This repository contains the code base for our publication Characterization of Industrial Smoke Plumes from Remote Sensing Data, presented at the Tackling Climate Change with Machine Learning workshop at NeurIPS 2020.

The major driver of global warming has been identified as the anthropogenic release of greenhouse gas (GHG) emissions from industrial activities. The quantitative monitoring of these emissions is mandatory to fully understand their effect on the Earth’s climate and to enforce emission regulations on a large scale. In this work, we investigate the possibility to detect and quantify industrial smoke plumes from globally and freely available multiband image data from ESA’s Sentinel-2 satellites. Using a modified ResNet-50, we can detect smoke plumes of different sizes with an accuracy of 94.3%. The model correctly ignores natural clouds and focuses on those imaging channels that are related to the spectral absorption from aerosols and water vapor, enabling the localization of smoke. We exploit this localization ability and train a U-Net segmentation model on a labeled subsample of our data, resulting in an Intersection-over-Union (IoU) metric of 0.608 and an overall accuracy for the detection of any smoke plume of 94.0%; on average, our model can reproduce the area covered by smoke in an image to within 5.6%. The performance of our model is mostly limited by occasional confusion with surface objects, the inability to identify semi-transparent smoke, and human limitations to properly identify smoke based on RGB-only images. Nevertheless, our results enable us to reliably detect and qualitatively estimate the level of smoke activity in order to monitor activity in industrial plants across the globe. Our data set and code base are publicly available.

The full publication is available on arxiv.

The data set is available on zenodo.

classification/: Resnet-50 classifier code, training and evaluation routines segmentation/: U-Net segmentation model code, training and evaluation routines

Download this repository as well as the data and decompress the latter. For both model training and evaluation, you will have to modify the directory paths appropriately so that they point to the image and segmentation label data.

It is expected that the data are split into separate data sets for training, validation, and evaluation. For our publication, this has been done in such a way that all observations of a single location are contained in a single data set. Other strategies are possible and will be left to the user.

Either model can be trained by invoking:

python train.py

with the following optional parameters:

• -bs <int> to define a batch size,
• -ep <int> to define the number of training epochs,
• -lr <float> to define a starting learning rate, and
• -mo <float> to define a momentum value.

The models can be evaluated on the test data set by calling the corresponding eval.py script.

If you use this code for your own project, please cite the following conference contribution:

Mommert, M., Sigel, M., Neuhausler, M., Scheibenreif, L., Borth, D.,
"Characterization of Industrial Smoke Plumes from Remote Sensing Data",
Tackling Climate Change with Machine Learning Workshop,
NeurIPS 2020.