Artificial neural network modeling of dissolved oxygen in reservoir

The water quality of reservoirs is one of the key factors in the operation and water quality management of reservoirs. Dissolved oxygen (DO) in water column is essential for microorganisms and a significant indicator of the state of aquatic ecosystems. In this study, two artificial neural network (ANN) models including back propagation neural network (BPNN) and adaptive neural-based fuzzy inference system (ANFIS) approaches and multilinear regression (MLR) model were developed to estimate the DO concentration in the Feitsui Reservoir of northern Taiwan. The input variables of the neural network are determined as water temperature, pH, conductivity, turbidity, suspended solids, total hardness, total alkalinity, and ammonium nitrogen. The performance of the ANN models and MLR model was assessed through the mean absolute error, root mean square error, and correlation coefficient computed from the measured and model-simulated DO values. The results reveal that ANN estimation performances were superior to those of MLR. Comparing to the BPNN and ANFIS models through the performance criteria, the ANFIS model is better than the BPNN model for predicting the DO values. Study results show that the neural network particularly using ANFIS model is able to predict the DO concentrations with reasonable accuracy, suggesting that the neural network is a valuable tool for reservoir management in Taiwan.     

ANFIS-based modelling for coagulant dosage in drinking water treatment plant: a case study

Coagulation is the most important stage in drinking water treatment processes for the maintenance of acceptable treated water quality and economic plant operation, which involves many complex physical and chemical phenomena. Moreover, coagulant dosing rate is non-linearly correlated to raw water characteristics such as turbidity, conductivity, pH, temperature, etc. As such, coagulation reaction is hard or even impossible to control satisfactorily by conventional methods. Traditionally, jar tests are used to determine the optimum coagulant dosage. However, this is expensive and time-consuming and does not enable responses to changes in raw water quality in real time. Modelling can be used to overcome these limitations. In this study, an Adaptive Neuro-Fuzzy Inference System (ANFIS) was used for modelling of coagulant dosage in drinking water treatment plant of Boudouaou, Algeria. Six on-line variables of raw water quality including turbidity, conductivity, temperature, dissolved oxygen, ultraviolet absorbance, and the pH of water, and alum dosage were used to build the coagulant dosage model. Two ANFIS-based Neuro-fuzzy systems are presented. The two Neuro-fuzzy systems are: (1) grid partition-based fuzzy inference system (FIS), named ANFIS-GRID, and (2) subtractive clustering based (FIS), named ANFIS-SUB. The low root mean square error and high correlation coefficient values were obtained with ANFIS-SUB method of a first-order Sugeno type inference. This study demonstrates that ANFIS-SUB outperforms ANFIS-GRID due to its simplicity in parameter selection and its fitness in the target problem.     

Application of ANN and ANFIS models for reconstructing missing flow data

Hydrological yearbooks, especially in developing countries, are full of gaps in flow data series. Filling missing records is needed to make feasibility studies, potential assessment, and real-time decision making. In this research project, it was tried to predict the missing data of gauging stations using data from neighboring sites and a relevant architecture of artificial neural networks (ANN) as well as adaptive neuro-fuzzy inference system (ANFIS). To be able to evaluate the results produced by these new techniques, two traditionally used methods including the normal ratio method and the correlation method were also employed. According to the results, although in some cases all four methods presented acceptable predictions, the ANFIS technique presented a superior ability to predict missing flow data especially in arid land stations with variable and heterogeneous data. Comparing the results, ANN was also found as an efficient method to predict the missing data in comparison to the traditional approaches.     

Comparison of Groundwater Level Estimation Using Neuro-fuzzy and Ordinary Kriging

Water level in aquifer plays the main role in groundwater modeling as one of the input data. In practice, due to aspects of time and cost, data monitoring of water levels is conducted at a limited number of sites, and interpolation technique such as kriging is widely used for estimation of this variable in unsampled sites. In this study, the efficiency of the ordinary kriging (OK) and adaptive network-based fuzzy inference system (ANFIS) was investigated in interpolation of groundwater level in an unconfined aquifer in the north of Iran. The results showed that ANFIS model is more efficient in estimating the groundwater level than OK.     

