine forecasts for surface water temperature are skillful up to about two weeks into the future, and the forecasts for bottom temperature, surface and bottom salinity, and density stratification are skillful for all or the majority of the forecast period. Bottom oxygen forecasts are skillful when compared to the model hindcast, but not when compared with observations. We also find that skill for all variables in the estuary can be improved by taking the mean of multiple estuarine forecasts driven by an ensemble of atmospheric forecasts. Finally, we examine the forecasts in detail using two case studies of extreme events, and we discuss opportunities for improving the forecast skill.
Ni, Wenfei, Ming Li, Andrew C Ross, and R G Najjar, November 2019: Large Projected Decline in Dissolved Oxygen in a Eutrophic Estuary Due to Climate Change. Journal of Geophysical Research: Oceans, 124(11), DOI:10.1029/2019JC015274. Abstract
Climate change is known to cause deoxygenation in the open ocean, but its effects on eutrophic and seasonally hypoxic estuaries and coastal oceans are less clear. Using Chesapeake Bay as a study site, we conducted climate downscaling projections for dissolved oxygen and found that the hypoxic and anoxic volumes would increase by 10–30% between the late 20th and mid‐21st century. A budget analysis of dissolved oxygen in the bottom water revealed differing physical and biogeochemical responses to climate change. Sea level rise and larger winter‐spring runoff led to stronger stratification and large reductions in the vertical oxygen supply to the bottom water. On the other hand, warming led to earlier initiation of hypoxia, accompanied by weaker summer respiration and more rapid termination of hypoxia. Decreasing solubility due to warming accounted for about 50% of the reduction in the bottom‐water oxygen content.
Ross, Andrew C., and Charles A Stock, May 2019: An assessment of the predictability of column minimum dissolved oxygen concentrations in Chesapeake Bay using a machine learning model. Estuarine, Coastal and Shelf Science, 221, DOI:10.1016/j.ecss.2019.03.007. Abstract
Subseasonal to seasonal forecasts have the potential to be a useful tool for managing estuarine fisheries and water quality, and with increasing skill at forecasting conditions at these time scales in the atmosphere and open ocean, skillful forecasts of estuarine salinity, temperature, and biogeochemistry may be possible. In this study, we use a machine learning model to assess the predictability of column minimum dissolved oxygen in Chesapeake Bay at a monthly time scale. Compared to previous models for dissolved oxygen and hypoxia, our model has the advantages of resolving spatial variability and fitting more flexible relationships between dissolved oxygen and the predictor variables. Using a concise set of predictors with established relationships with dissolved oxygen, we find that dissolved oxygen in a given month can be skillfully predicted with knowledge of stratification and mean temperature during the same month. Furthermore, the predictions generated by the model are consistent with expectations from prior knowledge and basic physics. The model reveals that accurate knowledge or skillful forecasts of the vertical density gradient is the key to successful prediction of dissolved oxygen, and prediction skill disappears if stratification is only known at the beginning of the forecast. The lost skill cannot be recovered by replacing stratification as a predictor with variables that have a lagged correlation with stratification (such as river discharge); however, skill is obtainable in many cases if stratification can be forecast with an error of less than about 1 kg m−3. Thus, future research on hypoxia forecasting should focus on understanding and forecasting variations in stratification over subseasonal time scales (between about two weeks and two months).
