Machine learning approaches for improving precipitation forecasting in the Ambica River basin of Navsari District, Gujarat

Abstract The article examines machine learning models for precipitation forecasting in the Ambica River basin, addressing the important requirement for accurate hydrological forecasts in water resource management. Using a comprehensive collection of meteorological variables such as temperature, humidity, wind speed, and precipitation, four separate models are used: Support Vector Regression (SVR), Random Forest (RF), Decision Tree (DT), and Multiple Linear Regression (MLR). These models’ performance is rigorously evaluated using various assessment indicators. The cross-correlation function (XCF) is used in this study to evaluate the correlations between climatic variables and precipitation. The XCF analysis reveals several noteworthy trends, such as a high link between maximum temperature and precipitation, with maxima consistently found at months across all four sites. Furthermore, relative humidity and wind speed have significant connections with precipitation. The findings highlight the value of machine learning approaches in improving precipitation forecast accuracy. The RF and SVR models typically outperform, with values ranging from 0.74 to 0.91. This impressive accuracy underlines their effectiveness in precipitation forecasting, beating competing models in both the training and testing stages. These findings have significant consequences for hydrological processes, notably in the Ambica River basin, where accurate precipitation forecasting is critical for sustainable water resource management.