Benchmarking Historical Simulations of Leaf Area Index from CMIP6 Using Satellite Climate Data Records

The change of the Earth system is assessed with a suite of models developed by international agencies to understand climate trends and plan for mitigation actions. There is a need to assess the quality of vegetation status simulated by these Earth system models to ensure global carbon flux is accurately estimated. Leaf Area Index (LAI), the one-sided green leaf area per unit ground surface area, is a key variable characterizing vegetation status and has strong impacts on multiscale biogeochemical and biogeophysical interactions. Comparisons of the historical simulations of LAI provided by the latest Coupled Model Intercomparison Project (CMIP6) to satellite-based observations were made with monthly data for the period 1981-2014 at three common model spatial resolutions: 1, 2.5, and 5 degree. A total of 49 CMIP6 models were evaluated individually and as ensembles when multiple realizations were available. Two long-term global satellite time series, NOAA’s LAI Climate Data Record and Global Inventory Modeling and Mapping Studies (GIMMS) LAI3g, were used for benchmarking CMIP6 model accuracy. Basic comparison metrics of root mean squared error (RMSE), mean bias error (MBE), and R-squared were evaluated in addition to careful examinations of density distributions, seasonal patterns, and geospatial patterns of differences. This comprehensive analysis serves the larger Earth system modeling community by identifying where improvements to the simulations of terrestrial components are needed. It also provides insights on how to benchmark future CMIP experiments using long term satellite observations.