Guest post by David Middleton
My State is currently in the grip of a very severe drought…
Professor Andrew Dessler, an atmospheric sciences professor at our nation’s greatest university, recently authored a column about our drought in the Bryan-College Station Eagle…
Published Tuesday, August 30, 2011 12:05 AM
Paying the price for climate change
By ANDREW DESSLER
Special to The Eagle
Texas Gov. Rick Perry stirred up controversy on the campaign trail recently when he dismissed the problem of climate change and accused scientists of basically making up the problem.
As a born-and-bred Texan, it’s especially disturbing to hear this now, when our state is getting absolutely hammered by heat and drought. I’ve got to wonder how any resident of Texas — and particularly the governor who not so long ago was asking us to pray for rain — can be so cavalier about climate change…
[…]
I know that climate change does not cause any specific weather event. But I also know that humans have warmed the climate over the past century, and that this warming has almost certainly made the heat wave and drought more extreme than it would otherwise have been.
[…]
Dr. Roy Spencer had an interesting take on Dr. Dessler’s column in his blog…
Dessler vs. Rick Perry: Is the 2011 Texas Drought Evidence of Human-Caused Climate Change?
September 5th, 2011 by Roy W. Spencer, Ph. D.
One of the most annoying things about the climate change debate is that any regional weather event is blamed on humans, if even only partly. Such unscientific claims cannot be supported by data — they are little more than ambiguous statements of faith.
[…]
Andy Dessler recently made what I’m sure he thought was a safe claim when faulting Texas Gov. Rick Perry for being “cavalier about climate change” (as if we could stop climate from changing by being concerned about it).
Dessler said, “..warming has almost certainly made the (Texas) heat wave and drought more extreme than it would otherwise have been.”
This clever tactic of claiming near-certainty of at least SOME effect of humans on weather events was originally invented by NASA’s James Hansen in his 1988 Senate testimony for Al Gore, an event that became the turning point for raising public awareness of “global warming” (oops, I’m sorry, I mean climate change).
The trouble is that climate change theory predicts changes, up and down, in just about anything you can imagine. So, anything unusual that happens anywhere, anytime, is deemed “consistent” with global warming.
[…]
According to Dr. Spencer the current national drought conditions are not exceptional; nor is there any statistically significant trend…
And, while Texas is experiencing a record-setting drought; the “record” is just over a century-long and there is no trend at all…
The lack of a trend in the precipitation data made me wonder… Just how often should we be setting precipitation records if the annual variation is random?
The record only goes back to 1895. Does anyone know how often record highs and record lows should be broken in such a short time series?
At a record length of 117 years, there was a 1% chance of setting a new record high in the 117th year…
The probability, pn(1), that the nth observation of a series xm= x1, x2, … xn has a higher value than the previous observations [pn(1) = Pr(xn > xi |i < n)] can be expressed as:
pn(1)= 1/n (1)
provided the values in series are iid random variables.
(Benestad, 2003)
The cumulative probability says that 5 records should have been set between 1895 and 2011.
So, let’s have a look at the data. I downloaded the summer precipitation data for Texas from NCDC’s “U.S. Climate at a Glance” page…
In order to analyze the frequency of record excursions, I plotted the absolute value of the annual summer precipitation anomaly along with an “expected records” curve…
There have been 5 record excursions from the average annual summer precipitation – Exactly what there should have been in a random series of numbers. And the records have occurred with the expected frequency of a random series of numbers. The fifth record excursion should have occurred between 1945 and 2030 – It occurred in 2007.
I calculated the absolute value of the anomaly… The records were for maximum excursions from average. So the five “records” were for extremes in both directions.
If you parse it by maxima and minima, there have been five records in each direction… Which is exactly what you would expect from a 117 year long record of random numbers.
“Precipitation: The balance of evidence does not support the assertion that the rainfall deficit since October 2010 was made larger or more likely by global warming.”
David Middleton:
Thankyou for your excellent, concise and useful article.
Dessler’s comment seems to be an example of a recent meme because there have been warmists recently making similar comments (e.g. with respect to hurricanes) on WUWT. Their assertions have been countered by pointing out that statistics do not support their claims, but they have failed to understand the point. Warmists (especially warmist trolls) have little knowledge and understanding of science and mathematics – especially statistical analyses – so it is difficult to point them to a clear explanation they can comprehend. Your article povides an example of such a clear explanation.
Again, thankyou.
Richard
How do you define gross ignorance? 144 Aggies in the same room! 😀
I don’t think that is true. For instance, if you’re in Scotland – go outside, right now. It’s raining.
Having lived in Scotland I know I stand a pretty good chance of being right, no matter when I say that. Does that make the statement “It’s always raining in Scotland” accurate? I think not.
What is the difference between the Aggies and Rice Crispies? Rice Crispies know what to do in a bowl.
Why do Aggies like smart women? Opposites attract.
Why did the Aggie get fired from the M&M plant as a quality control inspector? He kept throwing out all the W&W’s!
CAGW has not a thing to do with this. 1930’s level volcanic activity…
Returning to “normal” rainfall will never break a drought….
….floods break droughts
…you just can’t win
Mike Bromley the Canucklehead says:
September 13, 2011 at 8:32 pm
“Warming has almost certainly made the (Texas) heat wave and drought more extreme than it would otherwise have been.”
Almost Certainly. And how do you know this, Dr. Dessler? I have a general question concerning the general heat problem: How does a fraction of a degree of warming over a century translate into many tens of degrees in a heat wave? I never quite understood the non-answer to this question lurking in that CC-causes-extremes camp. And what fraction of that degree is “normal”?
##################################
Pos feedback. One is the main one which is water vapor. If it concentrates more infrared is reraditated back to the earth increasing the local temperature.
Texas had a pos feedback from drying out its land. A little higher temperature evaporates all the more water vapor sooner. A high is created around the really dry area of land and it becomes self perpetuating. This would of happened with AGW, but adding more temperature doesn’t help.
You can have negative rainfall. Its called pan evaporation rate.
Rainfall – sum( evaporation rate)
Pan evaporation rates are interesting in and of themselves from a climatological perspective. It takes ALOT of energy flow to evaporate water and this is indicative of solar flux and surface temps.
And the evaporation rate has been going down for some time. Now there may be lots of “surface station” like problems with the measuring sites.
http://en.wikipedia.org/wiki/Pan_evaporation
“Over the last 50 or so years, pan evaporation has been carefully monitored. For decades, nobody took much notice of the pan evaporation measurements. But in the 1990s scientists spotted something that at the time was considered very strange; the rate of evaporation was falling. This trend has been observed all over the world except in a few places where it has increased.
As the global climate warms, all other things being equal, evaporation will increase and as a result, the hydrological cycle will accelerate.The downward trend of pan evaporation has been linked to a phenomenon called global dimming. In 2005 Wild et al. and Pinker et al. found that the “dimming” trend had reversed since about 1990 ”
In fact, and in corroboration, a number of pluvial lakes that dried up 8000 years ago would be reborn if not for human use of the surface waters.
http://en.wikipedia.org/wiki/Lake_Lahontan
“Surprisingly, the watershed feeding Lake Lahontan is not thought to have been significantly wetter during its highstand than it is currently. Rather, its desiccation is thought to be mostly due to increase in the evaporation rate as the climate warmed. Recent computer simulations (using the DSSAM Model and other techniques) indicate that if precipitation and evaporation rates within the watershed were maintained at their historical yearly maximum and minimum, respectively and if diversions of the Truckee River ceased, the Ice Age extent of Lake Lahontan could return.”
TomB says:
I believe you are correct that Scotland will have a different local climate than other places.
#############
Climate
http://en.wikipedia.org/wiki/Climate
Climate encompasses the statistics of temperature, humidity, atmospheric pressure, wind, rainfall, atmospheric particle count and other meteorological elemental measurements in a given region over long periods. Climate can be contrasted to weather, which is the present condition of these elements and their variations over shorter periods.
Interesting, David. Thank-you for the response. So 2011 wasn’t a record for precipitation in Texas?
Rattus says:
“Y’all might want to read what one of Anthony’s co-authors has to say on this topic.” (http://blog.chron.com/climateabyss/2011/09/texas-drought-and-global-warming/)
Yes, we all might want to do exactly that. You probably wouldnt want to read it, which perhaps explains why you apparently did not. It contains the following conclusions, which clash dramatically with your alarmist world view:
Precipitation: The balance of evidence does not support the assertion that the rainfall deficit since October 2010 was made larger or more likely by global warming.
So, 0% the Texas drought was caused by ‘global warming’, from any source.
Temperature: Compared to long-term averages of summer temperature, the rainfall deficit accounted for about 4 F of excess heat and global warming accounted for about 1 F of excess heat. Warmer temperatures lead to greater water demand, faster evaporation, and greater drying-out of potential fuels for fire. Thus, the impacts of the drought were enhanced by global warming, much of which has been caused by man.
