Enhancing rainfall estimation for water resource management through multiple linear regression approach in Khordha District, Odisha, India

Abstract

Rainfall estimation holds paramount significance for optimizing water resource utilization and effective management, particularly in countries like India, characterized by an agro-based economy and limited irrigation infrastructure. This paper introduces a multiple linear regression (MLR) model tailored for estimating annual rainfall in Khordha district, Odisha, India. The model is designed to predict rainfall for a given year by leveraging data from the preceding three years.Through rigorous testing on a century-long dataset of annual rainfall, the model demonstrated exceptional performance. By setting the intercept of the MLR model to zero, the derived equation yielded highly accurate results. The predicted values exhibited a robust correspondence with observed data, resulting in a R² value of 0.963. This remarkable R² value underscores the model's suitability for application across the study area. The robustness and accuracy of this model offers valuable insights for enhancing hydro-meteorological assessments and informing strategic water resource management decisions in the region.