Search

Recent advances in machine learning (ML) have revolutionized an extensive range of research and industry fields by successfully addressing intricate problems that cannot be resolved with conventional approaches. However, low interpretability and incompatibility make it challenging to apply ML to complicated bioprocesses, which rely on the delicate metabolic interplay among living cells. This overview attempts to delineate ML applications to bioprocess from different perspectives, and their inherent limitations (i.e., uncertainties in prediction) were then discussed with unique attempts to supplement the ML models. A clear classification can be made depending on the purpose of the ML (supervised vs unsupervised) per application, as well as on their system boundaries (engineered vs natural). Although a limited number of hybrid approaches with meaningful outcomes (e.g., improved accuracy) are available, there is still a need to further enhance the interpretability, compatibility, and user-friendliness of ML models.

The degradation of chloroform in a liquid phase plasma (LPP) reactor was examined according to the operating parameters and other influencing factors. The operating parameters included the applied voltage, frequency, and pulse width. The effects of titanium dioxide, hydrogen peroxide, and their combination on the degradation of chloroform were evaluated in terms of the influencing factors. Increasing the operating parameters up to the maximum allowed in the LPP process made chloroform decompose much faster than that of the minimum. On the other hand, increasing either the titanium dioxide or hydrogen peroxide concentrations in the reactant solution did not always destroy the chloroform efficiently compared to degradation without their loading. Although the impact of titanium dioxide was slightly higher than that of hydrogen peroxide, the synergistic effect between them enhanced the breakdown of chloroform further. Overall, both the operating parameters and influencing factors should be determined carefully to maximize the decomposition efficiency in the given treatment system, particularly for processes involving the LPP reaction.

Understanding the performance and mechanism of the degradation of various contaminants is essential for the effective design of wastewater treatment systems. This study examined the degradation of 1,4-dichlorobenzene in a photocatalytic system, which embedded TiO2 particles (as a thin film) on alumina balls (as a solid support). Decomposition of the parent compound was assessed by introducing a TiO2 photocatalyst, microwave irradiation, and ultraviolet irradiation jointly or separately (in terms of the system configuration) with and without an adjustment for the microwave intensity, pH, and hydrogen peroxide levels (in terms of the operating conditions) to the photocatalytic system. The results showed that the photocatalytic system adopting multiple processes concomitantly improved the degradation performance of the parent compound that was resistant to the single-unit processes. On the other hand, the decomposition efficiency was increased further with the sophisticated control of the operating conditions, specifically the pH and hydrogen peroxide levels. A dechlorination process as well as substitution and addition reactions involving the hydroxyl radical appeared to occur in the degradation pathway to produce four degradation products from the parent compound. The significance of this study lies in the introduction of a systematic approach to optimizing the operating conditions that facilitate the decomposition efficiency in a given photocatalytic system at no extra cost.

This study aimed to investigate the applicability of deep learning algorithms to (monthly) surface water quality forecasting. A comparison was made between the performance of an autoregressive integrated moving average (ARIMA) model and four deep learning models. All prediction algorithms, except for the ARIMA model working on a single variable, were tested with univariate inputs consisting of one of two dependent variables as well as multivariate inputs containing both dependent and independent variables. We found that deep learning models (6.31–18.78%, in terms of the mean absolute percentage error) showed better performance than the ARIMA model (27.32–404.54%) in univariate data sets, regardless of dependent variables. However, the accuracy of prediction was not improved for all dependent variables in the presence of other associated water quality variables. In addition, changes in the number of input variables, sliding window size (i.e., input and output time steps), and relevant variables (e.g., meteorological and discharge parameters) resulted in wide variation of the predictive accuracy of deep learning models, reaching as high as 377.97%. Therefore, a refined search identifying the optimal values on such influencing factors is recommended to achieve the best performance of any deep learning model in given multivariate data sets.

