Thursday, February 16, 2012

Data Doesn't Lie

An editorial in the current issue of Nature magazine ($) is so poorly written and twisted in its logic that I am shocked a respected journal such as Nature would publish it. The editorial is about methane emissions from fracking operations. Methane is widely recognized as a greenhouse gas - even by climate change deniers. But all of this is beside the point. I'm not discussing fracking, greenhouse gases or climate change today. Instead I'm focusing on two aspects that every scientist and engineer deals with daily: data, and making conclusions from data. Here's how not to do it:

"How clean is natural gas? Although it is often lumped in with coal and oil, many in the energy industry are at pains to point out that burning gas to generate electricity produces fewer greenhouse-gas emissions than does burning other fossil fuels. Certainly, countries claim reductions in carbon emissions when they switch from coal to gas, as Britain did on a large scale in the 1990s...Industry maintains that the problem has been exaggerated, and many scientists agree. Sorting fact from fiction has been difficult, however, because nobody had any independent data — until now.

As discussed on page 139, a study led by scientists from the US National Oceanic and Atmospheric Administration (NOAA), headquartered in Washington DC, and the University of Colorado in Boulder looked at methane and other emissions from a natural-gas field north of Denver, where fracking methods are used to open up sand formations. They estimated cumulative emissions from the field using not industry reports or conceptual models, but concentrations of pollutants in air samples. This is important because the atmosphere does not misrepresent data or make mistakes; nor does it bend to ideology or political will.

The data suggest that methane emissions from natural-gas operations could be substantially higher — and so be worse for global warming — than was thought. At works in the Denver-Julesburg Basin, methane emissions were roughly double the official estimate.

This will by no means settle the debate. The NOAA scientists had to make assumptions to convert atmospheric data to cumulative emissions from a vast energy complex. They readily acknowledge substantial uncertainty in their calculations, and estimate that between 2% and 8% of the methane produced from wells in the Denver-Julesburg Basin is lost to the atmosphere, with a best guess of 4%."
(emphasis added)

Wasn't that a perfect setup? "This is important because the atmosphere does not misrepresent data or make mistakes; nor does it bend to ideology or political will." So then how come "this will by no means settle the debate"? We have ideologically/politically neutral data? What's the problem?

This is a perfect example of the Achilles Heel of science: making a conclusion from data. It is such a fragile endeavor because it involves human beings and their thoughts and their biases. That is when assumptions are made and logic is applied. Worse yet, there is no guarantee that any conclusion reached is correct. This is why we have climate change deniers - not because of data, but because of the conclusions made from it.

How can Nature publish an editorial like this that is so removed from an understanding of how science works?


Eric F. Brown said...

Ahem. The data don't lie.

milkshaken said...

The result of our impartial analysis from a carefully selected dataset are unexpectedly in full agreement with our predetermined conclusions.

John said...

Busted! I'd offer up 100 mea culpas, but as lousy as my Latin is, I'm not sure if it shouldn't be 100 mea cuplae.