Mr. Optimistic,
That link was to a set of a powerpoint slides, with no further explanation.
As the original language is English, I assume the simulation was likely prepared by an entity in the UK or the U.S., possibly Woods Hole.
I also assume that the drift was calculated based on the differing degrees of immersion, and the variable contributions over time of wind (both direction and velocity) and current (both direction and velocity) in steering an object under various immersion scenarios.
As the different drift paths all converge on the same line in the period before recovery, I assume that reflects better on-site data regarding wind and current in the search area, and thus reducing the error, or spread, that is likely present in the calculations for the period immediately following AF 447 crashing into the sea.