Climate Change Attribution

Main Image

It is not unusual to hear of attempts to ascribe a certain extreme climate event to global warming. Much commentary has occurred in recent weeks on the unusual behaviour of the hurricane that devasted the Bahamas and how that behaviour was due to climate change. Similarly, impacts of our recent drought are linked by some to the warming effects of greenhouse gases over the last 50 or more years.

It is very difficult to make such claims with any scientific confidence given that similar natural extreme weather events are likely to have occurred in the pre-historical past, if not the historical past. But what is happening is that there is a concerted effort being made by many climate scientists to detect and attribute a climate change signal induced by anthropogenic processes in recent climate information and use that knowledge to improve our capacity to understand future climate change.

In July this year, a multi-authored consortium produced a paper in Nature Geoscience that addressed this issue by analysing a vast global collection of temperature-sensitive palaeoclimate records. The title of the paper is “Consistent multidecadal variability in global temperature reconstructions and simulations over the Common Era”. This is an important paper in that it looks at the relative contributions of various climate forcings and thereby assesses the ability of climate models to accurately simulate observed climate phenomena.

What the 17 consortium authors from a range of countries attempt to do is compile palaeoclimate proxy-based observations of temperature and climate forcings for the last 2000 years, the Common Era (CE). This enables them to reconstruct a global mean surface temperature (GMST) record showing fluctuations at a multidecadal time scale. This provides context for recent warming. Seven different statistical methods are used to reconstruct GMST. They compared an ensemble of 23 climate model simulations with the temperature reconstructions. Various uncertainties in the methods are outlined in the paper.

The study showed that the GMST during the first millennium of the CE was warmer than during the second, excluding the 20th century. Periods of reduced temperature variability occurred during both the overall cold seventeenth century and the relatively warm eleventh century. These are captured by both the reconstructions and simulations. They also showed the significant cooling trend before 1850 was followed by rapid industrial era warming.  However, “the warmest 10 year period of the past two millennia falls within the second half of the twentieth century”.

Their work suggests the dominant influence of external forcings on multidecadal GMST variability.  A “formal detection and attribution (D&A) analysis” was applied to “disentangle the influence of different forcing factors”. Three different factors were considered: volcanic, solar and greenhouse gas (GHG). The period 1300-1800 CE was selected for D&A application because it was a period when anthropogenic forcing is negligible and uncertainties are small compared with earlier centuries. By extending this analysis further towards the present it became clear that solar forcing is not detectable at multidecadal timescales. However, volcanic eruptions and GHG forcing produced responses that are consistent with both the reconstructions and simulations. Volcanic eruptions are found to coincide with “strong multidecadal cooling trends” that was followed by strong warming trends, noting that such trends occur more frequently than would be expected by chance. Temperature trends during the industrial era are shown to be outside the range of pre-industrial variability, especially the modern period of strong warming from the mid-1970s to today. All this suggests to these authors that climate models are “skilful in simulating the natural range of multidecadal GMST variability”.

What is apparent from this analysis of temperature-sensitive proxy data combined with simulation of global surface temperature trends, is the degree to which anthropogenic factors such as GHG emissions can influence present-day temperatures. The warming trend of more recent decades is quite distinctive in this record and thus can be distinguished from pre-historical climate shifts known to have occurred over the last 1000-2000 years. There is no doubting that these pre-historical climate shifts have had significant regional environmental and social impacts (e.g. so-called Little Ice Age or Medieval Warm Period).  But this study of climate detection and attribution more clearly defines the degree to which global warming is driven by anthropogenic factors. Yet it still does not answer questions as to how much any given extreme event can be attributed to such warming. The likelihood of a GHG contribution to natural variability becomes more apparent as confidence grows in what these authors highlight as the “extraordinary rate of the industrial era temperature increase” that greatly exceeds values expected from chance alone.

In coastal areas, the problem becomes one of how to disentangle impacts of driving forces as observed in pre-historical and historical records, from that which may be attributed to global warming (e.g. sea level rise; change in storm regimes). The non-linear nature of coastal systems makes this a complicated process given changes in sediment budgets and tidal dynamics affecting estuaries. Continuation of this “extraordinary rate” of temperature increase as documented by the authors of this paper in Nature Geoscience must make us more alert to how best we communicate that knowledge to coastal managers who are faced with the challenge of assessing present-day and future risk.

Bruce Thom

Words by Prof Bruce Thom. Please respect the author's thoughts and reference appropriately: (c) ACS, 2019, for correspondence about this blog post please email