This study aimed to evaluate classification algorithms to predict largemouth bass (Micropterus salmoides) occurrence in South Korea. Fish monitoring and environmental data (temperature, precipitation, flow rate, water quality, elevation, and slope) were collected from 581 locations throughout four major river basins for 5 years (2011–2015). Initially, 13 classification models built in the caret package were evaluated for predicting largemouth bass occurrence. Based on the accuracy (&gt, 0.8) and kappa (&gt, 0.5) criteria, the top three classification algorithms (i.e., random forest (rf), C5.0, and conditional inference random forest) were selected to develop ensemble models. However, combining the best individual models did not work better than the best individual model (rf) at predicting the frequency of largemouth bass occurrence. Additionally, annual mean temperature (12.1 °C) and fall mean temperature (13.6 °C) were the most important environmental variables to discriminate the presence and absence of largemouth bass. The evaluation process proposed in this study will be useful to select a prediction model for the prediction of freshwater fish occurrence but will require further study to ensure ecological reliability.

In this study, we conducted a numerical simulation to evaluate stability in the seawater reverse osmosis (SWRO) process against seasonal variations of feed water. The impact of element configuration on SWRO operation was analyzed using a process model consisting of the solution–diffusion model. Seawater quality and temperature data collected from the intake of the Fujairah desalination plant (United Arab Emirates) were used to identify optimum internally staged design (ISD) configurations. Monitoring data were also used to simulate changes in SWRO performance. The process performance of the ISD configurations as compared to conventional design configurations in terms of specific energy consumption, permeate total dissolved solids concentration and recovery ratio. The results showed that SWRO processes with four ISD configurations met the performance requirements over 12 months despite the fluctuations in seawater quality and temperature. The pre-diction results of the ISD configurations showed stable operation during the 12 month simulation period. These configurations had a significant advantage with respect to membrane maintenance. This study’s findings could be used to optimize the element configuration of the SWRO process by predicting the process performance while accounting for seasonal variations of seawater quality and temperature.

Climate change is a primary driver that alters water pollution hotspots in watersheds, which complicates the use of Best management practices (BMPs) for diffuse pollution mitigation strategies. The objective of this study is to evaluate future changes in BMPs on total phosphorus (TP) loads to the river system, depending on climate change. Current weather data were collected from 2000 to 2010 and future weather data from 2040 to 2050 were obtained from regional climate model simulations based on Representative Concentration Pathways (RCP) 2.6, 4.5, and 8.5 scenarios. This study describes various BMP implementation plans at affordable cost that are adapted to different weather conditions, current (for 2000–2010) and future climate scenarios (for 2040–2050). The Soil and Water Assessment Tool (SWAT), i.e., one of the deterministic models for watershed management, was used to estimate total phosphorus (TP) removal efficiency of various types of BMP at two crop fields (rice paddy and soybean fields) in the Yeongsan River (YR) watershed of Korea. The Non-dominated Sorting Genetic Algorithm II (NSGA-II), i.e., one of the artificial intelligence (AI) models for decision support framework, was applied to obtain allow for trade-off analysis between two conflict objectives, the maximum TP load reduction against implementation costs. The SWAT simulation results showed that the model performance based on the Nash-Sutcliffe efficiency and percent bias was acceptable for simulating three parameters; daily flow discharge, monthly sediment loads, and TP loads in the river. As a result, monthly sediment loads and TP loads increase remarkably under all future climate change scenarios except for July, when a comparable amount of precipitation is recorded during the present and future conditions. New implementation of BMPs, therefore, will be required for future climate change scenarios to achieve 50% TP load reduction in the given watershed condition. Interestingly, parallel terraces which showed good efficiency in reducing TP loads at both fields under the current weather condition cannot be effective any more under the worst-case weather scenario in the future climate. In such a case, detention ponds can be proposed as BMP alternative to parallel terraces. Overall, this study not only demonstrates that watershed management plans using BMPs should be adjusted according to climate change scenarios, but also that the hybrid use of SWAT model and AI can help in refining existing and future BMPs at affordable cost.

