A comparative study of artificial neural networks,Bayesian neural networks and adaptive neuro-fuzzy inference system in groundwater level prediction

Predictive modeling of hydrological time series is essential for groundwater resource development and management. Here, we examined the comparative merits and demerits of three modern soft computing techniques, namely, artificial neural networks (ANN) optimized by scaled conjugate gradient (SCG) (ANN.SCG), Bayesian neural networks (BNN) optimized by SCG (BNN.SCG) with evidence approximation and adaptive neuro-fuzzy inference system (ANFIS) in the predictive modeling of groundwater level fluctuations. As a first step of our analysis, a sensitivity analysis was carried out using automatic relevance determination scheme to examine the relative influence of each of the hydro-meteorological attributes on groundwater level fluctuations. Secondly, the result of stability analysis was studied by perturbing the underlying data sets with different levels of correlated red noise. Finally, guided by the ensuing theoretical experiments, the above techniques were applied to model the groundwater level fluctuation time series of six wells from a hard rock area of Dindigul in Southern India. We used four standard quantitative statistical measures to compare the robustness of the different models. These measures are (1) root mean square error, (2) reduction of error, (3) index of agreement (IA), and (4) Pearson’s correlation coefficient (R). Based on the above analyses, it is found that the ANFIS model performed better in modeling noise-free data than the BNN.SCG and ANN.SCG models. However, modeling of hydrological time series correlated with significant amount of red noise, the BNN.SCG models performed better than both the ANFIS and ANN.SCG models. Hence, appropriate care should be taken for selecting suitable methodology for modeling the complex and noisy hydrological time series. These results may be used to constrain the model of groundwater level fluctuations, which would in turn, facilitate the development and implementation of more effective sustainable groundwater management and planning strategies in semi-arid hard rock area of Dindigul, Southern India and alike.     

Health risk assessment of nitrate exposure in groundwater of rural areas of Gonabad and Bajestan,Iran

Prolonged exposure to excessive levels of nitrate through drinking water is a potential risk for human health. The current research reports the analytical results and associated health risk for water quality in term of nitrate in 39 groundwater samples during January 2018 in rural areas of Gonabad and Bajestan, Iran. Nitrate concentrations ranged from 1.8 to 82.2 and from 5.5 to 84.3 mg/L for Gonabad and Bajestan, respectively. In this work, the potential risk to human health was determined using the hazard quotient (HQ) for three age groups including adults, children and infants. Comparison of HQs among the 39 sampling sites showed that the rural areas in Bajestan had higher HQs than Gonabad. Among the studied groups, infants exposed to a higher risk than children and adults. The results also indicated that the health of individuals from nitrate exposure in most of the groundwater studied was not acceptable and most of the consumers were in danger from current nitrate concentrations. Therefore, there is an urgent need for enforcing effective plans to improve groundwater quality and to better manage and control probable nitrate contaminated sources.     

Statistical models for the assessment of nitrate contamination in urban groundwater using GIS

The purpose of this study is to develop statistical models for groundwater quality assessment in urban areas using Geographic Information Systems (GIS). To develop the models, the concentrations of nitrate (expressed as nitrogen, NO₃-N), which are different according to the type of land use, well depth and distribution of rainfall, were analyzed in the Seoul (the capital of South Korea) area. Data such as land use, location of wells and groundwater quality data for nitrate contamination were collected and a database constructed within GIS. The distribution of NO₃-N concentrations is not normal, and the results of the Mann-Whitney U-test analysis show the difference of NO₃-N concentration by well depth and by distribution of rainfall. In both the shallow and deep wells, the radius of influence is 200 m in the dry season and 250 m in the rainy season, showing the tendency to increase in the rainy season. The results of correlation and regression analysis indicate that mixed residential and business areas and cropped field areas are likely to be the major contributor of increasing NO₃-N concentration. Land uses are better correlated with NO₃-N in deep wells than in shallow wells.     

滹沱河冲洪积扇地下水硝酸盐的污染特征及污染源解析

为研究滹沱河冲洪积扇地区地下水硝酸盐污染机制,对滹沱河冲洪积扇地区地下水和地表水进行了采样监测,运用环境健康风险评价模型对研究区硝酸盐进行评价,采用水化学和多元统计方法研究了滹沱河冲洪积扇地区地下水硝酸盐污染问题。结果表明:研究区地表水NO-3污染较轻,NO-3均值为19.54 mg/L,所有水样均未超出我国地表水环境质量标准（45 mg/L）;但是,地下水已经受到了NO-3的严重污染,NO-3均值为75.84 mg/L,且有30.43%水样超出我国地下水质量标准（88. 6 mg/L）。研究区3个水文地质单元地下水硝酸盐的平均个人年健康风险分别为4.94×10-8、1.99×10-8和2.61×10-9,低于国际辐射防护委员会（ICRP）推荐的最大可接受风险水平（5.0×10-5/a）,因此,认为不会对人群构成严重危害。水文地质单元和地下水埋深对硝酸盐污染有显著影响,但是,土地利用类型对硝酸盐浓度的影响不显著。滹沱河冲洪积扇地区地下水硝酸盐的主要污染来源是生活污水和化肥。此外,强烈开采地下水也是该地区NO-3污染的诱因。     

Nitrogen circulation and nitrate in groundwater in an agricultural catchment in Southern India

High contents of nitrate in groundwater, ranging up to 1,500 mg/l, have been found. High concentrations are more common in village wells than in irrigation wells situated in the fields. The losses of nitrogen from the soil zone through deep leaching into the groundwater are small (260 kg N/km²); however, due to a small net infiltration (29 mm/year), the median content in groundwater still approaches the permissible limit of 50 mg/l NO ₃ ^? . In villages about 10–20% of the nitrogen from excreta are leached into the groundwater. Mineralization of soil nitrogen during a dry period, followed by heavy rains, caused extremely high contents of nitrate in groundwater.     

Combined use of 15N and 18O of nitrate and 11B to evaluate nitrate contamination in groundwater

Isotopic composition of NO₃ (δ¹⁵N_NO3 and δ¹⁸O_NO3) and B (δ¹¹B) were used to evaluate NO₃ contamination and identify geochemical processes occurring in a hydrologically complex Basin and Range valley in northern Nevada with multiple potential sources of NO₃. Combined use of these isotopes may be a useful tool in identifying NO₃ sources because NO₃ and B co-migrate in many environmental settings, their isotopes are fractionated by different environmental processes, and because wastewater and fertilizers may have distinct isotopic signatures for N and B. The principal cause of elevated NO₃ concentrations in residential parts of the study area is wastewater and not natural NO₃ or fertilizers. This is indicated by some samples with elevated NO₃ concentrations plotting along δ¹⁵N_NO3 and NO₃ mixing lines between natural NO₃ from the study area and theoretical septic-system effluent. This conclusion is supported by the presence of caffeine in one sample and the absence of samples with elevated NO₃ concentrations that fall along mixing lines between natural NO₃ and theoretical percolate below fertilized lawns. Nitrogen isotopes alone could not be used to determine NO₃ sources in several wells because denitrification blurred the original isotopic signatures. The range of δ¹¹B values in native ground water in the study area (−8.2‰ to +21.2‰) is large. The samples with the low δ¹¹B values have a geochemical signature characteristic of hydrothermal systems. Physical and chemical data suggest B is not being strongly fractionated by adsorption onto clays. δ¹¹B values from local STP effluent (−2.7‰) and wash water from a domestic washing machine (−5.7‰) were used to plot mixing lines between wastewater and native ground water. In general, wells with elevated NO₃ concentrations fell along mixing lines between wastewater and background water on plots of δ¹¹B against 1/B and Cl/B. Combined use of δ¹⁵N and δ¹¹B in the study area was generally successful in identifying contaminant sources and processes that are occurring, however, it is likely to be more successful in simpler settings with a well-characterized δ¹¹B value for background wells.     

Pesticide transport and transformation modeling in soil column and groundwater contamination prediction

Pesticide transport and transformation were modeled in soil column from the soil surface to groundwater zone. A one dimensional dynamic mathematical and computer model is formulated to simulate two types of pesticides namely 2,4-dichlorophenoxy acetic acid and 1,2-dibromo 3-chloro propane in soil column. This model predicts the behavior and persistence of these pesticides in soil column and groundwater. The model is based on mass balance equation, including convective transport, dispersive transport and chemical adsorption in the phases such as solid, liquid and gas. The mathematical solution is obtained by finite difference implicit method. The model was verified with experimental measurements and also with analytical solution. The simulation results are in good agreement with measured values. The major findings of this research are the development of the model which can calculate and predict the concentration of pesticides in soil profiles, as well as groundwater after 4, 12, 31 days of pesticide application under steady state and unsteady water flow condition. With the results of this study, the distribution of various types of pesticides in soil column to groundwater table can be predicted.     

