Ensuring Efficient and Robust Offshore Storage - Use of Models and Machine Learning Techniques to Design Leak Detection Monitoring

The use of machine learning techniques to identify CO 2 seeps to marine waters is assessed. These techniques require a large amount of data for training, here obtained through model predictions on how CO 2 seeps behave in the water column. Goldeneye, off the coast of Scotland, has been used as area of study. It is shown that Convolutional Neural Networks (CNN) are able to, with high confidence, to classify time series from the model simulations into leak and no-leak situations. CNN in data analysis can increase the detectability of CO 2 seeps, and thus the optimization of sensor deployment and monitoring design.