This excerpt from Dr. Roger Pielke Jr. on his blog likely will cause Bill McKibben and the McKibbenites at 350.org to ramp up the rhetoric over the post Sandy “new normal” yet again, because as we’ve learned, new factual information doesn’t dent their resolve.
Pielke writes:
Earlier this year, Jessica Weinkle, Ryan Maue and I published a paper in the Journal of Climate on trends in global landfalling hurricanes (a PDF can be found here as well). At the global level the data is good from 1970. Our analysis covered through 2010. With 2012 almost in the books I recently asked Ryan if he could provide an initial tabulation of the 2012 data (note that the data could be revised from these initial estimates, and 2012 is still not quite over). […]
Below is the dataset from 1970 first presented in our paper, updated with 2011 and 2012 included. In short, 2012 is just about an average year with 14 total landfalls (15.4 is average) of which 4 (initially, but could change, 4.6 is the average) characterized as major.
Here are some updated factoids summarized from the data:
- Over 1970 to 2012 the globe averaged about 15 TC landfalls per year (Category 1-5)
- Of those 15, about 5 are intense (Category 3, 4 or 5)
- 1971 had the most global landfalls with 30, far exceeding the second place, 25 in 1996
- 1978 had the fewest with 7
- 2011 tied for second place for the fewest global landfalls with 10 (and 3 were intense, tying 1973, 1981 and 2002)
- Five years share the most intense TC landfalls with 9, most recently 2008.
- 1981 had the fewest intense TC landfalls with zero
- The US is currently in the midst of the longest streak ever recorded without an intense hurricane landfall
- The past 4 years have seen 12 major landfalling hurricanes, very low but not unprecedented — 1984-1987 had just 11. The most is 35 (2005-2008).
- The past 4 years have seen 51 total landfalling hurricanes, on the low end — the least is 41 (1978-1981) and the most is 80 (four periods, most recently 2004-2007).
- There have been frequent four-year periods with more than 25 landfalling major hurricanes, or more than a 100% increase of what has been observed over the past 4 years.
Anyone who’d like to argue that the world is experiencing a “new normal” with respect to tropical cyclones is simply mistaken. Over the past 4 years, the world is actually in the midst of a very low period in tropical cyclone landfalls — at least as measured over the past 43 years.
===============================================================
Read his full post here
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Those aren’t factOIDS (‘like’ a fact) – rather, just plain old proper FACTS. Don’t sell the data short. ;-D
I have and always have had a problem with this “new normal” term and way of thinking. 1. Those presenting such things are almost always ignorant of long time history, things like geologic time. 2. Those wishing to postulate a new anything tend to have no real clue what the old normal was/is. 3. This term and its use is simply more RP baffle gab without sufficient history it is meaningless. Time and time again people like Dr. Pielke Jr. point out the shortcoming but it apparently falls on the ears of the believers and gets rearranged by their cognitive dissonance.
“RP baffle gab” ?
So what about the number of named storms in the Atlantic Basin? Up or down?
Bill McKibben, where are you?
This article disputes your “new normal” meme. Better still, IPCC AR5 SOD also says you are wrong. Isn’t it about time you published a retraction? Or are you prepared to assert your opinion of the science in opposition to both the skeptics and the IPCC at the same time?
Or will you just slink away in silence?
Nevertheless, things tend toward the normal value, so one should expect that a period of low landfalls (like the one we are in now) will be followed by a higher rate -nearer to the normal or even above normal for some period of time. Just look at the chart and you can see it. But that will be normal too.
If you search: Arctic Hurricanes Play Large Role in Climate
you will see some additional information on how previous climate models have badly failed in some critical features of why the higher Northern latitudes warm or cool, and newer better models show a likely cooling period coming. Combining the lack of reasonable prediction of magnitude of solar effects (the new IPCC leak) and the lack of previous understanding of large Arctic Hurricanes effects on climate models make all previous models totally worthless.
I hope that Gore et al continue with their extreme weather alarmism.
It is so completely and unequivocally rebutted by the actual data, it can only destroy what is left of their reputation for honesty
The temperature records are far less clear cut. I’m firmly in the skeptical camp but I recognize there is at least some scope for reasonable people to disagree over many details.
Having said that, there’s really no data to demonstrate that the proposed remediation measures are an effective use of resources. And guess what? Joe Public is rapidly realizing this.
It’s a long, slow haul. But the truth will out!
“Over the past 4 years, the world is actually in the midst of a very low period in tropical cyclone landfalls”
=====
So when it returns to normal, warmists will really think it’s the end of the world.
But,.. but this is climatology, therefor users of empirical data that contradicts the models, are required to correct that data to match the gospel.
The conclusions suggested by actual measurements can never be more correct than the modelling output.