Groundwater nitrate contamination and risk assessment in an agricultural area,South Korea

The nitrate of groundwater in the Gimpo agricultural area, South Korea, was characterized by means of nitrate concentration, nitrogen-isotope analysis, and the risk assessment of nitrogen. The groundwaters belonging to Ca–(Cl + NO₃) and Na–(Cl + NO₃) types displayed a higher average NO₃ ⁻ concentration (79.4 mg/L), exceeding the Korean drinking water standard (<44.3 mg/L NO₃ ⁻). The relationship between δ¹⁸O–NO₃ ⁻ values and δ¹⁵N–NO₃ ⁻ values revealed that nearly all groundwater samples with δ¹⁵N–NO₃ ⁻ of +7.57 to +13.5‰ were affected by nitrate from manure/sewage as well as microbial nitrification and negligible denitrification. The risk assessment of nitrate for groundwater in the study area was carried out using the risk-based corrective action model since it was recognized that there is a necessity of a quantitative assessment of health hazard, as well as a simple estimation of nitrate concentration. All the groundwaters of higher nitrate concentration than the Korean drinking water standard (<44.3 mg/L NO₃ ⁻) belonged to the domain of the hazard index <1, indicating no health hazard by nitrate in groundwater in the study area. Further, the human exposure to the nitrate-contaminated soil was below the critical limit of non-carcinogenic risk.     

Soil and groundwater quality with reference to nitrate in a semiarid agricultural region

Nitrate is a common pollutant in surface water and groundwater of agricultural areas. It is essential to monitor this pollutant in groundwater, especially when it is used for drinking purposes without treatment. The present study was carried out in an intensively irrigated area which forms a part of Nalgonda district, Andhra Pradesh, India where groundwater meets all the water needs of the rural population living in this area. The objective was to assess the spatiotemporal variation in the concentration of nitrate in groundwater and soil. Based on the analysis of 496 groundwater samples collected from 45 wells over a period of 2 years from March 2008 to January 2010 by sampling every 2 months, it was observed that groundwater in 242 km² of the total 724 km² area had nitrate above the maximum permissible limit of 45 mg/l for drinking purposes. Nitrate concentration in groundwater showed a positive relation with potassium, chloride, and sulfate, indicating their source from fertilizers. Reasons for the high concentration of nitrate in domestic areas were the dumping of animal wastes and leakage from septic tanks. The pH of the soil samples showed that most of the area had basic soil. Apart from pH, organic carbon, available phosphorous, available potassium, ammoniacal nitrogen, and nitrate nitrogen were also analyzed in the 97 soil samples.     

Neural network prediction of nitrate in groundwater of Harran Plain, Turkey

Monitoring groundwater quality by cost-effective techniques is important as the aquifers are vulnerable to contamination from the uncontrolled discharge of sewage, agricultural and industrial activities. Faulty planning and mismanagement of irrigation schemes are the principle reasons of groundwater quality deterioration. This study presents an artificial neural network (ANN) model predicting concentration of nitrate, the most common pollutant in shallow aquifers, in groundwater of Harran Plain. The samples from 24 observation wells were monthly analysed for 1 year. Nitrate was found in almost all groundwater samples to be significantly above the maximum allowable concentration of 50 mg/L, probably due to the excessive use of artificial fertilizers in intensive agricultural activities. Easily measurable parameters such as temperature, electrical conductivity, groundwater level and pH were used as input parameters in the ANN-based nitrate prediction. The best back-propagation (BP) algorithm and neuron numbers were determined for optimization of the model architecture. The Levenberg–Marquardt algorithm was selected as the best of 12 BP algorithms and optimal neuron number was determined as 25. The model tracked the experimental data very closely (R = 0.93). Hence, it is possible to manage groundwater resources in a more cost-effective and easier way with the proposed model application.     