Hydrometeorological variables predict fecal indicator bacteria densities in freshwater: data-driven methods for variable selection

Statistical models of microbial water quality inform risk management for water recreation. Current research focuses on resource-intensive, location-specific data collection and water quality modeling, but this approach may be cost-prohibitive for risk managers responsible for numerous recreation sites. As an alternative, we tested the ability of two data-driven models, tree regression and random forests with conditional inference trees, to select readily available hydrometeorological variables for use in linear mixed effects (LME) models predicting bacterial density. The study included the Chicago Area Waterway System (CAWS) and Lake Michigan beaches and harbors in Chicago, Illinois, at which Escherichia coli and enterococci were measured seasonally in 2007–2009. Tree regression node variables reduced data dimensionality by >50 %. Variable importance ranks from random forests were used in a forward-step selection based on R ² and root mean squared prediction error (RMSPE). We found two to three variables explained bacteria densities well relative to random forests with all variables. LME models with tree- or forest-selected variables performed reasonably well (0.335?<?R ²?<?0.658). LME models for Lake Michigan had good prediction accuracy with respect to the single sample maximum standard (72–77 %), but limited sensitivity (23–62 %). Results suggest that our alternative approach is feasible and performs similarly to more resource-intensive approaches.     

Ordinal time series model for forecasting air quality index for ozone in Southern California

Air quality index (AQI) for ozone is currently divided into six states depending on the level of public health concern. Generalized linear type modeling is a convenient and effective way to handle the AQI state, which can be characterized as non-stationary ordinal-valued time series. Various link functions which include cumulative logit, cumulative probit, and complimentary log-log are considered, and the partial maximum likelihood method is used for estimation. For a comparison purpose, the identity link, which yields a multiple regression model on the cumulative probabilities, is also considered. Random time-varying covariates include past AQI states, various meteorological processes, and periodic components. For model selection and comparison, the partial likelihood ratio tests, AIC and SIC are used. The proposed models are applied to 3 years of daily AQI ozone data from a station in San Bernardino County, CA. An independent year-long data from the same station are used to evaluate the performance of day-ahead forecasts of AQI state. The results show that the logit and probit models remove the non-stationarity in residuals, and both models successfully forecast day-ahead AQI states with almost 90 % of the chance.     

Monitoring sediment oxygen demand for assessment of dissolved oxygen distribution in river

Sediment oxygen demand (SOD) has become an integral part of modeling dissolved oxygen (DO) within surface water bodies. Because very few data on SOD are available, it is common for modeler to take SOD values from literature for use within DO models. SOD is such an important parameter in modeling DO that this approach may lead to erroneous results. This paper reported on developing an approach for monitoring sediment oxygen demand conducted with undisturbed sediment core samples, where the measured results were incorporated into a water quality model for simulating and assessing dissolved oxygen distribution in the Xindian River of northern Taiwan. The measured results indicate that a higher freshwater discharge results in a lower SOD. Throughout a 1-year observation in 2004, the measured SOD ranged from 0.367 to 1.246 g/m(2)/day at the temperature of 20°C. The mean values of the measured SOD at each station were adopted in a vertical two-dimensional water quality model to simulate the DO distribution along the Xindian River. The simulating results accurately depict the field-measured DO distribution during the low and high flow conditions. Model sensitivity analyses were also conducted with increasing and decreasing SOD values for the low and high flow conditions and revealed that SOD had a significant impact on the DO distribution along the Xindian River. The present work combined with field measurements and numerical simulation should assist in river water quality management.     

