Thanks!! We also evaluated the method on carbon monoxide plumes; it performs similarly there. The challenge of in-depth analyses is that the mechanisms leading to false positives can be different in every gas (depends on the retrieval algorithm). The carbon monoxide plume problem is easier, though: lower background levels, afaik not as many artefact issues.
I would also love to evaluate the approach with nitrogen dioxide plumes, there is already some work on ML-based NO2 plume detection, but I couldn't get my hands on a dataset.
Thanks!! We also evaluated the method on carbon monoxide plumes; it performs similarly there. The challenge of in-depth analyses is that the mechanisms leading to false positives can be different in every gas (depends on the retrieval algorithm). The carbon monoxide plume problem is easier, though: lower background levels, afaik not as many artefact issues.
I would also love to evaluate the approach with nitrogen dioxide plumes, there is already some work on ML-based NO2 plume detection, but I couldn't get my hands on a dataset.