Evaluation of an intrinsic and a specific vulnerability assessment method in comparison with groundwater salinisation and nitrate contamination levels in two agricultural regions in the south of Portugal

The applicability of two vulnerability assessment methods in evaluating the impact of agricultural activities on groundwater quality, is tested in two areas in the south of Portugal with modest results. Intensive citri- and horticulture require large amounts of fertiliser and water supplied by irrigation, which induces groundwater salinisation and contamination by nitrates. The degree of contamination varies highly within and between the study areas and is related to hydrogeological factors as well as intensity of agricultural practices. Vulnerability mapping is performed with the intrinsic DRASTIC method and the specific Susceptibility Index (SI), which is an adaptation of DRASTIC. These methods can constitute useful groundwater management tools, for instance when designating new Nitrate Vulnerable Zones as defined in the European Directive 91/676/EEC. However, in the case of DRASTIC, little correspondence exists between the most vulnerable and the most contaminated areas. This is mainly a result of underestimating the dilution capacity and overemphasising the attenuating potential of the unsaturated zone and aquifer, as both chloride and nitrate prove to be very stable contaminants. By including a parameter for land use, SI manages to produce more reliable results, although in many areas the vulnerability is overestimated.

---

Resumen Se evalúa la aplicabilidad de dos métodos de estimación de vulnerabilidad en evaluar el impacto de actividades agrícolas en la calidad del agua subterránea para dos áreas en el sur de Portugal obteniendo resultados modestos. La horticultura y citricultura intensiva requiere grandes cantidades de fertilizantes y agua abastecida por riego, lo cual induce salinización de agua subterránea y contaminación por nitratos. El grado de contaminación varía fuertemente dentro y entre las áreas de estudio y se relaciona con factores hidrogeológicos así como con la intensidad de las prácticas agrícolas. El mapeo de vulnerabilidad se lleva a cabo con el método intrínsico DRASTIC y el Índice de Susceptibilidad específica (SI), el cual es una adaptación de DRASTIC. Estos métodos pueden constituir herramientas de manejo de aguas subterráneas útiles, por ejemplo al designar nuevas Zonas Vulnerables por Nitratos del modo que se definen en la Directiva Europea 91/676/EEC. Sin embargo, en el caso de DRASTIC, existen poca correspondencia entre las zonas más vulnerables y las áreas más contaminadas. Esto se debe principalmente a la subestimación de la capacidad de dilución y a al sobre énfasis del potencial de atenuación de la zona no saturada y el acuífero, ya que tanto cloruro como nitrato han probado ser contaminantes muy estables. Al incluir un parámetro del uso de la tierra, SI genera resultados más confiables, aunque en muchas áreas se sobrestima la vulnerabilidad.

---

Résumé Lapplication de deux méthodes de calcul de la vulnérabilité permettant dévaluer limpact des activités agricoles sur la qualité des eaux souterraines, est testée dans deux zones du Sud du Portugal, avec des résultats modestes. La citriculture et lhorticulture intensives nécessitent de grandes quantités e fertilisants et deau souterraine pour lirrigation, ce qui induit la salinisation et la contamination des eaux souterraines par les nitrates. Le degré de contamination varie grandement à lintérieur et entre les zones détudes, en fonction des facteurs hydrogéologiques et de lintensité des pratiques agricoles. La cartographie de la vulnérabilité est mise en oeuvre via la méthodologie DRASTIC et lIndex de Susceptibilité (SI) spécifique, qui est une adaptation de la méthode DRASTIC. Ces méthodes êuvent constituer des outils de management des eaux souterraines, par exemple lors de la désignation de nouvelles zones de vulnérabilité aux Nitrates selon la n Directive Européenne 91/676/EEC. Par ailleurs dans le cas de DRASTIC, de petites correspondances existent entre les zones les plus vulnérables et les plus contaminées. Ceci est principalement le résultat dune sous-estimation de la capacité de dilution et de la sur-accentuation du potentiel datténuation de la zone non-saturée de laquifère, car et le chlore et les nitrates sont des contaminants très stables. En incluant un paramètre dutilisation des sols, SI produit des résultats plus réalistes, bien que dans de nombreuses zones la vulnérabilité soit surestimée.

