January 5, 2016

Study: Low Bias in Previous Assessments of Climate Sensitivity

How much the climate will respond to a given increase in atmospheric CO2 is an essential question for climate science and policy. People skeptical of the case for CO2 reductions will often selectively analyze the recent literature to argue that the climate will not change as much under increased CO2 than previously thought or represented in climate models.  Such arguments assert that certain estimates of the climate’s sensitivity to CO2 are preferable (c.f. Global Warming Policy Foundation). But a new study led by scientists from the NASA Goddard Institute of Space Studies, published last month in Nature Climate Change, shows that the studies favored by skeptics estimate too low a sensitivity.

Scientists talk about this response in terms of climate sensitivity. One measure is the equilibrium climate sensitivity (ECS), which is the long-term warming that would result from a sustained doubling of atmospheric CO2. The last IPCC assessment reports that ECS is likely (2/3 chance) between 1.5 and 4.5 degrees C. Scientists have had difficulty being more precise than that. The present range is the same reached in 1979 by the National Research Council Climate Research Board. But the modern range of estimates is based on more lines of evidence and better methodology (analyzing recent observations, better models, and investigations into pre-human climate conditions), which give about the same picture for now.

Skeptics often advance estimates of climate sensitivity from studies of the instrumental period (roughly the last ~150 years) over other methods. This is the period when human activity began significantly altering the climate. It is also when we have increasingly good measures of atmospheric temperature, ocean heat content, and the various agents that warm and cool the climate. Skeptics are right to assert that these studies and their results should be carefully considered, because of the quality of data we have available. Many scientists that skeptics might not otherwise agree with would argue the same.

The idea behind these studies is that by knowing how much the climate was forced, and how much the climate has warmed, you can infer the climate sensitivity and quantify ECS by using simple heat balance arguments. Recent analyses of modern observations have estimated ECS values the lower end of the IPCC range with median values of 2 degrees C or less for ECS (c.f. Otto et al. 2013, Lewis 2013).

Such studies, however, make simplifying assumptions and may leave out significant factors. For instance, the spatial and temporal extent of different forcings (external or internal factors that that influence climate) matters. The Northern Hemisphere, for instance, has a lot of land, where temperature responds more quickly to forcing than it would over the ocean. So a given forcing will more rapidly change temperature when applied in the Northern Hemisphere than the Southern, which is almost all ocean.

This matters because only some forcings are evenly distributed. CO2 concentrations are relatively well-distributed in the atmosphere, because its atmospheric lifetime is long compared to the time it takes to mix the atmosphere. So even though it is primarily emitted in the Northern Hemisphere, human-enhanced CO2 forcing is pretty uniform globally. Aerosols (small particulates in the atmosphere that have a net cooling effect), on the other hand, have a short atmospheric lifetime and aerosol emissions from fossil fuel burning are concentrated in the Northern Hemisphere. Since the forcings act over different areas, where the climate response will be different, the climate sensitivity one infers from each forcing is different, even if the forcings themselves are of similar magnitude.

In this new study, the authors use a climate model to individually simulate the temperature and climate impact of anthropogenic aerosols, volcanic aerosols, ozone, solar insolation, land surface changes, and greenhouse gases. Previous studies have not accounted for the efficacy of different forcings in this way.  The study measured the “efficacy” (that is, how much each forcing affects heat accumulation in the climate for a given strength) of each forcing by comparing the global change in temperature when only that forcing was acting to change temperature with the change in global temperature when only CO2 is acting on the climate.

This new study found that accounting for the efficacy of historical forcings changes the best estimate of climate sensitivity from analysis of the instrumental period. Previous methods give estimates of sensitivity to CO2 that are biased low. The authors find that the best estimate for ECS increases from about 2 degrees C to about 3 degrees C when these effects are considered. That difference represents a substantial reconciliation with the climate sensitivities shown by climate models and other lines of evidence.

In the supplementary material, the authors also show how their revisions alter the probability distribution of ECS. Their results substantially increase the high-end, low-probability, estimates of ECS. For instance, previous methods estimated the 95th percentile for ECS between 3.4 and 5.4 degrees C (depending on the details of the calculation). This study has estimates ranging from 7.2 to 19.4 degrees C. The sensitivity of that upper bound in this study has important implications for risk-based climate change policy.

One study is rarely the final word on any scientific issue, and the authors of this study are careful to point out that their result depends on one climate model and its own set of assumptions. Accordingly, I expect we will see more work along these lines in the future. Nevertheless, this new paper challenges the idea that modern observations are evidence of a significantly lower climate sensitivity.

[Banner Image: Haze over China via NASA GISS]