The paper “Systematic Benchmarking of Climate Models: Methodologies, Applications, and New Directions,” to which our colleague Valerio Lembo of CNR-ISAC contributed, was published on the American Geosciences Union’s journal Reviews of Geophysics. 

This work is part of the activities of the CMIP7 Model Benchmarking Task Team, which aims to define a standardized protocol for the analysis of climate simulations that will be used in the next IPCC assessment report.

Global and regional climate models are becoming increasingly advanced and complex. The observational data used to evaluate the ability of models to reproduce a realistic climate are becoming increasingly diverse and available over longer time intervals. The increasing complexity of climate models and the diversification and extension of observational data make the comparison between the former and the latter an increasingly complex task. 

Analytical or diagnostic methods have been developed to evaluate and compare model data both in the presence and absence of observations. These diagnostics have highlighted numerous improvements in the ability of the current generation of models to reproduce the current climate, but some issues remain where model results and observations do not agree adequately.

This paper describes the characteristics of some of the packages provided freely by scientists to the global community to quantify the disagreement between models and observations. Good practices on how to interpret the results obtained are discussed, and it is explained why the choice of observations and the selection of appropriate diagnostics are fundamental to making correct assessments.

Link https://doi.org/10.1029/2025RG000891Digital Object Identifier (DOI)
Picture 10.5281/zenodo.13934005 (Lembo et al. 2024)