On the reliability of computer-based climate models

Abstract

Since 1850 the global surface temperature of the Earth has warmed by about 0.9°C. The computer climate models adopted by the Intergovernmental Panel on Climate Change (IPCC), such as the General Circulation Models of the Coupled Model Intercomparison Project Phase 5 (CMIP5), projected that the global surface temperature could rise more than 1.5°C by 2030 and more than 4-5°C by 2100 relative to the pre-industrial period (1850-1900) because of anthropogenic greenhouse gas emissions. These computer projections are being used to justify expensive mitigation policies to drastically reduce CO2 emissions due to the use of fossil fuel. However, these models must be validated before their interpretation of climate change could be considered reliable. Herein, I summarize recent scientific research pointing out that these GCMs fail to properly reconstruct the natural variability of the climate throughout the entire Holocene and at multiple time scales such as: (1) the Holocene Climatic Optimum (9000-6000 years ago) with the subsequent cooling from 5000 years ago to now; (2) the large millennial oscillations observed throughout the Holocene that were responsible, for example, for the Medieval Warm Period; (3) several shorter climatic oscillations with periods of about 9.1, 10.4, 20, 60 years; (4) the climate change trend after 2000 to date, which the models greatly overestimate; and many other patterns. These different pieces of evidence imply two main facts: (1) the models’ equilibrium climate sensitivity (ECS) to radiativeforcing, such as to an atmospheric CO2 doubling, is overestimated at least by a factor of 2, which implies a more realistic ECS between 1°C and 2°C; (2) there are a number of solar and astronomical forcings that are still missing in the models or are poorly understood yet. Consequently, these GCMs are not physically reliable for properly interpreting past and future climatic changes. Alternatively, semi-empirical climatic models should be used. Data analysis found that the climatic natural variability is made of several oscillations from the decadal to the millennial scales (e.g. periods of about 9.1, 10.4, 20, 60, 115, 1000 years) and others. These oscillations are coherent with solar, lunar and astronomical oscillations. A semi-empirical climate model that makes use of these oscillations plus a reduced ECS reconstructs with great accuracy the climate variability observed since 1850 and projects a very moderate warming until 2040 and a warming lower than 2°C from 2000 to 2100 using the same anthropogenic emission scenarios used for the 21st-century climate simulations of the CMIP5 models. This result suggests that climatic adaptation policies, which are far less expensive than the mitigation ones, could be sufficient to address the consequences of climatic changes that could occur during the 21st century.