An efficient strategy for predicting river dissolved oxygen concentration: application of deep recurrent neural network model.

Dissolved oxygen (DO) concentration in water is one of the key parameters for assessing river water quality. Artificial intelligence (AI) methods have previously proved to be accurate tools for DO concentration prediction. This study presents the implementation of a deep learning approach applied to a recurrent neural network (RNN) algorithm. The proposed deep recurrent neural network (DRNN) model is compared with support vector machine (SVM) and artificial neural network (ANN) models, formerly shown to be robust AI algorithms. The Fanno Creek in Oregon (USA) is selected as a case study and daily values of water temperature, specific conductance, streamflow discharge, pH, and DO concentration are used as input variables to predict DO concentration for three different lead times ("t + 1," "t + 3," and "t + 7"). Based on Pearson's correlation coefficient, several input variable combinations are formed and used for prediction. The model prediction performance is evaluated using various indices such as correlation coefficient, Nash-Sutcliffe efficiency, root mean square error, and mean absolute error. The results identify the DRNN model ([Formula: see text]) as the most accurate among the three models considered, highlighting the potential of deep learning approaches for water quality parameter prediction.

Estimation of sodium adsorption ratio in a river with kernel-based and decision-tree models.

The control of surface water quality plays an important role in the management of water resources. In this context, the estimation and assessment of sodium adsorption ratio (SAR) are required which is one of the significant water quality parameters in the agricultural production sector. Chemical analysis might not, however, be feasible for a longer period of time in all the country-scale rivers. Therefore in this study, a support vector regression (SVR) model with different kernel functions; K nearest neighbour algorithm; and four decision-tree models, namely, Hoeffding tree, random forest, random tree, and REPTree, were used to estimate the SAR value with minimal parameters in the Aladag River in Turkey. In alternative scenarios, a correlation matrix and sensitivity analysis were used to ascertain the model inputs from among the 15 distinct parameters. All 15 parameters were utilized as model inputs in the first scenario, and only the sodium (Na) parameter was utilized as the model input in the final scenario. The accuracy of the aforesaid models was then assessed making use of correlation coefficient, Nash-Sutcliffe model efficiency coefficient, root mean square error, mean absolute error, and Willmott index of agreement. The results indicate that the SVR model with the poly kernel function provides the best estimates of SAR among the considered models. According to the findings, there is no considerable difference between the results acquired in the first and last scenarios, and one can determine the SAR value while making use of machine learning approaches taking into account only Na parameter.