Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Selection optimization”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 631 records · Page 35

Towards a 𝜈 𝜇 CC 1𝜇 + 1p exc. cross section measurement at the NOvA Near Detector

In the recent past, the NOvA near detector has been used to perform several neutrino-matter cross section measurements as a function of final state muon or electron kinematics around $E_\nu = 2$ GeV. In this talk, I show how the muon neutrino charged-current measurement in particular can be extended into a more exclusive region of phase space in order to probe nuclear effects. This can be done by selecting a final state with 1$\mu$, 1 proton, and nothing else, and measuring the cross section in terms of transverse kinematic imbalance (TKI) variables. The measurement I describe here would also be sensitive to quasi-elastic (QEL), two-particle two-hole (2p2h), and resonant (RES) neutrino interactions. Currently, I am optimizing the selection cuts for this analysis and developing a proton momentum estimator, which is necessary for reconstructing the TKI variables.

Weber, Colin [Minnesota U.]↗

Design and Optimization of Processes for Recovering Rare Earth Elements from End-of-Life Hard Disk Drives

In this conference paper, we propose a superstructure-based approach to finding the optimal pathways for recovering rare earth elements in their commercialized rare earth oxide form from end-of-life HDDs. The proposed superstructure was modeled as a MILP optimization problem, selecting the net present value as the objective function. Whenever possible, costing data taken from the literature was used to inform this mode. However, due to the novelty of this research area data were often not available thus requiring the generation of flowsheets that were implemented in Aspen Plus. To establish the base case optimal result, projections for the number of EOL HDDs in the U.S. available for recycling and estimates of the projected rare earth oxide prices over the lifetime of the plant were used to inform the model. The model was then expanded to include the recycling of EOL HDDs generated prior to the beginning of plant production (period ranging from 2006 through 2024).

Laliwala, Chris↗

A hybrid nonlinear programming method for design optimization

Solutions to engineering design problems formulated as nonlinear programming (NLP) problems usually require the use of more than one optimization technique. Moreover, the interaction between the user (analysis/synthesis) program and the NLP system can lead to interface, scaling, or convergence problems. An NLP solution system is presented that seeks to solve these problems by providing a programming system to ease the user-system interface. A simple set of rules is used to select an optimization technique or to switch from one technique to another in an attempt to detect, diagnose, and solve some potential problems. Numerical examples involving finite element based optimal design of space trusses and rotor bearing systems are used to illustrate the applicability of the proposed methodology.

Rajan, S. D.↗

Parametric studies of stitching effectiveness for preventing substructure disbond

A methodology is desired that will allow a designer to select appropriate amounts of through-thickness reinforcement needed to meet design requirements. The goal is to use a relatively simple analysis to minimize the amount of testing that needs to be performed, and to make test results from simple configurations applicable to more general structures. Using this methodology one should be able to optimize the selection of stitching materials, the weight of the yarn, and the stitching density. The analysis approach is to treat substructure disbond as a crack propagation problem. In this approach, the stitches have little influence until a delamination begins to grow. Once the delamination reaches, or extends beyond a stitch, the stitch serves to reduce the strain-energy-release-rate (G) at the crack tip for a given applied load. The reduced G can then be compared to the unstitched materials toughness to predict the load required to further extend the crack. The current model treats the stitch as a simple spring which responds to displacements in the vertical (through-thickness) direction. In concept, this approach is similar to that proposed by other authors. Test results indicate that the model should be refined to include the shearing stiffness of the stitch. The strain-energy-release-rate calculations are performed using a code which uses interconnected higher-order plates to model built-up composite cross-sections. When plates are stacked vertically, the interfacial tractions between the plates can be computed. The plate differential equations are solved in closed-form. The code, called SUBLAM, was developed as part of this section in one dimension. Because of this limitation, rows of stitches are treated as a two-dimensional sheet. The spring stiffness of a row of stitches can be estimated from the stitch material, weight, and density. As a practical and conservative approach, we can assume that the stitch is bonded until a crack passes the stitch location. After the crack passes, it is fully bonded. A series of tests were performed to exercise this methodology and incorporated an attached flange such that the sudden change in thickness initiated a delamination. The analysis was used to estimate the materials' critical G from that of the unstitched specimens. With this data, a prediction was made for the load required to delaminate the stitched specimens. Using the methodology, design charts have been created for simplified geometries. These charts give stitch force, along with G(sub 1) and G(sub 2) as as function of the stitch spring stiffness. Using the charts, it should be possible to determine the stitch spring stiffness and strength required to reduce the G to a desired level. From these parameters, the actual stitching material, weight, and density can be computed.

