Texas Tall Tales and Global Warming

Last week the national media was full of stories about how global warming has made Texas heat waves TWENTY TIMES more probable.  We are talking about hundreds of stories in respected national media outlets (including the NY Times, the Washington Post, and even the Seattle Times).   These stories were all based on an article in the Bulletin of the American Meteorological Society (DID HUMAN INFLUENCE ON CLIMATE MAKE THE 2011 TEXAS DROUGHT MORE PROBABLE?  with lead authors David E. Rupp and Philip W. Mote of the Oregon Climate Change Research Institute and some British colleagues...found here...scroll down to page 12).

The trouble is that this study is flawed (I will explain why) and its scary conclusions are insupportable.   This is important story:  about the hyping of global warming, about poor research being published, about the media jumping on sexy, scary global-warming stories.  And most worrisome of all..this is not an isolated incident.

Before I go further, let me stress that I believe that human-induced increases in greenhouse gases will cause the planet to warm significantly over the next century.  The impacts could be both profound and serious.   But exaggerating the impact of human-induced warming today and in the past only serves to weaken the efforts of the meteorological community to provide information society needs to make rational decisions.  If you cry wolf too many times and are proven wrong it is bad for your credibility.

So lets consider the Rupp/Mote et al. study.   Texas had an extraordinary six-month  heat wave and drought in 2011...no doubt about it.   The question is whether we can ascribe this event to global warming..human-induced or otherwise.

Now to examine this issue, the authors of this article examined temperatures and precipitation for March through August and June through August over Texas using both observations (from the National Climatic Data Center) and simulations by the UK Meteorological Office’s Hadley Center Atmospheric General Circulation Model 3P (HadAM3P).   HadAM3P is a global atmospheric model used to simulate climate.  Specially, they ran the climate model many times for the decades of 1960-1970 and 2000-2010.   This is called an ensemble.  Each ensemble member is started with a slightly different initial state in order to get some handle on the uncertainty of the forecast. Totals of 171, 1464, 522, and 1087 ensemble members were analyzed for 1964, 1967, 1968, and 2008, respectively.  Why they used different number of ensemble members for each year is never explained.  Furthermore, they selected those specific years because all were La Nina years.  The idea was that La Nina/El Nino variability is an important natural source of climate/weather change and could skew the results, so they wanted to insure that they were comparing apples to oranges.  It turns out they forgot some other fruit (more later!)

The following graph is from figure 8 of their paper, showing a plot of temperature versus precipitation over Texas for March through August using both observations (National Climate Data Center, NCDC, 1895-2011) and the climate model (HadAM3P) ensembles for 1964 and 2008.  For both observations and the model, there is a tendency for drier years to be warmer.  But there are real warning signs that the climate model is out to lunch (or out to whatever climate models do when they are not doing their job!).

First, the climate model (blue and red circles) is MUCH warmer and drier than reality...and the observations included the dry/warm conditions of the 1930s.  A serious bias.  Furthermore, the relationship between temperature and precipitation in the model and observations are VERY different...very different slope, with the model warming up much more quickly as lesser precipitation than the observations.  Clearly, the model is not simulating Texas climate very well.

Rupp, Mote et al., Figure 8

With this flawed GCM simulation, the authors should have been hesitant in going further in the analysis, but they decided to use the biased/flawed modeling results to determine whether the chances of heat waves are increasing over Texas.

Their next figure shows a return time analysis of the model temperatures over Texas.

Specifically, using the collection of simulations for each of four years (1964, 1967, 1968, and 2008) they calculated the typical number of years one would have to wait until a certain mean March-August mean temperature occurs.  So a mean of 25C would be expected to occur every 1-2 years in a 2008 climate and every 3-4 years for the 60s.  27C is expected to happen every 10 years for the climate of 2008 and perhaps once in 500 years (extrapolating the graph) for a 1960s climate.

 Furthermore, 100-yr return period MAMJJA and JJA heat events under 1964 conditions (roughly 26.5C)  had only 5- and 6-yr return periods, respectively, under 2008 conditions. It is this graph that was the basis of their statement:

"extreme heat events were roughly 20 times more likely in 2008 than in other La Niña
years in the 1960s"

It is this statement that has made headlines across the country.   Headlines you shouldn't believe.

Let me explain why.

Now I already have shown you that the model "climate" was way too warm and dry, and its simulated relationship between temperature and precipitation was all wrong.   But it is worst than that.  Looking at their figure, you can see the average model temperatures (March-August) in 1964 (blue circles) are roughly 24.5C, while the model 2008 temperatures (red circles) are approximately 26.25C...so about 1.75C warmer (give or take .25C for my reading of the graph).   (This kind of information SHOULD have been explicitly stated in the paper).

 So what is really happening in Texas?   How correct was the model?  Mark Albright of the University of Washington acquired and plotted the NCDC observations over Texas and plotted the average Texas temperatures for March-June, and July-August (see below) for 1895-2011.  In March through May there is a weak upward trend (perhaps .5F, .3C) over the entire period. The trend over June to August is much less.  The second figure also shows how anomalous 2011 was...it was an extreme year that was completely outside the envelope of variability of the previous decades.  In this figure one does not see a trend consistent with global warming, which slowly increased starting during  the mid-70s.

March-May

June-August

The bottom line:  the actual observations show the temperatures over Texas have warmed by a perhaps a few tenths of a degree C since the mid-1960s, while the GCM model used by Rupp/Mote et al had major warming (1.5-2 C).  Clearly, one can not trust the model and the conclusions reached in this paper are unsupportable.

And folks it is even worse than this.  There are other modes of natural variability in the atmosphere, such as the Atlantic Multidecadal Oscillation (AMO).  The AMO, which is associated with the temperature of the Atlantic Ocean, has a substantial impact on the weather of eastern N. America., including heat waves and droughts.   During the mid to late1960s this cycle was in the negative (cool) phase, while in the 2000s it has been in the warm phase (associated with heat waves and droughts over the Midwest)--see graph.

Thus, the authors picked dates that would maximize the warming signal associated with natural variability, irrespective of global warming due to greenhouse gas emissions.


Moral of this Sad Story

This situation is so disappointing on so many levels.   It is disappointing the peer review process has allowed this paper to be published in a well known and prestigious journal.  I have learned from personal experience that articles noting major global warming effects fly through the review process with only cursory examination, while papers with a more nuanced view of the issue are given a hard time.

It is disappointing that the media distributed these results so widely...with headlines...throughout the nation and world.  The faults noted above were easy to find..is there a way for media folks to evaluate the materials they headline?  Even worse, sometimes the media publicizes materials that are not even published in peer-review journals or are made available only in press releases.  They need to act more responsibly and secure the resources (e.g., trained science journalists) needed  insure the rigor of the materials they spotlight.

This is only one if series of weak global warming scare articles--I can cite many more.  My own sensitivity to the issue came five years ago when certain folks (including a coauthor of the Texas article) were saying that global warming was resulting in the rapid loss of the Cascade snowpack (which has not declined in 30 years by the way).  These folks think they are doing society a favor by hyping global warming impacts now and in the past.  They aren't.  Most of the impacts of global warming due to anthropogenic greenhouse gas increases are in the future and society will not believe us if you cry wolf now.

The Rupp/Mote et al paper  will only hammer the credibility of the scientific community at a time when society needs to be taking global warming seriously.

PS:  Hard to believe but on the Sunday evening NPR All Things Considered they talked about the Texas study.