Biomonitoring of testate amoebae (protozoa) as toxic metals absorbed in aquatic bryophytes from the Hg-Tl mineralized area (China)

A new method has been developed for the determination of chemical oxygen demand (COD) in pulping effluent using chemometrics-assisted spectrophotometry. Two calibration models were established by inducing UV-visible spectroscopy (model 1) and derivative spectroscopy (model 2), combined with the chemometrics software Smica-P. Correlation coefficients of the two models are 0.9954 (model 1) and 0.9963 (model 2) when COD of samples is in the range of 0 to 405 mg/L. Sensitivities of the two models are 0.0061 (model 1) and 0.0056 (model 2) and method detection limits are 2.02?C2.45 mg/L (model 1) and 2.13?C2.51 mg/L (model 2). Validation experiment showed that the average standard deviation of model 2 was 1.11 and that of model 1 was 1.54. Similarly, average relative error of model 2 (4.25%) was lower than model 1 (5.00%), which indicated that the predictability of model 2 was better than that of model 1. Chemometrics-assisted spectrophotometry method did not need chemical reagents and digestion which were required in the conventional methods, and the testing time of the new method was significantly shorter than the conventional ones. The proposed method can be used to measure COD in pulping effluent as an environmentally friendly approach with satisfactory results.     

Occurrence and sorption properties of arsenicals in marine sediments

The content of total arsenic, the inorganic forms: arsenite (As(III)) and arsenate (As(V)), the methylated forms: monomethylarsonic acid and dimethylarsinic acid (DMA), trimethylarsenic oxide, tetramethylarsenonium ion and arsenobetaine was measured in 95 sediment samples and 11 pore water samples from the Baltic Sea near the island of Bornholm at depths of up to 100 m. As(III+V) and DMA were detected in the sediment and As(III+V) was detected in the sediment pore water. Average total As concentration of 10.6?±?7.4 mg/kg dry matter (DM) in the sediment corresponds to previously reported values in the Baltic Sea and other parts of the world. Existing data for on-site measurements of sorption coefficients (Kd) of arsenicals in marine and freshwater sediments show large variability from <1 to >1,000 L/kg. In this work, calculated sorption coefficients (Kd and Koc) for As(III+V) showed significant correlation with depth, dissolved oxygen (DO), salinity and sediment classification; for depths <70 m, salinity <11 %, DO >9 mg/L and sand/silt/clay sediments the Kd was 118?±?76 L/kg DM and for depths >70 m, salinity >11 %, DO?<?9 mg/L and muddy sediments the Kd was 513?±?233 L/kg DM. The authors recommend using the found Kd value for arsenic in marine sediments when conditions are similar to the Baltic Sea. At locations with significant anthropogenic point sources or where the local geology contains volcanic rock and sulphide mineral deposits, there may be significantly elevated arsenic concentrations, and it is recommended to determine on-site Kd values.     

Application of fuzzy sets and cognitive maps to incorporate social science scenarios in integrated assessment modelsA case study of urbanization in Ujung Pandang, Indonesia

Decision–support systems in the field of integrated water management could benefit considerably from social science knowledge, as many environmental changes are human-induced. Unfortunately the adequate incorporation of qualitative social science concepts in a quantitative modeling framework is not straightforward. The applicability of fuzzy set theory and fuzzy cognitive maps for the integration of qualitative scenarios in a decision–support system was examined for the urbanization of the coastal city of Ujung Pandang, Indonesia. The results indicate that both techniques are useful tools for the design of integrated models based on a combination of concepts from the natural and social sciences. This revised version was published online in July 2006 with corrections to the Cover Date.     

Carbon, nitrogen, and phosphorus stoichiometry of plankton and the nutrient regime in Cabo Frio Bay, SE Brazil

This long-term study, performed during the years 2003–2005 and 2008–2009, investigated the carbon (C), nitrogen (N), and phosphorus (P) contents of the phyto- and zooplankton communities and the nutrient regime of Cabo Frio Bay, SE Brazil. The information intends to serve as baseline of the plankton C, N, and P stoichiometry for the calibration of biogeochemical and ecological models in support to future findings related to the local and regional phenomena of climatic change. Cabo Frio Bay is a small semienclosed system set adjacent to a region subject to sporadic coastal upwelling. Zooplankton exhibited average annual C, N, and P contents of 11.6?±?6.9 %, 2.8?±?1.8 %, and 0.18?±?0.08 %, and phytoplankton (>20 μm) 6.8?±?6.0 %, 1.6?±?1.5 %, and 0.09?±?0.08 %, respectively. The C/N/P ratios correspond to the lowest already found to date for a marine environment. The low C contents must have been brought about by a predominance of gelatinous zooplankton, like Doliolids/ Salps and also Pteropods. Average annual nutrient concentrations in the water were 0.21?±?0.1 μM for phosphate, 0.08?±?0.1 μM for nitrite, 0.74?±?1.6 μM for nitrate, and 1.27?±?1.1 μM for ammonium. N/P ratios were around 8:1 during the first study period and 12:1 during the second. The plankton C/N/P and N/P nutrient ratios and elemental concentrations suggest that the system was oligotrophic and nitrogen limited. The sporadic intrusions of upwelling waters during the first study period had no marked effect upon the systems metabolism, likely due to dilution effects and the short residence times of water of the bay.     

