Marc Morano forwarded my post Lies, Damned Lies And Statistics to Dr. Desai at the University of Wisconsin. Below is his response.
From: Ankur Desai
Sent: Wednesday, April 13, 2011 12:54 PM
To: Marc Morano
Sent: Wednesday, April 13, 2011 12:54 PM
To: Marc Morano
Subject: Re: Your claims challenged on temp trend…Lies, Damned Lies And Statistics | Real Science
I apologize for the delay in response, and perhaps my reply is old news by now, but I hope I can still have something useful to say.
As a caveat, allow me to first note that I answered the reporter’s questions using numbers provided by the Wisconsin Initiative on Climate Change Impacts (http://www.wicci.wisc.edu ) and produced by a detailed analysis of temperature station trends by graduate student Shawn Serbin (Forest and Wildlife Ecology) and Profs. Chris Kucharik (Agronomy Dept) and Dan Vimont (Atmospheric Science). My own research focuses more on understanding how ecosystems respond to climate variability (and vice versa) and I answered the reporter’s question since I happened to be around and as a climate scientist, am well versed in much of this research.
If you’d like I’d be happy to get you a copy of their paper – Serbin, S.P. and C.J. Kucharik, (2009). Spatio-temporal mapping of temperature and precipitation for the development of a multi-decadal climate dataset for Wisconsin. J. Applied Meteorology and Climatology, Vol. 48, 742-757.
So that’s not a cop-out, but just a note that while I support the claims made by WICCI based on my own understanding of temperature trends regionally and globally, that I did not directly produce this analysis and so would need more time to try replicating some of the trends reported in the blog. If you think that would be worthwhile, I could attempt to do so, but it will have to be in several weeks as the end of the semester tends to be a pretty busy time.
My short answer is that – yes indeed it is possible to “cherry-pick” trends and find time periods of a decade or two that stand in contrast to longer-term trends (and vice versa, trends greater than the long-term trend). Natural climate variability on the scale of a decade is well observed and even modeled reliably (in magnitude if not in exact timing given the random nature of much of how these variations are initiated) and reflects a variety of process such as changes in ocean mixing and atmospheric pressure patterns, even fluctuations in solar output. That said, models and observations agree that with continued forcing on the climate system (from any cause), trends become apparent once you average out past a couple of decades (i.e., over several cycles of natural internal variability). A recent paper ran a climate model, for example, over I think a thousand years with continually increased forcing (an unrealistic scenario, but more of a test of the model), and the model never goes more than a decade or so out with significant periods of no temperature increase or cooling.
Indeed, it is also true that the 1930s (Dust Bowl) was above normal, and possibly even similar in temperatures to today for Wisconsin. There is good evidence to support a anomaly in Pacific Ocean temperatures during that time period that shifted regions of high pressure, which was further exacerbated by the large-scale expansion of agriculture occurring at that time. I can send you a paper about that if you’re interested (Cook et al., 2009, Proc. Nat. Acad. Sci.). That said, as Pacific ocean waters cooled, the warm temperatures and dry conditions in the Great Plains receded.
Also, even with a warmer mean temperature, cold, even exceptionally cold seasons can occur. They just do so with lower probability. Gerry Meehl, a scientist at the National Center for Atmospheric Research, has shown quite nicely for example that the ratio of record highs set to record lows set has moved from 1:1 to about 2:1 over the past 50 years, even when correcting for the decrease in total number of highs and lows that would naturally occur with time and changes in number of stations. See: http://www2.ucar.edu/news/1036/record-high-temperatures-far-outpace-record-lows-across-us
Here’s an analogy that might make more sense (but one I just made up, so bear with me if it’s not perfect): Imagine a bathtub being filled with water. With time the level of water will rise. Now imagine we put a excited toddler (not a bad description of the climate system) on the other end who splashes about and causes waves to form back and forth. The level over any area of the tub goes up and down, but eventually goes up. The splashing also removes some of the water and over short time, may actually cause the water level to fall, but if the splashing is not inordinately violent and the spigot is open wide enough, the water level will rise. Certainly we can debate the ratio between the spigot and the splashing intensity!
The climate system is even more fascinating. There is some evidence that the exceptionally cold winters in the eastern U.S. this winter were triggered by a shift in the Atlantic pressure patterns caused by an unusually warm and open (less ice) Arctic. Indeed, while it was a cold winter (3-6 C below average winter) for many populated regions of the Northern Hemisphere (Europe and eastern North America), it was exceptionally warm further north (they were up to 21 C above average in northern Canada !).
Some sources for this last claim:
Our climate system never ceases to fascinate me and it’s nice, despite the regrettable rancor from many people (us scientists included) on the climate policy debate, that we have the resources and general interest from everyone to explore this system that sustains us!
So the short answer is that it is not trivial to define the averaging period (in time or space) for a trend. That said, once you get beyond 30-years, the trend becomes more about forcings (such as greenhouse gases or aerosols or solar variation) and less about internal variability.
As a final point, I should also note that in the short newspaper article, I did not ascribe a singular cause to the regional temperature trend in Wisconsin, but rather noted that it is quite similar to global trends; I believe above those trends for winter. I am perfectly willing to accept that the temperature trends in Wisconsin from 1950 are a mixture of local weather variability, land use change, changes in storm tracks as well as anthropogenic change, but the science is pretty strong that the global trends are mostly a response of increasing greenhouse gases (modulated by increasing aerosols) and so it doesn’t surprise me that those trends can be seen on smaller scales, too.
Now how these trends continue in the future, whether these trends are a cause for concern for humans or ecosystems, and whether mitigation policies would have an influence or are worth the cost are other worthwhile debates to have, but not the point of the specific question asked by the reporter or the issues brought up by the blog.
I hope this reply helps a bit and you are welcome to share it with others. I am happy to answer more questions (as long as you give me enough time 🙂 ). If you ever have specific questions especially related to issues around the carbon cycling, land ecosystem responses to climate variation, and how vegetation influences climate, please feel free to contact me.
Ankur R Desai
Asst. Professor, Atmospheric & Oceanic Sciences Dept.
Faculty affiliate, Center for Climatic Research, Nelson Institute for Environmental Studies
University of Wisconsin – Madison