Predicting Surface Temperature at Indian Ports Using Machine Learning Models

Surface temperature measurement is essential for several reasons, including climatology, agriculture, and land use. This study aims to predict the surface temperature at three major Indian ports—Chennai, Kolkata, and Kochi—using multiple machine learning models: ANNs, RNNs, RF, and SVR. An integrated high-resolution atmospheric data set named ERA5 was used for the present study. The data preprocessing part involved getting the temperature for each place and then averaging it for one month, organizing the features and the targets needed to train the model. All the models were built and evaluated using the evaluation metrics, which include the Root mean square error (RMSE), Mean absolute errors (MAE), the coefficient of determination (R-squared), and system bias. As demonstrated in the study, the Random Forest model recorded the lowest average RMSE and MAE while at the same time recording the highest average R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values across all locations. Based on the results, it can be concluded that ANN and RNN gave moderately fruitful results, but SVR gave the worst results. They evaluate that the Random Forest model has high accuracy in predicting the surface temperature and improves the climate-based decision-making system. The results emphasize the value of ensemble approaches in climate modeling and offer several directions for further study to increase forecast precision and broaden the range of climate applications. this is the strong point of this research paper.