NEURAL NETWORK MODEL FOR WATER PARAMETER PREDICTIONAND EVALUATION: A CASE STUDY OF PIPAR CITY, RAJASTHAN,INDIASangeeta Parihar, Jai Singh Kachhwaha, Raina Jadhav, Tarun Gehlot andKrishan Kumar Saini
This study presents the model of the neural network for sodium adsorption prediction sodium carbonateresidual, magnesium adsorption, Kelly ratio and sodium percentage in groundwater samples of Pipar city,Jodhpur, Rajasthan. Diverse physicochemical parameters including pH, EC, TDS, Ca, Mg, Na, K, Cl, HCO3,CO3, SO4 and NO3 were tested for 50 groundwater samples for the pre-monsoon season of 2021. The ANNmodel, then contrasted with MS-Excel, is created with R programming. The right algorithm and neuronalnumbers have been determined for optimizing the model architecture Via a responsive analysis,sevenneurons were optimized through the resilient back propagation algorithm for weight-back monitoring. Itwas found that the prediction of irrigation appropriateness indices with a seven-neuron network was highlyaccurate. The relative mean squared error, coefficient of decision (R2) and mean absolute relative error iscalculated for the experimental outcomes and model outputs. There is a close harmonization between actualdata and ANN groundwater outputs for irrigation suitability indices for the planning and analysis ofdatasets. For measured and projected values, spatial distribution maps have been created. The ANN modeltherefore seems to be a valuable tool in Pipar city, Jodhpur, Rajasthan, India to forecast groundwatersuitability.