Do CMIP5 Models Skillfully Match Actual Warming?

A comparison the global mean surface temperatures (GMST) from 1970 to 2020 to the climate models from the fifth Coupled Model Inter-comparison Project (CMIP5) historical and projection shows good agreement. At first look, this appears to contradict the statement “The world has warmed significantly less than predicted by IPCC“ in CLINTEL’s ‘World Climate Declaration’. Nic Lewis published an article with the subtitle “Why matching of CMIP5 model-simulated to observed warming does not indicate model skill”. Lewis wrote that the good agreement “is perhaps unsurprising given that modelers knew when developing and tuning their models what the observed warming had been over most of this period.” Climate models on average have higher climate sensitivity than indicated by empirical observations, even when assuming that none of the observed warming was caused by the urban heat island effect or natural climate change. The set of CMIP6 climate models used for the 2021 IPCC working group 1 report are even more sensitive to increasing greenhouse gases than the CMIP5 models. The match of GMST to the climate model average would indicate model skill only if changes in the climate forcings of greenhouse gases and aerosols match the actual changes in those climate forcings. The climate forcings changes are the effect on global radiative flux at the top of the Earth’s atmosphere that causes a temperature change.

Lewis compares the effective radiative forcing (ERF) given in Annex III of the IPCC AR6 report to that used in the climate models over 1970-2020. From 1970 to 1988 the models’ ERF was too high, and after 2007 it was too low. The linear trend change of the actual ERF over the period was 2.66 W/m2 while that of the climate models was 1.92 W/m2. That is, the models used only 72% of the actual climate forcing to fudge the surface temperatures to roughly match the observed warming. Based on the first and last decade, the ratio of actual to model forcing is 1.46. The discrepancy is largely due to the models’ incorrect aerosol forcing both before 1988 and after 2007. The models’ over-sensitivity was canceled out by increasing the ERF much less than the IPCC estimated ERF increase.