Paying More Attention to Uncertainties

Mike Hulme's picture

The recent leaking of a draft of the Working Group I report of the IPCC’s Fifth Assessment Report has drawn attention to the changing levels of confidence and certainty the IPCC attach to various knowledge claims about the changing climate.  Some future climate risks have been downgraded (e.g. droughts, large-scale reorganisation of the Atlantic circulation) and others have been upgraded (e.g. ice melt from large ice-sheets).  

One of the consequences of the climate science controversies of the winter of 2009/10 (Climategate and the various criticisms of the IPCC) has been to encourage much more careful articulation by scientists of the uncertainties associated with climate science and climate predictions.  This is evident in communications from climate scientists themselves, as well as by some media commentators and reporters.  

Since 2009/10, my informal monitoring of climate change reporting suggests there has been more frequent depictions of new scientific studies in terms of ‘climate change may be less serious than previously thought’, balancing the previous dominance of the tag-line ‘climate change is worse than previously thought’.  At the very least, many predictions are now couched more carefully using the languages of uncertainty.  For example, the IPCC’s Special Report of Weather Extremes published in 2011 is much more cautious about attributing weather extreme trends to human influences than was the IPCC’s 2007 Report.

But is there empirical evidence for a more systematic adjustment in scientific practice and communication?  Using the Scopus database I searched all peer-review journal articles in the physical sciences for those dealing with ‘climate change’ (in title, keywords and abstract) for the period 1996 to 2012.  The overall number of such articles continues to rise, from about 2,000 in 2002 to about 6,500 in 2012.

But what proportion of these articles dealt specifically with uncertainty?  I searched this article corpus for those which included the words ‘uncertainty’ or ‘uncertainties’ in their title, keywords or abstract.  After remaining stable at around 6 per cent from 1996 to 2005, the proportion rose slightly to around 6.5 per cent between 2006 and 2008.  But by 2012 the percentage had risen to 9.1.

Comparing the two years immediately before and after Climategate (2008-2009 with 2011-2012; ignoring 2010) reveals the change more clearly.  The total number of ‘climate change’ articles increased by about 30 per cent, from 10,047 in 2008-2009 to 13,111 in 2011-2012.  But the number of these articles dealing with ‘uncertainty’ or ‘uncertainties’ increased by about 73 per cent (from 692 to 1,197).  In other words, articles dealing explicitly with climate change uncertainties increased at more than twice the rate of general climate change articles.

(One finds similar trends in ‘climate change’ articles which include the term ‘unknown’ in their title, keywords or abstracts, although the absolute number of such articles is an order of magnitude fewer.  In the life sciences, health sciences and social sciences and humanities, a smaller proportion of ‘climate change’ articles address ‘uncertainties’ compared to the physical sciences: proportionately only about half as many.  Nevertheless in these fields of study too, the climate change uncertainty articles have trended the same way post-Climategate.)  

This is clear evidence of a reflexive reaction by climate scientists, and other scholars, following Climategate to engage more directly with uncertainties in their research and to communicate this engagement in their professional publications.  Whether it means we are more or less certain about the causes of present climate risks or about the credibility of future climate predictions is another matter.  But it certainly shows how public controversy about climate science can alter scientific practice and communication.  How science is done cannot be detached from its wider cultural settings.