Construction of fuzzy membership functions for urban vehicular exhaust emissions modeling

This paper presents a method for constructing a membership function (MF) for the fuzzy sets that expert systems deal with. This paper introduces a Bezier curve-based mechanism for constructing MFs of convex normal fuzzy sets. The mechanism can fit any given data set with a minimum level of discrepancy. In the absence of data, the mechanism can be intuitively manipulated by the user to construct MFs with the desired shape. MFs have been developed using the proposed mechanism for urban vehicular exhaust emission modeling. It has been observed that all meteorological and vehicular parameters have either S-shaped MFs or Z-shaped MFs. Gaussian MF has been mostly applied for modeling air quality. The present study explored the application of fuzzy MF to analyze air pollution data from vehicular emission. The study reveals that S-shaped and Z-shaped MF can be used in addition to Gaussian MF.     

Modeling seasonal variations of long-term soil CO<Subscript>2</Subscript> emissions in an orchard plantation in a semiarid area,SE Turkey

Emissions of soil CO₂ under different management systems have a significant effect on the carbon balance in the atmosphere. Soil CO₂ emissions were measured from an apricot orchard at two different locations: under the crown of trees (CO₂-UC) and between tree rows (CO₂-BR). For comparison, one other measurement was performed on bare soil (CO₂-BS) located next to the orchard field. Analytical data were obtained weekly during 8 years from April 2008 to December 2016. Various environmental parameters such as air temperature, soil temperature at different depths, soil moisture, rainfall, and relative humidity were used for modeling and estimating the long-term seasonal variations in soil CO₂ emissions using two different methods: generalized linear model (GLM) and artificial neural network (ANN). Before modeling, data were randomly split into two parts, one for calibration and the second for validation, with a varying number of samples in each part. Performances of the models were compared and evaluated using means absolute of estimations (MAE), square root of mean of prediction (RMSEP), and coefficient of determination (R²) values. CO₂-UC, CO₂-BR, and CO₂-BS values ranged from 11 to 3985, from 9 to 2365, and from 8 to 1722 kg ha^?1 week^?1, respectively. Soil CO₂ emissions were significantly correlated (p?<?0.05) with some environmental variables. The results showed that GLM and ANN models provided similar accuracies in modeling and estimating soil CO₂ emissions, as the number of samples in the validation data set increased. The ANN was more advantageous than GLM models by providing a better fit between actual observations and predictions and lower RMSEP and MAE values. The results suggested that the success of environmental variables for estimations of CO₂ emissions using the two methods was moderate.     

Different hydrodynamic processes regulated on water quality (nutrients,dissolved oxygen,and phytoplankton biomass) in three contrasting waters of Hong Kong

The subtropical Hong Kong (HK) waters are located at the eastern side of the Pearl River Estuary. Monthly changes of water quality, including nutrients, dissolved oxygen (DO), and phytoplankton biomass (Chl-a) were routinely investigated in 2003 by the Hong Kong Environmental Protection Department in three contrasting waters of HK with different prevailing hydrodynamic processes. The western, eastern, and southern waters were mainly dominated by nutrient-replete Pearl River discharge, the nutrient-poor coastal/shelf oceanic waters, and mixtures of estuarine and coastal seawater and sewage effluent of Hong Kong, respectively. Acting in response, the water quality in these three contrasting areas showed apparently spatial–temporal variation pattern. Nutrients usually decreased along western waters to eastern waters. In the dry season, the water column was strongly mixed by monsoon winds and tidal currents, which resulted in relatively low Chl-a (<5 μg l^?1) and high bottom DO (>4 mg l^?1), suggesting that mixing enhanced the buffering capacity of eutrophication in HK waters. However, in the wet season, surface Chl-a was generally >10 μg l^?1 in southern waters in summer due to halocline and thermohaline stratification, adequate nutrients, and light availability. Although summer hypoxia (DO <2 mg l^?1) was episodically observed near sewage effluent site and in southern waters induced by vertical stratification, the eutrophication impacts in HK waters were not as severe as expected owing to P limitation and short water residence time in the wet season.     