Successful modification of surface properties of a nanocomposite electrode is prerequisite to enhancing the overall performance of electrochemical supercapacitors. The present study was designed to describe the microstructural and electrochemical characteristics of a new composite electrode assembled by activated carbon (AC) powder (as a host) and copper precursor (as a guest) using liquid phase plasma. The fabrication processes were conducted by changing plasma discharge time from 30 to 90 min in the presence and absence of (thermal) oxidation. We observed that merging plasma and oxidation treatments raised the content of copper oxide nanoparticles precipitated (evenly) on the AC surface, along with oxygen. A mixed valence state of copper oxides (in the forms of Cuo, Cu2O, and CuO) was found in different composites with and without oxidation, where CuO and Cuo affected a specific capacitance in positive and negative ways, respectively. This led to the difference of electrochemical stability and resistance among the assembled composites. For instance, the best cycling performance was observed in the plasma-treated composite for 90 min with oxidation, whereas that of 60 min without oxidation recorded the lowest resistance. Therefore, a proper balance between the capacitance and resistance appears to be required for effective fabrication of the supercapacitor electrode, specifically in cases involving copper oxides.

The unavailability of thermal infrared satellite observations in cloudy conditions has limited the spatial distribution and temporal coverage of remotely sensed evapotranspiration (ET) data. As a result, a number of approaches have been developed to reconstruct remotely sensed daily ET data in cloudy-sky conditions. Despite the wide application of these approaches, no work has been conducted to compare their performance over a wide variety of climatic and vegetative conditions. In this study, three commonly used ET reconstruction approaches namely, reference ET fraction (EToF), land surface temperature reconstruction (LSTR), and variational data assimilation (VDA) are used to obtain daily ET data under cloudy conditions. The abovementioned three approaches are applied to the Heihe River Basin (HRB) in the northwest of China during the growing season in 2015. The HRB covers an area of approximately 1,432,000 km2 and contains various land covers. The large aperture scintillometer (LAS) and eddy covariance (EC) measurements are used to evaluate performance of the abovementioned three methods over the grassland, cropland, and riparian forest sites in the HRB. The results show that the EToF approach underestimates ET at the grassland and riparian sites in cloudy days because of the negatively biased ET/ETo (where ETo is the reference evapotranspiration) in clear-sky days. The VDA approach overestimates ET in cloudy-sky conditions because of the overpredicted evaporative fraction values. The LSTR approach overestimates ET due to the under-reconstructed LST in cloudy day. The mean absolute percentage difference (MAPD) and root mean square error (RMSE) statistical metrics are used to compare performance of the three approaches The LSTR (MAPD = 43.1% and RMSE = 1.3 mm/day) and VDA (MAPD = 46.7% and RMSE = 1.4 mm/day) approaches slightly outperform EToF (MAPD = 46.6% and RMSE = 1.5 mm/day). The RMSEs of reconstructed ET values by EToF, VDA, and LSTR increase respectively from 1.6, 1.2, and 1.5 (mm/day) to 2.2, 1.9, and 2.0 (mm/day) as the number of consecutive cloudy days increases from 1 to 3. These outcomes suggest that synergistic use of space-borne microwave and thermal infrared LST observations into remote sensing-based ET methods can improve the reconstruction of ET data under cloudy conditions.

