Supervised segmentation of NO2 plumes from individual ships using TROPOMI satellite data

Starting from 2021, the International Maritime Organization significantly tightened the $\text{NO}_\text{x}$ emission requirements for ships entering the Baltic and the North Sea waters. Since all methods currently used for the ships' compliance monitoring are costly and require proximity to the ship, the performance of global and continuous monitoring of the emission standards' fulfillment has been impossible up to now. A promising approach is the use of remote sensing with the recently launched TROPOMI/S5P satellite. Due to its unprecedentedly high spatial resolution, it allows for the visual distinction of $\text{NO}_\text{2}$ plumes of individual ships. To successfully deploy a compliance monitoring system that is based on TROPOMI data, an automated procedure for the attribution of $\text{NO}_\text{2}$ to individual ships has to be developed. However, due to the extremely low signal-to-noise ratio, interference with the signal from other - often stronger - sources, and the absence of ground truth, the task is very challenging. This is the first study proposing an application of supervised learning for the segmentation of emission plumes produced by individual ships. As such, it is the first step towards an automated procedure for global ship compliance monitoring using remote sensing data. To this end, we developed a feature construction method allowing the application of multivariate models on spatial data. We applied several supervised-learning models and benchmark them towards existing unsupervised solutions of ship-plume segmentation with TROPOMI satellite data. We showed that the proposed approach leads to significant plume segmentation improvement and a high correlation with the theoretically derived measure of the ship's emission potential.