Kulwinder Singh Parmar

River Water Prediction Modeling Using Neural Networks, Fuzzy and Wavelet Coupled Model

In this paper, new prediction model introduced by coupling of neural networks model, fuzzy model and wavelet model for the water resources management. Artificial neural network (ANN), fuzzy, wavelet and adaptive neuro-fuzzy inference system (ANFIS) are found to be a sturdy tool to model many non-linear hydrological processes. Wavelet transformation will improve the ability of a prediction model by capturing valuable information on different resolution levels. The target of this research is to compare our model with other famous data-driven models for monthly forecasting of water quality parameter chemical oxygen demand (COD) level monitored at Nizamuddin station, New Delhi, India of river Yamuna based on the past history. The data has been decomposed into wavelet domain constitutive sub series using Daubechies wavelet at level 8 (Db8). Statistical behavior of wavelet domain constitutive series has been studied. The foretelling performance of the wavelet coupled model has been compared with classical neuro fuzzy, artificial neural network and regression models. The result shows that the wavelet coupled model produces considerably higher leads to comparison to neuro fuzzy, neural network, regression models.

Fractal, predictability index and variability in trends analysis of river-water dynamics

ABSTRACT Statistical modelling, analysis of physico-chemical parameters chemical oxygen demand (COD), biochemical oxygen demand (BOD), dissolved oxygen (DO), water temperature (WT), free ammonia (AMM), total Kjeldahl nitrogen (TKN), total coliform (TC), fecal coliform (FC) and potential of hydrogen (pH) monitored at the Hathnikund barrage (Haryana) sample site of river Yamuna in India have been studied. It has been observed that water-quality parameters such as COD-BOD, AMM-TKN, WT-pH and TC-FC are positively correlated whereas COD-DO, BOD-DO, TKN-FC and DO-WT are negatively correlated. For water-quality parameters such as pH, AMM, TC and FC no seasonal pattern is observed. Parameters such as COD, BOD, TKN, DO and WT follow a six-month seasonal pattern. All the parameters except DO and WT follow a positive trend for monthly and annual variations. BOD, AMM and TKN have anti-persistence behaviour for both monthly and yearly variations. For parameters COD (+27.83%), BOD (+42.36%), AMM (+49.63%), TKN (+22.71%), TC (+141.80%) and FC (+42.89%) the future trend remains positive with high variability. WT (−7.47%) follows a negative trend with low variation and DO (−17.12%) has a negative trend with lofty variation. Using fractal, predictability index and variability in trend analysis, it is concluded that all parameters, except pH and WT, cross the prescribed limits of WHO/EPA and if the same trend should be followed, then in the future the quality of water shall continuously deteriorate and water may not be fit for drinking, agriculture and industrial use.

Trend, Time Series, and Wavelet Analysis of River Water Dynamics

Time series, trend, wavelet and statistical analysis of water quality parameters Chemical Oxygen Demand (COD), Biochemical Oxygen Demand (BOD), Dissolved Oxygen (DO) monitored for river Yamuna in India have been studied. It is observed that COD is highly correlated with BOD. For all auto regressive integrated moving average model (p,d,q) value of “d,” i.e. middle value is zero thus process is stationary. It is also observed that RMSE values are comparatively very low, thus dependent series is closed with the model predicted level. MAPE, MaxAPE, MAE, MaxAE, Normalized BIC are calculated and have low value for all parameters. Trend is calculated by using auto correlation function, partial auto correlation function, and lag. Thus the predictive model is useful at 95 % confidence limits. 1-D discrete and continuous Daubechies Wavelet analysis explains that the parameters COD, BOD, DO have the maximum value 120, 50, 8 and amplitude (a5) varies between 52 to 78, 10 to 30, 0.2 to 1.4, respectively. The scale values of Db5, i.e. d5, d4, d3, d2, and d1 range between − 20 and + 20 for all parameters. All parameters cross the prescribed limits of WHO/EPA, thus water is not fit for drinking, agriculture, and industrial use.