Observation‐based detection and attribution of 21st century climate change

Abstract

The beginning of the 21st century has been widely designated a “global warming hiatus” or “pause” or “slowdown” because the rate of change of global surface temperature is a factor of two to three smaller than in the last half of the twentieth century. Indeed, observed global surface temperature (shown as black symbols in the top panel) warmed minimally. The rate of change of global surface temperature (shown as black symbols in the second panel) reached its lowest value in the 12‐year interval centered on 2007–2008. But while the terminology “hiatus” and “pause” may imply a reduction or cessation of the warming of the Earth by increasing concentrations of greenhouse gases, statistical models constructed from the observations over two different time periods (the space era, 1979 to 2017, shown as orange lines and the historical era, 1882 to 2017, shown as the blue lines) readily reproduce the observed surface temperature anomalies as simply the mitigation of ongoing anthropogenic warming by natural influences. The bottom panels show that the rates of change in temperature anomalies due to El Niño and La Niña (third panel) and solar irradiance (fifth panel), in particular, have negative values during the “pause” which counter much of the positive anthropogenic influence (sixth panel). Even though natural influences mitigated the anthropogenic warming globally, this was typically not the case regionally. On the right are the different spatial patterns of surface temperature rates of change that the space‐era statistical model attributes to each influence.

CMIP5 Climate Models Overestimate Cooling by Volcanic Aerosols

“…In that hemisphere the solar variability shows the second strongest observed effect on the southern hemisphere temperature, similar to ENSO on seasonal time scales, and about a quarter of the anthropogenic influence. A similar underestimation of solar variability in models was pointed out by Lean (2018).…”

“…The current set of climate models (CMIP5) are tuned (Golaz et al, 2013;Stevens, 2015;Suzuki et al, 2013) to reproduce reasonably well the past climate variability, including short-term cooling by volcanic forcing. It has been suggested that the models' cooling response to volcanic and anthropogenic aerosols may have been overestimated (Canty et al, 2013;Driscoll et al, 2012;Lean, 2018;Marotzke & Forster, 2015;PAGES 2k Consortium, 2019;Sato et al, 2018). While it has been shown that models fail to capture the dynamical response following volcanic eruptions (Driscoll et al, 2012), others suggest that the overestimate might be due to a too strong radiative forcing (Marotzke & Forster, 2015).…”

“…While it has been shown that models fail to capture the dynamical response following volcanic eruptions (Driscoll et al, 2012), others suggest that the overestimate might be due to a too strong radiative forcing (Marotzke & Forster, 2015). Climate models have also been shown to underestimate solar variability (Lean, 2018) and overestimate Arctic warming (Huang et al, 2019).…”

Recommended temperature metrics for carbon budget estimates, model evaluation and climate policy

“…Refs. 67,68 and Refs. 69,70 quantify the contribution of tropical Pacific variability to GBST using multiple linear regression.…”

Pacific variability reconciles observed and modelled global mean temperature increase since 1950

“…In regression-based approaches (Folland et al 2018;Foster and Rahmstorf 2011;Hu and Fedorov 2017;Johansson et al 2015;Lean 2018;Rypdal 2018;Saenko et al 2016;Schmidt et al 2014) the observed GMST, or its rate of change, is modelled by a number of predictors that are linearly combined. These studies differ in which predictors they include, whether the predictors are filtered, for example with a long-memory response (Rypdal 2018), an e-folding response profile (Folland et al 2018), or by using a mixedlayer model (e.g., Thompson et al 2008), and the temporal resolution of the predictors, i.e., monthly or annual.…”