A paper published today in Nature Climate Change finds climate models have greatly exaggerated global warming over the past 20 years, noting the observed warming is “less than half” of the modeled warming.
The authors falsify the models at a confidence level of 90%, and also find that there has been no statistically significant global warming for the past 20 years. According to the authors, “The evidence, therefore, indicates that the current generation of climate models …do not reproduce the observed global warming over the past 20 years, or the slowdown in global warming over the past fifteen years.” The paper follows another recent paper falsifying climate models at a confidence level of greater than 98% for the past 15 years.
In terms of reasons for model underestimation, the apparent ‘preferred’ explanation of ‘the ocean ate it’ does not get any play here, other than in context of a brief consideration of natural internal variability. Their conclusion This difference might be explained by some combination of errors in external forcing, model response and internal [natural] climate variability is right on the money IMO, although I don’t think their analysis of why the models might be wrong was particularly illuminating. If you would like further illumination on why the climate models might be wrong, I refer you to my uncertainty monster paper.
Recent observed global warming is significantly less than that simulated by climate models. This difference might be explained by some combination of errors in external forcing, model response and internal [natural] climate variability.
At a glance
Global mean surface temperature over the past 20 years (1993–2012) rose at a rate of 0.14 ± 0.06 °C per decade (95% confidence interval)1. This rate of warming is significantly slower than that simulated by the climate models participating in Phase 5 of the Coupled Model Intercomparison Project (CMIP5). To illustrate this, we considered trends in global mean surface temperature computed from 117 simulations of the climate by 37 CMIP5 models (see Supplementary Information). These models generally simulate natural variability — including that associated with the El Niño–Southern Oscillation and explosive volcanic eruptions — as well as estimate the combined response of climate to changes in greenhouse gas concentrations, aerosol abundance (of sulphate, black carbon and organic carbon, for example), ozone concentrations (tropospheric and stratospheric), land use (for example, deforestation) and solar variability. By averaging simulated temperatures only at locations where corresponding observations exist, we find an average simulated rise in global mean surface temperature of 0.30 ± 0.02 °C per decade (using 95% confidence intervals on the model average). The observed rate of warming given above is less than half of this simulated rate, and only a few simulations provide warming trends within the range of observational uncertainty (Fig. 1a)
The inconsistency between observed and simulated global warming is even more striking for temperature trends computed over the past fifteen years (1998–2012). For this period, the observed trend of 0.05 ± 0.08 °C per decade is more than four times smaller than the average simulated trend of 0.21 ± 0.03 °C per decade (Fig. 1b). It is worth noting that the observed trend over this period — not significantly different from zero — suggests a temporary ‘hiatus’ in global warming2, 3, 4. The divergence between observed and CMIP5-simulated global warming begins in the early 1990s, as can be seen when comparing observed and simulated running trends from 1970–2012 (Fig. 2a and 2b for 20-year and 15-year running trends, respectively).
The evidence, therefore, indicates that the current generation of climate models (when run as a group, with the CMIP5 prescribed forcings) do not reproduce the observed global warming over the past 20 years, or the slowdown in global warming over the past fifteen years. This interpretation is supported by statistical tests of the null hypothesis that the observed and model mean trends are equal, assuming that either: (1) the models are exchangeable with each other (that is, the ‘truth plus error’ view); or (2) the models are exchangeable with each other and with the observations (see Supplementary Information). Differences between observed and simulated 20-year trends have p values (Supplementary Information) that drop to close to zero by 1993–2012 under assumption (1) and to 0.04 under assumption (2) (Fig. 2c). Here we note that the smaller the p value is, the stronger the evidence against the null hypothesis. On this basis, the rarity of the 1993–2012 trend difference under assumption (1) is obvious. Under assumption (2), this implies that such an inconsistency is only expected to occur by chance once in 500 years, if 20-year periods are considered statistically independent. Similar results apply to trends for 1998–2012 (Fig. 2d). In conclusion, we reject the null hypothesis that the observed and model mean trends are equal at the 10% level.