Flanagan, Gerry↗

Homogenizing atomic dynamics by fractional differential equations

In this paper, we propose two ways to construct fractional differential equations (FDE) for approximating atomic chain dynamics. Taking harmonic chain as an example, we add a power function of fractional order to Taylor expansion of the dispersion relation, and determine the parameters by matching two selected wave numbers. This approximate function leads to an FDE after considering both directions for wave propagation. As an alternative, we consider the symbol of the force term, and approximate it by a similar function. It also induces an FDE. Both approaches produce excellent agreement with the harmonic chain dynamics. The accuracy may be improved by optimizing the selected wave numbers, or starting with higher order Taylor expansions. When resolved in the lattice constant, the resulting FDE's faithfully reproduce the lattice dynamics. When resolved in a coarse grid instead, they systematically generate homogenized algorithms. Numerical tests are performed to verify the proposed approaches. Moreover, FDE's are also constructed for diatomic chain and anharmonic lattice, to illustrate the generality of the proposed approaches. - Highlights: • Match the dispersion relation of linear chain by power function of fractional order, added on low order Taylor expansion. • Design FDE by approximating dispersion relation, or the symbol for the force term. • Solve the FDE by algorithm of spectral accuracy. • Obtain faithful and systematical homogenization when resolved at coarser grid. • Applies to general lattices: complex, nonlinear and multidimensional.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Laser-pulse compression in a collisional plasma under weak-relativistic ponderomotive nonlinearity

We present theory and numerical analysis which demonstrate laser-pulse compression in a collisional plasma under the weak-relativistic ponderomotive nonlinearity. Plasma equilibrium density is modified due to the ohmic heating of electrons, the collisions, and the weak relativistic-ponderomotive force during the interaction of a laser pulse with plasmas. First, within one-dimensional analysis, the longitudinal self-compression mechanism is discussed. Three-dimensional analysis (spatiotemporal) of laser pulse propagation is also investigated by coupling the self-compression with the self-focusing. In the regime in which the laser becomes self-focused due to the weak relativistic-ponderomotive nonlinearity, we provide results for enhanced pulse compression. The results show that the matched interplay between self-focusing and self-compression can improve significantly the temporal profile of the compressed pulse. Enhanced pulse compression can be achieved by optimizing and selecting the parameters such as collision frequency, ion-temperature, and laser intensity.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Smoothing Lexis diagrams using kernel functions: A contemporary approach

Lexis diagrams are rectangular arrays of event rates indexed by age and period. Analysis of Lexis diagrams is a cornerstone of cancer surveillance research. Typically, population-based descriptive studies analyze multiple Lexis diagrams defined by sex, tumor characteristics, race/ethnicity, geographic region, etc. Inevitably the amount of information per Lexis diminishes with increasing stratification. Several methods have been proposed to smooth observed Lexis diagrams up front to clarify salient patterns and improve summary estimates of averages, gradients, and trends. In this article, we develop a novel bivariate kernel-based smoother that incorporates two key innovations. First, for any given kernel, we calculate its singular values decomposition, and select an optimal truncation point—the number of leading singular vectors to retain—based on the bias-corrected Akaike information criterion. Second, we model-average over a panel of candidate kernels with diverse shapes and bandwidths. The truncated model averaging approach is fast, automatic, has excellent performance, and provides a variance-covariance matrix that takes model selection into account. We present an in-depth case study (invasive estrogen receptor-negative breast cancer incidence among non-Hispanic white women in the United States) and simulate operating characteristics for 20 representative cancers. The truncated model averaging approach consistently outperforms any fixed kernel. Our results support the routine use of the truncated model averaging approach in descriptive studies of cancer.