     

Spatial and statistical assessment of nitrate contamination in groundwater:Case of Sais Basin,Morocco

The objective of this study is to evaluate the nitrate contamination in the plioquaternary aquifer of Sais Basin based on a statistical approach. A total of 98 samples were collected in the cultivated area during the spring and autumn period of 2018. The results show that 55% and 57% of the samples in spring and autumn respectively exceed the threshold fixed by WHO(50 mg/L). However, nitrate concentrations do not show seasonal and spatial variation(p0.05). The results of the correlation matrix, principal component analysis(PCA), and hierarchical cluster analysis(HCA) suggest that nitrate pollution is related to anthropogenic source. Moreover, multiple linear regression results show that NO_3 is more positively explained in the spring period by Ca and SO_4 and negatively explained by pH and HCO_3. Regarding the autumn period, nitrate pollution is positively explained by Ca and negatively by pH. This study proposes a useful statistical platform for assessing nitrate pollution in groundwater.     

农村家庭利用自酿米酒去除地下水中硝酸盐的初步研究

为去除农村家庭饮用地下水中的硝酸盐,以农村自酿米酒为碳源,利用简易的沙桶装置开展了反硝化去除硝酸盐的实验,对比了不同乙醇浓度下的反硝化效果。实验结果表明,以沙桶为实验装置,自酿米酒为碳源,在家庭中异位反硝化去除抽取地下水中的硝酸盐方法是有效果,易操作的。硝酸盐的去除率和C/N质量比有直接关系,当C/N质量比大于1.99时,硝酸盐去除率达99%,亚硝酸盐零积累,但易积累乙酸盐;当C/N质量比为0.89时,硝酸盐去除率达99%,生成物亚硝酸盐浓度远高于国家饮用水限值(1 mg/L);当C/N质量比为0.43时,反硝化过程不彻底,硝酸盐去除率不高,且易生成溶度较高的亚硝酸盐。溶解氧的存在不会对反硝化产生显著影响。米酒和硝酸盐之间的C/N最佳比例,宜大于本次实验的0.89,小于1.99,利用农村自酿米酒作为碳源去除地下水中硝酸盐是可行的。     

Superiority of neural networks for pillar stress prediction in bord and pillar method

Estimation of pillar stress is a crucial task in underground mining. This is used to determine pillar dimensions, room width, roof conditions, and general mine layout. There are several methods for estimating induced stresses due to underground excavations, i.e., empirical methods, numerical solutions, and currently artificial intelligence (AI). AI based techniques are gradually gaining popularity especially for problems involving uncertainty. In this paper, an attempt has been made to predict stresses developed in the pillars of bord and pillar mining using artificial neural network. A comparison has also been done to compare the obtained results with the boundary element method as well as measured field values. For this purpose, a multilayer perceptron neural network model was developed. A number of architectures with different hidden layers and neurons were tried to get the best solution, and the architecture 5-20-8-1 was found to be an optimum solution. Sensitivity analysis was also carried out to understand the influence of important input parameters on pillar stress concentration.     

Soil contamination by metals with high ecological risk in urban and rural areas

Industrial development, intensive agriculture and fast urbanization have caused the metal contents of soils to increase to many times the allowable limits. To assess this impact on urban and rural soils, we quantified the Cd, Cr, Cu, Pb, Ni and Zn contents of 258 soil samples from the Recife metropolitan region (RMR). The objectives of the study were to estimate the probability of ecological risk, to determine the spatial pattern of the metals’ accumulation in the soil and to identify potential sources for the metals using a multivariate geostatistical approach. Mean concentrations of Zn, Cr, Pb, Cu, Ni and Cd in soils were 65.2, 17.9, 16.5, 12.8, 6.3 and 1.5 mg kg^?1, respectively. The results demonstrated that the Cd was anthropogenic in origin, the Cr and Ni were lithogenic (natural) in origin and the Cu, Pb and Zn were mixed in origin. Cd contaminated 91% of the samples; the median content of Cd (1.4 mg kg^?1) was three times the quality reference value for soil. The Cd contents of sugarcane fields exceeded the allowable limit (3.0 mg kg^?1) for agricultural areas. The spatial variability of the metals in the RMR showed that metallurgy, cement production, vehicle exhaust and vehicular traffic were the main sources of metals in urban areas, while phosphate-based fertilizers were the main sources in rural areas. More than 80% of the metropolitan region surveyed in the study was at moderate to high ecological risk.     

