Environmental assessment based surface water quality prediction using hyper-parameter optimized machine learning models based on consistent big data

Abstract The environmental assessment of the surface water quality of the Western Bug River has been made using the system of classification quality of land surface water of Ukraine in accordance with the approved methodology, which allows comparing water quality of separate areas of water objects of different regions. The calculation of the environmental assessment of water quality has been carried according to three blocks: block of salt composition, block of trophic and saprobic (ecological and sanitary) indicators and block of indicators of content of specific toxic substances. The results are presented in the form of a combined environmental assessment, based on the final conclusions of the three blocks and consists in calculating the integral ecological index. Comprehensive studies of changes in the water quality of the Western Bug River have been conducted within the territory of Ukraine for a long-term period. The water quality of the river on the final values of the integral indicators of the ecological condition corresponded mainly to 4nd category of the 3rd class – the water is “satisfactory” by condition and “little polluted” by degree of purity (except for points of observation that located within the Volyn region, where the water quality corresponded to 3rd category and the 2nd class. It is “good” by condition and “fairly clean” by the degree of purity). Visualization and part of the analysis are performed using GIS technologies in the software of the ArcGIS 10.3.

Download Full-text

Assessing the Performance of Deep Learning Algorithms for Short-Term Surface Water Quality Prediction

Sustainability ◽

10.3390/su131910690 ◽

2021 ◽

Vol 13 (19) ◽

pp. 10690

Author(s):

Heelak Choi ◽

Sang-Ik Suh ◽

Su-Hee Kim ◽

Eun Jin Han ◽

Seo Jin Ki

Keyword(s):

Water Quality ◽

Deep Learning ◽

Surface Water ◽

Learning Algorithms ◽

Arima Model ◽

Surface Water Quality ◽

Percentage Error ◽

Data Sets ◽

Learning Models ◽

Dependent Variables

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.

Download Full-text

Hybrid Non-Dominated Sorting Genetic Algorithm

Advancements in Applied Metaheuristic Computing - Advances in Data Mining and Database Management ◽

10.4018/978-1-5225-4151-6.ch011 ◽

2018 ◽

pp. 264-286 ◽

Cited By ~ 9

Author(s):

Sankhadeep Chatterjee ◽

Sarbartha Sarkar ◽

Nilanjan Dey ◽

Amira S. Ashour ◽

Soumya Sen

Keyword(s):

Public Health ◽

Machine Learning ◽

Genetic Algorithm ◽

Water Quality ◽

Quality Prediction ◽

Learning Methods ◽

Nsga Ii ◽

Water Quality Prediction ◽

Machine Learning Methods ◽

Precautionary Measures

Water pollution due to industrial and domestic reasons is highly affecting the water quality. In undeveloped and developed countries, it has become a major reason behind a number of water borne diseases. Poor public health is putting an extra economic liability in order to deploy precautionary measures against these diseases. Recent research works have been directed toward more sustainable solutions to this problem. It has been revealed that good quality of water supply can not only improve the public health, it also accelerates economic growth of a geographical location as well. Water quality prediction using machine learning methods is still at its primitive stage. Besides, most of the studies did not follow any national or international standard for water quality prediction. In the current work, both the problems have been addressed. First, advanced machine learning methods, namely Artificial Neural Networks (ANNs) supported by a well-known multi-objective optimization algorithm called the Non-dominated Sorting Genetic Algorithm-II (NSGA-II) has been used to classify the water samples into two different classes. Secondly, Indian national standard for water quality (IS 10500:2012) has been utilized for this classification task. The hybrid NN-NSGA-II model is compared with another two well-known meta-heuristic supported ANN classifiers, namely ANN trained by Genetic Algorithm (NN-GA) and by Particle Swarm Optimization (NN-PSO). Apart from that, the support vector machine (SVM) has also been included in the comparative study. Besides analysing the performance based on several performance measuring methods, the statistical significance of the results obtained by NN-NSGA-II has been judged by performing Wilcoxon rank sum test with 5% confidence level. Results have indicated the ingenuity of the proposed NN-NSGA-II model over the other classifiers under current study.

Download Full-text