In this study, Internally Staged Design (ISD) was numerically optimized in order to demonstrate the long-term performance of full-scale seawater reverse osmosis (SWRO) membrane systems in the presence of colloidal foulants. To this end, a numerical model based on a finite difference method was developed and optimized with various feed water qualities, such as fouling potential and total dissolved solids (TDS). As a result, the optimized ISD exhibited greater water flux and higher energy efficiency in long-term operation (90 days) compared to conventional designs, where single-type membrane elements were employed over a pressure vessel while achieving potable TDS concentration in permeate (

A conventional photocatalytic system is a viable tool to purify wastewaters, whereas poor degradation performance due to diverse pollutants under various conditions still leaves it behind commercial markets. This study aimed to determine the degradation efficiency and mechanism of 4-chlorophenol in a hybrid system integrating a series of unit processes such as a conventional TiO 2 photocatalytic reactor as well as microwave and ultraviolet (UV) radiation. The decomposition (rate) of 4-chlorophenol was assessed with respect to the microwave intensity, pH, circulating fluid velocity, hydrogen peroxide level, and oxygen gas injection, along with a mix of individual processes. Results showed that there were favorable conditions for improving the degradation performance within certain limits. However, a significant synergy effect was also observed in the combined processes, which was substantially larger than any single process. Out of them, the best degradation performance was achieved using microwave irradiation and hydrogen peroxide in the photocatalytic reactor. The four intermediates (i.e., benzoquinone, hydroquinone, 4-chlorocatechol, and hydroxyhydroquinone) appeared to be generated through hydroxyl radicals-mediated hydroxylation and dechlorination in the proposed photocatalytic pathway. The present research is expected to provide new insights into designing the high performance photocatalytic system with affordable cost.

The presence of flow control devices and/or obstacles in the fluid flow such as baffles and spacers complicates the solid-fluid mixing pattern, which remains difficult to describe by classical analytical solutions. In this study, the removal of fine particles in a tertiary sedimentation tank mounting an adjustable baffle was investigated using the computational fluid dynamics code-COMSOL. The solid-fluid motion was solved by consecutively applying the equations of the continuity and momentum using the finite element method. The experiment was conducted by the sedimentation tank with the adjustable baffle inclined at 30° in a pilot scale plant. It's used as the reference data set for numerical simulations that were run on a 2-dimensional domain by modifying the configuration settings of angles for an adjustable baffle (i.e., 30°, 45°, and 60°) and without one. Results showed that the simulation results matched well with the experimental data for an adjustable baffle at 30° (NSE=0.97). The sedimentation tank with the adjustable baffle at different angles had a lower overflow rate (in the area of flow rebound) and mixing intensity (in the area of flow curve) than without one, eventually leading to enhanced particle removal efficiency. This tendency became more pronounced as the particle motion stabilized over time. The sedimentation tank mounting the adjustable baffle at 30° provided the best settling efficiency among the four different flow patterns. However, the conventional index that represents the mixing properties did not correctly address their relative efficiency for fine particle removal. Therefore, a numerical simulation tailored to a given geometry should be conducted to fully elucidate the fluid dynamics in the sedimentation tank with complex devices or obstacles.

Providing high-resolution monitoring data is essential in promoting decision-making activities for surface water quality protection such as scientific modeling and data analysis. In this study, statistical models using multiple linear regression (MLR) were developed to estimate Chlorophyll-a (Chl-a) and total suspended solid (TSS) concentrations in a mesotrophic reservoir, the Yeongam Reservoir in Korea, from satellite observations. Two types of satellite data that covered different spectral regions, for wavelengths in 412–865 nm (for the Geostationary Ocean Color Imager) and those in 405–14,385 nm (for the Moderate Resolution Imaging Spectroradiometer), were used as inputs for statistical models, after bias correction. The MLR models for Chl-a and TSS were initially constructed and evaluated with 39 image data-sets obtained during 2011–2014. Subsequently, they compared with their corresponding algorithms that were developed under different environmental settings as well as the CE-QUAL-W2 model, a...

Statistical models play an important role in elucidating the dynamic behaviors of surface water quality, given limited data on a large scale. In this study, we examine alternative approaches to develop regression models that predict fecal coliform (FC) concentrations in a river using different methods for selecting important variables provided by a self-organizing map (SOM). The raw data used as input to the SOM included 11 water quality, 6 meteorological, and 7 land use parameters that were monitored along the Yeongsan River in Korea on various time scales (from daily to half a decade) during 1996–2008. In both test and validation data sets, (multiple) regressions using backward elimination were compared against regression models via forced entry, which included a set of ranked variables simultaneously based on four indices in the SOM (i.e. structuring index, relative importance, cluster description, and Spearman’s rank correlation). Results showed that the SOM effectively illustrated the complex...

