Indeed, different assumptions yield different data. The simulation I ran assumed 1.1 million lb of material released during the controlled burn on Feb 6 (this caused the famous "mushroom cloud" photos), while the Reddit sim seems to assume a constant release rate (quantity unknown) starting at the time of the initial derailment and fire. The actual plume will be some combination.
https://www.reddit.com/r/dataisbeautiful/comments/113wyj3/hy...
Looks like the plume didn't blow west much, which probably mitigates agricultural impact significantly, good point!