CyINSAT: Cyclone Dataset from Indian National Satellite for Forecasting

Tropical cyclones in India are annual natural disasters that have taken a toll of 1 lakh casualties in total to date and established the North Indian Ocean (NIO) as the deadliest basin historically. Moreover, it causes property damage and can lead to soil erosion, having adverse effects wherever it makes landfall. Effective forecasting models are needed to minimise the destructive aftermath of such a hazardous phenomenon. The first step in doing so is a consolidated dataset that is publicly available for research to take place. This study aims to combine satellite image data with cyclone track record data, which is currently not publicly available in a usable format. Satellite images containing four channels-two infrared, mid-wave infrared, and water vapour-from the Indian National Satellite System (INSAT) are obtained from the Meteorological and Oceanographic Satellite Data Archival Centre. Additionally, the track record data of cyclones with wind speed, pressure, category of the cyclone on the Indian Meteorological Department scale, and the latitude and longitude of the storm eye are retrieved from the Regional Specialised Meteorological Centre, New Delhi. The cyclone research community can use this dataset to develop a holistic forecasting model that can predict the track and intensity of any depression forming in the NIO basin, subsequently improving disaster management systems. Two applications of this dataset are explored later in this study: forecasting wind speed using time series forecasting models and predicting the cyclone’s eye track.