Anthropogenic influence on global warming for effective cost-benefit analysis: a machine learning perspective

In the climate domain, attribution is the process of determining the external forcings which are more likely to be responsible of the climate change which, in turn, affects global economic growth. These factors influence the climatic system by altering its properties including, for instance, the radiative balance. In this context, investigating the role of anthropogenic forcings toward natural factors in the global warming of the last decades is of paramount importance. Global climate models (GCMs) applied to attribution studies showed that the temperature increase in the second half of the twentieth century can be mainly imputable to the emissions of anthropogenic greenhouse gases. In this study we resort to a data-driven approach based on machine learning with the aim of analyzing the relationship between global temperature anomalies and natural and anthropogenic forcings. Our empirical findings fully agree with the results of GCMs attribution studies, and further shed light on the natural and anthropogenic drivers that, on a multivariate basis, exert the major influence on the global temperature.