My conclusions:
* A carbon tax would be nice to incentivize accountability around net- greenhouse gas contributions.
* This paper does quite a bit of probabilistic inference here. They conclude, for example, that the ~two orders of magnitude difference in emissions implies the model is working. ( https://www.elementascience.org/article/10.1525/elementa.358... ) This isn't super surprising since they're trying to measure methane emissions ~a km away from the source, but should measure our expectations.
Also, the whole dataset is available here (https://oneshare.cdlib.org/stash/dataset/doi:10.15146/R3WT2N) if anyone is interested. Props to the authors for including it.
(Disclosure: My company buys quite a bit of inorganic fertilizer every year. None of it is US sourced though as we're operating in subsaharan Africa.)