A transient model of seawater reverse osmosis (SWRO) system enables systematic assessment of membrane performance in response to changes in time-series parameters, operating conditions, or ancillary equipment. In this study, we describe the effects of energy recovery device (ERD) and feed pressure control on the SWRO plant in terms of energy consumption (reduced) and water quantity (increased) using a numerical model at pilot scale. In the simulation, two types of isobaric ERD, i.e. pressure exchanger (PX) and dual work exchanger energy recovery (DWEER), were used to quantify changes of the mass flow rates of inflow and outflow in the system. Also, temporal variation in the raw feedwater quality was addressed in the model with adaptive feed pressure control to maintain the amount of produced water under fouled membrane conditions. Results showed that the observed recovery and rejection rates in the pilot-scale plant had an excellent agreement with their predicted values under different seawater fe...

In a present study, tungsten oxide was employed to improve the removal efficiency of the parent compound diethyl phthalate as well as to reduce the band gap energy of the original photocatalyst titanium dioxide. The modified photocatalyst was prepared by depositing tungsten oxide onto titanium dioxide through the liquid phase plasma. The decomposition of the parent compound was examined by varying the loading of tungsten oxide in the photocatalyst surface under different light sources, UV and blue light sources. Results showed that the band gap energy in the modified photocatalyst was lower than that of the original photocatalyst, which was attributed to increased loading of tungsten oxide in the modified photocatalyst surface resulted from an increase in the initial precursor concentrations. Tungsten doped in the photocatalyst surface existed in the form of the WO3, which led to a decrease in the specific surface area for the modified photocatalyst. The degradation performance of the parent compound increased in response to an increase in the amount of tungsten oxide under blue light, but decreased with increasing the loading of tungsten oxide under UV light. Two intermediates observed during the photocatalytic degradation appeared to be involved in the proposed degradation pathway.

The influence of air pollutants originating from the Chinese region on air quality over South Korea has been a major concern for policymakers. To investigate the inter-annual trends of the long-distance transport of air pollutants from China to South Korea, multi-year trend analysis was carried out for Aerosol Optical Depth (AOD, as a proxy of particulate matter), and CO (a water-insoluble air pollutant) and SO2 (a partially water-soluble air pollutant), over three regions in Northeast Asia. Air pollutants are typically long-range transported from the highly polluted parts of China to South Korea through the Yellow Sea. Taking advantage of this geographical merit, we carried out the multi-year trend analysis with a special focus on the Yellow Sea region. Decreasing trends of about 5–10%, 13–17% and 55–61% during the last decade were observed in surface CO, AOD and tropospheric SO2 columns over the North China Plain (NCP), Yellow Sea (YS), and South Korea (SK), respectively. Such decreasing trends were also found consistently during the last three, five, and seven years, indicating that the changes in pollution levels are likely in response to recent policy measures taken by the Chinese and Korean governments to improve air quality over the regions. Due to these efforts, the amounts of air pollutants transported from China to South Korea are expected to decrease in future years, to the likely rates of 1.50 ppb yr−1, 0.05 DU yr−1, and 0.56 μg m−3 yr−1 over the YS region for CO, SO2, and PM2.5, respectively. Given the ambitious plans recently announced by the Chinese government for the 21st meeting of Conference of Parties (COP21) and its co-control effects, the suggested percentage rates may even be conservative numbers. This analysis is expected to provide South Korean policymakers with valuable information to establish new air pollution policies in South Korea.

