Thumbnail Image

Theses

Browse

20221
Reset filters
Subject: Civil Engineering × End date: 2024 × Start date: 2020 × Type: Thesis × Thesis Type: Ph.D ×
Reset filters

Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    ThesisItemOpen Access
    Spring water quality evaluation and prediction of water quality index using Artificial Neural Network in Bageshwar block of Bageshwar district, Uttarakhand state
    (G. B. Pant University of Agriculture and Technology, Pantnagar, 2022-03) Mohd Azam; Prasad, Jyothi
    Springs are a critical resource for the people of Bageshwar block of Bageshwar district, but are currently facing threats as a result of rapid urbanization and climate change. Here the people mainly depend on springs as the primary source of water. As a result, majority of people in Bageshwar become reliant on springs to satisfy their daily demand for water. The water quality of these springs is declining because of numerous anthropogenic activities, such as construction, deforestation, etc. The current study evaluated the drinking water quality of Bageshwar block springs, which is located in Uttarakhand's eastern Kumaon region between latitudes 29°41'60"N to 29°53'60"N and longitudes 79°35'60"E to 80°5'60"E and has an average elevation of 882.30 m above mean sea level. Spring water samples were collected and measured discharge between 2019- 2021 and checked for various physico-chemical and bacteriological parameters from eleven springs namely Banri (BS1), Manikhet (BS2), Darsu Aare (BS3), Kukudagaad (BS4), Kamedi (BS5), Shri Naula Dhaara (BS6), Bilauna (BS7), Bhaniya Dhaar-1 (BS8), Bhaniya Dhaar-2 (BS9), Bhitaal Gaon (BS9) and Nye Basti Chaurasi (BS11) in Bageshwar block which are conveniently accessible. Data on the spring inventory was obtained on site by interviewing the local spring users. In this study, all spring discharges were measured using the volumetric method and then various physiochemical and bacteriological water quality parameters tests were performed using BIS standard methods. The DEM of the study area was analyzed to generate the channel network map were created using QGIS 3.16. The Weighted Arithmetic Water Quality Index (WAWQI) method was applied to the eleven springs of the study area. Prediction of Water Quality Index is also done using Artificial neural Network (ANN) model of a fully-connected, feed-forward backpropagation MLP network in MATLAB. During the study period, it was found that the most of the springs were slightly alkaline. The EC, TDS, Chloride, Potassium, sodium, Residual Free Chlorine, turbidity, Iron, Nitrate and calcium concentration of Bageshwar block for all springs were within the acceptable limits for all the springs. The total hardness (200 mg/l), fluoride (1.0 mg/l) and magnesium concentration (30 mg/l) in most of the springs exceeded the acceptable limits with few springs falling below the acceptable limit. However, the total hardness (200-600 mg/l), fluoride (1.0-1.5 mg/l), total alkalinity (200-600 mg/l) and magnesium concentration (30-100 mg/l) in all the springs were within the permissible limit of BIS 10500 (2012) and WHO (2011). After confirmatory test of MPN index method it was found that the E coli was present in BS11 spring. After evaluating Water Quality Index, with Weighted Arithmetic Water Quality Index (WAWQI) method, it has been found that the overall quality of spring water for all spring sites found to potable. It could be concluded that the statistical correlation analysis was a suitable technique for finding correlation between different physico-chemical parameters. The WQI variable was modelled through ANNs using 15 water quality parameters as input variables. Subsequently, the models with fewer variables were tested with different scenarios and reduced up to the five water quality parameters (i.e., Q, Total Alkalinity, fluoride, iron, and magnesium). The overall performance of the ANN models was evaluated based on RMSE and R2. It was observed that the number of independent input parameters were reduced from fifteen to five with less effect on the model performance. Hence it is recommended that the best-performing MLP ANN model was M11 with five input parameters having RMSE of 0.7349 and reasonably high coefficient of determination-R2 (0.9625). The regression equation, y = 0.9141x + 1.1273, represents the relationship between the predicted and actual water quality index values. In purview of the present study, it is therefore suggested that the five-input parameter based multi-layer perceptron (MLP) artificial neural network (ANN) model will be the best suited to evaluate the water quality of the Bageshwar block springs.