60 APPLIED LIFE SCIENCES↗

SU-F-E-20: A Mathematical Model of Linac Jaw Calibration Integrated with Collimator Walkout

Purpose: Accurate jaw calibration is possible, but it does not necessarily achieve good junctions because of collimator rotation walkout. We developed a mathematical model seeking to pick an origin for calibration that minimizes the collimator walkout effect. Methods: We use radioopaque markers aligned with crosshair on the EPID to determine the collimator walkout at collimator angles 0°, 90° and 270°. We can accurately calibrate jaws to any arbitrary origin near the radiation field centre. While the absolute position of an origin moves with the collimator walkout, its relative location to the crosshair is an invariant. We studied two approaches to select an optimal origin. One approach seeks to bring all three origin locations (0°–90°–270°) as close as possible by minimizing the perimeter of the triangle formed by these points. The other approach focuses on the gap for 0°–90° junctions. Results: Our perimeter cost function has two variables and non-linear behaviour. Generally, it does not have zero-perimeter-length solution which leads to perfect jaw matches. The zero solution can only be achieved, if the collimator rotates about a single fixed axis. In the second approach, we can always get perfect 0°–0° and 0°–90° junctions, because we ignore the 0°–270° situation. For our TrueBeams, both techniques for selecting an origin improved junction dose inhomogeneities to less than ±6%. Conclusion: Our model considers the general jaw matching with collimator rotations and proposes two potential solutions. One solution optimizes the junction gaps by considering all three collimator angles while the other only considers 0°–90°. The first solution will not give perfect matching, but can be clinically acceptable with minimized collimator walkout effect, while the second can have perfect junctions at the expense of the 0°–270° junctions. Different clinics might choose between these two methods basing on their clinical practices.

60 APPLIED LIFE SCIENCES↗

Eastern North Atlantic (ENA) Aerosol Supplementary Site (S1) Data Analysis Report

We present strategies to identify time periods when local aerosol sources impact the data collected at the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) user facility in the Eastern North Atlantic (ENA). Data from the aerosol observing system (AOS) at the Central Facility and a temporary aerosol Supplementary Site (S1) were evaluated. S1 instrumentation was deployed within 0.75 km of the ENA Central Facility (C1) for approximately one year. Two time periods, one during the summer and one during the winter, were examined in support of the Aerosol and Cloud Experiments in the Eastern North Atlantic (ACE-ENA) intensive operating periods (IOPs) in 2017 and 2018. As the criteria for identifying regionally representative data varies based on the specific needs of the user, we present and discuss different strategies using measured aerosol properties and their associated meteorological parameters to isolate periods in AOS data that were impacted by local aerosol at ENA. We identify high-concentration aerosol events at ENA using submicron number concentrations, size distributions, high-time-resolution measurements, and diurnal profiles. An improved understanding of local aerosol to select the optimal aerosol mask was obtained with an understanding of local meteorology during the different seasons. We developed a mathematical algorithm that employs AOS data to mask high-concentration aerosol events at ENA. We validated the ENA Aerosol Mask with aerosol data from C1 and S1 and associated meteorological data. The Supplementary Site was ideal to understand and constrain local aerosol sources within the region. In its absence, collocated data from the ENA AOS could be used to select appropriate parameters to develop an aerosol mask that would limit over- or underestimating the regional baseline aerosol data. When available, associated metadata, such as airport flight logs and visual identification with AOS cameras should also be used to validate the application of an aerosol mask at a new site. Our results confirm the importance of optimizing an aerosol mask prior its application. In the future, the application of an aerosol mask at ARM observatories and during mobile deployments could be used to increase the data quality of continuous AOS data and to reduce uncertainties when parameterizing regional aerosol processes and impacts on clouds for incorporation in models.

