Study: Space Rock Impacts Not Random

sciencehabit writes When it comes to small space rocks blowing up in Earth's atmosphere, not all days are created equal. Scientists have found that, contrary to what they thought, such events are not random, and these explosions may occur more frequently on certain days. Rather than random occurrences, many large airbursts might result from collisions between Earth and streams of debris associated with small asteroids or comets. The new findings may help astronomers narrow their search for objects in orbits that threaten Earth, the researchers suggest.

Comments

Can you say meteor shower ?

[Crashmarik]

http://en.wikipedia.org/wiki/M...

http://stardate.org/nightsky/m...

I knew you could

Re:bar-room statisticians

[AthanasiusKircher]

It's even worse than that. Going by this quote, they're using it to mean even or homogeneous:

contrary to what they thought, such events are not random, and these explosions may occur more frequently on certain days.

You know, like if a coin comes up heads four times in a row that's "not random".

Actually, that's not a very good analogy. The main pattern that they noticed is clustering of events over long periods of time . It would be more like if you had a coin that was weighted in such a way that it only came up heads about 1 time out of a 100 or something. You flipped it once per day.

According to normal probability, if the only thing that's influencing the coin is just its weight that produces a 1 in 100 chance of heads, the pattern of heads should look relatively homogeneous over a long time span.

Instead, what they tended to find was a lot of clustering of events -- so it would be like going for hundreds of days and then suddenly getting heads on 2 or 3 days in a row, then going again for hundreds of days without any heads again.

In that case, it would be fair to say that there is something else influencing the distribution -- it's not just a "random" distribution you'd expect for a 1 in 100 chance of getting heads. Some other factor is leading to clustering.

Just from looking briefly at the article, it doesn't seem to me that they have a long-enough timespan or enough events to claim strong evidence for a pattern. They basically come up with a 2% stat that this pattern could occur by chance -- sure, that's better than the standard 95% confidence interval for exploratory studies, but there are various statistical features of their study that could be giving them a false-positive here. But it's enough that further study may be warranted.

Hold on

[barakn]

To all the commenters claiming we've already known this for centuries... no, we haven't. There's no reason to presume a priori that large objects occur in "showers" like the smaller (ash particle to pea sized) objects that make up familiar meteor showers. And astrostatisticians are very unhappy with the quality of the statistics in this paper, and they are suggesting the null hypothesis can't be rejecting using better statistical tools: https://astrostatistics.wordpr...