Tuesday, August 30, 2011

Climate forecasting: "A pinch of probability is worth a pound of perhaps"

Yesterday, Blake raised the issue of how weather events are used to argue for or against climate change. This gives me an opportunity to share some insight from Dr. David Pierce, a climate scientist from Scripps Institution of Oceanography who graciously gave his time to guest lecture in my climate change law class last week. I'm sure he dumbed things down for us, and I'm not sure I will accurately represent his explanation here -- but I'll do my best.

Normal curve Dr. Pierce very usefully framed his presentation in terms of climate change myths, calling out myths and showing how the science refutes them. One of the myths he called out was that “Someplace or other got colder … global warming can’t be true!” In response, he explained that forecasting a change in climate (i.e. a rise in global temperature, a rise in rainfall, a loss of glaciers) involves predicting a probability distribution. Take rainfall, as shown in the first graphic. If the prediction is that the amount of rainfall will remain unchanged in the future, then that implicitly means that about a third of places observed will experience dryer than normal years, a third will experience normal years, and a third will experience wetter than normal years. This happens because of natural variability; weather is a chaotic system. If only natural variability is acting, you'd expect about the same number of places to experience wet winters as dry. The fact that any one place experiences increased dryness or wetness does not contradict the forecast that rainfall will remain the same overall. The wetter and dryer places are part of the predicted distribution and are completely consistent with the forecast.

Offset curve Now consider a climate forecast that predicts less rainfall, as in La Nina years for San Diego. In this second graphic, the probability distribution is shifted to the left, indicating that more years than before will be drier.  But, importantly, the climate forecast still contains within it the prediction that a certain (small) number of years will be wetter. It is the same with climate change.  The forecast that the average global temperatures will increase means that more places will experience higher temperatures.  However, there will still be places where temperatures decrease, and that is part of the climate forecast. Those observations do not refute the forecast, they are consistent with it, as long as you observe many more warming places than cooling places.

The final graphic is a chart of temperature trends over the last 50 years, calculated in 5 degree latitude by 5 degree longitude boxes over the Earth.  It shows that most places Avg temps have warmed, but some have cooled.  And that is exactly what climate scientists would have predicted. In fact, you can see this shift in the distribution in record high and low temperatures as well. In the 1950s, the U.S. experienced about the same number of daily record high and low temperatures. But by the 2000s, there were twice as many record highs as record lows. We will still get record low temperatures decades from now, but there will be many record highs for each record low.

 -- “A pinch of probability is worth a pound of perhaps” -James Thurber (American Writer, 1894-1961)

 

 

-Lesley McAllister (Note: Thanks to Dr. Pierce for lecturing to my class and sharing these graphics. Any errors of explanation and interpretation are mine.)

http://lawprofessors.typepad.com/environmental_law/2011/08/climate-forecasting-is-probabilistic.html

| Permalink

TrackBack URL for this entry:

http://www.typepad.com/services/trackback/6a00d8341bfae553ef01539121841b970b

Listed below are links to weblogs that reference Climate forecasting: "A pinch of probability is worth a pound of perhaps":

Comments

Mahtso -- Yes, he actually addressed that issue really well, with particular reference to surfacestations.org. He explained that the stations are rated for their reliability (quality of siting), and the measurements from high quality stations correlate extremely well with measurements from poor quality stations. See here for a good discussion of this: http://www.skepticalscience.com/On-the-reliability-of-the-US-Surface-Temperature-Record.html . Thanks so much for your comment!

Posted by: Lesley | Aug 31, 2011 1:58:57 PM

Did he address how the temperatures were actually measured? I have seen the information at surfacestations.org and, based on the information shown there, I don’t see how the temperature data could be deemed accurate. But, I would be glad to hear why I am wrong.

Posted by: mahtso | Aug 30, 2011 12:07:45 PM

Post a comment