Prediction of Oceanic Heat Content in the Northern Indian Ocean Using Machine Learning

Oceanic heat content is a very important parameter of circulation of the Indian Ocean. The main purpose of the study is to analyze the OHC for different drought years. To calculate the OHC, the temperature and salinity play a very important role which has been taken from Asia Pacific Data Research Centre-SODA datasets for 45 years. The OHC is then calculated using an integration formula where heat content is integrated from the surface to the 20° C isothermal depth layer. The spatial and temporal variation in OHC for drought year 2004 and good monsoon year 2005 is calculated to analyze the oceanic heat content where is found that the OHC is more in 2005 and less in 2004 for all regions. In the temporal variation in OHC, the OHC is found in May or June and the minimum OHC is found in December or January. The OHC in the Arabian sea is varying from 7.5x 106 J/m2 to 8.3x 106 J/m2, 6.6x 106 J/m2 to 7.43x 106 J/m2 in the Bay of Bengal. At first, we investigated regression-based models, utilizing Random Forest, XGBoost, KNN, and Adaboost models to predict oceanic heat content using past meteorological data. Ensemble learning methods, such as RFR and XGBoost, have been crucial in demonstrating the ability to enhance predicting accuracy by integrating several models. The literature demonstrates a consistent progression in the utilization of machine learning for predicting OBC, with an increasing focus on tackling obstacles and enhancing the dependability of Ocean energy systems.