Paper Title
Comparative Analysis of Multiple Linear Regression and Generalized Regression Neural Network for Water Temperature Estimation in Fontaine Des Gazelles Reservoir Dam-Biskra, Algeria

Abstract
Our research centers on estimating the water temperature of Fontaine de Gazelles Reservoir Dam by analyzing air temperature, relative humidity, solar radiation, atmospheric pressure, wind speed, and precipitation. These variables collectively impact water temperature, reflecting the thermal environment, water vapor content, solar energy, air density, wind-induced processes, and precipitation cooling. We employ Multiple Linear Regression (MLR) and Generalized Regression Neural Networks (GRNN) models for accurate estimates. MLR captures linear dependencies among climate variables, while GRNN model complex nonlinear relationships. Trained on historical data and real-time measurements, both MLR and GRNN demonstrate strong capabilities. MLR achieves high Nash-Sutcliffe Efficiency (0.991 to 0.997) and low Root Mean Squared Error (0.406 to 0.625), while GRNN achieves similar values. Both models consistently exceed a coefficient of determination R2equal 0.99, indicating a robust correlation, and display low Mean Absolute Error (0.236 to 0.391), affirming their accuracy. This attests to MLR and GRNN's reliability in estimating water temperature for the Fontaine de Gazelles Reservoir Dam. Keywords - Water Temperature, Multiple Linear Regression, Generalized Regression Neural Network, Reservoir-Dam, Biskra-Algeria.