54 ENVIRONMENTAL SCIENCES↗

National Microalgae Biofuel Production Potential and Resource Demand

Microalgae continue to receive global attention as a potential sustainable "energy crop" for biofuel production. An important step to realizing the potential of algae is quantifying the demands commercial-scale algal biofuel production will place on water and land resources. We present a high-resolution national resource and oil production assessment that brings to bear fundamental research questions of where open pond microalgae production can occur, how much land and water resource is required, and how much energy is produced. Our study suggests under current technology microalgae have the potential to generate 220 billion liters/year of oil, equivalent to 48% of current U.S. petroleum imports for transportation fuels. However, this level of production would require 5.5% of the land area in the conterminous U.S., and nearly three times the volume of water currently used for irrigated agriculture, averaging 1,421 L water per L of oil. Optimizing the selection of locations for microalgae production based on water use efficiency can greatly reduce total water demand. For example, focusing on locations along the Gulf Coast, Southeastern Seaboard, and areas adjacent to the Great Lakes, shows a 75% reduction in water demand to 350 L per L of oil produced with a 67% reduction in land use. These optimized locations have the potential to generate an oil volume equivalent to 17% of imports for transportation fuels, equal to the Energy Independence and Security Act year 2022 "advanced biofuels" production target, and utilizing some 25% of the current irrigation consumptive water demand for the U. S. These results suggest that, with proper planning, adequate land and water are available to meet a significant portion of the U.S. renewable fuel goals.

02 PETROLEUM↗

Optimal well placement and brine extraction for pressure management during CO 2 sequestration

Large-scale pressure increases resulting from carbon dioxide (CO 2 ) injection in the subsurface can potentially impact caprock integrity, induce reactivation of critically stressed faults, and drive CO 2 or brine through conductive features into shallow groundwater. Pressure management involving the extraction of native fluids from storage formations can be used to reduce such pressure increases. However, dealing with large volumes of extracted brine can be technically challenging and expensive. Selection of optimal well locations and pumping rates are critical for maximizing CO 2 storage and minimizing brine extraction during geologic CO 2 sequestration (GCS). Robust and efficient computerized algorithms combining reservoir models and optimization methods are needed to make proper decisions on well placement and pumping rates. This work presents a constrained differential evolution (CDE) algorithm for solving global optimization problems involving pressure management of GCS projects. Application of the CDE optimization methodology was demonstrated for a hypothetical CO 2 storage scenario in a deep sandstone reservoir in the Southern San Joaquin Basin in California, USA. Industrial-scale storage of CO 2 would generate significant pressure buildup in this formation, which in turn would raise concerns about induced seismicity due to presence of multiple faults surrounding the injection site. Through the CDE optimization algorithm coupled to a vertically-averaged reservoir simulator, we successfully estimated optimal solutions for brine extraction wells in the reservoir that would limit the local pressure along the faults to a prescribed threshold. Multiple realizations of the reservoir permeability field were created to understand the impact of reservoir heterogeneity on optimization results. Our results indicate that the reservoir slope and heterogeneity have significant impact on optimum extraction rates. Reservoir heterogeneity is also a significant factor for extraction well locations, suggesting that in practice decisions about extraction well placement through optimization should be made at later project stages when data from a few years of CO 2 injection have allowed iterative updating and refining of the reservoir forward models. Although the study focused on optimization of brine extraction, the CDE optimization methodology presented in this paper has also potential to solve other complex optimization problems related to GCS, such as increasing storage efficiency by enhancing injectivity and capillary and dissolution trapping.

58 GEOSCIENCES↗

Optimal Operations Management of Mobility-on-Demand Systems

The emergence of the sharing economy in urban transportation networks has enabled new fast, convenient and accessible mobility services referred to as Mobilty-on-Demand systems (e.g., Uber, Lyft, DiDi). These platforms have flourished in the last decade around the globe and face many operational challenges in order to be competitive and provide good quality of service. A crucial step in the effective operation of these systems is to reduce customers' waiting time while properly selecting the optimal fleet size and pricing policy. In this paper, we jointly tackle three operational decisions: (i) fleet size, (ii) pricing, and (iii) rebalancing, in order to maximize the platform's profit or its customers' welfare. To accomplish this, we first devise an optimization framework which gives rise to a static policy. Then, we elaborate and propose dynamic policies that are more responsive to perturbations such as unexpected increases in demand. We test this framework in a simulation environment using three case studies and leveraging traffic flow and taxi data from Eastern Massachusetts, New York City, and Chicago. Our results show that solving the problem jointly could increase profits between 1% and up to 50%, depending on the benchmark. Moreover, we observe that the proposed fleet size yield utilization of the vehicles in the fleet is around 75% compared to private vehicle utilization of 5%.

