Guest commentary from Richard Millar (U. Oxford)
The recent Lewis and Curry study of climate sensitivity estimated from the transient surface temperature record is being lauded as something of a game-changer – but how much of a game-changer is it really?
The method at the heart of the new study is essentially identical to that used in the much discussed Otto et al. (2013) study. This method uses a simple equation of the energy balance of the climate and observations of global temperature change and estimated ocean heat uptake anomalies along with a time series of historical radiative forcing (code), in order to make inferences about the equilibrium climate sensitivity (ECS – the ultimate equilibrium warming resulting from doubling carbon dioxide concentrations) and its shorter-term counterpart the transient climate response (TCR – the warming at point of doubling after carbon dioxide concentrations are increased at 1% per year). [Ed. An overview of different methods to calculate sensitivity is available here. The L&C results are also discussed here].
Lewis and Curry use an updated radiative forcing estimate over that used in Otto et al along with slightly different assumptions over the periods used to define the observational anomalies. They use the latest IPCC numbers for radiative forcing and global temperature changes, but not the latest IPCC ocean heat content data. Their result is a 5 – 95% confidence interval on ECS of 1.1–4.1K and for TCR is 0.9-2.5K. These confidence intervals are very consistent with other constraints, from paleo or emergent observations and with the range of GCM estimates. For the TCR, arguably the more important measure of the climate response for policy makers as it is a better predictor of cumulative carbon budgets, the 5-95% confidence intervals are in fact almost identical to the AR5 likely range and similar to the CMIP5 general circulation model (GCM) estimated 5–95% range (shown below).
Figure 1: The 5-95% confidence ranges for transient climate response (TCR) taken from various studies as in Fig. TS.TFE6.2 of IPCC AR5 WG1. The green bordered bar at the top of figure is the estimated 5-95% range from the CMIP5 GCMs. blue bordered bar at the top of the figure is the 5-95% range from the Lewis and Curry (2014) study. The grey shading represents the AR5 consensus likely range for TCR.
There is a difference between the Lewis and Curry 17-83% confidence intervals and the IPCC likely ranges for TCR and ECS. However, for all quantities that are not directly observable, the IPCC typically interprets the 5-95% confidence intervals as likely ranges to account for the possibility that the model used to derive the confidence intervals could be missing something important (i.e. non-linearity that would not be captured by the simple models used in Otto et al and Lewis and Curry, which can particularly be a problem for ECS estimates using this method as the climate feedback parameter is assumed to be constant in time) [IPCC AR5 WG1 Ch10.8.2]. In this case, accounting for more complete surface temperature changes (Cowtan and Way, 2013), or the hemispheric imbalance associated with aerosol forcing (Shindell, 2014), or updates in the OHC changes, may all shift the Lewis and Curry distribution. [Ed. This expert judgement related to structural uncertainty was also applied to the attribution statements discussed here before].
The median estimate of the TCR from Lewis and Curry (1.3K) is towards the lower end of the IPCC likely range and lower than the CMIP5 median value of around 1.8K. A simple way to understand the importance of the exact TCR value for mitigation policy is via its impact on the cumulative carbon budget to avoid crossing a 2K threshold of global surface temperature warming. Using the Allen and Stocker relationship between TCR and TCRE (the transient climate response to cumulative emissions) we can scale the remaining carbon budget to reflect different values for the TCR. Taking the IPCC CO2-only carbon budget of 1000 GtC (based on the CMIP5 median TCR of 1.8K) to have a better than 2 in 3 chance of restricting CO2-induced warming to beneath 2K, means that emissions would have to fall on average at 2.4%/year from today onwards. If instead, we take the Lewis and Curry median estimate (1.3K), emissions would have to fall at 1.2%/year. If TCR is at the 5th percentile or 95th percentiles of the Lewis and Curry range, then emissions would need to fall at 0.6%/year and 7.1%/year respectively.
Non-CO2 emissions also contribute to peak warming. The RCP scenarios have a non-CO2 contribution to the 2K peak warming threshold of around 0.5K [IPCC AR5 WG1 – Summary for Policymakers]. Therefore, to limit total warming to 2K, the CO2-induced contribution to peak warming is restricted to around 1.5K. This restricts the remaining carbon budget further, meaning that emissions would have to fall at 4.5%/year assuming a TCR of 1.8K or 1.9%/year taking TCR to be equal to the Lewis & Curry median estimate of 1.3K (assuming no mitigation of non-CO2 emissions).
While of some scientific interest, the impact for real-world mitigation policy of the range of conceivable values for the TCR is small (see also this discussion in Sci. Am.). For targets like the 2 K guide-rail, a TCR on the lower end of the Lewis and Curry and IPCC ranges might just be the difference between a achievable rate of emissions reduction and an impossible one…
- N. Lewis, and J.A. Curry, "The implications for climate sensitivity of AR5 forcing and heat uptake estimates", Climate Dynamics, vol. 45, pp. 1009-1023, 2014. http://dx.doi.org/10.1007/s00382-014-2342-y
- A. Otto, F.E.L. Otto, O. Boucher, J. Church, G. Hegerl, P.M. Forster, N.P. Gillett, J. Gregory, G.C. Johnson, R. Knutti, N. Lewis, U. Lohmann, J. Marotzke, G. Myhre, D. Shindell, B. Stevens, and M.R. Allen, "Energy budget constraints on climate response", Nature Geoscience, vol. 6, pp. 415-416, 2013. http://dx.doi.org/10.1038/ngeo1836
- K. Cowtan, and R.G. Way, "Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends", Quarterly Journal of the Royal Meteorological Society, vol. 140, pp. 1935-1944, 2014. http://dx.doi.org/10.1002/qj.2297
- D.T. Shindell, "Inhomogeneous forcing and transient climate sensitivity", Nature Climate Change, vol. 4, pp. 274-277, 2014. http://dx.doi.org/10.1038/nclimate2136
- P.J. Durack, P.J. Gleckler, F.W. Landerer, and K.E. Taylor, "Quantifying underestimates of long-term upper-ocean warming", Nature Climate Change, vol. 4, pp. 999-1005, 2014. http://dx.doi.org/10.1038/nclimate2389
- M.R. Allen, and T.F. Stocker, "Impact of delay in reducing carbon dioxide emissions", Nature Climate Change, vol. 4, pp. 23-26, 2013. http://dx.doi.org/10.1038/nclimate2077