Evaluation of intrinsic vulnerability to nitrate contamination of groundwater: appropriate fertilizer application management

In agricultural areas, fertilizer application is the main source of nitrate contamination of groundwater. To develop fertilizer management strategies to combat this problem, arable land in Hokkaido, Japan was evaluated using geographic information system techniques for intrinsic groundwater vulnerability to nitrate contamination. The DRASTIC method was modified to adapt it to the Hokkaido environment and used for the evaluation. Of the seven original DRASTIC factors, the depth to water (D), net recharge (R), soil media (S), topography (T), and impact of vadose zone media (I) were selected and used to explain the vertical movement of contaminants to the aquifer. The rating for the net recharge factor was also modified to a dilution factor for contaminants, rather than as a transporter. The frequency of wells with nitrate concentrations exceeding the Japanese environmental standard (10 mg/L) was reasonably explained by vulnerability evaluation results (GLM: logit-link, quasi-binomial distribution, Y = [1 + exp(6.873765 − 0.045988 × X)]⁻¹, p < 0.001). However, in the paddy fields and pastures, vulnerability did not exhibit a clear relationship with the frequency of wells exceeding the standard. This suggests that the modified DRASTIC method is applicable for fertilizer application management in upland fields. In addition, under the ongoing policy for acreage allotment for rice production, this method will be useful for deciding the arrangement of arable land and crop rotation taking into consideration the potential risk of fertilizer-induced nitrate contamination of groundwater.     

Support vector machines and feed-forward neural networks for spatial modeling of groundwater qualitative parameters

The present study attempts to model the spatial variability of three groundwater qualitative parameters in Guilan Province, northern Iran, using artificial neural networks (ANNs) and support vector machines (SVMs). Data collected from 140 observation wells for the years 2002–2014 were used. Five variables, X and Y coordinates of the observation well, distance of the observation well from the shoreline, areal average 6-month rainfall depth, and groundwater level at the day of water quality sampling, were considered as primary input variables. In addition, nine qualitative variables were also considered as auxiliary input variables. Electrical conductivity (EC), sodium concentration (Na⁺), and sulfate concentration (SO₄ ^2?) of the groundwater in the region were estimated using ANNs and SVMs with different input combinations. The results showed that both ANNs and SVMs work well when the only primary input variable is the well location. The ANN yielded an RMSE of 1.03 mEq/l for SO₄ ^2?, 1.05 mEq/l for Na⁺, and 203.17 μS/cm for EC, using the X and Y coordinates of the observation wells in the study area. In the case of SVM, these values were, respectively, 0.87, 0.87, and 176.68. Considering the auxiliary input variables (pH, EC, and the concentrations of Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl^?, SO₄ ^2?, and HCO₃ ^?) resulted in a significant decrease in the RMSE of both ANNs (0.22, 0.30, and 33.04) and SVMs (0.26, 0.34, and 36.23). Comparing these RMSE values with those of cokriging interpolation technique (0.59, 0.98, and 177.59) indicated that ANNs and SVMs produced more accurate estimates of the three qualitative parameters. The relative importance of auxiliary input variables was also determined using Gamma test. The output uncertainty of ANNs and SVMs were determined using p-factor and d-factor. The results showed that SVMs have less uncertainty than ANNs.     

NARX neural network approach for the monthly prediction of groundwater levels in Sylhet Sadar,Bangladesh

Groundwater is important for managing the water supply in agricultural countries like Bangladesh. Therefore, the ability to predict the changes of groundwater level is necessary for jointly planning the uses of groundwater resources. In this study, a new nonlinear autoregressive with exogenous inputs(NARX) network has been applied to simulate monthly groundwater levels in a well of Sylhet Sadar at a local scale. The Levenberg-Marquardt(LM) and Bayesian Regularization(BR) algorithms were used to train the NARX network, and the results were compared to determine the best architecture for predicting monthly groundwater levels over time. The comparison between LM and BR showed that NARX-BR has advantages over predicting monthly levels based on the Mean Squared Error(MSE), coefficient of determination(R~2), and Nash-Sutcliffe coefficient of efficiency(NSE). The results show that BR is the most accurate method for predicting groundwater levels with an error of ± 0.35 m. This method is applied to the management of irrigation water source, which provides important information for the prediction of local groundwater fluctuation at local level during a short period.