Forecasting future estuarine hypoxia using a wavelet based neural network model

Muller, Andrew C.; Muller, Diana Lynn

doi:10.1016/j.ocemod.2015.11.003

Cited by 18 publications

(8 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Ref. [44] successfully trained neural networks while using a cross-wavelet analysis technique and several decades of yearly-averaged hypoxic volumes in Chesapeake Bay. Ref.…”

Section: Discussionmentioning

confidence: 99%

Machine Learning Based Predictions of Dissolved Oxygen in a Small Coastal Embayment

Valera

Walter

Bailey

et al. 2020

JMSE

View full text Add to dashboard Cite

Coastal dissolved oxygen (DO) concentrations have a profound impact on nearshore ecosystems and, in recent years, there has been an increased prevalance of low DO hypoxic events that negatively impact nearshore organisms. Even with advanced numerical models, accurate prediction of coastal DO variability is challenging and computationally expensive. Here, we apply machine learning techniques in order to reconstruct and predict nearshore DO concentrations in a small coastal embayment while using a comprehensive set of nearshore and offshore measurements and easily measured input (training) parameters. We show that both random forest regression (RFR) and support vector regression (SVR) models accurately reproduce both the offshore DO and nearshore DO with extremely high accuracy. In general, RFR consistently peformed slightly better than SVR, the latter of which was more difficult to tune and took longer to train. Although each of the nearshore datasets were able to accurately predict DO values using training data from the same site, the model only had moderate success when using training data from one site to predict DO at another site, which was likely due to the the complexities in the underlying dynamics across the sites. We also show that high accuracy can be achieved with relatively little training data, highlighting a potential application for correcting time series with missing DO data due to quality control or sensor issues. This work establishes the ability of machine learning models to accurately reproduce DO concentrations in both offshore and nearshore coastal waters, with important implications for the ability to detect and indirectly measure coastal hypoxic events in near real-time. Future work should explore the ability of machine learning models in order to accurately forecast hypoxic events.

show abstract

“…Ref. [44] successfully trained neural networks while using a cross-wavelet analysis technique and several decades of yearly-averaged hypoxic volumes in Chesapeake Bay. Ref.…”

Section: Discussionmentioning

confidence: 99%

Machine Learning Based Predictions of Dissolved Oxygen in a Small Coastal Embayment

Valera

Walter

Bailey

et al. 2020

JMSE

View full text Add to dashboard Cite

show abstract

“…Surprisingly however, Cerco and Noel (2013), comparing with themselves twenty years before (Cerco and Cole, 1993), note that the huge complexification of the model of the Chesapeake Bay has more succeeded in lightening some dark regions of the model by new state variables than in improving substantially the goodness of fit of the basic descriptors! This frustrating apparent limit to the accuracy of deterministic models has regularly pushed some authors to prefer purely empirical models, either statistical (Gowen et al, 1992;Edwards et al, 2003;Tamvakis et al, 2012) or, more recently, based on neural networks (Melesse et al, 2008;Muller and Muller, 2015). The lack of internal processes however limits the application of these models to scenarios or sites not very different from the case study where the calibration has been done, and prevents them to forecast situations where a hidden variable (not explicitly taken as an input of the model) has changed, for instance in the perspective of climate change.…”

Section: The Weakness Of Simulation: What Is Not Yet Modelled? Challenges and Gapsmentioning

confidence: 99%

Modelling the marine eutrophication: A review

Ménesguen¹,

Lacroix²

2018

Science of The Total Environment

View full text Add to dashboard Cite

In the frame of a national, joint scientific appraisal, 45 scientific French-speaking experts have been mandated in 2015-2016 by the French ministries of Environment and Agriculture to perform a global review of scientific literature dealing with the eutrophication phenomenon, in freshwater as well as in marine waters. This paper summarizes the main results of this review restricted to a sub-domain, the modelling approach of the marine eutrophication. After recalling the different aims pursued, an overview is given on the historical time course of this modelling effort, its world distribution and the various tools used. Then, the main results obtained are examined, highlighting the specific strengths and weaknesses of the present models. Needs for future improvement are then listed.

show abstract

“…A zero diffusion flux for particulate matters at the sediment-water interface has been assumed. The transfer of diffusion flux from sediment to water occurs under anoxic condition, and the threshold value of the bottom layer oxygen has been considered as 2 g/m 3 [53].…”

Section: Diffusion Flux At Sediment-water Interfacementioning

confidence: 99%

“…Discrepancies are considerably high at the surface owing to environmental forcing such as solar radiation and mix layer thickness. Severe hypoxia, which is typically defined as DO concentration below 2 g/m 3 [53], especially appears during summer in Tokyo Bay. Seasonal hypoxia at the three locations in Tokyo Bay has been well reproduced qualitatively, though it was underestimated to some extent (Figure 17).…”

Section: Water Qualitymentioning

confidence: 99%

Numerical Modeling of Long-Term Biogeochemical Processes and Its Application to Sedimentary Bed Formation in Tokyo Bay

Amunugama

Sasaki

2018

Water

View full text Add to dashboard Cite

Even though models of water quality have become increasingly detailed over time, their applicability to analyze long-term effect on sedimentary bed formation is yet to be clarified. Hence, an integrated, layer-resolved, process-based, sediment-water coupled, long-term robust, three-dimensional (3D) ecosystem model, including realistic sedimentary and pelagic processes, was developed. The constituents of the integrated model included a multi layered ecosystem model, a quasi-three dimensional hydrodynamic model, a wave hindcasting model, and a sediment deposition and erosion model. Because numerical modeling difficulties arise in setting the initial conditions, especially for a sedimentary bed, this novel modeling approach suggests to initiate the model with the simplest initial conditions of no particulate organic carbon content (POCC) and uniform inorganic sediment distribution. The model was applied to Tokyo Bay and quasi-steady state POCC in the sediment was obtained through long-term computation with realistic sedimentary and pelagic processes. Wave induced bed shear stress (WBSS) is critical for sediment erosion and spatial movement of sediment pollution, while a stable morphology is determined by the balance between the current induced bed shear stress (CBSS) and WBSS in Tokyo Bay. This novel modeling approach, with the simplest sedimentary initial conditions and realistic sedimentary and pelagic processes, provides a great tool for long-term ecosystem modeling in future studies.

show abstract

Forecasting future estuarine hypoxia using a wavelet based neural network model

Cited by 18 publications

References 76 publications

Machine Learning Based Predictions of Dissolved Oxygen in a Small Coastal Embayment

Machine Learning Based Predictions of Dissolved Oxygen in a Small Coastal Embayment

Modelling the marine eutrophication: A review

Numerical Modeling of Long-Term Biogeochemical Processes and Its Application to Sedimentary Bed Formation in Tokyo Bay

Contact Info

Product

Resources

About