This spring the US National Weather Service is rolling out a new “climate normal data set” – average temperatures, precipitation, highs and lows, and other variables including what is an “average” hurricane season based on a new 30 year reference period. At first glance the changes for daily weather may not look terribly significant, but they are quite consistent with the fact that average temperatures have been warming over the last century. For Savannah International Airport, the average temperature in May based on 1981-2010 data was 73.3 degrees F. The average temperature using 1991 to 2020 is 74.1 F, a 0.8 degree difference. Across the year that’s about the average increase. Only one month, October, saw a decrease in average temperature, from 59.3 F to 59.1 F. The biggest change was 1.5 F in December, followed by 1.2 F in January. Here’s a plot of the data …
So what does this have to do with hurricanes? As part of the assessment of the “new normal”, NOAA has re-evaluated what constitutes an “average” hurricane season. The old number, based on 1981-2010, was 12 named storms. The new definition says an average season will have 14 named storms. Here’s the summary:
While I think the overall reassessment is needed, I think the period is too short in both cases and biases the data, especially for hurricanes. I think a 50 year baseline of 1961-2010 makes more sense. The problem using the 1991-2020 period for hurricanes is that mostly covers the period of enhanced hurricane activity including the peak Atlantic Multidecadal Oscillation (AMO). On the flip side, NHC is counting storms that in past years would not have been given names (both due to better observations systems and changed procedures), so the high bias may be covering other sins in the statistics. Either way, if you hear that 2021 is an “average” year, keep in mind that’s two more storms than what an average year was in the past.
Which brings to mind one other point: averages don’t really tell you much about extremes. As as simple numerical example, the average of 30, 40, and 50 is of course 40, with extremes of plus or minus 10. But the average of 10,40, and 70 is also 40 – but with extremes of plus or minus 30! So you need both the average and another parameter like the variance to really understand a data set. This is a concept known in statistics as the (f)law of averages … so don’t be like this guy: