You need to have a good lateral view of any tornado; they sometimes have multiple votices rotating about a common center. Also the "wind wall" extends out much farther laterally than is indicted by the "dust pick-up cloud" in the center. While he was focused on the dust/debris pickup, the windwall came in from the left and whacked him...
To get a look at the lateral "reach" of that same tornado, look at this YouTube video of it as it crossed the Connecticut River (and hit two bridges). Notice how it pulls the surface water into itself (starting to form a waterspout) and how far up and down the river it reaches out and disturbs the water as it crosses diagonally. (The video appears to have been made from a "towercam" atop a building high up on the east bank...)
I haven't georeferenced this video to a map, but Donovan was probably somewhere on the far bank of the river somewhere near the bridge on the right in this video. I sure wouldn't have wanted to be on either of those two bridges!!
See what I mean about the danger of developing "tunnel vision" as a tornado is approaching...?
Wow! That funnel sucked the water up right down to the exposing the river bottom! Imagine how far away fish were falling as that monster dissipated.
I have located the skyscraper with the microwave dishes where the camera that made the video linked in #17 was mounted. The bridge on the right is the railroad bridge, and the second bridge hit (at left) is the Memorial Bridge. In the foreground is I-91, 5 parallels the river on the far bank.
I've already plotted locations for the camera, the approx. position of the tornado at the start of the tape, as well as my guess as to Donovan's location. I can "ballpark" kickstart's location on I-91, and, if he wishes, he can help me refine that.
If enough of you think it is worth the effort, I can create a set of KML files including those positions and an approximate track of the tornado center. It is a lot of work (mostly because the camera zooms [uses different focal lengths] and pans) but I may attempt overlaying a couple of the camera views on Google Earth.
If I do the above I could create a set of KML files that I can publish as a link. Clicking on the files would start you off in Google Earth, centered high over the "camera skyscraper" with north at the top, showing an overview of all features. Then I would rotate the view around to the camera's first view azimuth, drop down to the camera's elevation, and tilt over to approximate the view at the beginning of the video -- with the video view from the camera's viewpoint superimposed on the Google Earth view.
No promises, yet, but if several folks use Google Earth and are really interested, I may give it a try... Any takers?