So, 80% of the increased Texas temperature during the drought was not caused by ‘global warming’ from any source. Only 20%, less than 1 degree F, of the Texas summer heat was caused by ‘global warming’ from all sources. This means that only some fraction of one degree F of the Texas summer heat, if any at all, could possibly be attributed to ‘global warming’ from anthropogenic sources. Thus, the magnitude of the obligatory claim of possible ”enhancement” of the Texas drought by anthropogenic greenhouse gases is certainly very small, likely very very small, and may in fact be zero.
Thanks for directing us’all to that. It is nice to have corroberation for our view that anthropogenic ‘global warming’ doesn’t warrant a rats ass of genuine concern, even regarding extreme weather events.
@richard S Courtney
No, not a meme but a deliberate strategy. Get the lie out first – by the time it’s rebutted, the LSM has broadcast it world-wide, and moved on to other things.
‘A Lie Can Be Halfway Round the World Before the Truth Has Got Its Boots On’.
The CAGW spinmeisters strike me as being on the same level as ambulance-chasing lawyers when they go off linking natural disasters to their hobgoblin. There are real people dealing with real world problems and all these creeps can do is take advantage of their distress to try and scare them more. Pawns in their game to score political points. Since the lawsuit against companies for causing hurricane Katrina failed maybe the shyster scientists will encourage one for drought & wildfires. And a hockey stick will descend from the heavens proving this is the worst in 2000 years!
2011 set the record low for summer precipitation in Texas.
2007 was the record high for summer precipitation in Texas.
2007 was the record maximum excursion (absolute value) from the average summer precipitation in Texas.
All three records were the fifth records set for their respective time series.
I looked at the histogram, and while it does bear resemblance to a normal distribution, my opinion is that the tails look a little fat to be truly normal. In other words, extreme events occur more frequently than one might expect from a gaussian PDF. However, to settle the argument for this particular series of data, perhaps you should test the normality of this histogram. There are not many geophysical records of any sort that display a normal distribution. We encounter lognormal or hyperbolic distributions often.
Interestingly, using the same technique and same data source, there have been 8 new record highs for temperature and only 3 new record low temperatures in this same period. This would seem to be rather clear evidence SUPPORTING climate change.
Not only was this year a record high, it was “off-the-charts” high. 86.6 F is 3.8 standard deviations above the mean, which is beyond 0.001 chance for a normal distribution. (Yes, I know this will not necessarily be a normal distribution, but there is no way to consider this anything but an extreme event)
Gary Pearse says:
September 14, 2011 at 8:53 am
No, i believe it is a summation of the harmonic series, not a logarithm.
1 + 1/2 + 1/3 + 1/4 …..
That said, I disagree with David’s numbers a bit (although I like them better than the ln approximation). I think the numbers David gives are for the first year when a new record might be expected, not the typical year when a record might be expected. The mostly likely year should be the year the line on the graph crosses the various integers, not the year the line is at least 1/2 way to the next integer.
Here are the first, typical, and last years to expect a new record
1895-1895-1895
1896-1898-1900
1901-1905-1912
1913-1924-1944
1945-1977-2030
2031-2120-2266
David Middleton says:
September 14, 2011 at 12:33 pm
Actually, 2011 was the 6th record low for rainfall.
1895 10.12
1896 4.65
1910 4.29
1934 3.73
1956 3.48
2011 2.44
Not only that but it was an unusually large change. As more records get set, the change should be getting smaller (if nothing else, eventually the record will be close to zero and the next record can only be slightly closer to zero. Other than the very first change, this is the largest change in the record. This would suggest that 2011 was indeed an unusually bad summer drought.
@Tim Folkerts,
Even if this was the 6th record low, the 4th record low missed its expected year by an even wider margin.
Year of Record vs Expected Year
1895 10.12 1895 – On time.
1896 4.65. 1896 – On time.
1910 4.29. 1901 – 9 years late.
1934 3.73. 1913 – 23 years late.
1956 3.48. 1945 – 11 years late.
2011 2.44. 2031 – 20 years early.
There is no expectation for a pattern of diminishing variability in an iid random series… Or even in a short stochastic series of such a short record length.
@Kevin Kilty, re test for normality for histogram onTexas annual precipitation since 1896.
I’m using the standard statistics from Microsoft Excel (TM) so have the kurtosis and skewness. Kurtosis for this histogram at 0.298, rounded to 0.3, and skewness is 0.188, rounded to 0.2. Therefore, the data set is very close to a Normal Distribution, with the positive skewness indicating a slight emphasis to the right, compared to a Normal Distribution.
