ANN Model Based Optimization Along with Statistical Monitoring of Flow Forecasting: An Analysis on Longitudinal Data of Mayurakshi River, India

This research addresses the imperative need for enhanced flood management by deploying an Artificial Neural Network (ANN) model for forecasting upstream canal discharge. The model incorporates three crucial input variables: ‘Rainfall,’ ‘Upstream Pond Level,’ and ‘Upstream River Inflow.’ Furthermore, an extensive time-series analysis of these variables is utilized to assess both flood risk and stability. A longitudinal dataset spanning twenty-five years (1992–2016) is meticulously collected, with appropriate correction for seasonal variation. The performance of the ANN model is evaluated using R-square and RMSE metrics. Furthermore, Ordinary Least Squares (OLS) regression is applied to the four variables, with estimation of F ratio, p-value, and B value along with their levels of significance. The temporal evolution is partitioned into four distinct time blocks, facilitating an examination of stability and risk. Central tendencies and dispersion statistics are analyzed to provide a nuanced understanding of the flood situation over different time intervals. The findings contribute to advancing flood forecasting methodologies and optimizing flood management strategies for the Mayurakshi River Basin (MRB).