By UN-IPCC logic Bill McKibben is right, its reality that is mistaken.
I love the failure of logic that has the activists babbling about weather, its as old as human speech and as transparent as Chicken Little.
Ugh You bad, thunder come. Give me gift, chant and dance make lightning leave.
I don’t want to rain on anybody’s parade, but…
http://www.pbase.com/azleader/image/147905297.jpg
That link is to a line graph of Pielke’s data above I made with Excel, including 2011 and 2012. It plots only the number of “Intense” hurricanes (Cat 3,4 and 5) by year since 1970. Then I added the linear trend line in red.
As you can easily see, there is a linear increase in the number of “intense” hurricanes over time just like that AGW alarmists say.
The reason for the trend is that 2004-2008 had lots and lots of intense hurricanes. Though it is true that intense hurricanes are down over the last 4 years, the bottom line is that, in general, the alarmists are right on this one.
azleader, is your conclusion correct though? We get told to expect this sort of thing as the “new normal,” yet the number of extreme events, while supporting the Alarmist meme from 2004 to 2008, has fallen off considerably from 2009 onwards. While the trend you plot is upwards, the data is highly inconsistent, which is not what we should be expecting from the “new normal,” while CO2 is ramping up like a bitch still.
No, I don’t think you support the alarmists, but I just don’t think this data is saying anything other than we can see natural variability. The alarmists will see what they want to see, of course…
So, what does the p value and r2 look like for your trend line? Since you don’t say, It probably isn’t statistically significant which means the trend is statistically indistinguishable from a zero trend.
azleader
hate to rain on your parade, but for the decade 1991-2000 I count 58 severe hurricanes (red) and for the decade 2001-2010 I count 56 huricanes, for a decrease of 2 major hurricanes compared to the previous decade. Now, tell us about that bottom line once again, please and thank you.
Good question… I don’t know. 🙂
All I know for sure is the trend line shows a growth of over 1 “intense” hurricane making landfall per year since 1970 and that includes the drop off over the last 4 years.
TimiBoy says:
December 17, 2012 at 11:49 am
azleader, is your conclusion correct though?
Also great questions… I don’t know that either. 🙂
MattS, you are trying to make me actually understand stuff. LOL!!!!
My math is very rusty. However, I was able to muddle through to calculate a p2 value of 0.064497 using Excel’s RSQ function for hurricane numbers and years. I’m not even sure what that result means.
Excel has a TDIST() function to calculate the p-value, but I can’t figure out the input parameters and the tutorial I tried didn’t help. 🙁
————————————————-
MattS says:
December 17, 2012 at 11:52 am
So, what does the p value and r2 look like for your trend line?
“Bill McKibben, where are you?”
Why do you care what he says other than to example him as a lunatic in a pseudo-scientific political cult? Once you start laughing at them and pointing at their denialism of the lack of temps change, therefore their pr firm-guided obsessions with measurements other than from thermometers, you win, they lose the “debate” that they really don’t want to have.
If you really want to get under their skin, compare them to G. W. Bush’s economic policies–magical and insipid.
azleader,
The trend in intense landfalling global tropical cyclones is not significant at the 2-sigma level. The trend is 0.052 +- 0.031 TCs/yr with an R^2 of 0.064. It is just barely significant at the 90% level. The problem is that the data suffer from systematic issues. For example, the data start at a low point in the 60-year AMO cycle and end near a high point, so the trend is more likely an indicator of how much Atlantic TC activity affects global landfalling data.
The trend in overall global landfalling TCs is not significant either: -0.011 +- 0.047.
Mpainter… I hear ya, but here is the deal…
Pilke’s own data creates a trend line of greater than +1 in “intense” hurricanes since 1970. That is 42 data points. I didn’t manipulate anything at all… it is what Excel produced.
MattS makes a GREAT point about whether or not the result is statistically significant. That is what is important. If it isn’t then the trend line is meaningless. That part I don’t know. My rusty math skills aren’t good to figure it out yet.
But this much I do know, it can be proven one way of the other.
My sense is that a +1 when there is only a range of +8 “intense” hurricanes to start with over 42 data points sounds to me like it like it is probably statistically significant. That is yet to be determined.
azleader;
All I know for sure is the trend line shows a growth of over 1 “intense” hurricane making landfall per year since 1970 and that includes the drop off over the last 4 years.
>>>>>>>>>>>>>
That’s what happens when you try and cram a linear trend through cyclical data.
I walk up the side of a hill, recording my altitude from the foot of the hill to half way up every minute. At the half way point, I draw a linear line through my data and conclude that I will be leaving earth orbit in about 2 weeks. At the crest of the hill, I draw a new line through all my data and it turns out itz 3 weeks, not 2. I cross the top of the hill to the other side, assess my data again, looks like orbit in 4 weeks not 3. But I’m still proving that I’m going into orbit. Half way down the other side, I draw yet another line, and discover that orbit is now 6 weeks away, not 4.