Assessing coastal waters of American Samoa: territory-wide water quality data provide a critical “big-picture” view for this tropical archipelago

The coastal waters of American Samoa’s five high islands (Tutuila, Aunu’u, Ofu, Olosega, and Ta’u) were surveyed in 2004 using a probabilistic design. Water quality data were collected from the near-shore coastal habitat, defined as all near-shore coastal waters including embayments, extending out to 1/4 mile off-shore. Hydrography and water column samples were collected, and water quality data were compared to the Territorial water quality standards for pH, dissolved oxygen (DO), Enterococcus, chlorophyll a, water clarity, total nitrogen, and total phosphorus. All station measurements for pH, DO, and Enterococcus satisfied the local water quality standards, although some fraction of the Territory could not be assessed for either DO or Enterococcus. With respect to chlorophyll a, 66 ± 18% of Territory coastal waters complied with the standard, while 34 ± 18% failed to comply with the standard. For water clarity, 54 ± 18% of the Territorial waters complied with the standard while 42 ± 7% failed to comply. Territorial waters satisfied the standards for total nitrogen and phosphorus 72 ± 17% and 92 ± 10%, respectively. These data provide the first “big-picture” view of water quality in the near shore region around the high islands of American Samoa. While the picture is encouraging, these data suggest emerging water quality concerns.     

Calibration and verification of a dissolved oxygen management model for a highly polluted river with extreme flow variations in Pakistan

A dissolved oxygen (DO) model is calibrated and verified for a highly polluted River Ravi with large flow variations. The model calibration is done under medium flow conditions (431.5 m³/s), whereas the model verification is done using the data collected during low flow conditions (52.6 m³/s). Biokinetic rate coefficients for carbonaceous biochemical oxygen demand (CBOD) and nitrogenous biochemical oxygen demand (NBOD) (i.e, K _cr and K _n ) are determined through the measured CBOD and ammonia river profiles. The calculated values of K _cr and K _n are 0.36 day^?1 and 0.34 day^?1, respectively. The close agreement between the DO model results and the field values shows that the verified model can be used to develop DO management strategies for the River Ravi. The biokinetic coefficients are known to vary with degree of treatment (DOT) and therefore need to be adjusted for a rational water quality management model. The effect of this variation on level of treatment has been evaluated by using the verified model to attain a DO standard of 4 mg/L in the river using the biokinetic rate coefficients as determined during the model calibration and verification process. The required DOT in this case is found to be 96 %, whereas the DOT is 86 % if adjusted biokinetic rate coefficients are used to reflect the effect of wastewater treatment. The cost of wastewater treatment is known to increase exponentially as the removal efficiency increases; therefore, the use of appropriate biokinetic coefficients to manage the water quality in rivers is important.     

Application of Water Quality Indices and Dissolved Oxygen as Indicators for River Water Classification and Urban Impact Assessment

