TCI-Net: A Deep Learning Approach for Tropical Cyclone Intensity Prediction

One of the most affecting types of weather phenomena are tropical cyclones (TCs), commonly referred to as typhoons or hurricanes. Atlantic Ocean, North Pacific Ocean, Western Pacific Ocean, South Pacific Ocean, and Indian Ocean are the five oceans where hurricanes, typhoons, and tropical cyclones can form. Many techniques have been tried in the past to monitor cyclones and assess their intensity before they form, but they have always been shown to be ineffective. It is crucial to estimate TC intensity in order to forecast and decrease human loss and suffering. Recently, data scientists in addition to meteorologists have started to show interest in the task. This claim demonstrates an innovative approach based on image processing for predicting TC intensity given a meteorological geostationary satellite imagery. The Convolutional Neural Network (CNN) architecture which has to be modified in a number of ways, will be used to create the suggested cyclone intensity estimating model. The results obtained from experiments conducted on a widely accessible benchmark dataset reveal that the proposed approach stands out as one of the most precise models for estimating TC intensity. Moreover, it exhibits remarkable stability across various scenarios. These findings underscore the substantial promise of harnessing data science in the field of meteorological research.