At some point it may dawn on my that drawing a linear trend through cyclical data is just dumb. Unless I went to climaclownatology school….
Compare and contrast:
We say there are no significant trends in extreme weather & keep presenting the evidence.
They say that the weather is getting more extreme.
—————
Keeping your eye on the thimble: they present recent weather events as evidence but shy away from presenting trends, especially long-term.
Their message is not aimed at skeptics. It’s aimed squarely at the public, most of whom are unaware that they are being lied to.
http://wattsupwiththat.com/reference-pages/climatic-phenomena-pages/extreme-weather-page/
Thanks, Brian…
I was verified your R2 value with Excel, so I trust your conclusion that Excel’s trend line is not statistically significant. I wondered about than before my original comment. However, I’m gonna have to dig a little deeper to understand it better for myself.
—————————————————————-
Brian says:
December 17, 2012 at 12:38 pm
The trend in intense landfalling global tropical cyclones is not significant at the 2-sigma level. The trend is 0.052 +- 0.031 TCs/yr with an R^2 of 0.064.
Always need error bars on your trend line azleader. Random numbers will almost always show a trend also (it is unlikely that they will be exactly zero trend). It is the trend in comparison to the random variation that establishes if the trend is significant.
These days they name every squall in the Atlantic to ramp up their declining statistics.
This site could benefit from a set of reference pages on hypothesis testing with probability/likelihood estimates.
The “p2” metric, I believe, is the Pearson product-moment statistic, squared.
That metric is generally presented and believed to be one way to say the amount of change in one measure that is accounted for by another. A p2 of .06 would thus say that six percent of the observed variation in number of intense hurricanes per year is explained by variance in year.
Knowing that the trend is toward more hurricanes per year going from past to present, it means that there are more hurricanes as time goes forward, but that time as a predictor only explains six percent of the variation in the number per year.
Either other factors account for the remaining variability, i.e., explaining why any one year has more or fewer than any other year, or the categorication of hurricane severity is not measured well enough to get a decent measure, or both.
To consider this question, you are investigating a hypothesis: that severe hurricanes are getting more frequent across time. The contrary assertion would be: severe hurricanes are not getting more frequent across time. If the contrary were true, the Pearson correlation would be zero.
Now, in reality, values rarely ever turn out to be zero. There are many factors involved, and the result, if there truly is a zero relation, most likely won’t be zero, but a number very close, which is so small it is just reflecting random variation, and not some genuine relation.
That is where the significance test comes in. It uses variation in the number of severe hurricanes per year to see what numbers would be expected if there were only random variation around no changes from year to year, and tests how likely it is theat the obtained value shows a reliable difference from zero/normal variation, or is just normal variation around the usual annual number.
If you take the data and put it in a histogram, it gets more easy to grasp this. You can use an image search to see what a “histogram” is, then imagine plotting the number of severe hurricanes each year on the histogram, then seeing where the data points are, in the histogram’s distribution, for the recent few years.
If they are simply in the neighborhood of the rest of the points, you know you will not get a pearson r to be statistically ‘different’ from the rest of those data points.
Brian has results of a test of the slope, rather than a test of correlation between each year and each year’s number of severe hurricanes.
That tests whether the line, from left to right, is notably differently slanted or sloped upward, versus the “null hypothesis” of the line being flat.
Same logic applies. If there truly is no trend, your actual slope from almost any data will not be exactly zero point zero zero zero zero, but some insignificant number very close to zero, and the trend line, when graphed, may not be impressively sloped.
The floor in my home, we would all say, is not perfectly flat. If we measured it enough, we might find it slopes up from north to northwest, or something like that, but maybe by only a fraction of an inch across 40 feet. That rise is so small it is irrelevant.
Worse yet, the rise could be error in measurement.
My floor could actually be sloping the oher way, and msmt error is greater than the actual trend.
The likelihood test essentially asks this question, and declares how likely it is that the obtained data reflects a departure from the value of no-trend.
At or beyond two sigmas is fairly impressive, if the measures are good. That means the obtained data will be observed only about five percent of the time. That suggests that the trend would be notably divergent from flat – worth worrying about; otherwise, as Brian notes, the obtained data would be expected ten times of every hundred, so not very uncommon.
All of that is “internal” likelihood math.
At the same time, it is likely that, since the data start in 1970, tehse storms have been detected more reliably. If the likelihood of detecting a severe hurricane has increased across time, such as due to more consistent and better technology, better infrastructure, etc., then there is a surveillance bias that would make it appear as if more are occurring across time – we are looking more, hence we find more.