Saturday, December 3, 2011

CSU Hurricane Predictions: An Updated, Seven-Year Track Record

The 2011 Hurricane Season ended Wednesday. And, as last year, and as in most recent years, the annual CSU predictions were wildly off base. I’m not sure how accurate they claim to be, since few if any of the stories every Spring with their forecasts for the numbers of storms, hurricanes, and major hurricanes even mention the level of statistical confidence they believe those predictions to be.

Anyway, the Times-Picayune did run a story that included a chart tracking the CSU predictions for the past three years, But the chart isn’t included with the on-line version of the story, for some reason; so here's quick fuzzy scan of it:
Times Picayune Hurricane Seasons Predictions Retrospective

Last year I also made a chart of past hurricane season predictions because there were very few follow-up stories that followed up on these predictions after hurricane season, despite how widely covered the predictions are when they are announced each spring.

Here is my updated Track Record graphic of the CSU hurricane predictions for Hurricane Seasons 2005-2011:
Hurricane Predictions Track Record: 2005-2011

Again, wildly off, but as I said, I don’t know how accurate they purport their predictions to be.

My favorite numbers above are still the predictions for the 2006 and 2007 seasons. After under-predicting each figure for the 2005 season by over 50%, whatever factors they rely on apparently indicated that that season would not be a fluke. But that wasn’t the case and their predictions for the two following seasons were very high compared to what occured: by 66% or more for five of the six factors across the two seasons.

The bottom line: we can’t predict months in advance what one hurricane season is going to be like, so why should we believe predictions about what our climate is going to be like decades from now.

Full blog post...