| Summary: | Evaluation of vegetation structure and distribution simulations in Earth system models (ESMs) is the basis for understanding historical reconstruction and future projection of changes in terrestrial ecosystems, carbon cycle, and climate based on these ESMs. Such assessments can also provide important information of models’ merits and shortcomings or systematic biases, and so clues for model development. Vegetation structure and distribution in ESMs are primarily characterized by three variables: leaf area index (LAI), tree height, and fractional coverage of plant functional type (PFT). However, for the ongoing Coupled Model Intercomparison Project Phase 6 (CMIP6), only temporal variabilities of global-averaged LAI time series were evaluated, others remain largely uninvestigated. This study systematically investigates the spatial and/or temporal variability of the three critical variables from 27 ESMs in CMIP6 using satellite observations. Our results show that all models and the multi-model ensemble mean (MME) can generally reproduce the observed LAI spatial pattern but all of them overestimate the global mean LAI mainly due to overestimation of LAI in non-forested vegetated areas. Most CMIP6 models fail to capture the temporal variability in the annual LAI because of large biases in both the simulated trend magnitude and temporal pattern of interannual variability. The average LAI seasonal cycles in different latitude zones are roughly reproduced by the models, but 1–2 months delays in the LAI peak appear in the Arctic-boreal zone. Additionally, CMIP6 models overall overestimate tree height, and largely overestimate the global grass area but underestimate tree and shrub areas, especially in the middle and high latitudes. It should be kept in mind that such biases may have further impacts on the simulations of the related carbon and land–atmosphere interaction variables (e.g., ecosystem production, carbon storage, transpiration, and temperature) for global change research. Hence, bias-correction should be made to improve reliability of vegetation structure and distribution when future projections and historical reconstructions. They also underscore the urgent need in development and parameterization of dynamic vegetation within Earth system models, such as phenology, allocation, and morphology schemes.
|
|---|
