Cloud Detection and Nowcasting for Ship Navigation

Navigating through oceans is challenging. Sailors are disadvantaged in determining the most appropriate and safe path, free from hurdles and extreme weather conditions, especially on pitch-dark nights. Ships rely on their communication with satellites for navigation. When this connection breaks, the ship loses its ability to navigate and follow the required path. In such scenarios, the Geographical Information System (GIS) and Satellite Imagery-based algorithms are best suited to assist. It is necessary to focus on cloud behavior to ensure the success of ocean endeavors. Cloud Detection is performed by the methods of K-Means clustering and MRCNN. The approach of Mean Path Adjustment (MPA) is used to trace the center of mass to predict Cloud Motion; thereby, building a convolutional LSTM model for Cloud Nowcasting. Collectively, these methods help in detecting clouds, predicting their motion, and helping sailors decide the route to follow, based on the weather conditions. This system works well when the communication between the ship and satellite breaks. A high similarity index of 0.8363 is achieved with the test images.