Evapotranspiration (ET) is critical in linking global water, carbon and energy cycles. However, direct measurement of global terrestrial ET is not feasible. Here, we first reviewed the basic theory and state-of-the-art approaches for estimating global terrestrial ET, including remote-sensing-based physical models, machine-learning algorithms and land surface models (LSMs). We then utilized 4 remote-sensing-based physical models, 2 machine-learning algorithms and 14 LSMs to analyze the spatial and temporal variations in global terrestrial ET. The results showed that the ensemble means of annual global terrestrial ET estimated by these three categories of approaches agreed well, with values ranging from 589.6 mm yr⁻¹ (6.56×10⁴ km³ yr⁻¹) to 617.1 mm yr⁻¹ (6.87×10⁴ km³ yr⁻¹). For the period from 1982 to 2011, both the ensembles of remote-sensing-based physical models and machine-learning algorithms suggested increasing trends in global terrestrial ET (0.62 mm yr⁻² with a significance level of p<0.05 and 0.38 mm yr⁻² with a significance level of p<0.05, respectively). In contrast, the ensemble mean of the LSMs showed no statistically significant change (0.23 mm yr⁻², p>0.05), although many of the individual LSMs reproduced an increasing trend. Nevertheless, all 20 models used in this study showed that anthropogenic Earth greening had a positive role in increasing terrestrial ET. The concurrent small interannual variability, i.e., relative stability, found in all estimates of global terrestrial ET, suggests that a potential planetary boundary exists in regulating global terrestrial ET, with the value of this boundary being around 600 mm yr⁻¹. Uncertainties among approaches were identified in specific regions, particularly in the Amazon Basin and arid/semiarid regions. Improvements in parameterizing water stress and canopy dynamics, the utilization of new available satellite retrievals and deep-learning methods, and model–data fusion will advance our predictive understanding of global terrestrial ET.

The implementation of boundary conditions is a key aspect of climate simulations. We describe here how the Climate Model Intercomparison Project Phase 6 (CMIP6) forcing data sets have been processed and implemented in Version 6 of the Institut Pierre‐Simon Laplace (IPSL) climate model (IPSL‐CM6A‐LR) as used for CMIP6. Details peculiar to some of the Model Intercomparison Projects are also described. IPSL‐CM6A‐LR is run without interactive chemistry; thus, tropospheric and stratospheric aerosols as well as ozone have to be prescribed. We improved the aerosol interpolation procedure and highlight a new methodology to adjust the ozone vertical profile in a way that is consistent with the model dynamical state at the time step level. The corresponding instantaneous and effective radiative forcings have been estimated and are being presented where possible.

Global land cover is a key variable in the earth system with feedbacks on climate, biodiversity and natural resources. However, global land cover data sets presently fall short of user needs in providing detailed spatial and thematic information that is consistently mapped over time and easily transferable to the requirements of earth system models. In 2009, the European Space Agency launched the Climate Change Initiative (CCI), with land cover (LC_CCI) as 1 of 13 essential climate variables targeted for research development. The LC_CCI was implemented in three phases: first responding to a survey of user needs; developing a global, moderate-resolution land cover data set for three time periods, or epochs (2000, 2005, and 2010); and the last phase resulting in a user tool for converting land cover to plant functional type equivalents. Here we present the results of the LC_CCI project with a focus on the mapping approach used to convert the United Nations Land Cover Classification System to plant functional types (PFTs). The translation was performed as part of consultative process among map producers and users, and resulted in an open-source conversion tool. A comparison with existing PFT maps used by three earth system modeling teams shows significant differences between the LC_CCI PFT data set and those currently used in earth system models with likely consequences for modeling terrestrial biogeochemistry and land–atmosphere interactions. The main difference between the new LC_CCI product and PFT data sets used currently by three different dynamic global vegetation modeling teams is a reduction in high-latitude grassland cover, a reduction in tropical tree cover and an expansion in temperate forest cover in Europe. The LC_CCI tool is flexible for users to modify land cover to PFT conversions and will evolve as phase 2 of the European Space Agency CCI program continues.

The high-latitude regions of the Northern Hemisphere are a nexus for the interaction between land surface physical properties and their exchange of carbon and energy with the atmosphere. At these latitudes, two carbon pools of planetary significance – those of the permanently frozen soils (permafrost), and of the great expanse of boreal forest – are vulnerable to destabilization in the face of currently observed climatic warming, the speed and intensity of which are expected to increase with time. Improved projections of future Arctic and boreal ecosystem transformation require improved land surface models that integrate processes specific to these cold biomes. To this end, this study lays out relevant new parameterizations in the ORCHIDEE-MICT land surface model. These describe the interactions between soil carbon, soil temperature and hydrology, and their resulting feedbacks on water and CO₂ fluxes, in addition to a recently developed fire module. Outputs from ORCHIDEE-MICT, when forced by two climate input datasets, are extensively evaluated against (i) temperature gradients between the atmosphere and deep soils, (ii) the hydrological components comprising the water balance of the largest high-latitude basins, and (iii) CO₂ flux and carbon stock observations. The model performance is good with respect to empirical data, despite a simulated excessive plant water stress and a positive land surface temperature bias. In addition, acute model sensitivity to the choice of input forcing data suggests that the calibration of model parameters is strongly forcing-dependent. Overall, we suggest that this new model design is at the forefront of current efforts to reliably estimate future perturbations to the high-latitude terrestrial environment.

In this paper we review the current status for deriving land surface temperatures (LSTs) by remote sensing from satellites in the thermal infrared. Because of its widespread use and global applicability, we concentrate on the Advanced Very High Resolution Radiometer (AVHRR). The theoretical framework and methodologies used to derive LSTs are reviewed and amplified. Practical algorithms are described and their accuracy and application critically evaluated through sensitivity studies and by inter‐comparison. The important effects of the atmosphere, surface emissivity and instrument noise are considered and the current practice for removing these effects is specified. The accuracy currently attainable from the AVHRR for the LST algorithms studied lies between 1 and 2 K, depending critically upon the surface characteristics and the atmospheric structure. Suggestions about what improvements could be made to reduce the errors in LST estimation from space and the directions of future research are summarised.