I looked in detail at the data, and found that the data points less than the median were as close to a Normal Distribution as one could expect with such a low population (116 data points).
Compared to a perfect distribution (Normal Distribution), the number of data points below the mean are as follows:
Perfect . . . . . .Actual
Minus 1 Sigma . . . . . 39.5 . . . . . . .39
Minus 2 Sigma . . . . . 15.8 . . . . . . .17
Minus 3 Sigma . . . . . 2.4 . . . . . . . 2
And, the number of data points above the mean:
Perfect . . . . . .Actual
Minus 1 Sigma . . . . . 39.5 . . . . . . .44
Minus 2 Sigma . . . . . 15.8 . . . . . . .10
Minus 3 Sigma . . . . . 2.4 . . . . . . . 4
What is interesting to me, since this thread is about drought in Texas, is the 3-sigma value on the low end, the drought side. 3-Sigma is 12.8 inches of rain, compared to the lowest measured rainfall in the data set of 14.8 inches. There have been only 2 years out of the 116 years on record where rainfall fell between 2-Sigma and 3-Sigma on the drought side; a totally normal occurrence. For any alarm to be sounded legitimately, there must be either more years of 3-Sigma rainfall (less than 17.9 inches), or a year in the 4-Sigma (less than 7.75 inches), or both.
I do have some doubts about this data, though, and pose a question to the WUWT audience if anyone knows the answer. The first is how accurate is this data, since there is a prolonged and severe drought in Texas, yet the data from NCDC shows only one year below average since 2004? The second is, how can the data already show a value for 2011, with 3 -1/2 months yet to go in the year? NCDC has the annual rainfall for Texas at 27.95 inches for 2011. I suspect the answer to the second question is that the data shows the 12-month period starting in February and ending in January. Thus, the 2011 data actually is ending in January 2011
David Middleton says: “September 14, 2011 at 7:17 pm There is no expectation for a pattern of diminishing variability in an iid random series… Or even in a short stochastic series of such a short record length.”
Sure there is such an expectation in many data sets – although the size of the effect depends on the distribution. This effect is strongly obvious in a uniform distribution (which is indeed iid). For data that could fall between 0-1, the first change will on average be fairly large — around 1/3. After a few records, the record will be above 0.9 and it will be impossible to get change of 1/3. The change HAS TO get smaller and smaller.
For a normal distribution, there is no upper bound, so the expectation of diminishing returns is not so clear. But a check of 8 series of 10,000 random normal data (typically including ~ 8 new records) that shows:
* 1 set had a slight trend toward larger changes, while 7 had stronger trends toward larger values
* all sets had the smallest change in the second half of the records.
* all but 1 set had the largest change in the first 2 records.
Those all sound like trend to have the large changes early and the small changes later.
Austin quotes from wikipedia computer model simulations of historical precipitation and evaporation levels, etc. that “…the Ice Age extent of Lake Lahontan could return.” If those computer models incorporated a return of another Ice Age and its following interglacial, then they would definitely show Lake Lahontan returning; but, no one needs a computer model to predict that, as a general knowledge of climate change will suffice. Also, the memories left by my great, great grandfather indicate that Lake Lahontan was nowhere to be found during his time, which was well before the waters of the Truckee River were diverted, and Indians freely roamed the land.
Thanks, David, for this lucid and informative post. I’m sure you’re aware, but some readers here may not be, that the source you quote for the probability distribution equation (Benestad, 2003) is now a member of the Real Climate team. Some (although maybe not all) members of that team seem to have been quite capable of doing worthwhile and objective science. It’s sad that they are now devoting so much effort to the intolerant advocacy campaign centred round the demonstrably corrupt IPCC. What a waste of talent!
oops … In the last post I meant
* 1 set had a slight trend toward larger changes, while 7 had stronger trends toward
larger valuessmaller changes.The one trend toward larger values was not statistically significant. A couple of the trends toward smller values were statistically significant.
The point is that after you have already set a number of records, it becomes harder to break the record, so the change will tend to get smaller. (There is one statistical distribution where I suspect this might not be true, but it is a distribution that is mostly of academic interest – http://itl.nist.gov/div898/handbook/eda/section3/eda3663.htm)
Of course, if the distribution changes, then this will no longer be true. So, for example, if climate truly is getting warmer, then the new records might not show this diminishing trend.