The usefulness of water quality indices, as the indicators of water pollution, for assessment of spatial-temporal changes and classification of river water qualities was verified. Four water quality indices were investigated: WQI (considering 18 water quality parameters), WQI(min) and WQI(m) (considering five water quality parameters: temperature, pH, DO, EC and TSS) and WQI(DO) (considering a single parameter, DO). The water quality indices WQI(min), WQI(m) and WQI(DO) could be of particular interest for the developing countries because of the minimum analytical cost involved. As a case study, water quality indices were used to evaluate spatial and temporal changes of the water quality in the Bagmati river basin (Nepal) for the study period 1999-2003. The results allowed us to determine the serious negative effects of the city urban activity on the river water quality. In the studied section of the river, the water quality index (WQI) was 71 units (classified as good) at the entry station and 47.6 units (classified as bad) at the outlet station. For the studied period, a significant decrease in water quality (mean WQI decrease = 11.6%, p = 0.042) was observed in the rural areas. A comparative analysis revealed that the urban water quality was significantly bad as compared with rural. The analysis enabled to classify the water quality stations into three groups: good water quality, medium water quality and bad water quality. WQI(min) resulted in overestimation of the water quality but with similar trend as with WQI and is useful for the periodic routine monitoring program. The correlation of WQI with WQI(min) and DO resulted two new indices WQI(m) and WQI(DO), respectively. The classification of waters based on WQI(m) and WQI(DO) coincided in 90 and 93% of the samples, respectively.     

Biomonitoring with Gammarus pulex at the Meuse (NL), Aller (GER) and Rhine (F) rivers with the online Multispecies Freshwater Biomonitor

Biological early warning systems represent a set of tools that may be able to respond to certain chemical monitoring requirements of recent European legislation, the Water Framework Directive (WFD2000/60/EC), that aims to improve and protect water quality across Europe. In situ biomonitoring was performed along the rivers Meuse (NL), Aller (GER) and Rhine (F) within the frame of the European Union-funded Project SWIFT-WFD. Gammarus pulex was used as a test organism during the evaluation of the Multispecies Freshwater Biomonitor(R) (MFB), an online biomonitor to quantitatively record different behaviour patterns of animals. At the river Meuse G. pulex reacted to pulse exposure of either a mixture of trace metals or of several organic xenobiotics, by showing up to 20% decreased locomotory activity (already at the 1st pulse) and increased mortality (at 2nd or 3rd pulse only). G. pulex deployed within the MFB system were observed to survive well at the monitoring station on the Aller (100%) and monitoring did not result in the measurement of chemical irregularities. In contrast, deployment at the monitoring station on the Rhine river demonstrated that the test organism was able to detect chemical irregularities by up to 20% decreased locomotory activity in the animals. The MFB proved to be an alert system for water quality monitoring at sensitive sites and sites with accidental pollution.     

Monitoring diel dissolved oxygen dynamics through integrating wavelet denoising and temporal neural networks

Diel dissolved oxygen (DO) time series measured continuously using proximal sensors in situ for a temperate lake were denoised using discrete wavelet transform (DWT) with the orthogonal wavelet families of coiflet, daubechies, and symmlet with order of 10. Diel DO time series denoised were modeled using nine temporal artificial neural networks (ANNs) as a function of water level, water temperature, electrical conductivity, pH, day of year, and hour. Our results showed that time-lag recurrent network (TLRN) using denoised data emulated diel DO dynamics better than the best-performing TLRN using the original data, time-delay neural network (TDNN), and recurrent network (RNN). Daubechies basis dealt with diel DO data slightly better than the other bases given its coefficient of determination (r ²?=?87.1 %), while symmlet performed slightly better than the other bases in terms of root mean square error (RMSE?=?1.2 ppm) and mean absolute error (MAE?=?0.9 ppm).     

Study of heavy metal pollution in seawater of Kepez harbor of Canakkale (Turkey)

In this work, the heavy metals cadmium and lead in Kepez harbor were studied with regard to the determination of the environmental pollution levels. Seawater samples of the Dardanelles (Canakkale Strait) were collected from the nearest station to Kepez harbor in July 2007. Then, the concentrations of these metals were determined after simple pretreatment of samples by the proposed inductively coupled plasma?Catomic emission spectrometry (ICP-AES) method. The analysis of a given sample is completed in about 15 min by the ICP-AES method, and the concentrations of Cd and Pb were found to be 73.80 mg L^???1 (relative standard deviation [RSD], 1.07%) and 9.39 mg L^???1 (RSD, 0.70%), respectively. The accuracy of the applied ICP technique was checked by recovery studies, and good recoveries were obtained. The pollution levels of Cd and Pb recorded in this study were compared with those in other studies to analyze the water sample in Kepez harbor.