Wollenstein-Betech, Salomón↗

Enhanced Monte Carlo Simulations for Electron Energy Loss Mitigation in Real-Space Nanoimaging of Thick Biological Samples and Microchips

High-resolution imaging using Transmission Electron Microscopy (TEM) is essential for applications such as grain boundary analysis, microchip defect characterization, and biological imaging. However, TEM images are often compromised by electron energy spread and other factors. In TEM mode, where the objective and projector lenses are positioned downstream of the sample, electron–sample interactions cause energy loss, which adversely impacts image quality and resolution. This study introduces a simulation tool to estimate the electron energy loss spectrum (EELS) as a function of sample thickness, covering electron beam energies from 300 keV to 3 MeV. Leveraging recent advances in MeV-TEM/STEM technology, which includes a state-of-the-art electron source with 2-picometer emittance, an energy spread of 3 × 10 -5 , and optimized beam characteristics, we aim to minimize energy spread. By integrating EELS capabilities into the BNL Monte Carlo (MC) simulation code for thicker samples, we evaluate electron beam parameters to mitigate energy spread resulting from electron–sample interactions. Based on our simulations, we propose an experimental procedure for quantitively distinguishing between elastic and inelastic scattering. The findings will guide the selection of optimal beam settings, thereby enhancing resolution for nanoimaging of thick biological samples and microchips.

36 MATERIALS SCIENCE↗

Treyson Ricks - Intern Showcase Poster

Quinone-based sorbents offer a tunable, energy-efficient route to electrochemical CO2 capture, but systematic guidance for molecular design is lacking. Here, we report a high-throughput computational workflow that combines density functional theory (DFT) screening with machine-learning (ML) modeling to evaluate CO2 binding thermodynamics across several quinone derivatives, spanning benzoquinones, naphthoquinones, and anthraquinones. In addition to using solvents to stabilize the quinone anion and dianion, we studied the effect of ion-pairing on the reduction potentials and the CO2 binding energy. Automated Python scripts handled geometry optimizations and adduct-formation energies on an HPC cluster, reducing manual effort significantly. This integrated platform can uncover structure–property relationships and enables rapid in silico evaluation of untested candidates. We present one example from our workflow to showcase the capability of using quinones with ion-pairing to effectively capture CO2. Our approach paves the way for the rational selection of optimal quinone sorbents and can be extended with experimental thermochemical and kinetic data, alternative redox cycles, and stability assessments to accelerate development of next-generation electrochemical CO2 capture materials.

37 - INORGANIC, ORGANIC, PHYSICAL AND ANALYTICAL C↗

Rapid data acquisition and machine learning-assisted composition design of functionally graded alloys via wire arc additive manufacturing

Abstract The lack of high-quality datasets in materials science hinders artificial intelligence (AI)-driven alloy design. To address this challenge, wire arc additive manufacturing (WAAM) was employed to fabricate graded alloys, generating extensive data for machine learning (ML)-assisted property prediction. ML models were developed using high-throughput experiments, computational models, and genetic algorithm to optimize feature selection, successfully predicting hardness and porosity. The ML model demonstrated its efficacy by designing a gradient alloy with enhanced properties. However, scaling up revealed uncertainties in tensile property and porosity due to differences in size and thermal conditions between the designed alloy build and the gradient print used to construct the ML model. This underscores the need for uncertainty quantification and process optimization in WAAM-driven alloy design. Our work advances AI-integrated additive manufacturing, offering a rapid approach to exploring process–structure–property relationships and accelerating materials development.

Wang, Xin↗

Computational Redesign of Acyl-ACP Thioesterase with Improved Selectivity toward Medium-Chain-Length Fatty Acids

