Whups! You're all correct, I screwed it up. Now I gotta do it again. The reasoning behind my mistake was forgetting that it's in a collision that kinetic energy isn't conserved, and the collision hasn't happened yet.
OK, lemme see.
Simply because the dust would hit over a period of seconds or minutes, rather than all at once, I expect the effects would be less catastrophic. In addition, I would expect the dust would also be more spread out spatially. The change in space and time would make a great deal of difference; Crispy Duck's bullet-proof vest analogy is a good one. It's important to make the point that what happens to the atmosphere as the asteroid passes through it is far less important to the effects of a strike than what happens to the crust (and possibly ocean) when it hits.
The atmosphere is about 100 miles deep; at 40 miles a second (which is about medium speed for an asteroid) a whole asteroid will spend about two and a half seconds in the atmosphere. Within the first quarter second it will be heated to incandescence, and anyone looking at it will suffer permanent retinal damage. However, it won't matter much; if you're close enough to see the path through the atmosphere, you'll be dead within the minute no matter what else happens. Folks within something like a hundred miles will be vaporized by the heat of the strike, along with everything around them, a quarter to a half mile deep into the crust; rocks, soil, whatever, all vapor. A shockwave will move outward like a simultaneous level 10 on the Richter scale earthquake and thousand-mile-per-hour hurricane; this will destroy every living thing and every structure, and perhaps even mountains, for a thousand miles or more in every direction from the strike. Then the vapor cloud will deposit rock vapor for a lot farther than a thousand miles in every direction; any living thing touched by this vapor will be burned, and any breathing it will die. Huge gobbets of molten rock will splash outward from the impact, some of them flying out into space before crashing back down to cause wrack and ruin wherever they land, some of them half the world away. Some might even hit the Moon. The shockwave will circle the Earth multiple times, blowing down forests and all structures in its path, and its path will go everywhere. If the strike is at sea, a wall of water a thousand feet or more high will move outward at three to five hundred miles an hour, striking all the surrounding coasts and likely penetrating fifty to a hundred miles inland. Over a period of days or weeks, the crater will continue to blast heat and rock dust into the atmosphere, causing a "nuclear winter" that could last decades afterward.
Those are the effects of a major asteroid strike. Likely nothing larger than foraminifera will survive such an event in the same hemisphere with the strike. Note, however, that the Earth's crust is essentially uncompromised. The seas will recover, and if the strike was at sea, once the crater is cool enough, the sea will cover it. The continents will still exist, and still be substantially where they are now. The mantle will not be breached. The atmosphere will eventually clear, and be much as it is now. On the far side of the world, small fauna and flora will likely survive the effects of the long winter. Being quite persistent, and highly adaptable, it is even possible the human race will survive it, though in nothing like our current numbers. On the other hand, if nothing survives for us to eat, then our doom is sealed.
Many of these effects would be largely mitigated by the lack of a solid strike to land or sea. Yes, the dust would make a mess of the atmosphere; it might heat things up, or cloud the atmosphere. But the shockwaves, the gobbets of molten rock splashing down, the vaporized rock, the tsunami, the earthquakes- none of that would happen. I have no realistic idea what the effects on the atmosphere would be, but they would be considerably less than those of a solid strike. The megatons of rock vapor and vaporized water (assuming an ocean strike) would also be absent; the crust would likely be essentially unaffected. There might be a "nuclear winter" effect, but it would last months instead of decades; there simply isn't enough material in an asteroid to do more. I think overall that the effects would be strongly mitigated, and I'd be surprised if it were enough to kill us off. Billions might die, but billions more would likely survive.
That's my best guess.
