The result of this study is presented in three categories, namely; the descriptive statistics, the water quality test result and the ANN model and the model evaluation performance, respectively.
The descriptive statistics result is presented in Tables 1, 2, 3, 4. This describes the basic features of the data in this study. They provide simple summaries about the sample and the measures such as the mean, median, maximum, minimum and standard deviation, respectively.
The descriptive statistics in Tables 1,2, 3, 4 shows that the mean values of the data set ranges from 6.29 to 6.34, 1956.21 to 2458.19, 3.35 to 7.39 and 39.13 to 51.06 for Ph, TDS (mg/l), EC (dS/m) and Na (mg/l), respectively. The median values of the data set ranges from 6.31 to 6.39, 2010.00 to 2439.50, 3.14 to 4.24 and 39.13 to 51.06 for pH, TDS (mg/l), EC (dS/m) and Na (mg/l), respectively. The Maximum values data set ranges from 6.48 to 6.64, 2286.00 to 2742.00, 2.21 to 5.82, and 64.50 to 88.45 for Ph, TDS (mg/l), EC (dS/m) and Na (mg/l), respectively. The minimum values dataset ranges from 6.00 to 6.09, 1367.00 to 2199.00, 2.01 to 3.18, and 21.21 to 40.24 for Ph, TDS (mg/l), EC (dS/m) and Na (mg/l), respectively. The standard deviation values ranges from 0.08 to 0.16, 114.47 to 213.04, 0.23 to 31.49 and 14.06 to 8.16 for Ph, TDS (mg/l), EC (dS/m) and Na (mg/l), respectively. The low values of standard deviation recorded in this study shows that data set were very close to the mean of the dataset.
The water quality analysis test result indicates the level of concentrations of the TDS (mg/l), EC (dS/m) and Na (mg/l) in the Ele river in Nnewi, Anambra State Nigeria. The FAO standard for irrigation water quality for TDS, EC and Na are 0–2000, 0–3 and 0–40, respectively. The water quality results show that the pH values which ranges from 6.01 to 6.87 were within the FAO standard in all the points for both rainy and dry seasons, whereas the TDS (mg/l), EC (dS/m) and Na (mg/l) parametric values range from 2001 to 2506, 3.01 to 5.76, and 40.42 to 73.45 respectively, were above the FAO standard from point 1 to point 3 and falls within the FAO standard at point 4 with values ranging from 1003 to 1994, 2.01 to 2.78 and 31.24 to 39.44, respectively. However, during the dry season, the TDS, EC, and Na values range from 2002 to 2742, 3.04 to 5.82 and 40.14 to 88.45 respectively, were all above the FAO standard. Anthropogenic pollution emitted into water bodies has recently been identified as a significant source of pollutants that need immediate action in order to avoid serious environmental effects11.
The results equally revealed that the concentrations decrease along the sampling points going downstream. It is noteworthy that irrigation water with a pH outside the normal range may cause a nutritional imbalance or may contain a toxic ion which is harmful to crops19. The high concentrations of TDS as observed in this study are likely to increase the salinity of the river water, change the taste of the water, and as well decrease the dissolved oxygen level of the surface water making it difficult for the survival of plants and aquatic organisms7.
Moreover, these anions and cations which increase the electric conductivity in water affect irrigation adversely since salts settle at crop root zones making it difficult for infiltration, absorption of moisture and nutrients necessary for crop production.
The ANN model and forecast for the water quality parameters are shown from Figs. 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19. Considering the water quality permissible range, River quality modeling and forecast shows different variations seasonally such that the pollution level during dry season was higher than the rainy season.
(A and B): pH model and forecast graph at point 1.
(A and B): TDS model and forecast graph at point 1.
(A and B): EC model and forecast graph at point 1.
(A and B): Na model and Forecast graph at point 1.
(A and B): Ph model and Forecast graph at point 2.
(A and B): TDS model and Forecast graph at point 2.
(A and B): EC model and Forecast graph at point 2.
(A and B): Na model and Forecast graph at point 2.
(A and B): Ph model and Forecast graph at point 3.
(A and B): TDS model and Forecast graph at point 3.
(A and B): EC model and Forecast graph at point 3.
(A and B): Na model and Forecast graph at point 3.
(A and B): pH model and Forecast graph at point 4.
(A and B): TDS model and Forecast graph at point.
(A and B): EC model and Forecast graph at point 4.
(A and B): Na model and Forecast graph at point 4.
Generally, the artificial neural network model the actual data set very well. At various sampling points, the developed ANN models descriptively show insignificant values in deviation for the actual data set. There were continues variations in the developed models and forecasts over time.
The feed-forward Multilayer Neural Network (FFMNN) Model Performance Evaluation Results are shown in Table 5. The model performance evaluation was carried out based on the developed ANN model training, Testing and forecast, respectively. The model performance evaluation was carried out using the coefficient of multiple determination R2 and Root Mean Squared Error (RMSE).
The R2 values were generally observed to have varied in the second decimal place for the training, testing and forecast model, respectively.
The training performance evaluation shows that R2 values ranges from 0.981 to 0.990, 0.981 to 0.988, 0.981 to 0.989 and 0981 to 0.989, for pH, TDS, EC, and Na, respectively. The training results shows that the pH model have the best performance followed by EC, and Na.
Also, the testing performance shows that the R2 value ranges from 0.952 to 0.967, 0.953 to 0.970, 0.951 to 0.967 and 0.953 to 0.968, for pH, TDS, EC and Na, respectively. However, the testing performance evaluation shows that TDS had the best performance. The forecast performance evaluation shows that the R2 values ranges from 0.945 to 0.968, 0.946 to 0.968, 0.944 to 0.967 and 0.949 to 0.965 for pH, TDS, EC and Na respectively. It was however discovered that the TDS made best forecast followed by the pH. The water quality forecast performance was further evaluated using the Root Mean Squared Error (RMSE) which ranges from 0.022 to 0.088, 0.012 to 0.087, 0.015 to 0.085and 0.014 to 0.084 for pH, TDS, EC and Na, respectively. The ANN model performed very well as their coefficient of multiple determinations R2 were very close 1, which is in agreement with the study of Awu et al. (2017) and Abrahart et al., (2005). On comparing the performance of the training model to the testing model and forecast, it shows that the training set performed better than the testing set followed by the forecast as its coefficient of multiple determinations, R2, was much closer to 1.
Source: Ecology - nature.com