Enzyme and metabolic engineering offer the potential to develop biocatalysts for converting natural resources to a wide range of chemicals. To broaden the scope of potential products beyond natural metabolites, methods of engineering enzymes to accept alternative substrates and/or perform novel chemistries must be developed. DNA synthesis can create large libraries of enzyme-coding sequences, but most biochemistries lack a simple assay to screen for promising enzyme variants. Our solution to this challenge is structure-guided mutagenesis, in which optimization algorithms select the best sequences from libraries based on specified criteria (i.e., binding selectivity). We demonstrate this approach by identifying medium-chain (C8–C12) acyl-ACP thioesterases through structure-guided mutagenesis. Medium-chain fatty acids, which are products of thioesterase-catalyzed hydrolysis, are limited in natural abundance, compared to long-chain fatty acids; the limited supply leads to high costs of C6–C10 oleochemicals such as fatty alcohols, amines, and esters. Here, we applied computational tools to tune substrate binding of the highly active ‘TesA thioesterase in Escherichia coli. We used the IPRO algorithm to design thioesterase variants with enhanced C12 or C8 specificity, while maintaining high activity. After four rounds of structure-guided mutagenesis, we identified 3 variants with enhanced production of dodecanoic acid (C12) and 27 variants with enhanced production of octanoic acid (C8). The top variants reached up to 49% C12 and 50% C8 while exceeding native levels of total free fatty acids. A comparably sized library created by random mutagenesis failed to identify promising mutants. The chain length-preference of ‘TesA and the best mutant were confirmed in vitro using acyl-CoA substrates. Molecular dynamics simulations, confirmed by resolved crystal structures, of ‘TesA variants suggest that hydrophobic forces govern ‘TesA substrate specificity. Finally, we expect the design rules that we uncovered and the thioesterase variants that we identified will be useful to metabolic engineering projects aimed at sustainable production of medium-chain-length oleochemicals.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Sci-Fri AM: MRI and Diagnostic Imaging - 01: Estimating the Transverse Relaxation Time of Taurine Protons in Rat Brain at 9.4 T with Optimized PRESS Sequence Timings

Purpose: To investigate the response of taurine (Tau) protons as a function of PRESS echo times, TE{sub 1} and TE{sub 2}, to determine two TE combinations that can be employed to estimate the T{sub 2} (transverse relaxation) value of Tau at 9.4 T. The Tau protons are involved in J-coupling interactions; therefore, the two timing combinations should result in similar signal losses due to J-coupling. Methods: Experiments were performed with a 9.4 T animal MRI scanner. Numerical calculations of the response of Tau as a function of PRESS TE{sub 1} and TE{sub 2} were calculated and two TE combinations that yield a similar area for the 3.42 ppm Tau resonance were selected as optimal. The timings were verified on a 50 mM Tau/10 mM Cr (creatine) phantom. In-vivo experiments were performed on four rats. Spectra were acquired with the timings from a voxel placed in the rat brain and Tau peak areas were fit to monoexponentially decaying functions to obtain T{sub 2} values. Results: The PRESS TE combinations selected for Tau T{sub 2} determination are (TE{sub 1}, TE{sub 2}) = (17 ms, 10 ms) and (80 ms, 70 ms); the signal yield for the two timings differs by 5 % theoretically. The average Tau T{sub 2} for the four rats was found to be 106 ms with a standard deviation of 12 ms. Conclusions: We have demonstrated that acquiring PRESS spectra with (TE{sub 1}, TE{sub 2}) = (17 ms, 10 ms) and (TE{sub 1}, TE{sub 2}) = (80 ms, 70 ms) enables T{sub 2} corrected measures of Tau to be obtained at 9.4 T.

60 APPLIED LIFE SCIENCES↗

Can machine learning accelerate process understanding and decision‐relevant predictions of river water quality?

Abstract The global decline of water quality in rivers and streams has resulted in a pressing need to design new watershed management strategies. Water quality can be affected by multiple stressors including population growth, land use change, global warming, and extreme events, with repercussions on human and ecosystem health. A scientific understanding of factors affecting riverine water quality and predictions at local to regional scales, and at sub‐daily to decadal timescales are needed for optimal management of watersheds and river basins. Here, we discuss how machine learning (ML) can enable development of more accurate, computationally tractable, and scalable models for analysis and predictions of river water quality. We review relevant state‐of‐the art applications of ML for water quality models and discuss opportunities to improve the use of ML with emerging computational and mathematical methods for model selection, hyperparameter optimization, incorporating process knowledge into ML models, improving explainablity, uncertainty quantification, and model‐data integration. We then present considerations for using ML to address water quality problems given their scale and complexity, available data and computational resources, and stakeholder needs. When combined with decades of process understanding, interdisciplinary advances in knowledge‐guided ML, information theory, data integration, and analytics can help address fundamental science questions and enable decision‐relevant predictions of riverine water quality.

54 ENVIRONMENTAL SCIENCES↗