During the labeling of the training set, it was found that the temporal context was effective for the disambiguation of plumes and other plume-like patterns. It is believed that the model performance would improve by integrating that contextual information through the use of a 3D CNN, such as R-C3D (\citealt{saenko2017}) or by using L
For detection of temporal activity i.e. tracking of the plumes, the model will also build off of Region Convolutional 3D Network (R-C3D) which can be used to extract spatiotemporal features capturing activities, accurately localizing the start and end times of each plume .

Supplementary materials   

Methodology (Technical description)