Comments on: Notice Anything? http://bluecrabboulevard.com/2007/07/14/notice-anything-2/ Summum nec metuas diem, nec optes - Marcus Valerius Martialis Tue, 02 Aug 2011 22:27:18 +0000 hourly 1 http://wordpress.org/?v=3.0.4 By: Pirate’s Cove » >>Americans Never Quit » Sorta Blogless Sunday Pinup http://bluecrabboulevard.com/2007/07/14/notice-anything-2/comment-page-1/#comment-63129 Pirate’s Cove » >>Americans Never Quit » Sorta Blogless Sunday Pinup Sun, 15 Jul 2007 14:34:26 +0000 http://bluecrabboulevard.com/2007/07/14/notice-anything-2/#comment-63129 [...] Blue Crab Boulevard discusses when hurricanes attack [...] [...] Blue Crab Boulevard discusses when hurricanes attack [...]

]]> By: chuck http://bluecrabboulevard.com/2007/07/14/notice-anything-2/comment-page-1/#comment-63120 chuck Sun, 15 Jul 2007 04:14:04 +0000 http://bluecrabboulevard.com/2007/07/14/notice-anything-2/#comment-63120 If the storms are evenly distributed over the time from June 1 through November, and they probably aren't, then they would follow a Poisson distribution. The probability of k storms in a given period would then be exp(-L)*L^k/k!, where L is the average number of storms expected in that period. For example, if two storms were the average expected during the six weeks from June 1 to mid July, then the probability of zero storms (k = 0) is about exp(-2) ~= .135. Small, but not terribly small, about the same as getting three heads in a row flipping a coin. This approximation breaks down for large k because you can 'overlap' storms and have hundreds a day. Differently said, storms are forced to keep a certain distance from each other in time and space. Anyway, the same formula would also apply in the absence of trends for the yearly average of the number of storms in that period over a number of years. If el Nino and el Nina do affect the number of storms, then there are trends, so it all goes to H*ll ;) Sorry for the blather, but you tickled my thoughts. If the storms are evenly distributed over the time from June 1 through November, and they probably aren’t, then they would follow a Poisson distribution. The probability of k storms in a given period would then be

exp(-L)*L^k/k!,

where L is the average number of storms expected in that period. For example, if two storms were the average expected during the six weeks from June 1 to mid July, then the probability of zero storms (k = 0) is about exp(-2) ~= .135. Small, but not terribly small, about the same as getting three heads in a row flipping a coin.

This approximation breaks down for large k because you can ‘overlap’ storms and have hundreds a day. Differently said, storms are forced to keep a certain distance from each other in time and space. Anyway, the same formula would also apply in the absence of trends for the yearly average of the number of storms in that period over a number of years. If el Nino and el Nina do affect the number of storms, then there are trends, so it all goes to H*ll ;)

Sorry for the blather, but you tickled my thoughts.

]]>