To use an advanced oxidation process system for the degradation of 2,4-dichlorophenoxyacetic acid (2,4-D), a series of experiments were performed in which the effects of microwave and UV irradiation were evaluated. The decomposition rate of 2,4-D increased with increasing microwave intensity, UV intensity, and the auxiliary oxidant dosage. Excessive addition of some oxidants (H2O2 and O2), however, resulted in the reduction of the decomposition rate. The effect of addition of microwave irradiation was not significant unless the ozone addition was applied together. The decomposition rate constant obtained with microwave irradiation combined with ozone addition was considerably higher than those obtained with the combinations of UV and O3, of UV and photocatalyst, or of microwave, UV and photocatalyst. The rate constant obtained with the combination of microwave, UV, photocatalyst, and ozone was the highest, being 4.5 times that obtained with the microwave, UV and photocatalyst combination and more than 6 times that obtained injection of ozone only. This result suggests that there is a synergy effect when the constituent techniques, i.e., microwave irradiation, UV irradiation, ozone, and photocatalysis are applied together and that the irradiation of microwave can play an important role in the O3-assisted photocatalysis of organic pollutants in water.

Summary A regional screening tool—which is useful in cases where few site-specific parameters are available for complex vadose zone models—assesses the leaching potential of pollutants to groundwater over large areas. In this study, the previous pesticide leaching tool used in Hawaii was revised to account for the release of new volatile organic compounds (VOCs) from the soil surface. The tool was modified to introduce expanded terms in the traditional pesticide ranking indices (i.e., retardation and attenuation factors), allowing the estimation of the leaching fraction of volatile chemicals based on recharge, soil, and chemical properties to be updated. Results showed that the previous tool significantly overestimated the mass fraction of VOCs leached through soils as the recharge rates increased above 0.001801 m/d. In contrast, the revised tool successfully delineated vulnerable areas to the selected VOCs based on two reference chemicals, a known leacher and non-leacher, which were determined in local conditions. The sensitivity analysis with the Latin-Hypercube-One-factor-At-a-Time method revealed that the new leaching tool was most sensitive to changes in the soil organic carbon sorption coefficient, fractional organic carbon content, and Henry’s law constant; and least sensitive to parameters such as the bulk density, water content at field capacity, and particle density in soils. When the revised tool was compared to the analytical (STANMOD) and numerical (HYDRUS-1D) models as a susceptibility measure, it ranked particular VOCs well (e.g., benzene, carbofuran, and toluene) that were consistent with other two models under the given conditions. Therefore, the new leaching tool can be widely used to address intrinsic groundwater vulnerability to contamination of pesticides and VOCs, along with the DRASTIC method or similar Tier 1 models such as SCI-GROW and WIN-PST.

We developed a practical method to accelerate execution of Soil and Water Assessment Tool (SWAT) using open (free) computational resources. The SWAT source code (rev 622) was recompiled using a non-commercial Intel FORTRAN compiler in Ubuntu 12.04 LTS Linux platform, and newly named iOMP-SWAT in this study. GNU utilities of make , gprof , and diff were used to develop the iOMP-SWAT package, profile memory usage, and check identicalness of parallel and serial simulations. Among 302 SWAT subroutines, the slowest routines were identified using GNU gprof , and later modified using Open Multiple Processing (OpenMP) library in an 8-core shared memory system. In addition, a C wrapping function was used to rapidly set large arrays to zero by cross compiling with the original SWAT FORTRAN package. A universal speedup ratio of 2.3 was achieved using input data sets of a large number of hydrological response units. As we specifically focus on acceleration of a single SWAT run, the use of iOMP-SWAT for parameter calibrations will significantly improve the performance of SWAT optimization.

Manganese oxide/carbon composite electrode was synthesized using liquid-phase plasma method, to be applied to supercapacitor. Spherical manganese oxide nanoparticles smaller than 10 nm were dispersed uniformly on the surface of YP-50F. The increase in the amount of manganese oxide precipitated on YP-50F resulted in improved specific capacitance, charge–discharge rate, and Warburg impedance but specific surface area and total pore volume were decreased.