FluidFlow;
Thanks for posting your "least squares" polynomial regression curve deduced position for the crash location. It will be worthwhile to expand on the mathematical basis for the results, and how including the positions and times of as many objects as possible combines to reduce the error rate and increase the probability of the result being correct.
Perhaps, selecting a "notional crash site" at random, then taking the same "least squares" approach could help to validate the probability of the "oil slick" calculation.
mm43