Arctic Navigation
Abstract:
Abstract:
Melting glaciers add to rising sea levels, which in turn increases coastal erosion and elevates storm surges. The unusual frequency of Iceberg activities in a region is indicative of climate change. Ocean temperatures create more frequent and intense coastal storms like hurricanes and typhoons. Increasing global warming makes the condition worse. Millions of people will start to lose their homes and it will kill a lot of wildlife. In addition, icebergs are very dangerous for ships. A model is created using Convolutional Neural Networks (CNN) for detecting icebergs from live satellite images. Detecting icebergs from live satellite images will be able to help warn the ships in the area in time which could save billions of dollars annually for the shipping and logistics industry. Icebergs damage the oil rigs resulting in oil spills affecting the environment significantly. This algorithm can detect and thus prevent these accidents from happening. To create the algorithm live satellite image data are obtained. These data are from Sentinel-1 satellite. The data is around 1,500 elements containing pixels of satellite images and incidence angles in which these photos are taken of possible icebergs, ships or other objects in the deep ocean. The data is broken into 25% testing data and 75% training data for the algorithm. There are three situations of the position of an iceberg: icebergs in open water, icebergs in drifting ice, and icebergs near calving areas. On average, the testing data had an accuracy of 90% when run on the CNN model. This project can potentially revolutionize how climate change is detected and could potentially save billions of dollars for the shipping and oil industry.
Melting glaciers add to rising sea levels, which in turn increases coastal erosion and elevates storm surges. The unusual frequency of Iceberg activities in a region is indicative of climate change. Ocean temperatures create more frequent and intense coastal storms like hurricanes and typhoons. Increasing global warming makes the condition worse. Millions of people will start to lose their homes and it will kill a lot of wildlife. In addition, icebergs are very dangerous for ships. A model is created using Convolutional Neural Networks (CNN) for detecting icebergs from live satellite images. Detecting icebergs from live satellite images will be able to help warn the ships in the area in time which could save billions of dollars annually for the shipping and logistics industry. Icebergs damage the oil rigs resulting in oil spills affecting the environment significantly. This algorithm can detect and thus prevent these accidents from happening. To create the algorithm live satellite image data are obtained. These data are from Sentinel-1 satellite. The data is around 1,500 elements containing pixels of satellite images and incidence angles in which these photos are taken of possible icebergs, ships or other objects in the deep ocean. The data is broken into 25% testing data and 75% training data for the algorithm. There are three situations of the position of an iceberg: icebergs in open water, icebergs in drifting ice, and icebergs near calving areas. On average, the testing data had an accuracy of 90% when run on the CNN model. This project can potentially revolutionize how climate change is detected and could potentially save billions of dollars for the shipping and oil industry.