Do CMIP6 earth system models outperform their predecessors in simulating global vegetation changes?

• Latest satellite data benchmark CMIP6 ESMs’ vegetation simulation performance • CMIP6 ESMs surpass CMIP5 models with improved global vegetation simulations • Improved climate processes and vegetation responses boost model performance Earth System Models (ESMs) are crucial tools for projecting long-term, large-scale vegetation changes. Previous studies have evaluated the ability of Coupled Model Intercomparison Project Phase 5 (CMIP5) and Phase 6 (CMIP6) ESMs to simulate global vegetation changes; however, the accuracy of these models remains unclear. Long-term remotely sensed Leaf Area Index (LAI) products serve as vital benchmarks for assessing the performance of ESMs in simulating vegetation changes (e.g., the GIMMS LAI3g has been selected as reference data by the IPCC AR6). However, notable uncertainties among LAI products can significantly impact evaluation outcomes. The recently developed GIMMS LAI4g dataset demonstrates higher accuracy than its predecessor and other LAI products, thus offering a more reliable ESMs evaluation benchmark. Leveraging this dataset, we assessed 16 CMIP5 and 22 CMIP6 ESMs in reproducing observed vegetative changes and in expressing LAI sensitivities to climate factors. We found that CMIP6 ESMs show significant improvements ( p < 0.05) over CMIP5 in simulating vegetation changes during the historical period (1982—2005) when corroborated with GIMMS LAI4g. This enhanced performance can be attributed to better simulation of climate variables and more realistic representations of LAI response to climate variables. While CMIP5 and CMIP6 exhibit similar spatial patterns in LAI sensitivities, CMIP6 models generally outperform CMIP5 which overestimates the negative temperature sensitivity and positive precipitation sensitivity. Under all future scenarios, CMIP6 ESMs exhibit greater consistency in projecting vegetation distribution and smaller discrepancies in LAI sensitivities. After filtering for better-performance models, the CMIP6 projects a consistently larger LAI increase than the CMIP5 across all scenarios. Our comprehensive evaluation based on a reliable LAI product enhances the understanding of ESMs’ performance and global vegetation changes. These findings aim to guide better ESM selection and to improve projections of vegetation changes, thus having the potential to inform climate change mitigation strategies.