Integrating field observations and process-based modeling to predict watershed water quality under environmental perturbations

Lee, Raymond M.; Dwivedi, Dipankar; Son, Kyongho; Fang, Yilin; Graham, Emily B.; Stegen, James C.; Fisher, Joshua B.; Moulton, D.; Scheibe, Timothy D.

doi:10.1016/j.jhydrol.2020.125762

Cited by 30 publications

(18 citation statements)

References 314 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Over the past several decades, the complexity of rivers and their adjacent environments and the important roles that they play in watershed function have been increasingly recognized (Harvey and Gooseff, 2015;Chen et al, 2020;Dwivedi et al, 2018b). The large number of contributions to this Research Topic reflect continued high interest in understanding and quantifying the interactions between hydrological, microbial, and biogeochemical processes that underlie ecological health and water quality in these critical systems across spatiotemporal scales.…”

Section: Discussionmentioning

confidence: 99%

Editorial: Linking Hydrological and Biogeochemical Processes in Riparian Corridors

2021

Self Cite

View full text Add to dashboard Cite

The riparian corridor is a key component of the critical zone and an essential component of watershed systems. According to Merriam-Webster Dictionary, the word riparian is derived from the Latin word riparius, meaning "existing alongside a river." Riparian corridors typically extend from a few meters to hundreds of meters adjacent to a river and are marked by rich biodiversity, vegetation, and intense biogeochemical activity. They act as integrators of watershed processes and constitute the primary pathways for the subsurface geochemical exports from the watershed. Although riparian corridors comprise only 2-10% of a watershed's area, as much as 90-98% of biogeochemical processing in watersheds occurs in this region, thereby affecting the subsurface geochemical exports and downstream river water quality. Indeed, the riparian corridor is a good example of the Pareto principle (e.g., Dwivedi et al., 2018a;McClain et al., 2003;Bernhardt et al., 2017). This outsized contribution occurs at the interface between aquatic (river) and terrestrial (land) environments, where interactions between hydrologic and biogeochemical processes are intensified. Variations in the river corridor over time can thus also have outsize impacts. Therefore, it is important to understand the hydrological and biogeochemical linkages in riparian corridors to determine water availability and quality for sustainable management.Riparian corridors include various subsystems, such as hyporheic zones, meanders, wetlands, and lagoons, all of which impact river water quality (Figure 1). These subsystems demonstrate distinct biogeochemical potential depending upon their hydrologic connectivity to the main channel. However, several hurdles must be overcome to improve the predictive capability of riparian corridor processes across scales. This Research Topic aimed to enhance our understanding and predictive capability related to linked hydrological and biogeochemical processes in riparian corridors. We received contributions across a wide spectrum of topics, including hydrologic exchange and river connectivity as well as geochemical exports of carbon, nitrogen, colloids, and microbial dynamics (Figure 2). These topics also involved novel method development, new observational networks, advanced mechanistic modeling, and the use of artificial intelligence and machine learning approaches. Below, we briefly synthesize these contributions under two groups focused on dynamic hydrologic connectivity and microbial and physical controls on spatial patterns in river corridors.

show abstract

Section: Discussionmentioning

confidence: 99%

Editorial: Linking Hydrological and Biogeochemical Processes in Riparian Corridors

2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…The iteration between model and experiments (ModEx) is essential to improving the predictability of watershed models under both baseline and perturbed conditions 1 . The increase in model complexity and data volume has led to substantial increase in computational cost and exponential increase in data dimensionality that have both hampered ModEx.…”

Section: Rationalementioning

confidence: 99%

Building Intelligent Cyberinfrastructure to Learn Iteratively from both Observations and Models for Understanding Watershed Dynamics

Chen,

Mishra,

Fisher

et al. 2021

Self Cite

View full text Add to dashboard Cite

“…Permeability plays a crucial role in the predictive understanding of subsurface systems [1][2][3][4]. Characterizing spatially distributed permeability with reasonably good accuracy is a significant challenge in the Earth and energy sciences [5,6].…”

Section: Introductionmentioning

confidence: 99%

“…Characterizing spatially distributed permeability with reasonably good accuracy is a significant challenge in the Earth and energy sciences [5,6]. Good accuracy is needed because fluid flow and reactive-transport in applications such as hydrology [3], hydro-biogeochemistry [2,7,8], unconventional oil and gas [9][10][11], CO 2 sequestration [12][13][14], nuclear waste disposal and storage [15,16], and geothermal systems are controlled by permeability [17][18][19]. Due to the subsurface's heterogeneous nature, much uncertainty is involved in estimating the three-dimensional (3D) permeability field.…”

Section: Introductionmentioning

confidence: 99%

Deep Learning to Estimate Permeability using Geophysical Data

Mudunuru¹,

Cromwell²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Time-lapse electrical resistivity tomography (ERT) is a popular geophysical method to estimate three-dimensional (3D) permeability fields from electrical potential difference measurements. Traditional inversion and data assimilation methods are used to ingest this ERT data into hydrogeophysical models to estimate permeability. Due to ill-posedness and the curse of dimensionality, existing inversion strategies provide poor estimates and low resolution of the 3D permeability field. Recent advances in deep learning provide us with powerful algorithms to overcome this challenge. This paper presents a deep learning (DL) framework to estimate the 3D subsurface permeability from time-lapse ERT data. To test the feasibility of the proposed framework, we train DL-enabled inverse models on simulation data. Subsurface process models based on hydrogeophysics are used to generate this synthetic data for deep learning analyses. Training performed on limited simulation data resulted in the DL model overfitting. An advanced data augmentation based on mixup is implemented to generate additional training samples to overcome this issue. This mixup technique creates weakly labeled (low-fidelity) samples from strongly labeled (high-fidelity) data. The weakly labeled training data is then used to develop DL-enabled inverse models and reduce over-fitting. As the data samples are from a high-dimensional space, unsupervised learning based on principal component analysis is employed to reduce dimensionality. A deep neural network is then trained to map the encoded ERT to encoded permeability. This mixup training and unsupervised learning allowed us to build a fast and reasonably accurate DL-based inverse model under limited simulation data. Results show that proposed weak supervised learning can capture salient spatial features in the 3D permeability field. Quantitatively, the average mean squared error (in terms of the natural log) on the strongly labeled training, validation, and test datasets is less than 0.5. The R 2 -score (global metric) is greater than 0.75, and the percent error in each cell (local metric) is less than 10%. Finally, an added benefit in terms of computational cost is that the proposed DL-based inverse model is at least O(10 4 ) times faster than running a forward model. Note that traditional inversion may require multiple forward model simulations (e.g., in the order of 10 to 1000), which are very expensive. This computational savings ≈ O(10 5 ) − O(10 7 ) makes the proposed DL-based inverse model attractive for subsurface imaging and real-time ERT monitoring applications due to fast and yet reasonably accurate estimations of permeability field.

show abstract

Integrating field observations and process-based modeling to predict watershed water quality under environmental perturbations

Cited by 30 publications

References 314 publications

Editorial: Linking Hydrological and Biogeochemical Processes in Riparian Corridors

Editorial: Linking Hydrological and Biogeochemical Processes in Riparian Corridors

Building Intelligent Cyberinfrastructure to Learn Iteratively from both Observations and Models for Understanding Watershed Dynamics

Deep Learning to Estimate Permeability using Geophysical Data

Contact Info

Product

Resources

About